Embed Size (px)
Citation preview
EXPERIMENTAL AND TECHNO-ECONOMIC APPROACHES IN
IMPROVEMENT OF THE LIGNOCELLULOSIC-ETHANOL PROCESS
ZSOLT BARTA
DOCTORAL THESIS
2011
BUDAPEST UNIVERSITY OF TECHNOLOGY AND ECONOMICS
DEPARTMENT OF APPLIED BIOTECHNOLOGY AND FOOD SCIENCE
EXPERIMENTAL AND TECHNO-ECONOMIC APPROACHES IN
IMPROVEMENT OF THE LIGNOCELLULOSIC-ETHANOL PROCESS
ZSOLT BARTA
DOCTORAL THESIS
SUPERVISOR: DR. KATI RÉCZEY BUDAPEST UNIVERSITY OF TECHNOLOGY AND ECONOMICS DEPARTMENT OF APPLIED BIOTECHNOLOGY AND FOOD SCIENCE
Department of Applied Biotechnology and Food Science Faculty of Chemical Technology and Biotechnology Budapest University of Technology and Economics Budapest, Hungary © Zsolt Barta Department of Applied Biotechnology and Food Science
I
ACKNOWLEDGEMENTS
First of all, I would like to express my gratitude to Dr. Kati Réczey for herexcellent and straightforward guidance. Thank you for teaching me how todistinguish the important things from the non-important ones and how to finishtasks as soon as possible.
I also wish to express my appreciation to Professor Guido Zacchi, for giving methe opportunity to learn how to perform process simulation with Aspen, forinvolving me in the NILE project and giving challenging tasks, and for helping meto publish the results.
Special thanks to Dr. Per Sassner, who helped me a lot to learn the basics of Aspenand the techno-economic modelling.
I am grateful to Dr. Dóra Dienes for the thorough discussions of any kind ofresearch topic, and for reading and correcting my manuscripts.
Thanks to all my friends and colleagues at Lund University for colouring my lifeand exchanging ideas.
I am also very grateful to all my friends and colleagues in the group of Dr.Mercedes Ballesteros, at CIEMAT, for sharing their expertise and for their patiencewhen I wanted to speak Spanish with them.
I would like to thank my PhD fellows, Bálint Sipos and Miklós Gyalai-Korpos forthe time we spent together and to thank all my former colleagues in the Non-foodResearch Group of BME ABET for their assistance.
The New Hungary Development Plan (TÁMOP-4.2.1/B-09/1/KMR-2010-0002),the Hungarian National Research Fund (OTKA-K72710), the Foundation of ProProgressio, the Foundation of Varga József and the 6th Framework Programme ofthe European Commission (NILE project, contract No. 019882) are acknowledgedfor their financial support.
A great deal of gratitude goes to my family and my girlfriend, Eszter, for their loveand patience.
III
LIST OF ABBREVIATIONS
AD Anaerobic digestionAFEX Ammonia fibre/freeze explosionASE Accelerated solvent extractionCBP Consolidated bioprocessingCEF Cellulase enzyme fermentationCHP Combined heat and powerCIEMAT Centro de Investigaciones Energéticas, Medioambientales y
Tecnológicas (energy research centre in Madrid, Spain)COD Chemical oxygen demandDH District heatingDM Dry matterDP Degree of polymerisationFPU Filter paper unitGH61 Glycoside hydrolase family 61HHCF Hybrid hydrolysis and co-fermentationHMF Hydroxymethyl furfuralHP High-pressureHPLC High-performance liquid chromatographyIU International unitLHV Lower heating value (also referred to as net calorific value)LP Low-pressureME Monomer equivalentNREL National Renewable Energy LaboratoryODT/y Dry tonne per yearPWM Presaccharified wheat mealSEK Swedish kronorSHF Separate hydrolysis and fermentationSPWS Steam pretreated wheat strawSSCF Simultaneous saccharification and co-fermentationSSF Simultaneous saccharification and fermentationWIS Water-insoluble solidYC Yeast cultivation
V
ABSTRACT
Ethanol is one of the leading candidates with respect to replacing some fraction offossil fuels. Currently, it is produced from sugar- and starch-containing materials.However, the large-scale use of ethanol as fuel will require lignocellulosic biomassto be used as raw material. The conversion of lignocellulosic material to ethanol ismore complex than ethanol production from sugar or starch. Although pilot-scaleand demonstration plants have been brought into operation recently, the processconcept has not yet been introduced to an industrial scale.
The aim of my PhD work was to investigate the lignocellulose-to-ethanol processeither by performing experiments regarding certain process steps or by constructingand evaluating techno-economic models of the whole process. In the experimentalpart, steam pretreatment of non-impregnated hemp hurds was investigated at tworeactor scales by varying the temperature. The optimal temperature of steampretreatment of hemp hurds for releasing fermentable sugars in enzymatichydrolysis and for subsequent ethanol fermentation, is 210°C in terms of overallglucose (336 g/kg dry hurds, 75% of theoretical) and ethanol yield (141 g/kg dryhurds, 60%). However, the maximum sugar yield (414 g/kg dry hurds, 63%),considering glucose and xylose, can be obtained at lower temperature (200°C).
Besides steam pretreatment of hemp hurds, integrated simultaneoussaccharification and fermentation of steam pretreated wheat straw andpresaccharified wheat meal was investigated using various mixtures of the twosubstrates. By increasing the amount of presaccharified wheat meal, not only theethanol concentration, but also the ethanol yield increased. The maximum ethanolyield, 99% of theoretical based on glucose, was obtained for a mixture ofcontaining equal amounts of presaccharified wheat meal and steam pretreatedwheat straw (on a water-insoluble solids basis). Hence, mixing of the twosubstrates was found to be beneficial for the final ethanol concentration and alsofor the ethanol yield.
In the part of techno-economic modelling, on-site cellulase enzyme fermentation ina softwood-to-ethanol process was investigated. The effect of varying the carbonsource of enzyme fermentation was monitored through the whole process. Liquidfraction of steam pretreated material, the same supplemented with molasses and thewhole pretreated slurry served as carbon source. Capital cost was found to be themain cost contributor to enzyme fermentation. Productivity of cellulasefermentation was proved to be an important parameter in the cost of enzymeproduction. The lowest minimum ethanol selling prices were obtained in thosescenarios, where pretreated liquid fraction supplemented with molasses was usedas carbon source. On-site enzyme fermentation was found to be a feasiblealternative in some of the scenarios investigated.
VI
Regarding the downstream part of the lignocellulose-to-ethanol process, variousprocess configurations of anaerobic digestion of the stillage, with differentcombinations of co-products, were evaluated versus the reference case ofevaporation. Anaerobic digestion of the stillage showed a significantly higheroverall energy efficiency (87-92%), based on the lower heating values, than thereference case (81%). Although the amount of ethanol produced was the same inall scenarios, the production cost varied between 4.00 and 5.27 SEK/L, includingthe reference case. Anaerobic digestion of the stillage with biogas upgrading wasfound to be a favourable option for both energy efficiency and ethanol productioncost.
VII
LIST OF PUBLICATIONS
This thesis is based on the following scientific papers, which will be referred to inthe text by their roman numerals. The papers are appended at the end of the thesis.All the papers have been reprinted with the permission of the copyright holder.
I. Barta, Zs., Oliva, J.M., Ballesteros, I., Dienes, D., Ballesteros, M.,Réczey, K. (2010) Refining hemp hurds into fermentable sugars orethanol. Chemical and Biochemical Engineering Quarterly. 24 (3),331-339. IF: 0.35
II. Erdei, B., Barta, Zs., Sipos, B., Réczey, K., Galbe, M., Zacchi, G.(2010) Ethanol production from mixtures of wheat straw and wheatmeal. Biotechnology for Biofuels. 2010, 3:16, doi:10.1186/1754-6834-3-16. IF: 4.12
III. Barta, Zs., Sassner, P., Zacchi, G., Réczey, K. (2008) Techno-economic aspects of on-site cellulase production. Hungarian Journal ofIndustrial Chemistry Veszprém. 36 (1-2), 5-9.
IV. Barta, Zs., Kovács, K., Réczey, K., Zacchi, G. (2010) Process designand economics of on-site cellulase production on various carbonsources in a softwood-based ethanol plant. Enzyme Research. 2010,734182, doi:10.4061/2010/734182.
V. Barta, Zs., Réczey, K., Zacchi, G. (2010) Techno-economicevaluation of stillage treatment with anaerobic digestion in a softwood-to-ethanol process. Biotechnology for Biofuels. 2010, 3:21doi:10.1186/1754-6834-3-21. IF: 4.12
VIII
Oral presentations
Barta, Zs., Réczey, I., Zacchi G. What kind of co-products make a lignocellulosebased ethanol plant the most feasible? (Milyen melléktermékek esetén aleggazdaságosabb egy lignocellulóz alapú alkoholgyár?) Forum of PhD Students.Debrecen, November 4, 2010.
Barta, Zs., Dienes, D., Réczey, I. How can a sweet sorghum based small-scaleethanol plant be economically feasible? (Hogyan tehető gazdaságossá egycukorcirok alapú alkoholgyártó kisüzem?) LI. Georgikon Days. Keszthely, October1-2, 2009.
Barta, Zs., Oliva, J.M., Ballesteros, I., Dienes, D., Ballesteros, M., Réczey, I.Refining hemp hurds into fermentable sugars and ethanol (Kenderpozdorjafeldolgozása fermentálható cukrokká és etanollá) 336. Scientific Colloquy (KÉKI).Budapest, September 18, 2009.
Barta, Zs., Dienes, D., Réczey, K. Kinetic study on enzymatic cellulose hydrolysisof agricultural residues (Mezőgazdasági melléktermékeken végzett enzimescellulózhidrolízis kinetikai leírása) Polysaccharide-chemistry Workshop. Budapest,September 29, 2008.
Barta, Zs., Sassner, P., Zacchi, G., Réczey, K. Techno-econimic analysis ofenzyme production (Celluláztermelés technológiai és gazdasági elemzése)Technical Chemical Days’08. Veszprém, April 22-24, 2008.
Barta, Zs. Ethanol production from lignocelluloses – Pretreatment, fermentationand modelling the downstream process in Aspen Plus (Etanol előállításlignocellulóz szénforrásból - Előkezelés, fermentáció és a downstream műveletekmodellezése Aspen Plus-szal) Polysaccharide-chemistry Workshop. Budapest,November 7, 2007.
Poster presentations
Barta, Zs., Réczey, K., Zacchi, G. Improving the energy efficiency and economicsof a softwood-based ethanol process by producing biogas from the stillage. The 4thAnnual Workshop of COST FP0602, Biotechnical processing of lignocellulosic rawmaterials. Cesme, Turkey, September 21-24, 2010.
Barta, Zs., Réczey, K., Zacchi, G. Techno-economic evaluation of biogasproduction from stillage in a spruce-to-ethanol process. 32nd Symposium onBiotechnology for Fuels and Chemicals. Clearwater Beach, FL, USA, April 19-22,2010.
IX
Gyalai-Korpos, M., Mangel, R., Barta, Zs., Alvira, P., Dienes, D., Ballesteros, M.,Réczey, K. Cellulase production using different streams of wheat grain and wheatstraw based ethanol processes. 32nd Symposium on Biotechnology for Fuels andChemicals. Clearwater Beach, FL, USA, April 19-22, 2010.
Barta, Zs., Oliva, J.M., Ballesteros, I., Dienes, D., Ballesteros, M., Réczey, K.Steam pretreatment of hemp hurds for bioethanol production. Third EuropeanWorkshop on Biotechnology for Lignocellulose Biorefineries. Varenna, Italy,September 3-4, 2009.
Barta, Zs., Deák, A., Sipos, B., Réczey, K. Modelling a sweet sorghum processingbioethanol plant – whole crop utilization. Second European Workshop onBiotechnology for Lignocellulose Biorefineries. Biel, Switzerland, December 4-5,2008.
Sipos, B., Barta, Zs., Gyalai-Korpos, M, Sassner, P., Réczey, K. Can sweetsorghum be a feasible raw material for ethanol production in Hungary? 16th
European Biomass Conference & Exhibition. Valencia, Spain, June 2-6, 2008.
Gyalai-Korpos, M., Barta, Zs., Sipos, B., Réczey, K. Looking for feedstock –bioethanol potential in Hungary. 16th European Biomass Conference & Exhibition.Valencia, Spain, June 2-6, 2008.
Barta, Zs., Dienes, D., Réczey, K. Kinetic modelling of enzymatic hydrolysis ofagricultural residues. Era-chemistry workshop. Krakow, Poland, April 13-16, 2008.
Barta, Zs., Dienes, D., Réczey, K. Kinetic modelling of enzymatic hydrolysis ofagricultural residues. First European Workshop on Biotechnology forLignocellulose Biorefineries. Copenhagen, Denmark, March 27-28, 2008.
Conference proceedings
Barta, Zs., Dienes, D., Réczey, I. (2009) How can a sweet sorghum based small-scale ethanol plant be economically feasible? (Hogyan tehető gazdaságossá egycukorcirok alapú alkoholgyártó kisüzem?) Conference Proceedings of LI.Georgikon Days (ISBN: 978-963-9639-35-5) 73-82.
Barta, Zs., Sassner, P., Zacchi, G., Réczey, K. (2008) Techno-econimic analysis ofenzyme production (Celluláztermelés technológiai és gazdasági elemzése)Conference Proceedings of Technical Chemical Days ’08 (ISBN 978-963-9696-36-5), 31-36.
XI
TABLE OF CONTENTS
1 INTRODUCTION.............................................................................................11.1 AIM OF THE THESIS ................................................................................................2
2 BACKGROUND................................................................................................32.1 LIGNOCELLULOSES................................................................................................3
2.1.1 Cellulose...........................................................................................................42.1.2 Hemicellulose...................................................................................................42.1.3 Lignin ...............................................................................................................42.1.4 Hemp hurds ......................................................................................................52.1.5 Wheat straw......................................................................................................6
2.2 CELLULOLYTIC ENZYME SYSTEMS.........................................................................62.2.1 Cellulases .........................................................................................................62.2.2 Hemicellulases .................................................................................................82.2.3 Synergy of cellulolytic enzymes ......................................................................9
2.3 CONVERSION OF LIGNOCELLULOSES......................................................................92.3.1 Pretreatment ...................................................................................................102.3.2 Enzyme production.........................................................................................142.3.3 Enzymatic hydrolysis .....................................................................................152.3.4 Fermentation...................................................................................................172.3.5 Downstream operations ..................................................................................192.3.6 Process configurations....................................................................................20
2.4 PROCESS DESIGN AND ECONOMICS ......................................................................222.4.1 Process simulation..........................................................................................222.4.2 Economics ......................................................................................................222.4.3 Detailed description of some techno-economic evaluations...........................25
3 MATERIALS AND METHODS....................................................................293.1 RAW MATERIALS .................................................................................................293.2 ENZYMES.............................................................................................................303.3 COMPOSITIONAL ANALYSIS .................................................................................303.4 STEAM PRETREATMENT .......................................................................................31
3.4.1 Steam pretreatment of hemp hurds.................................................................313.4.2 Steam pretreatment of wheat straw ................................................................32
3.5 ENZYMATIC CELLULOSE HYDROLYSIS .................................................................323.6 SIMULTANEOUS SACCHARIFICATION AND FERMENTATION ..................................33
3.6.1 Shake flask experiments .................................................................................333.6.2 Experiments in laboratory fermentors ............................................................33
3.7 ANALYSIS OF SUGARS, ETHANOL AND BY-PRODUCTS ..........................................343.8 STATISTICAL ANALYSIS .......................................................................................343.9 METHODOLOGY OF TECHNO-ECONOMIC ANALYSIS .............................................34
4 RESULTS AND DISCUSSION OF EXPERIMENTAL PART..................374.1 HEMP HURDS AS POTENTIAL RAW MATERIAL FOR ETHANOL PRODUCTION (PAPERI) 37
XII
4.1.1 Steam pretreatment ........................................................................................ 374.1.2 Enzymatic hydrolysis and SSF ...................................................................... 39
4.2 INTEGRATING PRODUCTION TECHNOLOGIES OF FIRST AND SECOND GENERATIONSPRIOR TO SIMULTANEOUS SACCHARIFICATION AND FERMENTATION (PAPER II) .............. 43
4.2.1 Effect of PWM on the final ethanol concentration obtained in SSF.............. 434.2.2 Effect of mixing lignocellulosic and starch-containing substrates on ethanolyield of SSF ................................................................................................................ 43
5 RESULTS AND DISCUSSION OF TECHNO-ECONOMIC ANALYSIS475.1 BASE LIGNOCELLULOSE-TO-ETHANOL PROCESS ................................................. 47
5.1.1 Steam pretreatment ........................................................................................ 475.1.2 Yeast cultivation and simultaneous saccharification and fermentation ......... 485.1.3 Distillation and evaporation........................................................................... 495.1.4 Combined heat and power production ........................................................... 505.1.5 Other process areas ........................................................................................ 50
5.2 ON-SITE ENZYME PRODUCTION ........................................................................... 515.2.1 Preliminary model of enzyme production (Paper III).................................... 525.2.2 The effect of enzyme fermentation productivity on the cost of cellulaseproduction................................................................................................................... 525.2.3 Modified model of enzyme production (Paper IV)........................................ 535.2.4 The effect of on-site enzyme fermentation on the whole process.................. 555.2.5 Economics: specific enzyme cost, minimum ethanol selling price................ 56
5.3 ALTERNATIVE TREATMENT OF THE STILLAGE AND PRODUCTION OF VARIOUS CO-PRODUCTS (PAPER V) ...................................................................................................... 58
5.3.1 Anaerobic digestion ....................................................................................... 595.3.2 Aerobic treatment .......................................................................................... 605.3.3 Combined heat and power production in scenarios of alternative stillagetreatment ..................................................................................................................... 615.3.4 Energy aspects of the various scenarios ........................................................ 625.3.5 Economic aspects of the scenarios investigated ............................................ 64
6 SUMMARY......................................................................................................676.1 NOVEL SCIENTIFIC FINDINGS .............................................................................. 68
7 REFERENCES ................................................................................................71
1
1 INTRODUCTION
Increasing awareness that fossil fuel resources are limited has prompted theresearch for alternative fuels. Ideally, alternative fuels should not only beecologically sustainable, but also cheaper than fossil fuels. Ethanol is one of theleading candidates with respect to replacing some fraction of fossil fuels. It can beproduced in large quantities from renewable raw materials at a relatively low cost.Its introduction to the fuel market is facilitated by the following features: ethanolcan be handled and distributed similarly to petrol and diesel, it is completelymiscible with petrol, and ethanol-petrol mixtures with ethanol content up to 20%can be used in current spark-ignited combustion engines without any majormodification.
The majority of governments also support the replacement of conventional fossilfuels in the transportation sector with alternative ones by means of directives. TheEuropean Commission plans to progressively replace 20% (on energy basis) by2020 (Directive, 2003), and ethanol is expected to be one of the main means ofachieving this goal (Galbe et al., 2007).
The world annual ethanol production has shown a steady increase: 31, 51, 66million cubic meter in 2001, 2006, 2009, respectively (Biofuels, 2011). As theamounts used in beverages and industrial applications have remained almostconstant over this period, the growing demand for fuel ethanol is responsible forthe whole increase. The two leader countries in ethanol production are the UnitedStates and Brazil, contributing to 43 and 27% of the total production in 2009(Biofuels, 2011), respectively.
According to maturity and raw materials of the biochemical processes first- andsecond-generation ethanol production technologies can be distinguished (Brienset al., 2008). For the former, sugar substances and grains serve as feedstocks andthese technologies already exist on a large scale, while the latter utiliseslignocellulosic materials and involve more complex technologies that have not yetbeen demonstrated on an industrial scale. Whereas sugar and starch-containingfeedstocks are limited, relatively expensive and supply the food industry as well,lignocellulosic biomass is abundant, available at low cost and largely unused(Tomás-Pejó et al., 2008a). However, currently the two major feedstocks used on afull scale are corn grain and sugar cane (Biofuels, 2011).
Demonstration activities on the biochemical route of lignocellulosic ethanolproduction are pursued mainly in Europe (e.g. Abengoa, Spain; BioGasol,Denmark; Inbicon, Denmark; M&G/Chemtex, Italy; Procethol 2G/Futurol, France;SEKAB, Sweden), in the United States (e.g. BlueFire Ethanol, KL Energy,Mascoma, POET, QTeros, Verenium) and in Canada (Iogen). Except BlueFire
2
Ethanol, which applies acid hydrolysis, all the others, either in Europe or in NorthAmerica, hydrolyse the cellulose in enzymatic way (Gnansounou and Dauriat,2010).
1.1 Aim of the thesis
The general aim of my work was to add some novel knowledge to thelignocellulose-to-ethanol process intensively investigated recently. I chose twoways of gathering new information, i.e. by performing experiments regardingcertain steps of the process and by constructing and evaluating techno-economicmodels of the whole process with all the interactions thereof. The models takenfrom the literature or constructed by the author are described in present tenses.
Based on the two-way approach, the present work contains the results of thefollowing studies:
1) Optimisation of steam pretreatment of hemp hurds
2) Integrated ethanol production from wheat straw and wheat meal
3) Modelling of on-site enzyme fermentation
4) Modelling of alternative stillage treatment and production of various co-products
The work was carried out at the Non-Food Research Group of Department ofApplied Biotechnology and Food Science, Budapest University of Technology andEconomics in cooperation with Lund University (Lund, Sweden) and CIEMAT(Madrid, Spain). Steam pretreatment of hemp hurds was performed at the BiomassUnit of CIEMAT. Integrated ethanol production was carried out at the Departmentof Chemical Engineering of Lund University. Simulation softwares used in themodelling studies were available at the Department of Chemical Engineering(Lund University) and at the Department of Chemical and Environmental ProcessEngineering (Budapest University of Technology and Economics).
3
2 BACKGROUND
2.1 Lignocelluloses
Lignocelluloses constitute a number of materials including woody biomass,herbaceous crops, agricultural residues, etc. Although these materials may appearquite different, they consist of the same three main constituents: cellulose,hemicellulose and lignin (Figure 1). Cellulose is the fibre fraction, which providesthe strength and flexibility of the cell wall. Lignin has an impregnating role, andmakes the structure more resistant to biological and chemical attacks. Cellulose andlignin are chemically coupled by hemicellulose. Minor, but still important,constituents are extractives and ash.
Figure 1 Constituents of lignocelluloses:cellulose (yellow), hemicellulose (blue) and lignin (brown).
Typical compositions of some lignocellulosic materials are presented in Table 1.The two major polysaccharides are glucan and xylan at hardwood and agriculturalresidues, while at softwood they are glucan and mannan. Polysaccharidescontribute around two third to the dry matter of lignocelluloses.
Table 1 Composition of some lignocellulosic material expressed as % ofdry matter.
Raw material Glucan Mannan Galactan Xylan Arabinan Lignin Ref.Softwood (spruce) 43.3 11.7 5.3 6.1 1.6 32.0 1
Hardwood (willow) 43.0 3.2 2.0 14.9 1.2 26.4 2
Wheat straw 32.6 0.0 0.8 20.1 3.3 24.2 3
Corn stover 42.5 n.r. n.r. 16.1 n.r. 20.1 4
1 Monavari et al., 20092 Sassner et al., 20063 Linde et al., 20084 Kálmán et al., 2002n.r. not reported
4
2.1.1 Cellulose
Cellulose, the major component of the plant cell wall, is the most abundantrenewable polysaccharide on earth. It is a linear homopolymer of -1,4 boundglucose molecules, with the disaccharide cellobiose as repeating unit (Figure 2).The chain contains between 2 000 and 20 000 glucose molecules, which are rotated180° with respect to its neighbour. Cellulose has a strong tendency to form intra-chain and inter-chain hydrogen bonds (Sjöström, 1993). Intra-chain bonds arepresent between the hydroxyl group of C3 and the oxygen of C5 in the adjacentglucose molecule. Through inter-chain bonds bundles of cellulose macromoleculesaggregate to microfibrils, in which highly ordered crystalline and less orderedamorphous regions are combined (Hon, 1994). Microfibrils form fibrils, whichbuild up the cellulose fibres. The highly ordered structure and the hydrogen bondsmake the cellulose insoluble in water and alkalis (Fengel and Wegener, 1989).
Figure 2 Disaccharide cellobiose is the repeating unit of cellulose.
2.1.2 Hemicellulose
Hemicellulose is a branched heteropolymer with less resistance, more amorphouspart and lower degree of polymerisation (DP<200) than cellulose (Saha, 2003). Itmainly consists of various hexoses such as glucose, galactose, mannose andpentoses such as xylose and arabinose, bound by -1,4 and -1,3 linkages. Thecomposition of hemicellulose largely depends on the origin. Hemicelluloses ofsoftwood, hardwood and herbaceous plant mainly contain galactoglucomannan, O-acetyl-4-O-methyl-glucuronoxylan and arabinoxylan, respectively (Harman andKubicek, 1998). Especially in hardwood, the majority of xylose units in the xylanbackbone are acetylated, i.e. hydroxyl groups either at C2 or at C3 are replaced byO-acetyl groups, which are released as acetic acid, when the material is hydrolysed(Shiraishi, 1991). Hemicellulose is more hydrophilic and can be hydrolysed easierthan cellulose. It can be solubilised in acids and concentrated bases as well (Fengeland Wegener, 1989).
2.1.3 Lignin
Lignin is constituted from polymers of phenyl propane units (Lynd, 1996). Theprimary building units are 4-hydroxyphenylpropane, guaiacylpropane, andsyringylpropane molecules, linked by ether and carbon-carbon bonds. The 4-
5
hydroxyphenylpropane unit is not methoxylated, while guaiacyl and syringyl unitshave one and two methoxyl groups adjacent to the phenolic hydroxyl group,respectively. The building units are polymerised by random-coupling reactions intoa cross-linked, complex macromolecule, thus lignin lacks a general structure(Bruce and Palfreyman, 1998). The ratio of building units varies over the species.Softwood lignin contains mainly guaiacyl, whereas hardwood lignin contains bothguaiacylpropane and syringylpropane units. Lignin of herbaceous plants comprisesof all the three monomers. Lignin, together with cellulose, plays an important rolein strengthening the plant material (Sjöström, 1993).
2.1.4 Hemp hurds
Hemp hurds are the woody core of industrial hemp (Cannabis sativa L.), which isone of the raw materials of the fibre industry, where bast fibres are required.Conventionally, they are removed after dew-retting the hemp stalks in the field(Garcia et al., 1998) or after water-retting. In these processes pectinases producedby bacteria present on the material digest the pectin and enable fibre separation.The residue, containing less cellulose but more hemicellulose than the fibrefraction, is referred to as hemp hurds and has had several, minor applications sofar, such as animal bedding – due to its favourable properties: high water-absorbingability, easy handling and rapid composting after use –, garden mulch or acomponent of light-weight concrete. In 2002, 95% of the hemp hurds produced inthe European Union (40 000 tonnes) was utilised as animal bedding, mainly ashorse bedding, whereas 4% was used in the construction sector as pour-ininsulation, hurds board and additive to bricks or loam (Karus and Vogt, 2004).Hemp hurds constitute 70% of the stalk dry matter (Dang and Nguyen, 2006).
Industrial hemp is an attractive crop, because it requires little or no pesticides orfertilisers, replenishes the soil with nutrients and eliminates competing weeds(Moxley et al., 2008). In the 1960s, Hungary produced huge amounts of industrialhemp, which has dropped significantly since then; nevertheless the conditions(climate, soil etc.) are still appropriate for cultivation of large quantities.
Since hemp hurds are an agro-industrial by-product with a high carbohydratecontent, they are a potential candidate for second-generation fuel-ethanolproduction. Vignon et al. (1995) investigated impregnated hemp woody core in asteam explosion treatment to optimise fibre separation and delignification,primarily by means of optical and scanning electron microscopy and compositionalanalysis. Moxley et al. (2008) performed cellulose-solvent-based lignocellulosefractionation of hemp hurds and subsequent enzymatic hydrolysis of the residualcellulose fraction. At the best pretreatment condition (84% H3PO4 at 50 °C for 60min), the glucan digestibility was found to be 96% at 24 h at a cellulase loading of15 FPU/g glucan.
6
2.1.5 Wheat straw
Wheat straw is an abundant agricultural residue, and is considered as one of thepotential raw materials of lignocellulosic ethanol production. The main advantageof wheat straw in terms of ethanol production is its easily degradable structure,which requires milder pretreatment conditions than woody biomass. Wheat straw istested in several pilot facilities (Solomon et al., 2007) and is the feedstock of somedemonstration plants already, such as those of Iogen (Solomon et al., 2007),Abengoa (Gnansounou and Dauriat, 2010) and Inbicon (Persson, 2009).
The ratio of residue to grain for wheat is about 1.3:1.0 (w/w) (Milbrandt, 2005).Generally, 30-40% of the straw is left on the field for soil protection, hence themass of the straw available and the mass of grain harvested are approximately thesame. Kim and Dale (2004) assumed 60% ground cover, instead of 30%, due to theuncertainties of local situations, and estimated a global wheat straw availability of354 million tonnes annually. The European Union produced 140 million tonnes ofwheat straw in 2006 (Tomás-Pejó et al., 2008b). The Hungarian production was 7.3million tonnes in 2008 (KSH, 2010).
2.2 Cellulolytic enzyme systems
Cellulolytic enzymes systems play an important role in the carbon cycle, and areproduced by a number of microorganisms, including aerobes (Microbispora,Pseudomonas), facultative anaerobes (Bacillus, Cellulomonas), anaerobes(Clostridium, Acetivibrio) and a variety of fungal species (Trichoderma,Penicillium, Aspergillus, Phanerochaete, Fusarium) (Coughlan, 1992). Someanaerobic bacteria, e.g. Clostridium thermocellum, produce a multienzymecomplex referred to as cellulosome, comprising both cellulases and hemicellulasesorganised around a non-catalytic integrating protein on the cell surface. It has beenproven that cellulosomes mediate attachment to the substrate, and can efficientlybreak down crystalline cellulose (Xu and Smith, 2010). In contrast to anaerobes,aerobic bacteria and fungi excrete discretely acting, non-complexed cellulasesexisting as free entities in the culture medium (Ding et al., 2008).
2.2.1 Cellulases
In this section fungal cellulase enzyme systems will be discussed. Three majortypes of enzyme can be distinguished: endoglucanases (EC 3.2.1.4); exoglucanasesincluding cellobiohydrolases (EC 3.2.1.91) and glucohydrolases (EC 3.2.1.74); and-glucosidases (EC 3.2.1.21). In parenthesis the codes recommended by EnzymeCommission (EC) are given. Endoglucanases attack -1,4-glycosidic bonds insidethe cellulose chain in the amorphous region liberating gluco-oligomers of variouslength (Teeri, 1997). Exoglucanases act in a processive way on the ends of
7
cellulose chain and on the gluco-oligomers generated by endoglucanases (Figure3). Processive mode of action means that enzymes move along the chain cleavingmore than one glycosidic bond before being desorped from the cellulose (Jalak andValjamae, 2010). Cellobiohydrolases and glucohydrolases release cellobiose andglucose, respectively, and have high affinity to crystalline cellulose (Teeri, 1997).Final step of the cellulose-to-glucose conversion is the action of -glucosidases,which cleave primarly cellobiose, but also longer water-soluble gluco-oligomers, toglucose. Strictly speaking, -glucosidase does not belong to cellulases, as it doesnot act on cellulose chain, however it plays an undoubtedly important role in theefficient cellulose hydrolysis by converting the cellobiose that has strong inhibitoryeffect on both cellobiohydrolases and endoglucanases (Chauve et al., 2010).
Figure 3 Mechanistic scheme of enzymatic cellulose hydrolysis byTrichoderma non-complexed cellulase system (adapted fromZhang et al., 2006). Endos: endoglucanases, exos:exoglucanases, R: reducing end (•), NR: non-reducing end (o), -Gase: -glucosidase.
7
cellulose chain and on the gluco-oligomers generated by endoglucanases (Figure3). Processive mode of action means that enzymes move along the chain cleavingmore than one glycosidic bond before being desorped from the cellulose (Jalak andValjamae, 2010). Cellobiohydrolases and glucohydrolases release cellobiose andglucose, respectively, and have high affinity to crystalline cellulose (Teeri, 1997).Final step of the cellulose-to-glucose conversion is the action of -glucosidases,which cleave primarly cellobiose, but also longer water-soluble gluco-oligomers, toglucose. Strictly speaking, -glucosidase does not belong to cellulases, as it doesnot act on cellulose chain, however it plays an undoubtedly important role in theefficient cellulose hydrolysis by converting the cellobiose that has strong inhibitoryeffect on both cellobiohydrolases and endoglucanases (Chauve et al., 2010).
Figure 3 Mechanistic scheme of enzymatic cellulose hydrolysis byTrichoderma non-complexed cellulase system (adapted fromZhang et al., 2006). Endos: endoglucanases, exos:exoglucanases, R: reducing end (•), NR: non-reducing end (o), -Gase: -glucosidase.
8
The most widely studied cellulolytic fungus is Trichoderma reesei, which is alsothe most important cellulase producing microorganism in industrial point of view.The principal components of the excreted enzymes are cellobiohydrolases(CBHI/Cel7A, CBHII/Cel6A) contributing 70-80% to the total amount of enzymeprotein (Wood, 1992). In parenthesis names of enzyme component are givenaccording to the traditional and recent nomenclatures; the latter was proposed byHenrissat et al. in 1998. Cel7A and Cel6A act on reducing and non-reducing endsof cellulose chain, respectively. Beside the cellobiohydrolases, five majorendoglucanases, EGI/Cel7B, EGII/Cel5A, EGIII/Cel12A, EGIV/Cel61A, andEGV/Cel45A, are also present in the enzyme cocktail secreted (Sipos et al., 2010).Trichoderma reesei produces two -glucosidases, BGLI/Cel3A and BGL II/Cel1A.Both Cel3A and Cel1A have been isolated from culture supernatants, but a largefraction of these enzymes remains bound to the cell wall (Messner et al., 1990).
In the case of Trichoderma reesei, the production of enzymes for the utilisation ofcellulose, is induced e.g. in the presence of cellulose, products thereof, i.e.gluco-oligomers and cellobiose; lactose and sophorose (Persson et al., 1991). Thelatter is considered to be a strong inducer, because it is already active at very lowconcentrations (Mandels et al., 1962). Other important feature of the wild-typeTrichoderma reesei cellulase enzyme system is that the enzyme synthesis isrepressed by glucose, the end product of cellulose hydrolysis (Mach and Zeilinger,2003). However, the cellulase-hyperproducing Trichoderma reesei strain Rut C30has a cre1 mutation – the general carbon catabolite repressor protein CRE1represses the transcription of cellulase genes (Strauss et al., 1995) – due to thismutation the Rut C30 strain produces cellulases even in the presence of glucose.
2.2.2 HemicellulasesBeside cellulases, hemicellulases are also essential in the complete hydrolysis oflignocellulosic substrates. They comprise a number of enzymes that are responsiblefor the degradation of different types of hemicelluloses, primarily xylans andmannans.
Hemicellulases can be placed into three general categories (Wyman, 1996). Endo-acting enzymes attack the polysaccharide backbone internally producing oligomersof varying length, e.g. endo-xylanase (EC 3.2.1.8) and endo-mannanase (EC3.2.1.78). Exo-acting enzymes act processively at the chain ends, e.g. -xylosidase(EC 3.2.1.37) and -mannosidase (EC 3.2.1.25) release xylose and mannose,respectively. “Accessory” enzymes include a variety of acetylesterases (EC3.1.1.6) and arylesterases (EC 3.1.1.2), such as coumaric acid and ferulic acidesterases, and cleave the corresponding acids from the hemicellulose. Since theseacids serve as covalent cross-linker between hemicellulose and lignin, their releaseloosens the hemicellulose-lignin interaction (Tabka et al., 2006).
9
2.2.3 Synergy of cellulolytic enzymes
Cellulolytic enzymes display synergy when act on a lignocellulosic substrate. Theactivity of a mixture of enzymes is higher than the activity of the individualenzymes.
The first hypothesis for the synergy of cellulases was proposed by Reese et al.(1950). The C1/Cx concept assumed that cellulase complex consisted of at least twoconsecutive enzymes: C1 was capable to open up cellulose for Cx. In 1972, Woodand McCrae introduced the most widely accepted hypothesis, the endo/exocooperation. According to this hypothesis endoglucanases initiate degradation byhydrolysing amorphous domains in cellulose and thus generating new chain ends.On the other hand cellobiohydrolases, which produce predominantly cellobiosefrom insoluble cellulose, catalyse the recurrent removal of cellobiose units from thenon-reducing ends of cellulose chain created by endoglucanases. Fägerstam andPetterson (1980) made the first observation of the exo/exo cooperation between thetwo cellobiohydrolases of Trichoderma reesei. Some authors also observedsynergism between endoglucanases (Boisset et al., 2000, Tuka et al., 1992, Walkeret al., 1992, Zhou and Ingram, 2000). Nidetzky et al. (1994) proposed a new modelfor the synergy of cellulases, which involves a sequential attack of cellulose withno requirement for simultaneous action of different enzymes. This model provideda possible explanation for the exo/exo cooperation between the twocellobiohydrolases of Trichoderma reesei.
Other studies demonstrated the synergy of cellulases with non-cellulase enzymes,including primarily hemicellulases (Han et al., 2004, Koukiekolo et al., 2005,Morgavi et al., 2000, Murashima et al., 2003). The synergism between cellulasesand hemicellulases is believed to arise from the ability of hemicellulases to exposethe cellulose microfibrils, by either removing the hemicellulose or thehemicellulosic sidechains (Yu et al., 2003).
2.3 Conversion of lignocelluloses
The schematic overview of the enzymatic route of lignocellulosic ethanolproduction is illustrated in Figure 4. The raw material is pretreated and hydrolysedto monomer sugars that are fermented to ethanol by a microorganism, such asyeast. Both cellulase enzyme production and propagation of the fermentingorganism can be carried out using the sugar-containing liquid fraction of thepretreated material. Ethanol is recovered by means of distillation, and various typesof co-product can be produced, including pelletised solid fuel, biogas, electricityand district heat.
10
Figure 4 Enzymatic route of the lignocellulosic ethanol process. Boxes inthe dashed rectangle can be integrated in various ways, seeFigure 7.
2.3.1 Pretreatment
In the context of biological processing of lignocelluloses to sugars for fermentationto ethanol and other products, pretreatment generally refers to the disruption of theresistant carbohydrate-lignin shield that limits the accessibility of enzymes tocellulose and hemicellulose (Wyman et al., 2005). The considerable variation instructure and composition of raw materials makes it difficult to establish agenerally applicable pretreatment method with optimal conditions, therefore anempirical approach is necessary, basically for each raw material (Canettieri et al.,2007). The pretreatment step affects all subsequent process steps, even thewastewater treatment, as well as biomass production and handling of raw materialprior to pretreatment (Yang and Wyman, 2008). Pretreatment methods can bedivided into different categories: physical (e.g. milling, grinding and irradiation),biological, chemical (e.g. alkali, dilute acid), physicochemical (e.g. steampretreatment, hydrothermolysis and wet oxidation), or combinations of these (Weilet al., 1994).
11
In most cases of physical methods (e.g. milling, grinding) power consumption istoo high to reach appropriate cellulose digestibility in the enzymatic hydrolysis. Itcan be even higher than the theoretical energy content available in the biomass(Datta, 1981). Irradiation of cellulose by gamma rays cleaves the β-1,4-glycosidicbonds giving larger surface area and a lower crystallinity (Takács et al., 2000),however, is too expensive to be applied in a full-scale process.
Biological pretreatments employ wood-degrading fungi (soft, brown and white rot)to modify the chemical composition of the lignocelluloses. Generally, soft andbrown rot fungi primarily degrade the hemicellulose, while white-rot fungi canattack the lignin fraction (Ander and Eriksson, 1977). Major disadvantages ofbiological pretreatments are the long residence time (10–14 days), thus large spacerequirement, and the need for careful control of growth conditions (Wyman, 1996).Nevertheless, these methods could be used as first step followed by anotherpretreatment.
Alkaline pretreatment refers to as soaking of the material in an alkaline solution,such as NaOH, and then heating it for a certain time. The swelling increases theinternal surface area, and decreases the degree of polymerisation and crystallinity.Usually the major fraction of lignin is solubilised together with part of thehemicellulose that is recovered mainly as oligomer. Alkaline pretreatment provedto be more effective on agricultural residues and herbaceous crops than on woodymaterials (Galbe and Zacchi, 2007). Pretreatment using lime instead of sodiumhydroxide is especially suited for agricultural residues, e.g. corn stover (Kim andHoltzapple, 2005), or hardwood, such as poplar (Chang et al., 2001).
Dilute acid pretreatment is performed by soaking the material in dilute acidsolution, usually below 4%, and then heating to temperatures between 120 and200°C. The residence time varies from several minutes up to few hours. The mostwidely investigated approaches are based on dilute sulphuric acid (Lloyd andWyman, 2005, Torget et al., 1992), since sulphuric acid is inexpensive andeffective. However, nitric acid, hydrochloric acid (Goldstein and Easter, 1992), andphosphoric acid (Israilides et al., 1978) have also been tested. The hemicellulose ishydrolysed and the main part is usually obtained as monomer sugars. It has beenshown that materials that have been subjected to dilute acid pretreatment may beharder to ferment because of the inhibitors formed during the treatment (Larsson etal., 1991).
The use of ionic liquids (e.g. 3-methyl-N-butylpyridinium chloride, 1-n-butyl-3-methylimidazolium chloride and 1-allyl-3-methylimidazolium chloride) forpretreatment has recently received attention as green solvents. Ionic liquids aresalts with very low volatility and high thermal stability (up to 300°C), and candissolve carbohydrates and lignin simultaneously forming hydrogen bonds betweenthe non-hydrated chloride ions and the hydrogen atoms of the hydroxyl groups
12
while minimising formation of degradation products (Dadi et al., 2006). However,it is still uncertain whether ionic liquids can be a feasible option in full-scalepretreatment of lignocelluloses (Yang and Wyman, 2008).
Figure 5 Scheme of steam pretreatment unit(adapted from Monavari et al., 2010).
Combined methods presented below are the mixture of physical and chemicalpretreatments. Steam pretreatment (formerly referred to as steam explosion) is oneof the most widely investigated and used methods for pretreatment oflignocelluloses. The raw material is usually treated with high-pressure saturatedsteam at a temperature typically between 160 and 240°C (corresponding to asaturated steam pressure between 6 and 34 bars). At the time set, which varies fromfew seconds to several minutes, some steam is rapidly vented from the reactor toreduce the pressure, and the contents are discharged into a large vessel to flash coolthe biomass (Figure 5). During pretreatment some of the dry matter, predominantlyhemicellulose, is solubilised, and is present in the liquid phase as oligomer andmonomer sugars, meanwhile the cellulose in the solid phase becomes moreaccessible to the enzymes. It has been shown that the effect of steam pretreatment,which renders a material suitable for enzymatic hydrolysis, is more likely due toacid hydrolysis of the hemicellulose than the “explosive” action when the pressureis released (Galbe and Zacchi, 2007). Drawbacks of steam pretreatment are partialdegradation of hemicellulose, production of enzymes and fermentation inhibitors,and incomplete separation of lignin and cellulose (Hendriks and Zeeman, 2009,Sun and Cheng, 2002).
13
Since hemicellulose, in particular that of hardwood and agricultural residues,contains acetyl groups that are cleaved during the reaction forming acetic acid,catalysis occurs, which is referred to as autohydrolysis (Fengel and Wegener,1989). Nevertheless, impregnation with other acids such as H2SO4 or SO2 (catalystaddition) prior to steam pretreatment has been proven to have a positive effect onthe sugar recovery through pretreatment by enabling lower temperatures andshorter residence times (Sassner et al., 2005). Moreover, impregnation improvesenzymatic hydrolysis as well as ethanol fermentation by reducing the formation ofinhibitory compounds from sugar degradation (Duff and Murray, 1996, Saddleret al., 1993). In some cases it is difficult to find the conditions that result in highrecovery of hemicellulose sugars, and at the same time also create a cellulosefraction which can be easily attacked by cellulases. This may require two-stepsteam pretreatment, where hemicellulose sugars are recovered at milder conditions,while the separated cellulose fraction is subjected to pretreatment under harshercircumstances. However, the difficulty with separation and washing of thepretreated material (slurry) at high pressure between the two steps and theincreased capital cost are major disadvantages. Monavari et al. (2010) reported thatone-step steam pretreatment, using linear or stepwise temperature increase, resultedin high ethanol yield and reduced formation of degradation products. Theyachieved as high ethanol yield using stepwise temperature increase mode, asobtained with one-step constant temperature pretreatment but at a significantlyshorter residence time (7 min versus 12 min), which is favourable in terms ofcapital cost.
Ammonia fibre/freeze explosion (AFEX) operates at high pressure but at moderatetemperate (below 100°C), hence the formation of sugar degradation products isminimised (Teymouri et al., 2005). Only a small amount of the solid material issolubilised, however, the structure is changed. The hemicellulose is degraded tooligomer sugars and deacetylated (Gollapalli et al., 2002), which is a probablereason for that the hemicellulose remains in the solid phase. By reducing thepressure after pretreatment, the volatile ammonia can be easily recycled.Herbaceous and agricultural residues are well suited for AFEX. However, thismethod works only moderately well on hardwoods, and is not attractive forsoftwoods due to their high lignin content (Holtzapple et al., 1991). The continuousversion of AFEX is called fibre extrusion (FIBEX, Dale et al., 1999).
Pretreatment with aqueous ammonia in a flowthrough mode involves percolatingammonia solution (5–15%) through a column reactor packed with biomass atelevated temperatures (160–180°C). This method is also known as ammoniarecycled percolation (ARP) process since ammonia is separated and recycled.Under these conditions, aqueous ammonia causes depolymerisation of lignin andcleavage of lignin-carbohydrate linkages (Mosier et al., 2005). High degree ofdelignification has been reported in tests with hardwood (Yoon et al., 1995) andagricultural residues (Iyer et al., 1996).
14
Hot water under pressure can penetrate the cell structure of raw material, hydratecellulose, and remove hemicellulose and part of the lignin (Mosier et al., 2005).This method is referred to as hydrothermolysis, or liquid hot-water (LHW)treatment. The pKa of water is affected by temperature: the pH of pure water at200°C is nearly 5. In addition, hot water cleaves hemiacetal linkages and liberatesacetic acid that catalyses break down of ether linkages. An alternative approach isbased on maintaining the pH between 4 and 7 by adding KOH when the pH dropsbelow 4 (Weil et al., 1998). The control of pH limits the chemical reactionsoccurring during pretreatment. The base differs in function from chemicals addedas catalysts in chemical pretreatment methods. The goal is to minimise degradationby avoiding the formation of monosaccharides that degrade at high temperature,and enzymes may be added at lower temperature for hydrolysis of the obtainedcellulose and hemicellulose oligomers. The optimal temperature has been proven tobe 180-190°C for corn stover (Mosier et al., 2005) and wheat straw (Perez et al.,2008) and 150-160°C for corn fiber (Mosier et al., 2005).
2.3.2 Enzyme productionEnzyme fermentation (e.g. with Trichoderma) can be carried out on various typesof carbon sources: purified celluloses, lignocelluloses and soluble sugars. Purifiedcellulosic substrates have been used for cellulase production in many studies, ascan be seen in the review of Persson et al. (1991), however, industrial enzymeproduction requires inexpensive raw materials. Therefore, much work has beendevoted to enzyme production from lignocellulosic substrates. Due to theirrecalcitrant structure, lignocelluloses have to be pretreated before being used ascarbon source in enzyme fermentation. The pretreated material consists of liquidand solid phases, which contain water-soluble sugars and water-insolublepolysaccharides, respectively. Hence, in enzyme fermentation either the wholepretreated slurry (Kovács et al., 2008), or only a certain part thereof, i.e. theseparated liquid (Szengyel et al., 2000) or washed solid fractions (Juhász et al.,2005, Szengyel et al., 2000) can serve as carbon source. During pretreatment,inhibitory compounds are formed in the liquid phase, which can impair theefficiency of the enzyme-fermenting microorganism. However, Szengyel andZacchi (2000) observed that acetic acid alone did not influence the cellulaseproduction, however, β-glucosidase secretion was increased with increasing aceticacid concentration. Furfural alone proved to be an inhibiting agent resulting in asignificant decrease in both cellulase and β-glucosidase production. In the presenceof furfural, however, acetic acid, applied at low concentrations, had a positiveeffect, cancelling the inhibition of furfural.
It was presumed that propagation of the cellulolytic microorganism on a certainlignocellulosic substrate would result in an enzyme profile, which is especiallysuitable for the hydrolysis of that particular material (Olsson et al., 2003).Trichoderma proved to grow and excrete cellulases on several pretreated
15
lignocellulosic material such as wheat straw (Acebal et al., 1985, Maheswari et al.,1993), corn stover (Juhász et al., 2005), spruce (Szengyel et al., 2000), poplar(Hayward et al., 1999), willow (Réczey et al., 1996, Szengyel et al., 1997), sugarcane bagasse (Adsul et al., 2004, Kawamori et al., 1986).
The establishments of soluble-sugar-based cellulase production plant wouldfacilitate the practical application of catabolite-derepressed mutants ofTrichoderma for large-scale ethanol production (Ike et al., 2010). Carbon sourcescontaining soluble sugars have several advantages: (1) soluble sugars can beobtained in bulk in many ways (molasses, hydrolysis of starch and lignocelluloses),(2) prior to enzyme fermentation the sugar solution can be sterilised readily viafiltration, (3) in the case of batch fermentation, the concentration of carbon sourcecan be increased easily in order to increase productivity, without facing theproblems of mixing and mass transfer limitation.
In lignocellulosic ethanol production point of view, cellulase enzymes can bepurchased from enzyme producers or can be fermented on site. The former has theadvantage of large-scale production, maintenance of latest technology, while thelatter can provide with tailor-made solution, simplified logistics, quick and flexiblesupply. Moreover, on-site enzyme production eliminates costs of enzymestabilisation, formulation and transportation and there are potential synergies (e.g.sharing utilities) between the enzyme fermenting step and the main, lignocellulose-to-ethanol process (Ike et al., 2010).
2.3.3 Enzymatic hydrolysisEnzymatic hydrolysis of cellulose is generally rate limiting in the lignocellulose-to-ethanol process (Lynd, 1996). Starch hydrolysis can be about 100-fold faster thanhydrolysis of cellulose under conditions anticipated for industrial processes (Zhangand Lynd, 2004). This difference can be attributed to the smaller fraction ofaccessible bonds, smaller frequency of chain ends and a much smaller fraction ofbonds cleaved in the liquid phase during hydrolysis: whereas cellodextrins areessentially insoluble at DP > 6–10 (Miller, 1963), maltooligosaccharides aresoluble at DP up to 60 (John et al., 1982).
Main factors that influence the rate of cellulose hydrolysis are either related to thesubstrate, such as accessible surface area, particle size, lignin content anddistribution, hemicellulose content, porosity; or to the enzymes, e.g. irreversibleadsorption, end-product inhibition, inactivation (Mansfield et al., 1999). Degree ofpolymerisation and cellulose crystallinity have been considered as importantfactors in determining the hydrolysis rate of refined cellulose substrates, however,data from several independent investigations indicate that these parameters alonedo not explain the recalcitrance of lignocellulosic substrates (Alvira et al., 2010).
16
It has been widely observed that the heterogeneous structure of cellulose gives riseto a rapid decrease in rate as hydrolysis proceeds, although the effects of cellulasedeactivation, irreversible enzyme adsorption on lignin and product inhibition aretaken into account. The declining reactivity of residual cellulose might be a resultof less surface area and fewer accessible chain ends and/or adsorption of inactivecellulases on the cellulose, which block further hydrolysis (Zhang and Lynd, 2004).Addition of surfactants, in particular non-ionic surfactants such as Tween 20,Tween 80, and polymers e.g. poly(ethylene glycol) during hydrolysis is capable ofmodifying the surface property of the lignocellulosic substrate and minimising theirreversible binding of cellulases, primarily on lignin but also on cellulose(Börjesson et al., 2007).
It is not possible to define a single optimum for enzymatic hydrolysis since thismay shift depending on factors such as temperature, pH, residence time and drymatter content (Galbe and Zacchi, 2002). The optimum temperature and pH is notonly a function of the raw material and the enzyme source, but is also highlydependent on the hydrolysis time. It has often been suggested that optimaltemperature and pH are 50°±5°C and 4.0-5.0, respectively (Galbe and Zacchi,2002). However, Tengborg et al. (2001) found that for longer residence times (>24h), a temperature of 38°C was optimal. In general, lower solids concentrationsresult in higher hydrolysis yield, especially for dry matter concentrations below5%. Nevertheless in batch hydrolysis, the initial solid concentration should beincreased as much as possible in order to obtain high sugar concentrations and thenhigh ethanol titre in the fermentation broth, which is essential with respect todecreasing the cost of downstream operations (Zaldivar et al., 2001). Jørgensen etal. (2007) demonstrated that it is possible to perform enzymatic hydrolysis up to40% initial dry matter (DM).
The enzyme loading has a high impact on the conversion of the cellulose.Increasing the dosage of cellulases, to a certain extent, can enhance the glucanconversion yield and rate of the hydrolysis, however, it also increases the cost ofthe process significantly.
In 2010, Harris et al. showed that certain proteins of glycoside hydrolase family 61(GH61) lack measurable hydrolytic activity by themselves but in the presence ofvarious divalent metal ions can significantly reduce the enzyme loading required tohydrolyse lignocelluloses. By incorporating the gene for one GH61 protein into acommercial Trichoderma reesei strain producing high levels of cellulolyticenzymes, they were able to reduce by half the total enzyme loading.
Trichoderma-derived cellulase systems supplemented with -glucosidases fromexternal source showed better hydrolytic performance than those withoutsupplementation (Xin et al., 1993). In some cases, especially when a significantfraction of hemicellulose remains after a less severe pretreatment, hemicellulases
17
added to cellulases exhibited a significant increase in the extent of celluloseconversion (Berlin et al., 2005).
Recycling of cellulase enzymes would be favourable in terms of processeconomics, however, the presence of solid phase in the reaction mixture makes itdifficult. As mentioned above, part of enzymes irreversibly binds on the surface oflignocellulose particles, and this part is not easy to recover (Sun and Cheng, 2002).Another problem is that a full-scale process requires enzyme cocktail that areoptimised in terms of the ratio of individual enzyme component, and recyclingprobably results in a modified composition of enzyme cocktail.
2.3.4 FermentationThe sugars released in the pretreatment and enzymatic hydrolysis steps arefermented to ethanol by yeast or bacteria. Desirable features of a microorganismsuitable for industrial ethanol production are the following: 1) it ferments ethanolwith high yield and productivity at as low pH as possible, 2) it has high tolerance toethanol and fermentation inhibitors, 3) it is able to utilise various types of sugars.The low pH is favourable with respect to reducing the risk of contamination causedby other microorganisms, e.g. Lactobacilli.
Saccharomyces cerevisiae is almost exclusively employed in industrial ethanolproduction of first generation, since it is robust enough to perform well underlarge-scale conditions. It fulfils more or less all desired features of industriallysuitable ethanologenic organism. The main drawback of Saccharomyces cerevisiaeis that the spectrum of fermentable sugars is limited: ordinary baker’s yeast is ableto utilise only hexoses (Table 2). Galactose fermentation is inducible, however, theconsumption of glucose and mannose is superior to that of galactose (Palmqvistet al., 1996).
Although bacteria usually have a broad sugar spectrum, they do not produceethanol with high yields. Zymomonas mobilis is an exception, it is able to fermentethanol with yields even higher than that of baker’s yeast. It tolerates well the lowpH, the high sugar and high ethanol concentrations (Swings and Deley, 1977).Major disadvantages of the wild type of this bacterium are the inability to fermentpentoses and low tolerance towards fermentation inhibitors present in the liquidphase of pretreated material (Olsson and Hahn-Hägerdal, 1993).
In nature, there are several yeasts and bacteria that utilise xylose and convert it intoethanol (Table 2). Unfortunately none of these is suitable for large-scale ethanolproduction. Different approaches exist for obtaining appropriate xylose-utilisingorganism via genetic engineering. One option aims the insertion of xylose-fermenting pathways – either fungal (Eliasson et al., 2000) or bacterial (Jeffries andJin, 2004) – into ordinary baker’s yeast. The recombinant Saccharomyces
18
cerevisiae cells, however, have no dedicated xylose transporters, and glucosetransporters have 100-times higher affinity to glucose than to xylose, thus xyloseuptake is inhibited in the presence of glucose (Hahn-Hägerdal et al., 2001). Otherapproaches try to modify natural pentose-fermenting organisms. Escherichia colican ferment both hexoses and pentoses, but the main product is acetic acid. Ohta etal. (1991) constructed mutant strains of Escherichia coli, which produce ethanolwith high yields.
Table 2 Sugar-fermenting ability of wild-type ethanologenic microorganisms(adapted from Rudolf, 2007).
Cellobiose Glucose Mannose Galactose Xylose ArabinoseYeastSaccharomycescerevisiae - + + +/- - -
Pichiastipitis + + + + + -
Candidashehatae - + + + + -
BacteriumZymomonasmobilis - + - - - -
During pretreatment, at elevated temperature, various kinds of compounds form,such as sugar and lignin degradation products and organic acids, which inhibit thefermenting microorganism. Dehydration of hexose and pentose monomers resultsin hydroymethyl furfural (HMF) and furfural, respectively. The HMF can befurther broken down into levulinic and formic acids. The latter is also formed fromfurfural under acidic conditions (Ulbricht et al., 1984). The lignin degradationproducts are mainly phenolic compounds. Figure 6 illustrates known inhibitionmechanisms of fermentation inhibitors on baker’s yeast. These compounds affectyeast metabolism in various ways, including extension of lag phase and reductionof growth rate, ethanol yield, specific ethanol productivity and cell viability(Almeida et al., 2007). To decrease inhibition and to achieve fast and efficientfermentation several methods are available. Yeast is able to convert inhibitorycompounds to less toxic ones, hence in fed batch fermentation, controlled additionof inhibitors can be realised, which results in simultaneous detoxification, therebythe toxic levels can be avoided (Ballesteros et al., 2002). Yeast adaption, i.e.increasing the capacity of yeast to resist or degrade inhibitors, can also be anoption. Even a short cultivation phase on pretreated liquid fraction significantlyimproves fermentation performance (Alkasrawi et al., 2006). Detoxification, e.g.overliming, can considerably increase the fermentability of carbon sourcescontaining of inhibitors, however, adds extra costs to the process, and often causesfurther sugar degradation (Aden, 2008).
19
Figure 6 Schematic view of known inhibition mechanisms of furans, weakacids and phenolic compounds in baker’s yeast (adapted fromAlmeida et al., 2007). HMF: inhibition of ADH (alcoholdehydrogenase), PDH (pyruvate dehydrogenase) and ALDH(aldehyde dehydrogenase), inhibition of glycolysis (either enzymeand/or cofactors). Furfural: same as HMF, plus cell membranedamages. Weak acids: ATP depletion, toxic anion accumulationand inhibition of aromatic amino acids uptake. Phenoliccompounds: uncoupling, generation of reactive O2 species andmembrane damage.
2.3.5 Downstream operationsAt first generation ethanol production, the main method for recovery of ethanolfrom the fermentation broth is the distillation, and presumably it will be the samefor ethanol production of second generation. Essentially, the total amount ofethanol is recovered from the fermentation broth in a stripping or beer column.Alternatively, pervaporation can be applied instead of stripping (Kazi et al., 2010).Thereafter, ethanol is concentrated in a rectifying column up to a concentration ofthe ethanol-water azeotrope (about 95% ethanol by weight). If water-free ethanol isneeded, purification beyond the azeotrope point can be achieved by furtherdistillation in the presence of an entrainer (e.g. benzene, cyclohexane) that issubsequently recovered, molecular sieve adsorption in zeolite columns, orpervaporation or other membrane-based operations (Busche, 1983).
Lignin is likely to be burned as a process fuel in a large-scale industry, becausemarkets for most lignin-derived by-products are much smaller than the markets forfuels. The dewatered lignin-rich solid residue is a good boiler fuel. Designs forlarger plants often involve cogeneration of electricity. For woody raw materials,combustion of process residues is sufficient to provide all of the steam and powerrequired by the plant, with some electricity usually exported for sale. At present,
20
Rankine cycle-based biomass combustion and power generation is appliedcommercially (Lynd, 1996).
Liquid fraction of the stillage, containing non-volatile soluble components such asunfermented sugars, can be utilised also as boiler fuel after concentration byevaporation, or can be digested anaerobically producing biogas that can also beburnt on site (Wingren et al., 2008), or can be exported for sale after purification.
2.3.6 Process configurations
Various kinds of coupling of enzyme production, enzymatic hydrolysis, hexose andpentose fermentation are possible, as illustrated in Figure 7.
Figure 7 Process strategies for conversion of pretreated lignocelluloses.Grey boxes represent combined biochemical processes, whichoccur in the same bioreactor. Hydrolysis 1 in the hybrid hydrolysisand co-fermentation (HHCF) configuration is also referred to asliquefaction.
In case of separate hydrolysis and fermentation (SHF) the four process steps underconsideration are entirely independent from each other. The major advantage of
21
SHF is the possibility to carry out both hydrolysis and fermentation at their optimaltemperature, i.e. at 40-50°C (with common cellulases) and at 30°C (in the case ofbaker’s yeast), respectively. In addition, by including separation and washing stepsafter the hydrolysis, and by fermenting only the liquid fraction obtained, theethanologenic microorganism can be recycled, and used many times in thefermentation. However, the major disadvantage of SHF is that sugars releasedduring hydrolysis inhibit the enzymes (Galbe and Zacchi, 2002).
Simultaneous saccharification and fermentation (SSF) combines hydrolysis andfermentation in one step. As the number of vessels required is reduced, the capitalcost decreases. The sugar produced in the hydrolysis is immediately consumed bythe fermenting microorganism, hence the end-product inhibition of -glucosidasecaused by glucose can be avoided, which results in higher hydrolysis rate. The lowconcentration of sugars also reduces the risk of contamination. Nevertheless, reuseof the fermenting organism is not feasible due to the residual lignin-rich solidmaterial. Since the temperature optima of hydrolysis and fermentation differ, acompromise temperature (35-37°C) must be applied in SSF. Development ofthermotolerant ethanologenic strains is anticipated to improve SSF performanceconsiderably (Galbe and Zacchi, 2002).
When pentose fermentation is considered, two approaches exist, i.e. using twofermenting strains, from which one is able to metabolise pentoses and the other isan ordinary hexose-fermenter. Alternatively, with a co-fermenting strain only onefermentation step is required. Co-fermentation eliminates the cost of some processequipments, such as slurry separator, extra fermentors, and the cost associated withpropagation of the pentose-fermenting organism, however, at present results inpoorer xylose-to-ethanol yield (Kazi et al., 2010). When the whole hydrolysis andco-fermentation are merged, the configuration is called simultaneoussaccharification and co-fermentation (SSCF). In the case of hybrid hydrolysis andco-fermentation (HHCF), thermostable cellulases with a high temperature optimum(~65°C) liquefy the pretreated material in a separate step decreasing the viscosityrapidly, and the rest of saccharification is completed simultaneously with the co-fermentation carried out at lower temperature (Aden and Foust, 2009).
Consolidated bioprocessing (CBP), combining all the four step mentioned above,employs either thermophilic anaerobic bacteria producing cellulosome enzymes,such as Clostridium thermocellum, or the mesophilic fungus, Fusarium oxysporum(Lynd et al., 2002). Although CBP is an attractive option in terms of cost reduction,the ethanol yield and ethanol tolerance of the organisms investigated are low. Thelow yield is partially due to by-product formation, e.g. acetic and lactic acids.However, recombinant strains may improve the performance of CBP in the future.
22
2.4 Process design and economics
2.4.1 Process simulation
Process simulation requires a flowsheeting program such as Aspen Plus, HYSYSor ChemCad to perform rigorous material and energy balance calculations.Because of their flexibility, these programs have many advantages in comparingdifferent process configurations. They also serve as powerful tools in performingsensitivity analyses due to the ease of changing process parameters. Allflowsheeting programs are based on modular approach where each module (e.g.distillation column) is a mathematical model of a unit operation (e.g. distillation).The simulation is performed by arranging different modules into a completeflowsheet that represents the process. In most techno-economic evaluations of thelignocellulose-to-ethanol process that have been performed in the last decade,Aspen Plus has been used (Galbe et al., 2007). The main reasons have been thecapability of handling vapour-liquid equilibrium and solid components. The latterare not involved in vapour-liquid equilibrium, therefore they are characterised byfewer property parameters.
In techno-economic studies, mostly two process concepts have been investigatedregarding the biochemical route of lignocellulosic ethanol production. The maindifference between them is the way in which cellulose chain is broken down; eitherdilute sulphuric acid or cellulolytic enzymes are used to hydrolyse the cellulose(Table 3). Most evaluations performed on the enzymatic process have applieddilute acid or steam pretreatment with acid catalyst (Gnansounou and Dauriat,2010).
2.4.2 Economics
Thorough reviews are available in the literature about process economics oflignocellulosic ethanol. Sivers and Zacchi (1996) gathered and compared costsfrom 22 techno-economic evaluations carried out from 1982 to 1993. Galbe et al.(2007) present a review, in which they compare the ethanol production costestimation of 15 studies undertaken in the United States and in Europe, publishedbetween 1996 and 2007. The sources reviewed by Gnansounou and Dauriat (2010)partially overlap with those of Galbe et al. (2007), however, some new evaluationspublished recently are also included.
Table 3 summarises the techno-economic reports from 1995 until now. Ethanolproduction cost always relates to absolute ethanol, even if the final product of theprocess is not absolute ethanol. While in the review of Sivers and Zacchi (1996)enzymatic, dilute and concentrated acid had similar share (8, 7 and 7 cases,respectively), since 2000 almost exclusively enzymatic processes have beenassessed. Within the enzymatic way, the SSF configuration proved to be dominant
23
(12 cases) over SHF, SSCF and CBP (7, 1 and 1 cases, respectively). The lowestand the highest ethanol production costs were 0.21 and 1.16 US$/L, respectively.The largest contributors to the ethanol production cost, i.e. the most importantparameters for the economic outcome, were the cost of raw material (30 – 90US$/dry tonne) and the capital cost (48 – 604 million US$). The latter is stronglyinfluenced by the plant capacity, which also varied in a wide range (100 – 1 587ktonne dry raw material/year).
Even if the costs were based on the same year by using indices for cost of capital,chemicals and labour, the large differences in ethanol production cost would notdisappear. They can be explained by variations in the process design and in theassumptions underlying the evaluations. Process variations are due to raw materialsused or conversion technologies applied (Table 3). However, the differencesoriginating from various assumptions in many cases overshadow the processvariations. Typical examples are differences in raw material cost (even at the sameraw material), plant capacity and investment parameters, e.g. depreciation periodand interest rate. Figure 8 shows the correlation between plant capacity and ethanolproduction cost: higher capacities generally results in lower production cost.
Therefore, great care must be taken when comparing ethanol production costs fromdifferent studies. Nevertheless, the economic studies are still valuable; they giveimportant information about which parts of the process are most expensive, andwhere bottlenecks, which need to be addressed by further research, can be found.
Figure 8 Ethanol production cost as a function of plant capacity, adaptedfrom Galbe et al. (2007).
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0 500 1000 1500 2000 2500 3000 3500
Prod
uctio
n co
st (U
S$/L
)
Capacity (ktonne/year)
24
Table 3 Type of process, capacity and production cost reported in techno-economic evaluations published since 1995. Raw materials utilisedare given in parenthesis; A: agricultural residue, CS: Corn stover,H: Hardwood, S: Softwood, SG: Switchgrass. For conversion fromSEK to US$ a factor of 7 SEK/US$ was used. Costs were notupdated by index.
Reference Process Capacity(ODT/y)
Production Cost(US$/L)
Sivers and Zacchi, 1995 Enz-SHF 100 000 (S) 0.61
Sivers and Zacchi, 1995 Dilute acid 100 000 (S) 0.65
Sivers and Zacchi, 1995 Conc. acid 100 000 (S) 0.64
Lynd et al., 1996 Enz-SSF 658 000 (H) 0.31
So and Brown, 1999 Dilute acid 263 000 (H) 0.36
So and Brown, 1999 Enz-SSF 268 000 (H) 0.34
Wooley et al., 1999 Enz-SSF 700 000 (H) 0.38
Aden et al., 2002 Enz-SHF 700 000 (CS) 0.28
Wingren et al., 2003 Enz-SHF 196 000 (S) 0.76
Wingren et al., 2003 Enz-SSF 196 000 (S) 0.69
Hamelinck et al., 2005 Enz-SSF 620 000 (H) 0.51Eggeman and Elander,2005 Enz-SSF 700 000 (CS) 0.26
Bohlmann, 2006 Enz-SHF 700 000 (A) 0.34
Wingren et al., 2008 Enz-SSF 200 000 (S) 0.59
Sassner et al., 2008 Enz-SSF 200 000 (H) 0.78
Sassner et al., 2008 Enz-SSF 200 000 (CS) 0.78
Sassner et al., 2008 Enz-SSF 200 000 (S) 0.62
Sassner and Zacchi, 2008 Enz-SSF 200 000 (S) 0.55
Sendich et al., 2008 Enz-SSCF 772 000 (CS) 0.27
Sendich et al., 2008 Enz-CBP 772 000 (CS) 0.21
Aden, 2008 Enz-SHF 701 000 (CS) 0.64
Humbird and Aden, 2009 Enz-SHF 700 000 (CS) 0.69
Laser et al., 2009 Enz-SSF 1 587 000 (SG) 0.45
Kazi et al., 2010 Enz-SHF 700 000 (CS) 0.90
Kazi et al., 2010 Dilute acid 700 000 (CS) 1.16
25
2.4.3 Detailed description of some techno-economic evaluations
In this section some of the techno-economic evaluations given in Table 3 are
described in details.
The first remarkable techno-economic report of National Renewable Energy
Laboratory (NREL) was released by Wooley et al. (1999), shown in Figure 9. In
the base process, poplar is pretreated with dilute sulphuric acid at 190°C for 10
minutes. The liquid fraction is detoxified by ion exchange and overlimed, and then
an SSF step is performed on the whole slurry merged after detoxification. Part of
the pretreated slurry is used for enzyme production. In the SSF step both hexose
and pentose sugars are co-fermented to ethanol by a recombinant Zymomonas
mobilis bacterium strain. Ethanol is removed from the broth through stripping and
purified by rectification and molecular sieve adsorption to 99.7%. The stillage is
dewatered by means of centrifugation. The solids, together with the concentrated
liquid from the evaporation step, are transferred to the boiler of the combined heat
and power (CHP) plant.
The second techno-economic study of NREL (Aden et al., 2002) uses the same
framework with some modifications. The raw material is changed to corn stover,
on-site production of enzyme is replaced by purchasing commercial enzyme
preparation; instead of SSF, SHF is applied, and instead of centrifuges, separation
of the stillage stream is performed with a belt filter.
Figure 9 Process scheme designed by Wooley et al., 1999.
In the version of NREL model updated for 2007 (Aden, 2008) NH4OH is used as
alternative conditioning agent instead of lime. An experimental screening study
identified circumstances of conditioning, at which treatment with NH4OH resulted
in negligible sugar losses and improved fermentation yields. However, the
improved fermentation yields is not captured in the model, since relatively high
26
ethanol yields (85% glucose conversion, 76% xylose conversion) are alreadyapplied. When lime is used as a conditioning agent, it is applied to the liquidfraction only (Figure 9), since it does not disperse well in the slurry. However,NH4OH is added in solution, and has good dispersing ability in the slurry, hencethe expensive solid-liquid separation equipment could be eliminated from theprocess. Nevertheless, this elimination is not implemented in the model.
In the update for 2008, released by Humbird and Aden (2009) an “oligomer holdstep” (a secondary mild hydrolysis) is added after pretreatment to convert xylo-oligomers into xylose to reach the target xylan conversion of 75%, since pilot-scaleresults in the pretreatment area showed insufficient xylan conversion for thecontinuous horizontal reactor used. The addition of this extra step is relativelyinexpensive, and is compensated by an increase in conversion of xylose to ethanol(from 76% to 80%).
Kazi et al. (2010) published a detailed report based on the 2002 design of NREL.On equal basis, 7 process alternatives were compared in terms of ethanolproduction cost. In 3 scenarios various pretreatment methods (dilute acid, hot waterand AFEX) were compared, while other process steps remained the same as in the2002 design. In other 3 scenarios alternative technologies (separate fermentation ofC5 and C6 sugars, on-site enzyme production, and pervaporation) were assessed atthe same (dilute acid) pretreatment. One case with a two-step dilute acid hydrolysiswas also investigated, i.e. acid catalysed the hydrolysis of both hemicellulose andcellulose. The lowest production cost was obtained at the scenario of dilute acidpretreatment with the base design.
In the following part, some techno-economic evaluations from Lund University(Lund, Sweden) are presented.
Wingren et al. (2003) modelled a softwood-based ethanol process based on SO2-catalysed steam pretreatment. Two configurations, one based on SSF and the otheron SHF, were evaluated and compared. The process parameters of pretreatment,SHF and SSF steps were adjusted to experimental data. Both enzymes and yeast(hexose fermenting Saccharomyces cerevisiae) are assumed to be purchased. Inboth configurations filtration is applied prior to distillation: either between thehydrolysis and fermentation at SHF configuration, or after SSF. The 94 wt-%ethanol obtained by rectification is considered as final product. The stillage isconcentrated by evaporation, and the syrup, together with the solid fraction comingfrom the filtration, is dried. The solid residue obtained is incinerated to supply theprocess with steam and the excess is pelletised, and sold as solid fuel. Theconfiguration of SSF was found to be less expensive, primarily due to the lowercapital cost and the higher overall ethanol yield.
27
Wingren et al. (2008) also compared various downstream alternatives to decreasethe energy demand of the softwood-to-ethanol process. In the base case, somemodifications are introduced regarding the former process model. Yeast is assumedto be produced directly in the fermentors prior to the start of SSF, which is theexclusive fermentation configuration in the study. Solid-liquid separation isperformed after distillation: the stillage stream is sent to two decanter centrifugesoperating in parallel. The evaporation concentrate and the solid fraction of thestillage are co-dried. The alternative configurations include the following options:1) the number of effects in the evaporator is increased; 2) the number of strippers(from two to one) is reduced, and the stripper is integrated with the evaporator; 3)mechanical vapour recompression is used to increase the temperature of the latentheat leaving the last effect, hence a significant part of the primary steam can bereplaced, however, the recompression requires supplementary electrical energy; 4)anaerobic digestion of the liquid fraction of stillage is implemented, and the biogasfuels the steam boiler. Both anaerobic digestion and mechanical vapourrecompression options resulted in lower overall energy demand and ethanolproduction cost than the base case.
Sassner et al. (2008) compared the techno-economic performance of three differentlignocelluloses: willow, corn stover and spruce. The process model is a modifiedand extended version of that used by Wingren et al (2008). Yeast cultivation takesplace in separate propagation tanks, and separation of the stillage is carried outwith a filter press. A model of wastewater treatment was added, which consisted ofanaerobic digestion followed by an aerobic treatment step. During these stepssludge forms, however, further treatment of sludge is neglected in the techno-economic model. The evaporated syrup, part of the dried solid and the biogasproduced in anaerobic digestion are incinerated in the steam boiler to cover theheat demand of the process. Similar trends were observed for the three rawmaterials when varying important process parameters, which implied that theprocess configuration employed determined the outcome to a greater extent thanthe choice of raw material.
Sassner and Zacchi (2008) compared various options for the utilisation of solidresidue formed in a spruce-based process, such as production of pellets, electricityand heat for district heating, in terms of overall energy efficiency and ethanolproduction cost. In the model used in their study the overhead vapour of therectifier is dehydrated by means of molecular sieve adsorption to 99.8 wt%. Onlypart of the solid fraction, which is later pelletised, is dried. The heat demand of theethanol process is covered by a CHP plant. With the assumed prices of co-products, utilisation of the solid residue for heat and power production was foundto be an economically favourable option.
29
3 MATERIALS AND METHODS
3.1 Raw materials
In the experimental work hemp hurds, wheat straw and wheat meal were used aspolysaccharide containing raw materials. In the techno-economic studies spruce-based ethanol processes were investigated.
Industrial hemp (Cannabis sativa L.) was cultivated by Hungarohemp (Nagylak,Hungary). The 1.5-2.5 m-long hemp stalks were retted, dried, and then crushed byfinned rolls. The fibres remained more or less intact, whereas the hemp hurds werebroken into 1-2 cm-long pieces with a width of 1–5 mm. Hemp hurds were used forsteam pretreatment at that size. Hemp hurds investigated contained 40.1%cellulose, 19.6% hemicellulose and 21.7% lignin.
Wheat straw was provided by Lunds Civila Ryttarförening (Lund, Sweden). It waschopped in a hammer mill and sieved to obtain a fraction of 2-10 mm. It comprised38.8% cellulose, 31.3% hemicellulose and 18.5% lignin.
Wheat meal (dried-milled wheat grain) with an average particle size of 2.5 mm wasreceived from Sileco (Halland, Sweden). It consisted of 72.7% starch and 24.3%starch-free residue. The latter is composed of 17.5% cellulose, 24.5%hemicellulose, 18.2% lignin (on dry matter basis of starch-free residue).
Table 4 Assumed composition and potential ethanol yield of spruce chipsapplied in the techno-economic analyses.
On-site CEF(preliminary)
On-site CEF(modified)
Alternativestillage treatment
Glucan (%) 44.0 37.9 43.5Mannan (%) 13.0 9.9 12.8Galactan (%) 2.3 1.8 2.1Xylan (%) 6.0 4.3 5.1Arabinan (%) 2.0 1.3 1.5Lignin (%) 27.5 28.0 29.4Potential ethanol yield fromhexoses (L/dry tonne) 426 356 420
The assumed compositions of spruce applied in the techno-economic analyses arepresented in Table 4. They were determined at Lund University (Lund Sweden) byanalysing three batches of spruce chips in three different years according to thestandardised method of NREL described in Section 3.3. The reason of changing theinput of the process model (e.g. composition of raw material, sugar recovery of
30
steam pretreatment, conversion of enzymatic hydrolysis, yield of fermentation)between the techno-economic analyses is that the most up-to-date experimentalresults were used as input data. As spruce contains a significant amount of mannanin its hemicellulose, the potential ethanol yield from hexose sugars is relativelyhigh, e.g. compared to that of agricultural residues.
3.2 Enzymes
Amylolytic enzymes
α-Amylase (Termamyl SC; Novozymes A/S, Bagsværd, Denmark) andamyloglucosidase (Spirizyme Fuel; Novozymes) amylolytic enzymes were used forstarch liquefaction and saccharification, respectively. The amylolytic activity ofthese enzymes were not measured; the amylolytic enzymes were loaded based ontheir weight according industrial recommendation (Destexhe et al., 2004).
Cellulolytic enzymes
In enzymatic hydrolysis and SSF experiments, cellulase (Celluclast 1.5 L andNS50013) and β-glucosidase (Novozym 188 and NS50010) enzyme preparations(all from Novozymes) were used. Cellulase preparations had an activity of~65filter paper units (FPU)/g enzyme solution, determined by the IUPAC protocol(Ghose, 1987), and ~33 international unit (IU)/g enzyme solution β-glucosidaseactivity according to the method of Berghem and Petterson (1974). Novozym 188and NS50010 preparations had a β-glucosidase activity of ~350 IU/g enzymesolution.
3.3 Compositional analysis
The carbohydrate and lignin contents were determined according to the NRELstandardised method (Sluiter et al., 2008c). Oven-dried samples to be analysedwere milled and sieved to a mesh size of 1 mm. They were treated in a two-stepacid hydrolysis with 72% H2SO4 for 1 h at 30°C and then with 4% H2SO4 for 1 h at121°C. Sugar contents of the supernatants were analysed with HPLC (high-performance liquid chromatography). The dried residue consisted of acid-insolublelignin and acid-insoluble ash. Total ash and acid-insoluble ash contents weremeasured by ashing the sample at 575°C for at least 4 h until the sample weightwas constant (Sluiter et al., 2008a). To determine the total sugar content in theliquid fraction of the pretreated material, samples were treated with 4% H2SO4 forat least 30 min at 121°C (Sluiter et al., 2008b). Sugar contents were determined byHPLC.
31
Non-volatile water and ethanol extractives were determined by means ofaccelerated solvent extraction (ASE) according to the NREL protocol (Sluiteret al., 2008d). ASE was performed on the 0.18-0.85 mm fraction of the rawmaterial (Hames et al., 2008). Non-volatile extractives remaining in the flask afterevaporation were measured by weight after drying at 105°C.
In compositional analyses presented so far, at least duplicates were run.
Starch content was determined in two-step enzymatic hydrolysis consisting ofliquefaction and saccharification. The dry matter content was set to 35%. In thefirst step the material with 0.5 g/kg DM Termamyl SC was liquefied at 85°C, at pH5.5 for 3 h. In the second step Spirizym Fuel was added at the ratio of 0.5 ml/kgDM at pH 4.2 and the mash was treated at 60°C for 24 h to carry out total starchhydrolysis. The saccharified mash was filtered and the glucose content of thesupernatant was measured by HPLC. The washed solid residue is referred to asstarch free residue. By now the author is aware that at 35% DM significant glucosereversion occurs during the saccharification step, however, this fact was unknownfor him, when these measurements were carried out. The reason for setting the drymatter content to 35% was that industrial enzymatic starch liquefaction isperformed at 30-35% DM. Since the starch content was determined by measuringthe liberated glucose, it was underestimated due to glucose reversion. Wheat mealwas presaccharified similarly, except for the time of saccharification, which was 2h instead of 24 h.
3.4 Steam pretreatment
Steam pretreatment of hemp hurds and wheat straw was performed at the BiomassUnit of CIEMAT (Madrid, Spain) and at the Department of Chemical Engineering,Lund University (Lund, Sweden), respectively.
3.4.1 Steam pretreatment of hemp hurds
The steam pretreatment units have been described by Ballesteros et al. (2004). Fivepretreatments were performed for 10 min without impregnation at 220 and 230°Cin a 2-L reactor as well as at 200, 210 and 220°C in a 10-L reactor (Figure 10) tostudy the effect of temperature and reactor size. The two reactors investigateddiffered in their temperature profile; heating and cooling phases lasted longer in thecase of the 10-L reactor. Dry matter loading was 200 g in the 2-L reactor and 500 gin the 10-L reactor, and air-dry material was fed to the reactor. The efficiency ofpretreatment was evaluated in terms of sugar recoveries after pretreatment,polysaccharide conversions in enzymatic hydrolysis, and ethanol yields obtained inSSF.
32
Figure 10 Steam pretreatment unit at CIEMAT (Madrid, Spain):1) 10-L reactor, 2) flash cyclone.
3.4.2 Steam pretreatment of wheat straw
The wheat straw was immersed in aqueous solution containing 0.2% H2SO4 at aliquid:dry straw weight ratio of 20. It was stored in sealed buckets for 1 h and thenit was squeezed to an average dry matter content of 43%.
Steam pretreatment was performed in a unit, which has been described byPalmqvist et al. (1996), comprising a 10 L pressurised vessel, and a flash cyclone.Each batch fed into the reactor was 600 g in wet weight. Previously optimisedconditions for wheat straw were applied (Linde et al., 2008), i.e. the temperaturewas maintained at 190°C for 10 min. The total solids recovery obtained was 79%.Almost 100% of the glucose was recovered, based on the original amount in theraw material, of which 95% was recovered in the solid fraction. The recovery ofxylose, which is the main sugar in hemicellulose, was 65% and 2%, in the liquidand solid phases, respectively.
3.5 Enzymatic cellulose hydrolysis
Enzymatic hydrolysis of the washed solid fraction was carried out in shake flasksin a reaction volume of 25 mL at a water-insoluble solid (WIS) content of 5% at50°C and 150 rpm using NS50013 and NS50010 enzymes. Enzyme loading of NS50013 was 15 FPU/g WIS and the volumetric ratio of NS50010 to NS50013 was0.1. Sodium citrate buffer (0.05 M) was used to maintain pH at 5.0. Triplicateswere incubated for 72 h. Released sugars were analysed by HPLC.
1)
2)
33
3.6 Simultaneous saccharification and fermentation
3.6.1 Shake flask experiments
Non-sterile simultaneous saccharification and fermentation of washed fibre fractionof pretreated hemp hurds was performed in shake flasks in a volume of 50 mL at aWIS content of 10% by applying commercial enzyme preparations (NS50013,NS50010) and hexose fermenting Saccharomyces cerevisiae (DER-CIEMATCulture Collection No. 1701) at 32°C. Enzyme loading of NS 50013 was 25 FPU/gglucan. Volumetric ratio of NS50010 to NS50013 and agitation speed were 0.1 and150 rpm, respectively. Yeast extract, NH4Cl, KH2PO4 and MgSO4·7H2O wereadded in final concentrations of 5, 2, 1 and 0.3 g/L, respectively. Flasks wereinoculated with a 4% (v/w) S. cerevisiae culture. Sodium citrate buffer (0.05 M)was used to adjust the initial pH to 5.0. Triplicates were incubated for 72 h. Ethanolwas measured by gas chromatography.
3.6.2 Experiments in laboratory fermentorsSSF experiments were performed in 2-L laboratory fermentors with a total weightof 1.4 kg. Presaccharified wheat meal (PWM), slurry of steam pretreated wheatstraw (SPWS) and various mixtures of these were used as substrates in SSF at atotal WIS content of 5% (Table 5). The PWM to SPWS WIS ratio for theinvestigated four mixtures (A to D) was 0.15, 0.3, 0.4 and 0.5, respectively. WhenSSF was performed on pure PWM, the WIS content was set to 2.8% (Table 5).
Table 5 Details of substrates used in the SSF experiments carried out inlaboratory fermentors. Total weight: 1400 g, DM of PWM: 27%,WIS of PWM: 9%, DM of SPWS: 11%, WIS of SPWS: 8%.
Substrate Mass (g) Glucose equivalent (%)PWM SPWS PWM SPWS
WIS Liquid WIS LiquidSPWS - 921 - - 3.7 0.2Mixture A 124 783 0.1 2.5 3.2 0.2Mixture B 247 645 0.3 5.0 2.6 0.1Mixture C 329 553 0.4 6.6 2.2 0.1Mixture D 412 461 0.5 8.3 1.9 0.1PWM 465 - 0.5 9.4 - -
The SSF experiments were performed using Celluclast 1.5L and Novozym 188,added at a dosage of 15 FPU/g glucan and 17 β-glucosidase IU/g glucan,respectively. The fermentor loaded with the steam pretreated material and thenutrients were sterilised separately at 121°C for 20 min, while PWM was not
34
sterilised. S. cerevisiae (ordinary baker’s yeast) was suspended in sterilised waterand added to the fermentor at a concentration of 5 g DM/L. As nutrients(NH4)2HPO4, MgSO4·7 H2O and yeast extract were used at a concentration of 0.5,0.025 and 1.0 g/L, respectively. SSF was performed at 36.5°C and pH wasmaintained at 5±0.2. The experiments were run at least for 72 h.
3.7 Analysis of sugars, ethanol and by-products
Reducing sugar contents were measured colorimetrically using dinitrosalicylic acidaccording to Miller's method (1959). The liquid fractions were analyzed withHPLC equipped with a refractive index detector. Cellobiose, glucose, mannose,xylose, galactose and arabinose were separated using an Aminex HPX-87P (Bio-Rad Laboratories, Hercules, CA) column at 85°C. Ultrapure water was used aseluent at a flow rate of 0.6 mL/min.
In Paper I, furfural and HMF contents were analysed by HPLC, equipped with anUV-diode-array detector, and the separation was performed with a Aminex HPX-87H column (Bio-Rad Laboratories, Hercules, CA) at 55°C with an eluent of 5 mMH2SO4 (89%) and acetonitrile (11%) at a flow rate of 0.7 mL/min. Separation ofacetic and formic acids was carried out using an Aminex HPX-87H column at65°C. The mobile phase was 5 mM H2SO4 at a flow rate of 0.6 mL/min. Ethanolcontent was measured using a gas chromatograph equipped with a Carbowax 20Mcolumn (Agilent Technologies, Santa Clara, CA) operating at 85°C and a flameionisation detector.
In Paper II, lactic acid, glycerol, acetic acid, ethanol, HMF and furfural wereseparated using a Aminex HPX-87H column at 65°C. The eluent was 5 mM H2SO4at a flow rate of 0.5 mL/min.
3.8 Statistical analysis
Statistical evaluation was carried out using Statistica 8.0 software (Statsoft Inc.,Tulsa, OK). The significance level was 0.05, and the two-tailed t-test probability isdenoted by p.
3.9 Methodology of techno-economic analysis
Mass and energy balances were solved using the commercial flowsheeting programAspen Plus 2006.5 (Aspen Technology, Inc., Cambridge, MA). Physical propertydata for biomass components such as polysaccharides and lignin were derived fromthe NREL database (Wooley and Putsche, 1996). Aspen HX-Net 2006.5 (Aspen
35
Technology, Inc.) was used to design a near-optimal heat exchanger network thatwas implemented in the process model in Aspen Plus. Energy flows of rawmaterial, pellets and biogas were calculated by using lower heating values (LHV).
Fixed capital investment costs were estimated either with Aspen Icarus ProcessEvaluator 2006.5 (Aspen Technology, Inc.) based on costs for 2009 or from vendorquotation. Conventional equipments such as pumps, tanks, columns, heatexchangers etc. together with the auxiliary units like piping, instrumentation etc.and indirect costs arising from engineering, fees, freight etc. were cost estimated inIcarus. However, costing of special equipments, which were not included in Icarus,such as the pretreatment unit, dehydration system, filter press, belt filter, dryer,pellet machine, steam boiler and flue-gas condenser were based on vendorquotation. In Icarus, the construction material was assumed to be 304 stainless steelfor all process vessels.
In addition, capital has to be invested also in raw materials, products and creditallowance for customers. These costs are referred to as working capital. Peters andTimmerhaus (1991) recommend that the working capital should be estimated as thesum of raw materials needed for 30 days production, 30 days of product storage,accounts receivable and 30 days cash on hand for monthly payment of wages, rawmaterials etc. In this evaluation these data have been used with the exception thatthe time for storage of chemicals was 10 days. This modification was introduced byWingren et al. (2008).
The annual capital cost was determined by multiplying the fixed capitalinvestments by an annuity factor of 0.110, corresponding to a depreciation periodof 15 years and an interest rate of 7%. The annual working capital was a product ofthe working capital and the interest rate.
Since the economic analyses were based on Swedish market conditions, all costsare presented in Swedish kronor (SEK, 1 € ≈ 10.5 SEK, 1 US$ ≈ 7.3 SEK, 1 SEK ≈30 Hungarian Forint, in 2010). Other costs comprise the costs of labour, insuranceand maintenance. Table 6 summarises the economic data used in the studies.
It must be emphasised that the primary aim was not to obtain an absolute ethanolproduction cost, but to compare the production cost of various process alternativeson an equal basis, and to study the effect of different process and economicparameters by performing sensitivity analyses.
36
Table 6 Prices and costs used in the economic evaluation.
Raw materialSpruce 557 SEK/dry tonne
Co-product incomePellets 195 SEK/MWhUpgraded biogas 600 SEK/MWhElectricity, spot price 350 SEK/MWhElectricity certificate 200 SEK/MWhDistrict heat 280 SEK/MWh
CO2 0.03 SEK/kgChemicals
SO2 1.5 SEK/kg
H2SO4 0.5 SEK/kg
NH3 (25 %) 2 SEK/kg
H3PO4 (50 %) 5 SEK/kgDefoamer 20 SEK/kg
(NH4)2HPO4 1.5 SEK/kg
MgSO4.7 H2O 4.4 SEK/kg
Molasses 1 SEK/kg
Enzymesa 28.5 SEK/106 FPUSoy-meal 1.5 SEK/kg
(NH4)2SO4 1.1 SEK/kg
KH2PO4 1 SEK/kg
FeSO4.7H2O 1 SEK/kg
Utilities
Electricityb 450 SEK/MWh
Cooling water 0.1 SEK/m3
Process water 1.4 SEK/m3
Other costsLabour 600 000 SEK/employee/yearInsurance 1 % of annual fixed capitalMaintenance 2 % of annual fixed capital
a When enzyme preparation is purchased.b When electricity is purchased.
37
4 RESULTS AND DISCUSSION OF EXPERIMENTAL PART
4.1 Hemp hurds as potential raw material for ethanol production(Paper I)
Hemp hurds upstream-processing lines, consisting of steam pretreatment andsubsequent enzymatic hydrolysis or simultaneous saccharification andfermentation, were investigated to produce fermentable sugars or ethanol,respectively. The lines differed in pretreatment conditions. The efficiency ofpretreatment was assessed in terms of sugar recoveries, polysaccharide conversionsin enzymatic hydrolysis, and ethanol yields obtained in SSF.
4.1.1 Steam pretreatmentComparing the compositions of raw material and pretreated solid fractions,increase in glucan and lignin contents and decrease in xylan, mannan and galactancontents were observed. This can be attributed to solubilisation of thehemicellulose fraction through pretreatment. This phenomenon was also observedby the pretreatment of wheat straw.
At least 82% of the initial glucan was recovered in the slurry, mainly in the water-insoluble fraction (Figure 11). Lost glucan was the portion hydrolysed to glucose,however, later it was further degraded. Xylan recovery varied between 18-66% oforiginal. The conditions investigated were more severe than the optimum for xylan,hence xylan recovery decreased drastically by increasing the temperature. Thewater-soluble xylan made up a significant share of the recovered part. These poorrecoveries might be due to the rather long residence time (10 minutes) and the lowmoisture content of the raw hurds fed to the reactor (~10%). These conditions havebeen reported to have unfavourable effects on xylan recoveries (Sassner et al.,2005). Since arabinan was not detected in the raw hemp hurds, the furfural presentmust have originated from xylan. Based on mass balance calculations (see Paper Ifor data), it was also concluded, that a significant amount of the original xylan wasnot degraded to furfural but other compounds such as formic acid.
The aim of using 2-L and 10-L reactors was to investigate whether the sametemperature and residence time resulted in the same severity. It was observed that,at the same temperature (220°C), larger losses were obtained in case of the 10-Lreactor. Glucan and xylan losses obtained in the large reactor at 210 and 220°Cwere similar to those obtained in the small one at 220 and 230°C, respectively.These indicate that a given treatment temperature resulted in more severepretreatment in the 10-L reactor. This might be due to differences in the heatingand cooling profiles of the two reactors.
38
Figure 11 Glucan (A) and xylan (B) recoveries through steam pretreatment ofhemp hurds, expressed in % of original, as a function of reactorvolume and pretreatment temperature.
0102030405060708090
100
220°C 230°C 200°C 210°C 220°C
2 L 10 L
Water-insoluble Water-soluble LostG
luca
n re
cove
ry(%
of o
rigin
al)
A)
0102030405060708090
100
220°C 230°C 200°C 210°C 220°C
2 L 10 L
Water-insoluble Water-soluble Lost
Xyla
n re
cove
ry(%
of o
rigin
al)
B)
39
4.1.2 Enzymatic hydrolysis and SSF
Glucan conversion after pretreatment in the 10-L reactor at 210°C was 83% oftheoretical, which was significantly higher than that in case of the 200°C (p=10-5)and 220°C (p=0.0038) pretreatments (Figure 12). In the 2-L reactor, there was nosignificant difference between the two temperatures (p=0.3876).
Figure 12 Conversions of glucan and xylan in enzymatic hydrolysis,expressed in % of theoretical, as a function of pretreatment reactorvolume and pretreatment temperature.
Considering the two reactor sizes separately, it can be seen that xylan conversiondecreased monotonically with the elevation of temperature. In other studies, thesame trend has been observed, i.e., xylan conversions decreased when pretreatmentseverity was enhanced (Öhgren et al., 2005, Sassner et al, 2005). More severepretreatment solubilises the less recalcitrant regions of hemicellulose to a greaterextent, hence the proportion of resistant, less-accessible regions increases in thepretreated solid fraction, which might interpret the declining xylan conversions inenzymatic hydrolysis. Pretreating hemp hurds in the small reactor at 220°C resultedin significantly higher glucan and xylan conversion than in the large one at thesame temperature (p=0.0343 and 0.0008, respectively).
Figure 13 shows the SSF yields based on glucan and mannan contents of thesubstrate (washed, steam pretreated hurds). The highest ethanol yield from washedfibres, 70% of theoretical, was obtained from the pretreatment at 210°C. This wassignificantly higher than that at 200°C (p=0.0003); nevertheless, the difference was
80 82
62
8376
6558
96
83
46
0
20
40
60
80
100
220°C 230°C 200°C 210°C 220°C
2 L 10 L
GlucanXylan
Con
vers
ion
in e
nzym
atic
hyd
roly
sis
(% o
f the
oret
ical
)
40
not significant compared to that at 220°C (p=0.4925). No significant differenceswere obtained either in the 2-L reactor between the two investigated temperatures(p=0.2081) or between the two reactor sizes at 220°C (p=0.2360).
Figure 13 Ethanol yields in simultaneous saccharification and fermentation(SSF), expressed in % of theoretical, based on glucan and mannancontents of the substrate (steam pretreated hurds), as a function ofpretreatment reactor volume and pretreatment temperature.
After pretreatment, solubilised sugars were present mainly in oligomeric forms(Figure 14). On the contrary, by the pretreatment of impregnated wheat straw,sugars in the liquid phase were present in monomeric form. These can be attributedto the absence and presence of exogenous acid catalyst, respectively. Overallglucose yields varied between 63 and 75% of theoretical (based on thepolysaccharide contents of raw hurds), and they showed the same pattern as theglucan conversions in enzymatic hydrolysis, since the amounts of monomeric andoligomeric glucose in the liquid fraction after pretreatment were rather small.Overall xylose yields ranged between 13 and 63% of theoretical and decreasedmonotonically with higher temperatures in both reactors. Beside the xylosereleased in enzymatic hydrolysis, however, xylose solubilised during pretreatmentmade up a significant share.
64 67
38
70 68
0
20
40
60
80
100
220°C 230°C 200°C 210°C 220°C
2 L 10 L
SSF
yiel
d (%
of t
heor
etic
al)
41
Figure 14 Overall yields of glucose (A) and xylose (B) through steampretreatment and enzymatic hydrolysis, expressed in % oftheoretical (based on the polysaccharide contents of raw hurds).
0102030405060708090
100
220°C 230°C 200°C 210°C 220°C
2 L 10 L
Ove
rall
gluc
ose
yiel
d (%
)Hydrolysis monomers
Pretreatment monomers
Pretreatment oligomers
A)
0102030405060708090
100
220°C 230°C 200°C 210°C 220°C
2 L 10 L
Ove
rall
xylo
se y
ield
(%)
Hydrolysis monomers
Pretreatment monomers
Pretreatment oligomers
B)
42
The best pretreatment temperature in terms of overall glucose yields was proved tobe 210°C, where 336 g glucose was obtained from 1 kg of dry raw material,whereas with regard to total sugar yield (glucose + xylose overall yields) it wasshifted to 200°C. At this temperature, 414 g sugar was recovered from 1 kg of dryraw material. The highest overall ethanol yield, which was achieved with thepretreatment at 210°C, was found to be 141 g ethanol/kg dry raw material.
Raw hemp hurds contain a remarkable amount of xylan, however, xylan recoverythrough steam pretreatment was poor at the temperatures required for achievingproper cellulose digestibility. A two-step steam pretreatment is a possiblealternative for preventing the less-recalcitrant hemicellulose fraction from furtherdegradation, thus improving its recovery (Söderstrom et al., 2003). Nevertheless,the additional pretreatment step increases both the capital investment andoperational costs. Alternatively, acidic impregnation of the raw material beforesteam pretreatment results in closer temperature optima for the yields of glucoseand xylose (Sassner et al., 2005). Neither the two-step pretreatment nor theimpregnation has been applied on hemp hurds. The method of linear or stepwisetemperature increase (Monavari et al, 2010) can also be an interesting option,however, so far it has only been tested with softwood that has mannan-richhemicellulose.
Figure 15 Ethanol concentration expressed in g/L versus time duringsimultaneous saccharification and fermentation of steam pretreatedwheat straw (SPWS), presaccharified wheat meal (PWM) andmixtures of the two (A to D, see Table 5 for details).
0
10
20
30
40
50
60
0 20 40 60 80 100
Mixture D
PWM only
Mixture C
Mixture B
Mixture A
SPWS only
Etha
nol c
once
ntra
tion
(g/L
)
Time (h)
43
4.2 Integrating production technologies of first and secondgenerations prior to simultaneous saccharification andfermentation (Paper II)
To decrease the energy demand of the downstream process high ethanol titre isessential. Besides increasing the WIS concentration in batch SSF and running SSFin fed-batch mode, supplementing the pretreated lignocellulosic substrate withstarch hydrolysate can be a promising option. Therefore, the SSF performance ofmixtures of SPWS and PWM was evaluated. The effect of varying theirproportions on final ethanol concentration and ethanol yield was investigated.
4.2.1 Effect of PWM on the final ethanol concentration obtained in SSFSignificant differences, in terms of the initial rate of ethanol formation, wereobserved between SSF on pure SPWS and SSF on mixtures containing differentproportions of PWM. During the first 2 hours of SSF, the ethanol productivity was1.6 g/L/h in the case of pure SPWS, whereas it was ≥4.7 g/L/h for PWM alone orPWM mixed with SPWS. This could be due to the high water-soluble sugar contentof PWM (Table 5) present at the beginning of fermentation, mainly as glucose,which was consumed rapidly, resulting in an increased rate of ethanol formation. Inpure SPWS, the major part of the glucose is in polymeric form bound in the solidphase, and this had to be hydrolysed before fermentation. However, glucoseconcentration > 0 g/L was measured in the solution during the initial 8 hours,which means that hydrolysis was not the rate-limiting step in this case. Possiblereason for the low ethanol productivity occurred at the beginning of SSF carriedout on pure SPWS is that ethanol production was inhibited by the degradation offurfural and HMF to furfuryl alcohol and HMF alcohol, respectively. The mostrapid ethanol formation was obtained with pure PWM. Ethanol concentrationsincreased with increased glucose content in the SSF experiments (Figure 15). Thehighest ethanol concentration obtained was 56.5 g/L (at 96 h) when SSF wascarried out on mixture D.
4.2.2 Effect of mixing lignocellulosic and starch-containing substrates onethanol yield of SSF
The ethanol yield is usually reported as g EtOH/g DM of the raw material.However, this kind of yield representation was not appropriate in this case, becausemixtures of materials were used. Therefore, the yields are expressed as apercentage of the theoretical maximum, considering only the glucose available inthe substrates, as usually galactose and the pentoses are not fermented to ethanol bybaker’s yeast (Palmqvist et al., 1996). These sugars were not consumed in any ofthe SSF experiments, which validates this assumption.
44
Figure 16 shows the ethanol yields for the SSF of various mixtures of substrate.The yield from pure SPWS was 68%, which is in agreement with the resultsobtained by Linde et al. (2008). The yield observed for pure PWM was 91%, whichis also typical for fermentation of starch-containing materials (Destexhe et al.,2004). In mixtures of the two substrates, the yield increased as the proportion ofPWM was increased. The highest ethanol yield, 99%, was obtained for SSF ofMixture D. This yield is rather high and could be due to the error of determinationof cellulose and starch contents. However, this would affect all the trials, thus thetrend of yield would remain similar. Hence the most important conclusion is thathigher ethanol yield can be achieved with a certain mixture of SPWS and PWM(mixture D) than using PWM solely. This is highly favourable in terms of processeconomy.
Figure 16 Ethanol yield (grey bars), expressed as % of theoretical, based onglucose exclusively, in simultaneous saccharification andfermentation of pure substrates and their mixtures (see Table 5 fordetails). White bars represent potential ethanol yield aftercorrection of lactic acid.
Small amounts of lactic acid were produced after 40 h of fermentation in all cases.Lactic acid can be formed from both hexoses and pentoses as a microbial metabolicproduct. As the lactic acid yield is 1 g/g consumed sugar regardless of the type ofmonomeric sugar, the amount of additional ethanol that could have been producedfrom hexoses instead of lactic acid, can be estimated as follows (the factor of 0.46corresponds to 90% of the maximum theoretical ethanol yield for hexose sugars):mass = 0.46 ∙ (mass − mass )
0
20
40
60
80
100
SPWS A B C D PWM
Etha
nol y
ield
(% o
f the
oret
ical
)
45
Several fermentation inhibitors were present in the slurry during SSF in the case ofSPWS-containing mixtures. The most significant inhibitors were acetic acid andfurfural because of their high concentrations. However, acetic acid can alsostimulate ethanol production at low concentration, which can be interpreted asfollows: at the pH of fermentation acetic acid is partially in undissociated form,which can diffuse through the plasma membrane. At the higher pH of cytoplasm itdissociates, and to avoid intracellular pH drop, the cell must pump out protons,which is driven by ATP hydrolysis (Figure 6). This means that more ATP has to begenerated to maintain the intracellular pH, which is achieved in anaerobicconditions by producing more ethanol, resulting in increased ethanol yield(Taherzadeh et al., 1997). The ethanol yield as a function of the acetic acidconcentration in SSF is illustrated in Figure 17. The values given are the average ofthe concentrations at 0 and 24 h, because acetic acid was released from thehemicellulose fraction at the beginning of SSF. The maximum yield (99%) wasobtained at an acetic acid concentration of 1.0 g/L. The results are in goodaccordance with a previous study reported by Larsson et al. (1999); however, theseauthors expressed the ethanol yield as a function of the total acid concentration.Nevertheless, it must be emphasised that it was not aimed to study the effect ofacetic acid concentration on the ethanol yield in SSF. Other parameters, such asconcentration of other inhibitors changed together with acetic acid concentration,therefore it cannot be concluded that the yield increase was unambiguously due tothe effect of acetic acid.
Figure 17 Ethanol yield expressed as % of theoretical, as a function of aceticacid concentration (the average of acetic acid concentrations at 0and 24 h was taken, since acetic acid was released from thehemicellulose fraction at the beginning of SSF).
0
20
40
60
80
100
0.0 0.5 1.0 1.5 2.0 2.5 3.0
Etha
nol y
ield
(% o
f the
oret
ical
)
Average acetic acid (g/L)
46
The specific raw material demands and the straw to meal ratios for the processscenarios are listed in Table 7. In the pure wheat straw-based process, 5.4 kg dryraw material was needed to produce 1 L of ethanol, whereas the correspondingamount of wheat meal was only 2.4 kg. This was simply because of the differencein the proportions of carbohydrates in the raw materials. The straw to meal ratioswere in the range of 0.5-2.6 in the case of mixtures. Mixture B was the closest tothe typical straw to meal ratio at harvest (1:1), and even its ethanol yield washigher (87%) then the calculated overall yield if SPWS and PWM fermentationswere carried out separately (78%).
Table 7 Raw material demand and straw to meal ratio at the variousprocess scenarios.
Process scenario Raw material required (kg total DM/L EtOH) Straw:mealWheat straw only 5.4 -Mixture A 4.0 2.6Mixture B 3.1 1.1Mixture C 2.9 0.7Mixture D 2.5 0.5Wheat meal only 2.4 0
47
5 RESULTS AND DISCUSSION OF TECHNO-ECONOMIC ANALYSIS
5.1 Base lignocellulose-to-ethanol process
The process model in techno-economic analyses was based on SO2 catalysed steampretreatment followed by SSF. It was a modified and extended version of that usedby Sassner and Zacchi (2008). Although the concept remained the same, the modelwas continuously modified and updated throughout the work, and in this sectionthe most recent version is described.
Experimental results recently obtained at the Department of Chemical Engineering,Lund University, Sweden were used in steam pretreatment and in SSF. In otherprocess steps the input data were taken from the literature, or were based on rule ofthumb or assumptions.
The scheme for the base lignocellulose-to-ethanol process (also referred to asreference scenario) is shown in Figure 18. The proposed ethanol plant was assumedto be located in Sweden and process 200 000 dry tonne spruce chips annually. Itwas run by 28 employees and was assumed to be in operation for 8000 h per year.Live steam was assumed to be available at 20 and 4 bar (high and low pressuresteam, respectively), and secondary steam was used to replace live steam wheneverpossible.
Figure 18 Overall scheme for the base lignocellulose-to-ethanol process(also referred to as reference scenario).
5.1.1 Steam pretreatmentThe raw material is impregnated with 2.5% SO2 based on the water content of rawmaterial and then preheated to 95°C by direct injection of low pressure steambefore being fed to the continuous pretreatment reactor. The quotation was askedfor a StakeTech reactor (SunOpta, Inc., Brampton, ON, Canada). The temperature
Spruce chips
Raw mat.handling
Steampretreatment
Yeast cultivation,SSF
Distillation,dehydration
Evaporation Separation Drying
Anaerobic digestion,Aerobic treatment
Heat and powerproduction
SO2 Steam Ethanol
Pellets
SteamElectricity
Condensate
Liquidfraction
Syrup
BiogasSludge
Stillage
Solidfraction
Enzymes
Flashstreams
48
in the reactor was maintained at 210°C by injecting high pressure steam. Heatlosses were assumed to be 10% of the adiabatic heat demand. After pretreatmentthe slurry is flash-cooled by a two-step pressure reduction, to 4 and 1 bar, and theflash streams are condensed by heat exchange with other process streams.
Glucan, mannan and galactan are hydrolysed to hexoses, and are partially degradedto HMF, levulinic and formic acids as well as non-specific degradation products. Insimilar way, xylan and arabinan are hydrolysed to pentoses, and are degraded tofurfural and non-specific products. The acetyl groups in the hemicellulose arereleased as acetic acid.
The steam pretreatment in the model produces a slurry with much higher WIScontent than what has been obtained in the experimental work. This is due to lowerheat losses applied in the model compared to what is obtained in a small scaleequipment with a large surface area. Lower heat losses result in less steamcondensed and higher WIS content of the slurry.
5.1.2 Yeast cultivation and simultaneous saccharification and fermentationThe pretreated material is fed to the SSF step. Water needed to adjust the drymatter in the SSF step to 10% WIS is added before pressing the pretreated slurry.This water demand was covered by part of the evaporation condensate. It alsocontains ammonia to neutralise the slurry from pH ~1.2 to pH 5.
The pressed liquid supplemented with molasses containing 50% sucrose is used inyeast cultivation (YC) without adding extra fresh water, hence the inhibitorconcentrations in YC and SSF are approximately the same. The hexoses fed intoYC are consumed entirely. Yeast seed train consisting of three stages operating at30°C provides SSF with 7.5% inoculum. Only the first and second stages aredesigned to be sterile, i.e. those vessels are pressure-rated for steam sterilisation.
In SSF, the concentration of ordinary baker’s yeast and the enzyme dosage are3 g DM/L and 10 FPU/g WIS, respectively. As nutrients (NH4)2HPO3 andMgSO4·7H2O are added in a concentration of 0.5 and 0.03 g/L, respectively. Thehexosans and pentosans are partially hydrolysed to hexoses and pentoses,respectively. Since ordinary baker’s yeast is applied, exclusively the hexoses arefermentated to ethanol. From the hexoses glycerol is also produced, and smallamount remains unfermented.
The SSF takes place in twelve agitated non-sterile fermentors with a total volumeof 920 m3 each. An SSF cycle including filling, fermentation (48 h), draining, andcleaning lasts for 60 h. The number of the YC fermentors was calculated from thecycle time, which was assumed to be 15 h for all YC stages.
49
5.1.3 Distillation and evaporation
The ethanol concentration obtained after SSF is ~3.5 wt-%, which is rather low,compared to that of first generation technologies. Distillation and molecular sieveadsorption are used to produce pure (>99.8 wt-%) ethanol. The distillation stepconsists of two stripper columns and a rectifier (Figure 19), which are heatintegrated by operating at different pressures. Ethanol recovery is assumed to be99.5% in each column. The remaining water in the overhead vapour leaving therectifier is removed in the dehydration columns that are regenerated with pureethanol vapour. The regenerate is returned to the rectifier.
Figure 19 Distillation and evaporation (dehydration is not shown). Dottedlines represent vapour streams.
The stillage of the stripper columns is separated in a filter press (quoted byPneumapress Filter Corp., Richmond, CA, USA) resulting in a solid fraction with aWIS content of 40%. In any scenario in that pellets are produced, washing isincluded in the separation step to decrease the sulphur content of the solid fraction,and it was assumed that washing – with a soluble solid recovery of 90% in theliquid stream – was sufficient to meet the requirement relating to the sulphurcontent of pellets.
The liquid fraction of the stillage is concentrated to 60% DM in an evaporationsystem which contains five effects in a forward-feed arrangement, i.e. only the firsteffect is heated by live steam, the subsequent ones utilise the vapour from theprevious effect operating at higher pressure. Boiling point elevation was accountedfor according to Larsson et al. (1997), and overall heat transfer coefficients wereestimated to vary between 400 and 2000 W/m2°C, depending on the temperature
50
and concentration of the liquid. Based on the work of Olsson et al. (2001), it wasassumed that by applying a stripper column after evaporation, recycling of part ofthe evaporation condensate to dilute the whole slurry was possible. The rest of thecondensate together with the condensed flash streams originating frompretreatment and drying are sent to anaerobic digestion (AD) followed by anaerobic step (Figure 18). Description of AD is given in Section 5.3.1, and details ofthe aerobic step can be found in Section 5.3.2.
5.1.4 Combined heat and power productionSteam and electricity are generated by burning the biogas, the concentrated liquidfraction and part of the solid fraction of the stillage (Figure 20). Superheated steam(91 bar, 470°C) is generated in the boiler. The moisture content of the fuel is ~50%and the heat losses in the boiler were assumed to be 1%. The generated steam isallowed to expand to 4 bar through an HP turbine system, however, part of thesteam is withdrawn at 20 bar for pretreatment and drying. District heating is notincluded in the base case, hence the temperature of the flue gases after the airheater is fixed to 150°C to avoid condensation of water. The excess solid residue,i.e. the solid fraction not required for steam generation, is dried in a superheatedsteam dryer to 88% DM. The secondary steam generated by drying is utilised in theprocess. Electricity is generated in the turbines, which were assumed to have anisentropic efficiency of 90%. The mechanical efficiency of the generator wasassumed to be 97%.
5.1.5 Other process areasThere are some parts of the process that are not included in the Aspen Plus models,however, they are included in the cost analysis, e.g. storage tanks for ethanol andfor liquid chemicals (including enzymes when on-site enzyme production is notincluded). These tanks vary from 1 to 800 m3 in size. Biogas was either useddirectly in the boiler, or after upgrading it was supplied directly to the grid. Solidcomponents were assumed to be stored in sacks in the storage area.
Cost of equipments for cleaning in place and sterile filtration are not included inthe economic analysis, however, naturally they are necessary in the final design.This will increase the ethanol production cost to some extent, but it affects all theinvestigated scenarios in the same way, hence differences between the scenariosremain the same.
51
Figure 20 Plant of CHP for base case and all other scenarios without districtheating. Grey solid and dashed lines represent gas and vapourstreams, respectively.
5.2 On-site enzyme production
Instead of using commercial enzyme preparations in SSF, options of on-siteenzyme production were investigated in terms of process design and economics.Process data for cellulase enzyme fermentation (CEF) were obtained from theliterature (Doppelbauer et al., 1987, Esterbauer et al., 1991), however, some key-assumptions were also made. The applied Trichoderma strain was assumed to beable to produce cellulase enzymes in the presence of monosaccharides, i.e. it wasnot catabolite-repressed (e.g. T. reesei RUT C30). The mycelium, soluble proteinand activity yields were 0.27 g, 0.26 g, 185 FPU/g carbohydrate in anhydroequivalent (Esterbauer et al., 1991), respectively, which resulted in a specificactivity of 710 FPU/g protein. After complete hydrolysis of polysaccharides, all themonosaccharides are consumed entirely in CEF, while other compounds are notinvolved in any reaction. Based on the work of Szengyel et al (2000), it wasassumed that inhibition due to compounds present in the pretreated material, suchas furan derivatives and organic acids, did not occur. The whole broth could beadded to SSF since the SSF was carried out at 37°C and above 35°C the growth of
52
mycelia is entirely inhibited. Using the whole culture had several advantages: a)additional separation is not needed; b) the enzymes adsorbed on the surface of thelignin, the cells as well as the ones trapped in the cytoplasm will also be present inthe SSF.
The feed of CEF is cooled down in a heat exchanger before entering thefermentors, which operate at 30°C. The aeration rate was 0.5 VVM in any scenario,and the following nutrient concentrations were applied: soy-meal 0.5%, (NH4)2SO40.15%, KH2PO4 0.07%, FeSO4·7H2O 0.001%.
5.2.1 Preliminary model of enzyme production (Paper III)
The feed of CEF is part of the whole pretreated slurry, i.e. contains both solid andliquid phases. At the beginning of fermentation the content of carbohydrates is setto 2%, expressed in anhydro equivalent. The base enzyme fermenting case includes16 aerated agitated fermentors, each 340 m3, arranged in 4 lines. A power-to-brothvalue of 500 W/m3 is set. The cooling is performed by cooling coils. Thefermentors are not pressure-rated for steam sterilisation. The pretreated materialand the makeup water coming entirely from the evaporation step are considered tobe sterile, hence only cleaning-in-place is applied in the tanks. During fermentationpH 6 is maintained.
The 4 production fermentors in a line follow the same schedule, however, they areshifted. At 25 h 10% of the culture in the first vessel is transferred to the secondone, and is used there as inoculum. The fermentation lasts for 90 hours and isfollowed by a 10-h harvesting, cleaning, charging period, which results in a 100-hcycle time. After 100 h from the start of the first fermentation the fourth vessel isready to transfer inoculum to the first one, which closes the line to a loop. Thisoperation pattern is referred to as Kornuta process (Nystrom and Allen, 1976), andit has the advantage that seed train fermentors can be omitted. In case ofcontamination inoculum could be transferred from a vessel in another line and bothlines could continue uninterruptedly.
5.2.2 The effect of enzyme fermentation productivity on the cost of cellulaseproduction
Productivity of enzyme fermentation is measured in FPU per liter of cellulaseproduction capacity per hour of production time. Besides the base productivitythree hypothetical cases with improved productivities were investigated in thepreliminary model of enzyme production. In the first case the activity yield wasincreased by 50%, and the content of carbohydrates was kept constant, whichresulted in 1.5-fold productivity. In the second case the content of carbohydrateswas increased to 4% and the same yield was assumed, hence the productivity was2-fold. In the third case both activity yield and content of carbohydrates were
53
increased by 50% and 100%, respectively, which corresponded to 3-foldproductivity. Hence, four scenarios with different productivities were obtained, andenzyme production cost was determined for each scenario.
Figure 21 shows the cost of on-site enzyme production as a function ofproductivity. The curve fitted is close to hyperbola (see equation in Figure 21),which means that the same productivity increment at low initial values decreasesenzyme production cost to a greater extent than at higher initial values. Similarcost-productivity curve was reported by Himmel et al. (1999) for on-site cellulaseproduction in a hardwood-based process, using the whole slurry as carbon source.However, they performed simple sensitivity analysis on the cellulase cost as afunction of productivity, i.e. they adjusted only one productivity-affectingparameter. On the contrary, in the present study, increase in productivity wasachieved in three different ways. Therefore the main conclusion is that at a givenprocess configuration any kind of change resulting in improved productivity incellulase fermentation is favourable with respect to enzyme cost.
Figure 21 Specific enzyme cost expressed in SEK/L ethanol as a function ofproductivity.
5.2.3 Modified model of enzyme production (Paper IV)
In the modified model of enzyme production a seed train was designed to providethe production stage of CEF with inoculum, and various carbon source (feed)options were investigated.
The 5% inoculum in the production stage is received from the second stage of atwo-stage seed train. Both seed stages operate with 5% inoculum at a cycle time of30 h. The first stage receives inoculum from a stock culture, while the second isinoculated with the broth of the first. Concerning the composition, the seed stages
y = 43∙x-0.8
R² = 0.997
0.00.20.40.60.81.01.21.41.61.8
50 100 150 200
Enzy
me
cost
(SEK
/L E
tOH)
Productivity, FPU/(L∙h)
54
are assumed to be run on the same feed as the production stage, where 120 h cycletime is presumed. As this time is double of the cycle time of SSF, the number ofvessels are 24 in the enzyme production stage. Considering the ratio of the cycletimes of seed and production stages, the number of seed vessels is 6 in both stages.In all scenarios these numbers were kept constant, hence the total vessel volumevaried in a range of 37-121 m3 in the production stage. The fermentors of the seedtrain are pressure-rated, and can be sterilised at 120°C, however, it was assumed atthe production stage, that sterilisation was not necessary. Cleaning-in-place issufficient, since the evaporation condensate and the pretreated material wereconsidered to be sterile and the fresh water added before pressing the pretreatedslurry is sterile filtered beforehand. Furthermore, the nutrients and the molasseswere assumed not to cause any contamination. In consequence, the seed vessels canbe sterilised empty. At all stages pH 5 is maintained. The heat released duringfermentation is removed by cooling water that circulates in jackets at the first stageand in coils at later stages. The cooling jacket is favourable in terms of cleaning,however, it is not sufficient at larger volume. The pH is controlled using ammonia.Aeration was assumed to ensure sufficient agitation when the solid content was lessthen 1% WIS.
Three configurations (A-C) were investigated (Figure 22), which differed in thecarbon source of CEF: in scenario A, part of the liquid fraction of the diluted slurrywas used, while in scenario B, the liquid fraction was supplemented with molassesto increase the sugar content. In scenario C, a mixture of the liquid fraction and thepressed slurry was prepared and used as feed for CEF. Scenario C was divided intothree sub-cases, C1 to C3, depending on the WIS content of the mixture, i.e. 1, 2,3%. For scenarios A and B, sensitivity analysis was performed, denoted with +.The specific activity of the soluble proteins was enhanced 1.5-fold, resulting in anincrease of 50% in the productivity in terms of enzyme activity, while protein andmycelium yields remained the same.
The liquid fraction also contained water-insoluble particles, as a WIS retention of99% was assumed in the filtration of the slurry. In the CEF feed the totalcarbohydrate content expressed in monomer equivalent (ME) and WISconcentration, in parentheses, were the following: A: 4.6% (0.5%), A+: 4.7%(0.7%), B: 10% (0.8%), B+: 10% (0.9%), C1: 4.9% (1%), C2: 5.5% (2%), C3:6.0% (3%). In cases C1 to C3, aeration was assumed to be insufficient for ensuringhomogeneity, therefore agitators were built in, and power-to-broth value was set to40 W/m3. In scenario B, molasses served as a complex nutrient source, hencenutrient supplementation was omitted.
55
Figure 22 Lay-out of cellulase enzyme fermentation (CEF), yeast cultivation(YC) and simultaneous saccharification and fermentation (SSF) atthe various configurations (A-C).
5.2.4 The effect of on-site enzyme fermentation on the whole processIn this section process design of the modified CEF model and the correspondingethanol process is presented. On-site enzyme production has the advantage thatpentoses, not fermentable by ordinary baker’s yeast, can also be utilised.Nevertheless, in regard to hexoses, CEF competes with ethanol fermentation, sincethe more sugar is used for CEF, the less sugar is available for ethanol production.The highest mass flow of C5 sugars (pentosans and pentoses in ME) into CEF wasobtained in scenario A, which corresponded to 13% of C5 sugars recovered afterpretreatment, hence even at the best utilisation, 87% was burnt or dried. Regardingthe composition of sugars in the CEF feed, scenario A gave the highest C5proportion (21%, Table 8). In the case supplemented with molasses (scenario B),
56
the C5 ratio was only 9%, due to the high ratio of sugars of molasses (57% of totalsugars consumed).
Although CEF decreased the C6 flow (hexosans and hexoses) in SSF feed, themajority of C6 sugars were fermented to ethanol (Table 8). Even in the worstscenario (C3), the decrement of C6 sugars (in ME) was 6% of the C6 sugars (inME) fed to SSF in the reference case. The overall ethanol yield of the process inCEF cases also decreased compared to the reference. The decrements were lowestin scenarios B and owing to the large proportion of sugars coming from molasses.The yields in scenarios A and C1 to C3 were equal, which implied that the modelwas indifferent to the WIS content of CEF in the range investigated.
Table 8 Features of enzyme and ethanol productions. Carbon source A:pretreated liquid fraction, B: pretreated liquid fraction andmolasses, C: pretreated liquid fraction and pressed pretreatedslurry with a total WIS content of 1, 2, 3%.
Reference A B C1 C2 C3
Components into CEF, tonne/year
Hexosans - 430 298 776 1374 1861
Pentosans - 7 5 13 23 32
Hexoses - 5133 6242 4812 4256 3805
Pentoses - 1487 622 1394 1233 1102
C5 ratio in CEFa, % - 21 9 20 18 16
Components into SSF, tonne/year
Hexosans 62980 62550 62683 62205 61606 61120
Hexoses 32634 27521 30499 27841 28397 28848
Overall EtOH yieldb, L/dry tonne 270 254 263 254 254 254
Ethanol production cost, SEK/L 4.88 4.96 4.82 4.99 4.97 4.96a (pentosans in ME + pentoses)/total sugar in MEb Includes yeast and enzyme productions and ethanol losses in the process
5.2.5 Economics: specific enzyme cost, minimum ethanol selling priceIn this section economics of the modified CEF model and the correspondingethanol process are discussed. Cost elements of enzyme production expressed inSEK/L ethanol are shown in Figure 23. Capital and chemical costs were found tobe the main contributors in each scenario (on the contrary, in the preliminary modelof CEF utilities were predominant besides capital, since a very high power-to-brothvalue was applied in that model). Utilities refer to the electricity consumption ofthe compressor and the agitators (scenarios C1 to C3) and to the demand forcooling water. The lowest capital costs were obtained in scenarios B, due to high
57
carbohydrate concentration in CEF, which resulted in small fermentor volume.However, the cost of chemicals was the highest in these cases, owing to the extracost of molasses. In scenarios C1 to C3, increasing the WIS concentration in CEFresulted in a reduced fermentation volume, and in consequence, lower capital andchemical costs.
Figure 23 Cost contributors of enzyme production in SEK/L ethanol.Molasses was added to chemicals. Other refers to maintenanceand insurance. No extra labour was accounted for the enzymefermentation area. Carbon source A: pretreated liquid fraction, B:pretreated liquid fraction and molasses, C: pretreated liquid fractionand pressed pretreated slurry with a total WIS content of 1, 2, 3%;+: 1.5-fold specific activity.
Scenario B proved to have the highest total cost of enzyme production. On theother hand, in regard to ethanol production cost, scenario B was the mostfavourable, furthermore, this was the only case with on-site enzyme production inwhich the ethanol production cost was lower than that in the reference case, withpurchased enzymes (Table 8). In spite of the extra expenses, molasses couldimprove the process economics considerably, since CEF supplemented withmolasses reduced the overall ethanol yield, the most important parameter in theproduction cost of ethanol (Sassner et al., 2008), to a smaller extent. The 1.5-foldactivity yield resulted in a decrease of 16 and 19% in total enzyme production costcompared to the base scenarios A and B, respectively (Figure 23).
0.50
0.42
0.53
0.43
0.52 0.50 0.49
0.0
0.1
0.2
0.3
0.4
0.5
0.6
A A+ B B+ C1 C2 C3
Total
Capital
Chemicals
Utilities
Other
Enzy
me
prod
uctio
n co
st (S
EK/L
EtO
H)
58
On-site enzyme fermentation contributed to 9-11% of the ethanol production cost.The feasibility of including enzyme production in the lignocellulosic ethanolprocess highly depends on the full-scale price of commercial cellulase enzymepreparation, which is still very uncertain. At the premises of the present study,some scenarios proved to be more feasible than that with purchased enzymes,which implies that on-site enzyme production can be an alternative.
Figure 24 Alternative stillage treatment scenarios. Only the liquid fraction ofthe stillage is sent to anaerobic digestion (scenarios A), or thewhole stillage goes directly to anaerobic digestion (scenario B).The dotted parts are optional, see Table 9 for further details. Thewastewater streams, such as the condensed flash streams frompretreatment and drying, are also sent to anaerobic digestion,however, they are not shown. FP: Filter press, BF: Belt filter.
5.3 Alternative treatment of the stillage and production of various co-products (Paper V)
In the base process (Figure 18) the liquid fraction of the stillage from thedistillation is concentrated in a multiple-effect evaporation plant and the syrup isincinerated together with a part of the solid residue in the CHP plant. As the
59
evaporation is an energy-intensive process step, alternative stillage treatmentscenarios, summarised in Figure 24 and Table 9, were investigated. They differedmainly in the co-products that were produced from the stillage.
In scenarios A1 to A4 the stillage is separated in a filter press, which was describedin Section 5.1.3, and is taken to AD for production of biogas. The effluent of AD isfed to an aerobic treatment step. The aerobic sludge separation is performed with abelt filter, which increases the sludge DM up to 30%.
In scenario B the whole stillage is sent directly to AD. However, the effluent of ADis separated and only the liquid fraction is fed to the aerobic treatment step. For thisseparation the same type of filter press – as the one for stillage separation – isapplied with an assumed WIS recovery of 35%.
Table 9 Details of the alternative stillage treatment scenarios. SF: Solidfuel.
Scenario Washing&drying
Pelletproduction
Biogasupgrading
Turbinesystem DH
A1 Yes Part of SF Yes HP No
A2 Yes Total amount of SF No HP-LP Yes
A3 No - Yes HP-LP Yes
A4 No - No HP-LP Yes
B - - Yes HP-LP Yes
Scenario Co-products (besides electricity) Burnt in CHP (besides sludge)
A1 Upgraded biogas, pellets Part of SF, tail gas
A2 Pellets Biogas
A3 Upgraded biogas Solid fraction of stillage, tail gas
A4 - Solid fraction of stillage, biogas
B Upgraded biogas Solid fraction of AD effluent, tail gas
5.3.1 Anaerobic digestion
The AD step is performed at mesophilic conditions, hence the inlet flow is cooleddown to 37°C before being fed to the first digester. The features of AD are shownin Table 10. The digesters were continuous stirred tank reactors with a total volumeof around 3500 m3 each, hence their number varied between 2 and 12, includingthe scenarios with evaporation. The lignin was considered to be inert in terms of
60
biogas production, therefore the COD of lignin was not accounted for in the CODused for calculating the reactor volume and the demand of chemicals.
The biogas is assumed to consist of 50 wt-% methane, 46% CO2 and 4% water.Biogas upgrading is performed in some scenarios where the biogas is sold as a co-product supplied to the gas grid. In those scenarios the total amount of producedbiogas is upgraded. However, 5% is assumed to be retained in the tail gas that isburnt on-site.
Table 10 Assumed features of anaerobic digestion.
Degradation factorsSoluble sugars, organic acids, ethanol,
90%glycerol, enzyme, yeastPolysaccharides, extractives
50%degradation products, soluble ligninLignin 0%YieldsMethane 0.35 Nm3/kg COD removed
AD sludge 0.03 kg DM/kg CODDesignOrganic loading rate 10 kg COD/m3/day
Power of stirring 20a/40b W/m3
Chemical demandsNH3 (25%) 18 g/kg COD
H3PO4 (50%) 4 g/kg CODa When the liquid fraction of the stillage was sent to AD.b When the whole stillage was subjected to AD.
5.3.2 Aerobic treatment
The effluent of AD or the separated liquid stream of AD effluent is first treatedaerobically. The organic matter is removed almost entirely, while sludge isproduced – 0.3 kg sludge DM/kg organic matter – which is separated and burnt on-site. The next step is ozone treatment, to treat the phenolic compounds that havenot decomposed yet. The effluent from this step is “clean” water, which is recycledto the process partially or entirely depending on the scenario. The recycled waterstream is sterile filtered before being used for diluting the pretreated slurry,however, the cost of this filter was not included in the economic evaluation.
61
5.3.3 Combined heat and power production in scenarios of alternativestillage treatment
In scenario A1 the CHP plant, shown in Figure 20, only covers the heat demand ofthe process, while in scenarios A2 to A4 and B excess heat is produced that is usedfor generating electricity through an additional low-pressure (LP) turbine operatingwith an isentropic efficiency of 85% and a discharge pressure of 0.75 bar (Figure25). The other data concerning the superheated steam and data for the electricitygeneration are the same as for scenario A1. The condensation heat of the outletsteam of LP turbine and the heat obtained in the flue gas condensation are used fordistrict heating. Detailed description of the turbine system, flue gas condensationand district heating system was reported by Sassner and Zacchi (2008). The onlydifference is that the return temperature of the district heating stream is set to 45°Cinstead of 48°C.
In the case of solid drying (scenario A2), a condensate of 20 bar returns to the CHPplant, see the blue part of Figure 25. In other scenarios with HP-LP turbines dryingis not included in the process, hence there, instead of the 20-bar condensate, make-up water of 10°C is used to saturate the withdrawn superheated steam.
Return condensate4 bar
Saturated steam20 bar
Saturated steam4 bar
Return condensate20 bar
Feed
Flue gas150°C
Air10°C
Make-up water10°C
District heating, return45°C
To district heating90°C
Flue gas50°C
Combustion
Boiler
Air heater Feed water tank4 bar
~91 bar
4 bar 2 bar 0.75 bar
Flue gascondenser
Figure 25 Plant of CHP with HP-LP turbines and district heating. Grey solidand dashed lines represent gas and vapour streams, respectively.The blue part is only applicable in scenario A2.
62
Based on the Swedish conditions, it was assumed that heat is delivered to a districtheating system during a period of time equivalent to 4500 h. Cooling water is usedfor the remaining 3500 h to remove heat.
5.3.4 Energy aspects of the various scenariosEnergy-related details of the scenarios are given in Table 11. The largest biogasproduction was obtained in scenario B, due to the increased flow ofpolysaccharides, soluble sugars and other degradable components. In the referencecase the organic matter fed to AD consisted of only volatile substances originatingfrom the pretreatment, evaporation and drying steps. Removing the evaporationstep decreased the heat duty of the process considerably, from 27.8 MW to 17.1-17.9 MW. The ethanol production was the same in all scenarios (6177 L/h), hencethe specific heat demands were reduced from 16.2 to 9.9-10.4 MJ/L ethanol. Theelectricity produced on-site varied considerably, unlike the total power requirementof the process. The highest electricity production was obtained for scenario A4,where both the biogas and the solid residue are burnt in the CHP plant. In scenariosA1 and A2 the power produced was lower than the power demand in the process,thus requiring the purchase of electricity, while in all other scenarios a surplus ofelectricity was produced that could be sold. The largest pellet production wasobtained for scenario A2, where, unlike other scenarios, the entire solid residuewas pelletised. In the case of district heating, the more heat that can be delivered,the more difficult it is to find an appropriate plant location.
Table 11 Process details of the various scenarios. Values are given in MW.The positive or negative difference between power demand andelectricity produced indicates purchasing or selling of electricity,respectively. A summary of the scenarios is given in Table 9.
Reference A1 A2 A3 A4 B
Energy flow1 of raw biogas 3.8 19.9 19.9 19.5 19.5 29.4
Energy flow1 of upgraded biogas - 18.9 - 18.5 - 27.9
Process overall heat duty 27.8 17.6 17.6 17.9 17.9 17.1
Total process power demand 4.1 4.4 4.5 3.8 4.0 4.5
Electricity produced in CHP 5.9 3.0 4.0 11.7 17.1 9.8
District heat - - 4.7 34.1 48.9 31.1
Energy flow1 of pellets2 48.5 39.3 56.2 - - -1 Based on lower heating values.2 Only the sold part of solid fuel is given here, i.e. the part incinerated on-site is excluded.
63
The reference case had significantly lower overall energy efficiency (81%) than thealternative stillage treatment scenarios (87–92%), see Figure 26. This was due tothe difference in the overall heat demand, since the reduction in the heat demand ofthe ethanol process increased the energy output in the co-products. The energyefficiency obtained for ethanol was about 33%, which emphasises the importanceof co-products in an energy-efficient process. In addition, as demonstrated in PaperIV, co-products also have an important role in making the process economicallyviable.
Figure 26 Overall energy efficiency, based on lower heating values (LHV),expressed as % of the input. A summary of the scenarios is givenin Table 9. REF: reference case.
The overall energy efficiency of the scenarios including district heating varied onlyslightly, unlike their ethanol production cost (Figure 26). Therefore, it can beconcluded that high energy efficiency does not necessarily result in improvedeconomics. The energy efficiency would differ significantly, however, if thedistrict heating was excluded from these scenarios (for example, during thesummer, when the released heat has to be removed by cooling water). All thescenarios including district heating (A2 to A4, B) had a higher overall efficiencythan scenarios without district heating (reference case and A1), as it was alreadydemonstrated by Sassner and Zacchi (2008).
0
10
20
30
40
50
60
70
80
90
100
REF A1 A2 A3 A4 B
District heat
Electricity
Biogas
Pellets
Ethanol
LHV
ener
gy e
ffici
ency
(%)
64
5.3.5 Economic aspects of the scenarios investigated
The lowest and highest total capital investment costs were obtained for scenariosA1 (1199 million SEK) and A4 (1360 million SEK), respectively (Table 12). Themain contributors to the total direct cost were the pretreatment unit, thefermentation stage (yeast cultivation and SSF) and the CHP plant. The highest costfor CHP was obtained in scenario A4, where it constituted 44% of the total directcost. Anaerobic digestion had the highest direct cost in scenario B, where thewhole stillage stream was anaerobically digested.
Table 12 Breakdown of the total capital investment cost in million SEK.A summary of the scenarios is given in Table 9.
Reference A1 A2 A3 A4 BRaw material handling 10 10 10 10 10 10Pretreatment 115 115 115 115 115 115Yeast cultivation & SSF 119 119 119 119 119 119Distillation 47 47 47 47 47 47Separation 26 26 26 26 26 25Evaporation 47 - - - - -Drying & pellet production 42 45 47 - - -CHP 155 110 136 266 343 248Storage 33 28 28 28 28 29Heat exchanger network 18 11 11 11 11 12AD 15 75 75 74 74 111Total direct cost 628 586 613 696 773 716Total indirect cost 578 551 557 539 556 557Fixed capital investment 1206 1137 1170 1236 1329 1273Working capital 69 62 75 32 32 32Total capital investment 1275 1199 1246 1268 1360 1305
The major contributors to ethanol production cost were the cost of capital, rawmaterial and chemicals, which included enzyme cost, in descending order (Figure27). Concordant results were obtained in Paper IV, and were reported by Sassner etal. (2008). The favourable combinations for the co-products of the alternativestillage treatment scenarios were: (i) upgraded biogas and pellets (scenario A1);and (ii) upgraded biogas, electricity and district heat (scenarios A3 and B). Withrespect to the production cost of ethanol, the alternative stillage treatmentconfigurations proved to be good options in some cases. Scenarios A1, A3 and Bresulted in low ethanol production costs (4.00-4.24 SEK/L), while scenarios A2and A4 were less economical (5.27 and 5.02 SEK/L, respectively). Scenario A2had an even higher ethanol production cost than the reference case (5.14 SEK/L).
65
The difference in the production cost of ethanol between using the whole stillage(scenario B) or only the liquid fraction in anaerobic digestion (scenario A3) wasnegligible for the combination of co-products including upgraded biogas,electricity and district heat.
Figure 27 Breakdown of ethanol production cost in SEK/L. Chemicals includeenzymes. ‘Others’ refers to the cost of labour, insurance andmaintenance. A summary of the scenarios is given in Table 9.REF: reference case; Upgr: upgraded.
The sensitivity of ethanol production cost to changes in the prices of the co-products was monitored by changing the price of one co-product at a timefrom -40% to +40% (Figure 28). The effect of changing price of the co-productsdepends on the share of the given co-product with regard to the total income: thehigher the share the greater the effect. The electricity price had the highest impacton scenario A4, followed by scenarios A3 and B. However, scenarios A3 and Bhad the lowest ethanol production cost in almost the whole range investigated.Without selling electricity certificates (200 SEK/MWh) scenario A1 would be themost profitable if the spot price of electricity decreased below 260 SEK/MWh.Scenario B was that most affected by the price of biogas. Even with a 40% lowerprice of biogas the scenarios including biogas upgrading (A1, A3, B) resulted inlower ethanol production costs than the scenarios including biogas burning(reference case, A2, A4), although the difference was much smaller. The incomefrom biogas could also decrease due to a lower production of biogas than thatassumed. However, as long as the product of the amount of biogas produced and
5.14
4.24
5.27
4.01
-3
-2
-1
0
1
2
3
4
5
6
REF A1 A2 A3Etha
nol p
rodu
ctio
nco
st (S
EK/L
)
65
The difference in the production cost of ethanol between using the whole stillage(scenario B) or only the liquid fraction in anaerobic digestion (scenario A3) wasnegligible for the combination of co-products including upgraded biogas,electricity and district heat.
Figure 27 Breakdown of ethanol production cost in SEK/L. Chemicals includeenzymes. ‘Others’ refers to the cost of labour, insurance andmaintenance. A summary of the scenarios is given in Table 9.REF: reference case; Upgr: upgraded.
The sensitivity of ethanol production cost to changes in the prices of the co-products was monitored by changing the price of one co-product at a timefrom -40% to +40% (Figure 28). The effect of changing price of the co-productsdepends on the share of the given co-product with regard to the total income: thehigher the share the greater the effect. The electricity price had the highest impacton scenario A4, followed by scenarios A3 and B. However, scenarios A3 and Bhad the lowest ethanol production cost in almost the whole range investigated.Without selling electricity certificates (200 SEK/MWh) scenario A1 would be themost profitable if the spot price of electricity decreased below 260 SEK/MWh.Scenario B was that most affected by the price of biogas. Even with a 40% lowerprice of biogas the scenarios including biogas upgrading (A1, A3, B) resulted inlower ethanol production costs than the scenarios including biogas burning(reference case, A2, A4), although the difference was much smaller. The incomefrom biogas could also decrease due to a lower production of biogas than thatassumed. However, as long as the product of the amount of biogas produced and
4.01
5.02
4.00
A3 A4 B
Raw materialCapitalChemicalsUtilitiesOthersUpgr. biogasPelletsElectricityDistrict heatTotal
66
the price does not decrease by more than 40%, the scenarios with biogas upgrading
are more favourable from an economic point of view. The pellet price had the
highest impact on scenario A2. For a pellet price higher than 235 SEK/MWh
scenario A1 would result in a lower ethanol production cost than any of the other
scenarios. Finally, the district heat price had a considerable influence on scenarios
A4, A3 and B, in descending order. Scenarios A3 and B had the lowest ethanol
production cost when the price of district heat was higher than 195 SEK/MWh,
whereas below this value scenario A1 became the most economical. Hence,
scenario A1 would have the lowest ethanol production cost, in the case where there
was no income from district heating - for example, the plant location would not be
appropriate for district heating. From the results of sensitivity analysis, it can be
concluded that alternative stillage treatment with biogas upgrading proved to be a
favourable option in many respects.
Figure 28 Ethanol production cost (SEK/L) as a function of prices of (A) electricity, (B) upgraded biogas, (C) pellets and (D) district heat. A summary of the scenarios is given in Table 9. REF: reference case.
67
6 SUMMARY
The work presented in this thesis focuses on improving the lignocellulose-to-ethanol process. Two kinds of approaches, i.e. experimental methods and techno-economic evaluations were involved in this progress. In the experimental part,optimisation of steam pretreatment of hemp hurds, a novel potential raw materialfor ethanol production, and an integration possibility for simultaneoussaccharification and fermentation are presented. In the techno-economic part,options of on-site enzyme fermentation, alternative stillage treatment andproduction of various co-products are investigated in a wood-based ethanolprocess.
Steam pretreatment of hemp hurds
Hemp hurds, the woody core of hemp, is an agro-industrial by-product withrelatively high carbohydrate content. Steam pretreatment of non-impregnated hemphurds was investigated at two reactor scales (2 and 10-L) by varying thetemperature from 200-230°C. Glucan recoveries were relatively high (> 82% oforiginal content), while xylan recoveries ranged from 18-66% of original.Conversions of glucan and xylan in enzymatic hydrolysis varied between 62-83%and 46-96%, respectively. Ethanol yields in simultaneous saccharification andfermentation ranged from 38-70%, based on glucan and mannan contents of thesubstrate. To obtain adequately high ethanol yield a steam pretreatmenttemperature of 210°C ≤ was required. However, the highest overall sugar yield(glucose + xylose, 63% of theoretical) was obtained at 200°C.
Integrated ethanol production from wheat straw and wheat meal
Integration of a lignocellulosic ethanol production technology to a starch-basedtechnology currently existing on an industrial scale can facilitate thecommercialisation of the former. Steam pretreated wheat straw was mixed withpresaccharified wheat meal and converted to ethanol in simultaneoussaccharification and fermentation. With increasing amount of presaccharifiedwheat meal, not only the ethanol concentration, but also the ethanol yieldincreased. The maximum ethanol yield, 99% of theoretical based on glucose, wasobtained for a mixture of containing equal amounts of presaccharified wheat mealand steam pretreated wheat straw (on a water-insoluble solids basis) resulting in anethanol concentration of 56.5 g/L. This yield was higher than those obtained withpure steam pretreated wheat straw (68 %) or with pure presaccharified wheat meal(91 %), hence mixing of the two substrates was found to be beneficial for the finalethanol concentration and also for the ethanol yield.
68
Modelling of on-site enzyme fermentation
On-site cellulase enzyme fermentation in a softwood-to-ethanol process wasinvestigated from a techno-economic aspect. The effect of varying the carbonsource of enzyme fermentation was monitored through the whole process. Liquidfraction of steam pretreated material, the same supplemented with molasses and thewhole pretreated slurry served as carbon source. Enzyme production step decreasedthe overall ethanol yield (270 L/dry tonne of raw material in the case of purchasedenzymes) by 5-16 L/tonne. Capital cost was found to be the main cost contributorto enzyme fermentation. Productivity of cellulase fermentation was proved to be animportant parameter in the cost of enzyme production. The lowest minimumethanol selling prices were obtained in those scenarios, where pretreated liquidfraction supplemented with molasses was used as carbon source. In somescenarios, on-site enzyme fermentation was found to be a feasible alternative.
Modelling of alternative stillage treatment and production of various co-products
Replacing the energy-intensive evaporation of stillage by anaerobic digestion isone way of decreasing the energy demand of the lignocellulosic ethanol process.The biogas can be incinerated or upgraded and sold out. Various processconfigurations of anaerobic digestion of the stillage, with different combinations ofco-products, were evaluated in terms of energy efficiency and ethanol productioncost versus the reference case of evaporation. Anaerobic digestion of the stillageshowed a significantly higher overall energy efficiency (87-92%), based on thelower heating values, than the reference case (81%). Although the amount ofethanol produced was the same in all scenarios, the production cost varied between4.00 and 5.27 SEK/L, including the reference case. Anaerobic digestion of thestillage with biogas upgrading can be a favourable option for both energyefficiency and ethanol production cost.
6.1 Novel scientific findings
1. The optimal temperature of steam pretreatment (performed without catalystaddition, for 10 minutes, in a reactor volume of 10 L) of hemp hurds for releasingfermentable sugars in enzymatic hydrolysis and for subsequent ethanolfermentation, is 210°C in terms of overall glucose (336 g/kg dry hurds, 75% oftheoretical) and ethanol yield (141 g/kg dry hurds, 60%). However, the maximumsugar yield (414 g/kg dry hurds, 63%), considering glucose and xylose, can beobtained at lower temperature (200°C) (Paper I).
69
2. Simultaneous saccharification and fermentation of a mixture containingequal amounts of presaccharified wheat meal and steam pretreated wheat straw (ona water-insoluble solids basis) results in higher ethanol yield (99% of theoretical)than the simultaneous saccharification and fermentation of presaccharified wheatmeal alone (91%) (Paper II).
3. In the case of on-site enzyme production integrated in a lignocellulose-to-ethanol process, the productivity of cellulase fermentation affects the specificenzyme cost (related to unit ethanol produced). Providing the same enzymefermentation configuration, the decrease of enzyme cost is determined by theextent of productivity increment and not by the origin of the increment (Paper III).
4. In enzyme fermentation pentose sugars are utilised together with hexosesugars. In the model of Paper IV, even including on-site enzyme production, themajor part of pentose sugars (87%≤) is burnt or dried into solid fuel. The amount ofhexose sugars consumed during enzyme fermentation is small (≤6% of the totalamount present in the pretreated material).
5. The main contributor to specific enzyme cost is the capital investment cost.On-site enzyme fermentation contributes to 9-11% of the ethanol production cost(Paper IV).
6. Anaerobic digestion of the stillage decreases the heat duty of the processconsiderably, compared to the evaporation of the stillage (from 27.8 MW to 17.1-17.9 MW). Consequently, the energy efficiency increases (from 81% to 87-92%,based on lower heating values) (Paper V).
7. Anaerobic digestion of the stillage results in lower ethanol production costthan evaporation of the stillage in some cases. The favourable combinations for theco-products of the stillage treatment with anaerobic digestion are: (i) upgradedbiogas and pellets, and (ii) upgraded biogas, electricity and district heat. Thesecombinations remain favourable by changing the price of one co-product at a timebetween -40% and +40%. The difference in the production cost of ethanol betweenusing the whole stillage or only the liquid fraction in anaerobic digestion isnegligible for the combination of co-products including upgraded biogas,electricity and district heat (Paper V).
71
7 REFERENCES
Acebal, C., Castillon, M. P., Estrada, P., Mata, I., Costa, E., Aguado, J., Romero,D., Lacaba, F., 1985. Cellulase production by Trichoderma reesei on wheat straw.Biochemical Society Transactions 13 (2), 453–455.
Aden, A., 2008. Biochemical production of ethanol from corn stover: 2007 state oftechnology model. Technical Report NREL/TP-510-43205, National RenewableEnergy Laboratory, Golden, CO.
Aden, A., Foust, T., 2009. Technoeconomic analysis of the dilute sulfuric acid andenzymatic hydrolysis process for the conversion of corn stover to ethanol.Cellulose 16 (4), 535–545.
Aden, A., Ruth, M. F., Ibsen, K., Jechura, J., Neeves, K., Sheehan, J., Wallace, B.,2002. Lignocellulosic biomass to ethanol processing design and economicsutilizing co-current dilute acid prehydrolysis and enzymatic hydrolysis for cornstover. Technical Report NREL/TP-510-32438, National Renewable EnergyLaboratory, Golden, CO.
Adsul, M. G., Ghule, J. E., Singh, R., Shaikh, H., Bastawde, K. B., Gokhale, D. V.,Varma, A. J., 2004. Polysaccharides from bagasse: applications in cellulase andxylanase production. Carbohydrate Polymers 57 (1), 67–72.
Alkasrawi, M., Rudolf, A., Lidén, G., Zacchi, G., 2006. Influence of strain andcultivation procedure on the performance of simultaneous saccharification andfermentation of steam pretreated spruce. Enzyme and Microbial Technology 38 (1-2), 279–286.
Almeida, J. R. M., Modig, T., Petersson, A., Hahn-Hägerdal, B., Lidén, G., Gorwa-Grauslund, M. F., 2007. Increased tolerance and conversion of inhibitors inlignocellulosic hydrolysates by Saccharomyces cerevisiae. Journal of ChemicalTechnology and Biotechnology 82 (4), 340–349.
Alvira, P., Tomás-Pejó, E., Ballesteros, M., Negro, M. J., 2010. Pretreatmenttechnologies for an efficient bioethanol production process based on enzymatichydrolysis: A review. Bioresource Technology 101 (13), 4851–4861.
Ander, P., Eriksson, K. E., 1977. Selective degradation of wood components bywhite-rot fungi. Physiologia Plantarum 41 (4), 239–248.
Ballesteros, M., Oliva, J. M., Manzanares, P., Negro, M. J., Ballesteros, I., 2002.Ethanol production from paper material using a simultaneous saccharification and
72
fermentation system in a fed-batch basis. World Journal of Microbiology &Biotechnology 18 (6), 559–561.
Ballesteros, M., Oliva, J. M., Negro, M. J., Manzanares, P., Ballesteros, I., 2004.Ethanol from lignocellulosic materials by a simultaneous saccharification andfermentation process (SFS) with Kluyveromyces marxianus CECT 10875. ProcessBiochemistry 39 (12), 1843–1848.
Berghem, L. E. R., Pettersson, L. G., 1974. Mechanism of enzymatic cellulosedegradation - isolation and some properties of a beta-glucosidase fromTrichoderma viride. European Journal of Biochemistry 46 (2), 295–305.
Berlin, A., Gilkes, N., Kilburn, D., Bura, R., Markov, A., Skomarovsky, A.,Okunev, O., Gusakov, A., Maximenko, V., Gregg, D., Sinitsyn, A., Saddler, J.,2005. Evaluation of novel fungal cellulase preparations for ability to hydrolyzesoftwood substrates - evidence for the role of accessory enzymes. Enzyme andMicrobial Technology 37 (2), 175–184.
Biofuels, 2011. http://www.biofuels-platform.ch/en/infos/production.php?id=bio-ethanoln.
Bohlmann, G. M., 2006. Process economic considerations for production ofethanol from biomass feedstocks. Industrial biotechnology 2 (1), 14–20.
Boisset, C., Fraschini, C., Schulein, M., Henrissat, B., Chanzy, H., 2000. Imagingthe enzymatic digestion of bacterial cellulose ribbons reveals the endo character ofthe cellobiohydrolase Cel6A from Humicola insolens and its mode of synergy withcellobiohydrolase Cel7A. Applied and Environmental Microbiology 66 (4), 1444–1452.
Briens, C., Piskorz, J., Berruti, F., 2008. Biomass valorization for fuel andchemicals production - A review. International Journal of Chemical ReactorEngineering 6.
Börjesson, J., Engqvist, M., Sipos, B., Tjerneld, F., 2007. Effect of poly(ethyleneglycol) on enzymatic hydrolysis and adsorption of cellulase enzymes to pretreatedlignocellulose. Enzyme and Microbial Technology 41 (1-2), 186–195.
Bruce, A., Palfreyman, J. W., 1998. Forest products biotechnology. Taylor &Francis, London.
Busche, R. M., 1983. Recovering chemical-products from dilute fermentationbroths. Biotechnology and Bioengineering 13, 597–615.
Canettieri, E. V., Rocha, G. J. D. M., de Carvalho, J. A., Silva, J. B. D. A. E., 2007.Optimization of acid hydrolysis from the hemicellulosic fraction of Eucalyptus
73
grandis residue using response surface methodology. Bioresource Technology98 (2), 422–428.
Chang, V. S., Kaar, W. E., Burr, B., Holtzapple, M. T., 2001. Simultaneoussaccharification and fermentation of lime-treated biomass. Biotechnology Letters23 (16), 1327–1333.
Chauve, M., Mathis, H., Huc, D., Casanave, D., Monot, F., Ferreira, N. L., 2010.Comparative kinetic analysis of two fungal beta-glucosidases. Biotechnology forBiofuels 3 (3), (11 February 2010).
Coughlan, M. P., 1992. Enzymatic-hydrolysis of cellulose - An overview.Bioresource Technology 39 (2), 107–115.
Dadi, A. P., Varanasi, S., Schall, C. A., 2006. Enhancement of cellulosesaccharification kinetics using an ionic liquid pretreatment step. Biotechnology andBioengineering 95 (5), 904–910.
Dale, B. E., Weaver, J., Byers, F. M., 1999. Extrusion processing for ammoniafiber explosion (AFEX). Applied Biochemistry and Biotechnology 77-9, 35–45.
Dang, V., Nguyen, K. L., 2006. Characterisation of the heterogeneous alkalinepulping kinetics of hemp woody core. Bioresource Technology 97 (12), 1353–1359.
Datta, R., 1981. Energy-requirements for lignocellulose pretreatment processes.Process Biochemistry 16 (4), 16–&.
Destexhe, A., Peckous, L., Picart, L., 2004. Using enzymes in ethanol production.Novozymes A/S.
Ding, S. Y., Xu, Q., Crowley, M., Zeng, Y., Nimlos, M., Lamed, R., Bayer, E. A.,Himmel, M. E., 2008. A biophysical perspective on the cellulosome: newopportunities for biomass conversion. Current Opinion in Biotechnology 19 (3),218–227.
Directive, 2003. Directive 2003/30/EC of the European Parliament and of theCouncil of 8 May 2003 on the promotion of the use of biofuels or other renewablefuels for transport. Brussels, Belgium.
Doppelbauer, R., Esterbauer, H., Steiner, W., Lafferty, R. M., Steinmuller, H.,1987. The use of lignocellulosic wastes for production of cellulase by Trichodermareesei. Applied Microbiology and Biotechnology 26 (5), 485–494.
Duff, S. J. B., Murray, W. D., 1996. Bioconversion of forest products industrywaste cellulosics to fuel ethanol: A review. Bioresource Technology 55 (1), 1–33.
74
Eggeman, T., Elander, R. T., 2005. Process and economic analysis of pretreatmenttechnologies. Bioresource Technology 96 (18), 2019–2025.
Eliasson, A., Christensson, C., Wahlbom, C. F., Hahn-Hägerdal, B., 2000.Anaerobic xylose fermentation by recombinant Saccharomyces cerevisiae carryingXYL1, XYL2, and XKS1 in mineral medium chemostat cultures. Applied andEnvironmental Microbiology 66 (8), 3381–3386.
Esterbauer, H., Steiner, W., Labudova, I., Hermann, A., Hayn, M., 1991.Production of Trichoderma cellulase in laboratory and pilot scale. BioresourceTechnology 36 (1), 51–65.
Fagerstam, L. G., Pettersson, L. G., 1980. The 1.4-beta-glucan cellobiohydrolasesof Trichoderma reesei QM 9414 - A new type of cellulolytic synergism. FEBSLetters 119 (1), 97–100.
Fengel, D., Wegener, G., 1989. Wood: chemistry, ultrastructure, reactions. WalterDe Gruyter, Berlin, New York.
Galbe, M., Sassner, P., Wingren, A., Zacchi, G., 2007. Process engineeringeconomics of bioethanol production. Biofuels 108, 303–327.
Galbe, M., Zacchi, G., 2002. A review of the production of ethanol from softwood.Applied Microbiology and Biotechnology 59 (6), 618–628.
Galbe, M., Zacchi, G., 2007. Pretreatment of lignocellulosic materials for efficientbioethanol production. Biofuels 108, 41–65.
Garcia, C., Jaldon, Dupeyre, D., Vignon, M. R., Mar. 1998. Fibres from semi-retted hemp bundles by steam explosion treatment. Biomass & Bioenergy 14 (3),251–260.
Ghose, T. K., 1987. Measurement of cellulase activities. Pure and AppliedChemistry 59 (2), 257–268.
Gnansounou, E., Dauriat, A., 2010. Techno-economic analysis of lignocellulosicethanol: A review. Bioresource Technology 101 (13), 4980–4991.
Goldstein, I. S., Easter, J. M., 1992. An improved process for converting celluloseto ethanol. Tappi Journal 75 (8), 135–140.
Gollapalli, L. E., Dale, B. E., Rivers, D. M., 2002. Predicting digestibility ofammonia fiber explosion (AFEX)-treated rice straw. Applied Biochemistry andBiotechnology 98, 23–35.
75
Hahn-Hägerdal, B., Wahlbom, C. F., Gárdonyi, M., van Zyl, W. H., Otero, R.R. C., Jonsson, L. J., 2001. Metabolic engineering of Saccharomyces cerevisiae forxylose utilization. Advances in Biochemical Engineering Biotechnology, 53–84.
Hamelinck, C. N., van Hooijdonk, G., Faaij, A. P. C., 2005. Ethanol fromlignocellulosic biomass: techno-economic performance in short-, middle- and long-term. Biomass & Bioenergy 28 (4), 384–410.
Hames, B., Ruiz, R., Scarlata, C., Sluiter, A., Sluiter, J., Templeton, D., 2008.Preparation of samples for compositional analysis. Laboratory AnalyticalProcedure NREL/TP-510-42620, National Renewable Energy Laboratory, Golden,CO.
Han, S. O., Cho, H. Y., Yukawa, H., Inui, M., Doi, R. H., 2004. Regulation ofexpression of cellulosomes and noncellulosomal (hemi)cellulolytic enzymes inClostridium cellulovorans during growth on different carbon sources. Journal ofBacteriology 186 (13), 4218–4227.
Harman, G. E., Kubicek, C. P., 1998. Trichoderma and Gliocladium; Enzymes,biological control and commercial applications. Taylor & Francis, London.
Harris, P. V., Welner, D., McFarland, K. C., Re, E., Poulsen, J. C. N., Brown, K.,Salbo, R., Ding, H. S., Vlasenko, E., Merino, S., Xu, F., Cherry, J., Larsen, S.,Lo Leggio, L., 2010. Stimulation of lignocellulosic biomass hydrolysis by proteinsof glycoside hydrolase family 61: Structure and function of a large, enigmaticfamily. Biochemistry 49 (15), 3305–3316.
Hayward, T. K., Hamilton, J., Templeton, D., Jennings, E., Ruth, M., Tholudur, A.,McMillan, J. D., Tucker, M., Mohagheghi, A., 1999. Enzyme production, growth,and adaptation of T. reesei strains QM 9414, L-27, RL-P37, and Rut C-30 toconditioned yellow poplar sawdust hydrolysate - Scientific note. AppliedBiochemistry and Biotechnology 77-9, 293–309.
Hendriks, A. T. W. M., Zeeman, G., 2009. Pretreatments to enhance thedigestibility of lignocellulosic biomass. Bioresource Technology 100 (1), 10–18.
Henrissat, B., Teeri, T. T., Warren, R. A. J., 1998. A scheme for designatingenzymes that hydrolyse the polysaccharides in the cell walls of plants. FEBSLetters 425 (2), 352–354.
Himmel, M. E., Ruth, M. F., Wyman, C. E., 1999. Cellulase for commodityproducts from cellulosic biomass. Current Opinion in Biotechnology 10 (4), 358–364.
76
Holtzapple, M. T., Jun, J. H., Ashok, G., Patibandla, S. L., Dale, B. E., 1991. Theammonia freeze explosion (AFEX) process - a practical lignocellulosepretreatment. Applied Biochemistry and Biotechnology 28-9, 59–74.
Hon, D. N. S., 1994. Cellulose - a random-walk along its historical path. Cellulose1 (1), 1–25.
Humbird, D., Aden, A., 2009. Biochemical production of ethanol from corn stover:2008 state of technology model. Technical Report NREL/TP-510-46214, NationalRenewable Energy Laboratory, Golden, CO.
Ike, M., Park, J., Tabuse, M., Tokuyasu, K., 2010. Cellulase production onglucose-based media by the UV-irradiated mutants of Trichoderma reesei. AppliedMicrobiology and Biotechnology 87 (6), 2059–2066.
Israilides, G. J., Grant, G. A., Han, Y. W., 1978. Sugar level, fermentability, andacceptability of straw treated with different acids. Applied and EnvironmentalMicrobiology 36 (1), 43–46.
Iyer, P. V., Wu, Z. W., Kim, S. B., Lee, Y. Y., 1996. Ammonia recycledpercolation process for pretreatment of herbaceous biomass. Applied Biochemistryand Biotechnology 57-8, 121–132.
Jalak, J., Valjamae, P., 2010. Mechanism of initial rapid rate retardation incellobiohydrolase catalyzed cellulose hydrolysis. Biotechnology andBioengineering 106 (6), 871–883.
Jeffries, T. W., Jin, Y. S., 2004. Metabolic engineering for improved fermentationof pentoses by yeasts. Applied Microbiology and Biotechnology 63 (5), 495–509.
John, M., Schmidt, J., Wandrey, C., Sahm, H., 1982. Gel chromatography ofoligosaccharides up to DP 60. Journal of Chromatography 247 (2), 281–288.
Jørgensen, H., Vibe-Pedersen, J., Larsen, J., Felby, C., 2007. Liquefaction oflignocellulose at high-solids concentrations. Biotechnology and Bioengineering96 (5), 862–870.
Juhász, T., Szengyel, Z., Réczey, K., Siika-Aho, M., Viikari, L., 2005.Characterization of cellulases and hemicellulases produced by Trichoderma reeseion various carbon sources. Process Biochemistry 40 (11), 3519–3525.
Kálmán, G., Varga, E., Réczey, K., 2002. Dilute sulphuric acid pretreatment ofcorn stover at long residence times. Chemical and Biochemical EngineeringQuarterly 16 (4), 151–157.
77
Karus, M., Vogt, D., 2004. European hemp industry: Cultivation, processing andproduct lines. Euphytica 140 (1-2), 7–12.
Kawamori, M., Morikawa, Y., Ado, Y., Takasawa, S., 1986. Production ofcellulases from alkali-treated bagasse in Trichoderma reesei. AppliedMicrobiology and Biotechnology 24 (6), 454–458.
Kazi, F. K., Fortman, J., Anex, R., Kothandaraman, G., Hsu, D., Aden, A., Dutta,A., 2010. Techno-economic analysis of biochemical scenarios for production ofcellulosic ethanol. Technical Report NREL/TP-6A2-46588, National RenewableEnergy Laboratory, Golden, CO.
Kim, S., Dale, B. E., 2004. Global potential bioethanol production from wastedcrops and crop residues. Biomass & Bioenergy 26 (4), 361–375.
Kim, S., Holtzapple, M. T., 2005. Lime pretreatment and enzymatic hydrolysis ofcorn stover. Bioresource Technology 96 (18), 1994–2006.
Koukiekolo, R., Cho, H. Y., Kosugi, A., Inui, M., Yukawa, H., Doi, R. H., 2005.Degradation of corn fiber by Clostridium cellulovorans cellulases andhemicellulases and contribution of scaffolding protein CbpA. Applied andEnvironmental Microbiology 71 (7), 3504–3511.
Kovács, K., Megyeri, L., Szakács, G., Kubicek, C. P., Galbe, M., Zacchi, G., 2008.Trichoderma atroviride mutants with enhanced production of cellulase and beta-glucosidase on pretreated willow. Enzyme and Microbial Technology 43 (1), 48–55.
KSH, 2010. www.ksh.hu.
Larsson, M., Galbe, M., Zacchi, G., 1997. Recirculation of process water in theproduction of ethanol from softwood. Bioresource Technology 60 (2), 143–151.
Larsson, S., Palmqvist, E., Hahn-Hägerdal, B., Tengborg, C., Stenberg, K., Zacchi,G., Nilvebrant, N. O., 1999. The generation of fermentation inhibitors during diluteacid hydrolysis of softwood. Enzyme and Microbial Technology 24 (3-4), 151–159.
Laser, M., Jin, H. M., Jayawardhana, K., Lynd, L. R., 2009. Coproduction ofethanol and power from switchgrass. Biofuels Bioproducts & Biorefining-biofpr3 (2), 195–218.
Linde, M., Jakobsson, E. L., Galbe, M., Zacchi, G., 2008. Steam pretreatment ofdilute H2SO4-impregnated wheat straw and SSF with low yeast and enzymeloadings for bioethanol production. Biomass & Bioenergy 32 (4), 326–332.
78
Lloyd, T. A., Wyman, C. E., 2005. Combined sugar yields for dilute sulfuric acidpretreatment of corn stover followed by enzymatic hydrolysis of the remainingsolids. Bioresource Technology 96 (18), 1967–1977.
Lynd, L. R., 1996. Overview and evaluation of fuel ethanol from cellulosicbiomass: Technology, economics, the environment, and policy. Annual Review ofEnergy and the Environment 21, 403–465.
Lynd, L. R., Elander, R. T., Wyman, C. E., 1996. Likely features and costs ofmature biomass ethanol technology. Applied Biochemistry and Biotechnology 57-8, 741–761.
Lynd, L. R., Weimer, P. J., van Zyl, W. H., Pretorius, I. S., 2002. Microbialcellulose utilization: Fundamentals and biotechnology. Microbiology andMolecular Biology Reviews 66 (3), 506–577.
Mach, R. L., Zeilinger, S., 2003. Regulation of gene expression in industrial fungi:Trichoderma. Applied Microbiology and Biotechnology 60 (5), 515–522.
Maheswari, D. K., Jahan, H., Paul, J., Varma, A., 1993. Wheat straw, a potentialsubstrate for cellulase production using Trichoderma reesei. World Journal ofMicrobiology & Biotechnology 9 (1), 120–121.
Mandels, M., Reese, E. T., Parrish, F. W., 1962. Sophorose as an inducer ofcellulase in Trichoderma viride. Journal of Bacteriology 83 (2), 400.
Mansfield, S. D., Mooney, C., Saddler, J. N., 1999. Substrate and enzymecharacteristics that limit cellulose hydrolysis. Biotechnology Progress 15 (5), 804–816.
Messner, R., Hagspiel, K., Kubicek, C. P., 1990. Isolation of a beta-glucosidasebinding and activating polysaccharide from cell-walls of Trichoderma reesei.Archives of Microbiology 154 (2), 150–155.
Milbrandt, A., 2005. A geographic perspective on the current biomass resourceavailability in the United States. Technical Report NREL/TP-560-39181, NationalRenewable Energy Laboratory, Golden, CO.
Miller, G. L., 1959. Use of dinitrosalicylic acid reagent for determination ofreducing sugar. Analytical Chemistry 31 (3), 426–428.
Miller, G. L., 1963. Cellodextrins. Methods in Carbohydrate Chemistry 3, 134–139.
79
Monavari, S., Bennato, A., Galbe, M., Zacchi, G., 2010. Improved one-step steampretreatment of SO2-impregnated softwood with time-dependent temperatureprofile for ethanol production. Biotechnology Progress 26 (4), 1054–1060.
Monavari, S., Galbe, M., Zacchi, G., 2009. Impact of impregnation time and chipsize on sugar yield in pretreatment of softwood for ethanol production. BioresourceTechnology 100 (24), 6312–6316.
Morgavi, D. P., Beauchemin, K. A., Nsereko, V. L., Rode, L. M., Iwaasa, A. D.,Yang, W. Z., McAllister, T. A., Wang, Y., 2000. Synergy between ruminalfibrolytic enzymes and enzymes from Trichoderma longibrachiatum. Journal ofDairy Science 83 (6), 1310–1321.
Mosier, N., Wyman, C., Dale, B., Elander, R., Lee, Y. Y., Holtzapple, M., Ladisch,M., 2005. Features of promising technologies for pretreatment of lignocellulosicbiomass. Bioresource Technology 96 (6), 673–686.
Moxley, G., Zhu, Z., Zhang, Y.-H. P., 2008. Efficient sugar release by the cellulosesolvent-based lignocellulose fractionation technology and enzymatic cellulosehydrolysis. Journal of Agricultural and Food Chemistry 56 (17), 7885–7890.
Murashima, K., Kosugi, A., Doi, R. H., 2003. Synergistic effects of cellulosomalxylanase and cellulases from Clostridium cellulovorans on plant cell walldegradation. Journal of Bacteriology 185 (5), 1518–1524.
Nidetzky, B., Steiner, W., Hayn, M., Claeyssens, M., 1994. Cellulose hydrolysis bythe cellulases from Trichoderma reesei - A new model for synergistic interaction.Biochemical Journal 298, 705–710.
Nystrom, J. M., Allen, A. L., 1976. Pilot scale investigations and economics ofcellulase production. Biotechnology and Bioengineering (6), 55–74.
Öhgren, K., Galbe, M., Zacchi, G., 2005. Optimization of steam pretreatment ofSO2-impregnated corn stover for fuel ethanol production. Applied Biochemistryand Biotechnology 124, 1055–1068.
Ohta, K., Beall, D. S., Mejia, J. P., Shanmugam, K. T., Ingram, L. O., 1991.Genetic-improvement of Escherichia coli for ethanol-production - chromosomalintegration of Zymomonas mobilis genes encoding pyruvate decarboxylase andalcohol dehydrogenase-II. Applied and Environmental Microbiology 57 (4), 893–900.
Olsson, J., Zacchi, G., 2001. Simulation of the condensate treatment process in akraft pulp mill. Chemical Engineering & Technology 24 (2), 195–203.
80
Olsson, L., Christensen, T. M. I. E., Hansen, K. P., Palmqvist, E. A., 2003.Influence of the carbon source on production of cellulases, hemicellulases andpectinases by Trichoderma reesei Rut C-30. Enzyme and Microbial Technology33 (5), 612–619.
Olsson, L., Hahn-Hägerdal, B., 1993. Fermentative performance of bacteria andyeasts in lignocellulose hydrolysates. Process Biochemistry 28 (4), 249–257.
Palmqvist, E., Hahn-Hägerdal, B., Galbe, M., Larsson, M., Stenberg, K., Szengyel,Z., Tengborg, C., Zacchi, G., Nov. 1996. Design and operation of a bench-scaleprocess development unit for the production of ethanol from lignocellulosics.Bioresource Technology 58 (2), 171–179.
Perez, J. A., Ballesteros, I., Ballesteros, M., Saez, F., Negro, M. J., Manzanares, P.,2008. Optimizing liquid hot water pretreatment conditions to enhance sugarrecovery from wheat straw for fuel-ethanol production. Fuel 87 (17-18), 3640–3647.
Persson, I., Tjerneld, F., Hahn-Hägerdal, B., 1991. Fungal cellulolytic enzyme-production - A review. Process Biochemistry 26 (2), 65–74.
Persson, M., 2009. Inbicon - production of cellulosic bioethanol on an industrialscale in sight. International Sugar Journal 111 (1331), 709–712.
Peters, M. S., Timmerhaus, K. D., West, R., 1991. Plant design and economics forchemical engineers. McGraw-Hill, New York.
Réczey, K., Szengyel, Z., Eklund, R., Zacchi, G., 1996. Cellulase production byT. reesei. Bioresource Technology 57 (1), 25–30.
Reese, E. T., Siu, R. G. H., Levinson, H. S., 1950. The biological degradation ofsoluble cellulose derivatives and its relationship to the mechanism of cellulosehydrolysis. Journal of Bacteriology 59 (4), 485–497.
Rudolf, A., 2007. Fermentation and cultivation technology for improved ethanolproduction from lignocellulose. PhD thesis, Lund University, Lund, Sweden.
Saddler, J. N., Ramos, L. P., Breuil, C., 1993. Steam pretreatment oflignocellulosic residues. Biotechnology in Agriculture 9, 73–91.
Saha, B. C., 2003. Hemicellulose bioconversion. Journal of IndustrialMicrobiology & Biotechnology 30 (5), 279–291.
Sassner, P., Galbe, M., Zacchi, G., 2005. Steam pretreatment of Salix with andwithout SO2 impregnation for production of bioethanol. Applied Biochemistry andBiotechnology 121, 1101–1117.
81
Sassner, P., Galbe, M., Zacchi, G., 2006. Bioethanol production based onsimultaneous saccharification and fermentation of steam-pretreated Salix at highdry-matter content. Enzyme and Microbial Technology 39 (4), 756–762.
Sassner, P., Galbe, M., Zacchi, G., 2008. Techno-economic evaluation ofbioethanol production from three different lignocellulosic materials. Biomass &Bioenergy 32 (5), 422–430.
Sassner, P., Zacchi, G., 2008. Integration options for high energy efficiency andimproved economics in a wood-to-ethanol process. Biotechnology for Biofuels1 (4), (15 April 2008).
Sendich, E., Laser, M., Kim, S., Alizadeh, H., Laureano-Perez, L., Dale, B., Lynd,L., 2008. Recent process improvements for the ammonia fiber expansion (AFEX)process and resulting reductions in minimum ethanol selling price. BioresourceTechnology 99 (17), 8429–8435.
Shiraishi, N., 1991. Wood and cellulosic chemistry. Marcel Dekker, New York.
Sipos, B., Benkő, Z., Dienes, D., Réczey, K., Viikari, L., Siika-aho, M., 2010.Characterisation of specific activities and hydrolytic properties of cell-wall-degrading enzymes produced by Trichoderma reesei Rut C30 on different carbonsources. Applied Biochemistry and Biotechnology 161 (1-8), 347–364.
Sivers, M., Zacchi, G., 1995. A technoeconomic comparison of 3 processes for theproduction of ethanol from pine. Bioresource Technology 51 (1), 43–52.
Sivers, M., Zacchi, G., 1996. Ethanol from lignocellulosics: A review of theeconomy. Bioresource Technology 56 (2-3), 131–140.
Sjöström, E., 1993. Wood chemistry: Fundamentals and applications. AcademicPress, San Diego, CA.
Sluiter, A., Hames, B., Ruiz, R., Scarlata, C., Sluiter, J., Templeton, D., 2008a.Determination of ash in biomass. Laboratory Analytical Procedure NREL/TP-510-42622, National Renewable Energy Laboratory, Golden, CO.
Sluiter, A., Hames, B., Ruiz, R., Scarlata, C., Sluiter, J., Templeton, D., 2008b.Determination of sugars, byproducts, and degradation products in liquid fractionprocess samples. Laboratory Analytical Procedure NREL/TP-510-42623, NationalRenewable Energy Laboratory, Golden, CO.
Sluiter, A., Hames, B., Ruiz, R., Scarlata, C., Sluiter, J., Templeton, D., Crocker,D., 2008c. Determination of structural carbohydrates and lignin in biomass.Laboratory Analytical Procedure NREL/TP-510-42618, National RenewableEnergy Laboratory, Golden, CO.
82
Sluiter, A., Ruiz, R., Scarlata, C., Sluiter, J., Templeton, D., 2008d. Determinationof extractives in biomass. Laboratory Analytical Procedure NREL/TP-510-42619,National Renewable Energy Laboratory, Golden, CO.
So, K. S., Brown, R. C., 1999. Economic analysis of selected lignocellulose-to-ethanol conversion technologies. Applied Biochemistry and Biotechnology 77-9,633–640.
Söderstrom, J., Pilcher, L., Galbe, M., Zacchi, G., 2003. Two-step steampretreatment of softwood by dilute H2SO4 impregnation for ethanol production.Biomass & Bioenergy 24 (6), 475–486.
Solomon, B. D., Barnes, J. R., Halvorsen, K. E., 2007. Grain and cellulosicethanol: History, economics, and energy policy. Biomass & Bioenergy 31 (6), 416–425.
Strauss, J., Mach, R. L., Zeilinger, S., Hartler, G., Stoffler, G., Wolschek, M.,Kubicek, C. P., 1995. Crel, the carbon catabolite repressor protein fromTrichoderma reesei. FEBS Letters 376 (1-2), 103–107.
Sun, Y., Cheng, J. Y., 2002. Hydrolysis of lignocellulosic materials for ethanolproduction: A review. Bioresource Technology 83 (1), 1–11.
Swings, J., Deley, J., 1977. Biology of Zymomonas. Bacteriological Reviews41 (1), 1–46.
Szengyel, Z., Zacchi, G., 2000. Effect of acetic acid and furfural on cellulaseproduction of Trichoderma reesei Rut C30. Applied Biochemistry andBiotechnology 89 (1), 31–42.
Szengyel, Z., Zacchi, G., Réczey, K., 1997. Cellulase production based onhemicellulose hydrolysate from steam-pretreated willow. Applied Biochemistryand Biotechnology 63-5, 351–362.
Szengyel, Z., Zacchi, G., Varga, A., Réczey, K., 2000. Cellulase production ofTrichoderma reesei Rut C30 using steam-pretreated spruce - Hydrolytic potentialof cellulases on different substrates. Applied Biochemistry and Biotechnology 84-6, 679–691.
Tabka, M. G., Herpoel-Gimbert, I., Monod, F., Asther, M., Sigoillot, J. C., 2006.Enzymatic saccharification of wheat straw for bioethanol production by acombined cellulase xylanase and feruloyl esterase treatment. Enzyme andMicrobial Technology 39 (4), 897–902.
83
Taherzadeh, M. J., Niklasson, C., Lidén, G., 1997. Acetic acid - friend or foe inanaerobic batch conversion of glucose to ethanol by Saccharomyces cerevisiae?Chemical Engineering Science 52 (15), 2653–2659.
Takács, E., Wojnarovits, L., Földváry, C., Hargittai, P., Borsa, J., Sajó, I., 2000.Effect of combined gamma-irradiation and alkali treatment on cotton-cellulose.Radiation Physics and Chemistry 57 (3-6), 399–403.
Teeri, T. T., 1997. Crystalline cellulose degradation: New insight into the functionof cellobiohydrolases. Trends in Biotechnology 15 (5), 160–167.
Tengborg, C., Galbe, M., Zacchi, G., 2001. Influence of enzyme loading andphysical parameters on the enzymatic hydrolysis of steam-pretreated softwood.Biotechnology Progress 17 (1), 110–117.
Teymouri, F., Laureano-Perez, L., Alizadeh, H., Dale, B. E., 2005. Optimization ofthe ammonia fiber explosion (AFEX) treatment parameters for enzymatichydrolysis of corn stover. Bioresource Technology 96 (18), 2014–2018.
Tomás-Pejó, E., Oliva, J. M., Ballesteros, M., 2008a. Realistic approach for full-scale bioethanol production from lignocellulose: A review. Journal of Scientific &Industrial Research 67 (11), 874–884.
Tomás-Pejó, E., Oliva, J. M., Ballesteros, M., Olsson, L., 2008b. Comparison ofSHF and SSF processes from steam-exploded wheat straw for ethanol productionby xylose-fermenting and robust glucose-fermenting Saccharomyces cerevisiaestrains. Biotechnology and Bioengineering 100 (6), 1122–1131.
Torget, R., Himmel, M., Grohmann, K., 1992. Dilute-acid pretreatment of 2 short-rotation herbaceous crops - Scientific note. Applied Biochemistry andBiotechnology 34-5, 115–123.
Tuka, K., Zverlov, V. V., Velikodvorskaya, G. A., 1992. Synergism betweenClostridium thermocellum cellulases cloned in Escherichia coli. AppliedBiochemistry and Biotechnology 37 (2), 201–207.
Ulbricht, R. J., Northup, S. J., Thomas, J. A., 1984. A review of 5-hydroxymethylfurfural (HMF) in parenteral solutions. Fundamental and AppliedToxicology 4 (5), 843–853.
Vignon, M., Garcia-Jaldon, C., Dupeyre, D., 1995. Steam explosion of woodyhemp chenevotte. International Journal of Biological Macromolecules 17 (6), 395–404.
Walker, L. P., Wilson, D. B., Irwin, D. C., Mcquire, C., Price, M., 1992.Fragmentation of cellulose by the major Thermomonospora fusca cellulases,
84
Trichoderma reesei CBH I, and their mixtures. Biotechnology and Bioengineering40 (9), 1019–1026.
Weil, J., Brewer, M., Hendrickson, R., Sarikaya, A., Ladisch, M. R., 1998.Continuous pH monitoring during pretreatment of yellow poplar wood sawdust bypressure cooking in water. Applied Biochemistry and Biotechnology 70-2, 99–111.
Weil, J., Westgate, P., Kohlmann, K., Ladisch, M. R., 1994. Cellulosepretreatments of lignocellulosic substrates. Enzyme and Microbial Technology16 (11), 1002–1004.
Wingren, A., Galbe, M., Zacchi, G., 2003. Techno-economic evaluation ofproducing ethanol from softwood: Comparison of SSF and SHF and identificationof bottlenecks. Biotechnology Progress 19 (4), 1109–1117.
Wingren, A., Galbe, M., Zacchi, G., 2008. Energy considerations for a SSF-basedsoftwood ethanol plant. Bioresource Technology 99 (7), 2121–2131.
Wood, T. M., 1992. Fungal cellulases. Biochemical Society Transactions 20 (1),46–53.
Wood, T. M., McCrae, S. I., 1972. Purification and properties of C1 component ofTrichoderma koningii cellulase. Biochemical Journal 128 (5), 1183–&.
Wooley, R., Putsche, V., 1996. Development of an Aspen Plus physical propertydatabase for biofuel components. Technical Report NREL/MP-425-20685,National Renewable Energy Laboratory, Golden, CO.
Wooley, R., Ruth, M., Glassner, D., Sheehan, J., 1999. Process design and costingof bioethanol technology: A tool for determining the status and direction ofresearch and development. Biotechnology Progress 15 (5), 794–803.
Wyman, C. E., 1996. Handbook on bioethanol: Production and utilization. Taylor& Francis, Washington.
Wyman, C. E., Dale, B. E., Elander, R. T., Holtzapple, M., Ladisch, M. R., Lee,Y. Y., 2005. Coordinated development of leading biomass pretreatmenttechnologies. Bioresource Technology 96 (18), 1959–1966.
Xin, Z., Yinbo, Q., Peiji, G., 1993. Acceleration of ethanol-production from paper-mill waste fiber by supplementation with beta-glucosidase. Enzyme and MicrobialTechnology 15 (1), 62–65.
Xu, J. C., Smith, J. C., 2010. Probing the mechanism of cellulosome attachment tothe Clostridium thermocellum cell surface: computer simulation of the type II
85
cohesin-dockerin complex and its variants. Protein Engineering Design & Selection23 (10), 759–768.
Yang, B., Wyman, C. E., 2008. Pretreatment: The key to unlocking low-costcellulosic ethanol. Biofuels Bioproducts & Biorefining-biofpr 2 (1), 26–40.
Yoon, H. H., Wu, Z. W., Lee, Y. Y., 1995. Ammonia-recycled percolation processfor pretreatment of biomass feedstock. Applied Biochemistry and Biotechnology51-2, 5–19.
Yu, P. Q., McKinnon, J. J., Maenz, D. D., Olkowski, A. A., Racz, V. J.,Christensen, D. A., 2003. Enzymic release of reducing sugars from oat hulls bycellulase, as influenced by Aspergillus ferulic acid esterase and Trichodermaxylanase. Journal of Agricultural and Food Chemistry 51 (1), 218–223.
Zaldivar, J., Nielsen, J., Olsson, L., 2001. Fuel ethanol production fromlignocellulose: a challenge for metabolic engineering and process integration.Applied Microbiology and Biotechnology 56 (1-2), 17–34.
Zhang, Y. H. P., Himmel, M. E., Mielenz, J. R., 2006. Outlook for cellulaseimprovement: Screening and selection strategies. Biotechnology Advances 24 (5),452–481.
Zhang, Y. H. P., Lynd, L. R., 2004. Toward an aggregated understanding ofenzymatic hydrolysis of cellulose: Noncomplexed cellulase systems.Biotechnology and Bioengineering 88 (7), 797–824.
Zhou, S. G., Ingram, L. O., 2000. Synergistic hydrolysis of carboxymethylcellulose and acid-swollen cellulose by two endoglucanases (CelZ and CelY) fromErwinia chrysanthemi. Journal of Bacteriology 182 (20), 5676–5682.
87
STATEMENT
I the undersigned, hereby declare, that this PhD thesis was written by me,
and I only used the sources listed in the reference list.
NYILATKOZAT
Alulírott Barta Zsolt kijelentem, hogy ezt a doktori értekezést magam
készítettem és abban csak a megadott forrásokat használtam fel. Minden
olyan részt, amelyet szó szerint, vagy azonos tartalomban, de átfogalmazva
más forrásból átvettem, egyértelműen, a forrás megadásával megjelöltem.
Budapest, 2011. január 19. Barta Zsolt
I
Refining Hemp Hurds into Fermentable Sugars or Ethanol
Z. Barta,a,* J. M. Oliva,b I. Ballesteros,b D. Dienes,a M. Ballesteros,b and K. Réczeya
aBudapest University of Technology and Economics,Department of Applied Biotechnology and Food Science,Szt. Gellért tér 4, 1111 Budapest, HungarybCIEMAT, Renewable Energies Division, Biomass Unit,Avda. Complutense n° 22, 28040 Madrid, Spain
Steam pretreatment is one of the most efficient pretreatment technologies employedprior to enzymatic hydrolysis of lignocellulosics to obtain high polysaccharide conversion.In this study, steam pretreatment of non-impregnated hemp hurds was investigated at tworeactor scales (2 and 10 L) by varying the temperature from 200–230 °C. Glucan recover-ies were relatively high (> 82 % of original content), while xylan recoveries ranged from18–66 % of original. Conversions of glucan and xylan in enzymatic hydrolysis varied be-tween X � 62–83 % and 46–96 %, respectively, based on glucan and xylan contents of thesubstrate. Ethanol yields in simultaneous saccharification and fermentation ranged fromY � 38–70 %, based on glucan and mannan contents of the substrate. The highest overallglucose yield was 336 g kg–1 dry hurds obtained at 210 °C, whereas the maximum sugaryield (glucose � xylose), 414 g kg–1 dry hurds, was achieved at 200 °C. The highest etha-nol yield, 141 g kg–1 dry hurds, was obtained at 210 °C.
Key words:
Hemp hurds, steam pretreatment, enzymatic hydrolysis, SSF, ethanol production
Introduction
Ethanol has a rapidly expanding market pri-marily due to utilisation in vehicular fuel.1,2 Cur-rently, more than 96 % is produced from renewableresources, referred to as bioethanol.3 First- and sec-ond-generation production technologies can be dis-tinguished according to maturity and feedstocks.4
For the former, sugar substances and grains serve asfeedstocks and these technologies already exist on alarge scale, while the latter utilise lignocellulosicmaterials and involve more complex technologiesthat have not yet been introduced at an industrialscale. Whereas sugar and starch-containing feed-stocks are limited, relatively expensive and supplythe food industry as well, lignocellulosic biomass isabundant, available at low cost and largely un-used.5,6
Cellulose can be broken down into fermentableglucose using an acid catalyst or cellulase enzymecomplex.7 In order to obtain appropriate poly-saccharide conversion in enzymatic hydrolysis, pre-treatment of the raw material is required. Steampretreatment has been found to be one of the mostefficient pretreatment technologies.8,9 In this pro-cess, chopped biomass is subjected to high-pressuresteam in a reactor for a given time, and then thepressure is released instantaneously. Since hemi-cellulose contains acetyl groups that are cleaved
during the reaction forming acetic acid, catalysisoccurs, which is referred to as autohydrolysis.10
Nevertheless, impregnation with other acids (cata-lyst addition) prior to pretreatment has been provento have a positive effect on the sugar recoverythrough steam pretreatment by enabling lower tem-peratures and shorter residence times.11 Moreover,impregnation improves enzymatic hydrolysis aswell as ethanol fermentation by reducing the forma-tion of inhibitory compounds from sugar degrada-tion.12,13 Drawbacks of steam pretreatment are thepartial degradation of hemicellulose, production ofenzymes and fermentation inhibitors, and incom-plete separation of lignin and cellulose.8,14
Hemp hurds are the woody core of industrialhemp (non-drug varieties of Cannabis sativa L.)which is one of the raw materials of the fibre indus-try, where bast fibres are required. Conventionally,they are removed after dew-retting the hemp stalksin the field15 or after water-retting. In these pro-cesses, the pectinases produced by bacteria presenton the material digest the pectin and enable fibreseparation. The residue, containing less cellulosebut more hemicellulose than the fibre fraction, is re-ferred to as hemp hurds and has had only minor ap-plications so far, such as animal bedding due to itshigh water-absorbing ability, garden mulch or acomponent of light-weight concrete. Hemp hurdsconstitute 70 % of the stalk dry matter.16
Industrial hemp is an attractive crop, because itcan reach high biomass yield, possesses low sus-
Z. BARTA et al., Refining Hemp Hurds into Fermentable Sugars or Ethanol, Chem. Biochem. Eng. Q. 24 (3) 331–339 (2010) 331
*Corresponding author. Tel.: �36-1-463-2843, Fax: �36-1-463-2598,
e mail: [email protected]
Original scientific paperReceived: October 21, 2009
Accepted: May 14, 2010
ceptibility to pests and diseases, and suppressescompeting weeds.17 In the 1960s, Hungary pro-duced huge amounts of industrial hemp, which hassince dropped significantly; nevertheless the condi-tions (climate, soil etc.) are still appropriate for cul-tivation of large quantities. Since hemp hurds are anagro-industrial by-product with a high carbohydratecontent, they are a potential candidate for sec-ond-generation fuel-ethanol production. Vignon etal.19 investigated impregnated hemp woody core ina steam explosion treatment to optimise fibre sepa-ration and delignification, primarily by means ofoptical and scanning electron microscopy andcompositional analysis. Moxley et al.17 performedcellulose-solvent-based lignocellulose fractionationof hemp hurds and subsequent enzymatic hydroly-sis of the residual cellulose fraction.
In this study, the hemp hurds upstream-pro-cessing lines, consisting of steam pretreatment andsubsequent enzymatic hydrolysis or simultaneoussaccharification and fermentation, were investi-gated to produce fermentable sugars or ethanol, re-spectively. The lines differed in pretreatment condi-tions. The efficiency of pretreatment was assessedin terms of sugar recoveries, polysaccharide con-versions in enzymatic hydrolysis using commercialcellulases, as well as ethanol yields obtained in si-multaneous saccharification and fermentation (SSF)employing ordinary baker’s yeast.
Materials and methods
Raw material and chemicals
Hemp hurds were kindly provided by Hungaro-hemp (Nagylak, Hungary). The 1.5–2.5 m longhemp stalks were retted, dried, and then milled bymeans of finned rolls. The fibres remained more orless intact, whereas the hemp hurds were brokeninto 1–2 cm long pieces with a width of 1–5 mm.Hemp hurds were used for steam pretreatment atthat size. All chemicals were obtained from Merck(Darmstadt, Germany), otherwise it is given below.
Steam pretreatment
Five pretreatments were performed for 10 minwithout impregnation at 220 and 230 °C in a 2 L re-actor as well as at 200, 210 and 220 °C in a 10 L re-actor. Steam pretreatment units have been describedelsewhere.20 Dry matter loading was 200 g in the2 L reactor and 500 g in the 10 L reactor. The re-sulting slurry was weighed and separated into liquidand solid fractions. The solids were washed andanalysed for carbohydrate and lignin contents.Monomeric sugars and degradation products in theliquid fraction were measured by high-performance
liquid chromatography (HPLC). Post-hydrolysis ofoligomeric sugars in the liquid fraction was alsoperformed.
Estimation of furan derivatives and acidsin flash vapour
During expansion to ambient pressure, the vol-atile compounds partially vapourise. Since thesteam pretreatment unit is not equipped with a flashvapour condenser, these compounds were measuredonly in the slurry. A flash calculation was per-formed with Aspen Plus (Aspen Tech Inc. Cam-bridge, MA, USA) using NRTL-2 method to esti-mate the total amount of furfural, hydroxymethyl-furfural (HMF), and acetic and formic acids formedin the reaction. Before expansion, the vapour frac-tion in the reactor was set to zero, i.e., all steam in-jected was assumed to condense completely.
Compositional analysis
Oven-dried samples of raw hurds were groundin a laboratory hammer mill (type: SM 2000,Retsch, Haan, Germany) and separated with a sieveset (type: RP 20, CISA, Barcelona, Spain) into twofractions in accordance with the procedure ofNational Renewable Energy Laboratory (NREL,Golden, CO).21 Samples of oven-dried pretreatedsolid fractions were milled and sieved to a meshsize of 1 mm. Non-volatile water and ethanol ex-tractives of the raw hurds were determined bymeans of an accelerated solvent extraction (ASE)according to the NREL protocol.22 ASE was per-formed on the 0.18–0.85 mm fraction. Non-volatileextractives remaining in the flask after evaporationwere measured by weight after drying at 105 °C.The carbohydrate and lignin contents were de-termined according to the NREL standardisedmethod.23 Extracted material as well as pretreatedsolid fractions were treated in a two-step acid hy-drolysis with 72 % H2SO4 for 1 h at 30 °C and thenwith 4 % H2SO4 for 1 h at 121 °C. Sugar contentsof the supernatants were analysed with HPLC. Thedried residue consisted of acid-insoluble lignin andacid-insoluble ash. Total ash, structural inorganicsand acid-insoluble ash contents were measured byashing the sample at 575 °C for at least 4 h until thesample weight was constant.24 Total ash refers tothe inorganic part of the raw material or solid frac-tion after pretreatment, whereas structural inorganicsare measured as ash of the extracted material. Todetermine the total sugar content in the liquid frac-tion of the pretreated material, samples of 25 mLwere mixed with 0.72 mL of 98 % H2SO4, andautoclaved for 30 min at 121 °C.25 Sugar contentswere determined by HPLC.
332 Z. BARTA et al., Refining Hemp Hurds into Fermentable Sugars or Ethanol, Chem. Biochem. Eng. Q. 24 (3) 331–339 (2010)
Enzymatic hydrolysis
Enzymatic hydrolysis (EH) of the washed solidfraction was carried out in shake flasks in a reactionvolume of 25 mL at a water-insoluble solid (WIS)content of 5 % at 50 °C using commercial enzymepreparations (NS50013 – cellulase complex, NS50010– beta-glucosidase preparation, both from NovozymesA/S, Bagsværd, Denmark). Enzyme loading ofNS50013 was 15 FPU (filter paper unit) g–1 WISand the volumetric ratio (�) of NS50010 toNS50013 was 0.1. The rotary shaker (Certomat-R,B-Braun, Germany) was adjusted to 150 rpm. So-dium citrate buffer (c � 0.05 mol dm–3) was used tomaintain pH at 5.0. Triplicates were incubated for72 h. Released sugars were analysed by HPLC.
Simultaneous saccharification andfermentation
Non-sterile simultaneous saccharification andfermentation of washed fibres was performed inshake flasks in a volume of 50 mL at a WIS contentof w � 10 % by applying commercial enzyme prep-arations (NS50013, NS50010) and Saccharomycescerevisiae (DER-CIEMAT Culture CollectionNo. 1701) at 32 °C. Enzyme loading of NS 50013was 25 FPU g–1 glucan. Volumetric ratio ofNS50010 to NS50013 and agitation speed were thesame as for the EH experiments. Yeast extract,NH4Cl, KH2PO4 (Sigma-Aldrich, Munich, Ger-many) and MgSO4 · 7H2O were added in final con-centrations of � � 5, 2, 1 and 0.3 g L–1, respec-tively. Flasks were inoculated with a 4 % (v/w)S. cerevisiae culture grown for 16 h at 32 °C at150 rpm in the same growth medium supplemen-ted with glucose (� � 30 g L–1). Sodium citratebuffer (c � 0.05 mol dm–3) was used to adjust theinitial pH to 5.0. Triplicates were incubated for72 h. Ethanol was measured by gas chromato-graphy.
Chromatography
Sugar contents (glucose, xylose, arabinose,mannose, and galactose) were determined by a Wa-ters liquid chromatograph with a refractive indexdetector using an Aminex HPX-87P column(Bio-Rad, Hercules, CA) operating at 85 °C. Ultra-pure water was used as eluent at a flow rate ofQ � 0.6 mL min–1. Furfural and HMF contentswere analysed by a Hewlett Packard (HP) liquidchromatograph, equipped with an UV-diode-arraydetector, and the separation was performed with anAminex HPX-87H column at 55 °C with an eluentof c � 5 mmol dm–3 H2SO4 (89 %) and acetonitrile(11 %) at a flow rate of Q � 0.7 mL min–1. Quanti-fication of acetic and formic acids was carried outwith a Waters liquid chromatograph with a refractive
index detector using an Aminex HPX-87H columnat 65 °C. The mobile phase was c � 5 mmol dm–3
H2SO4 at a flow rate of Q � 0.6 mL min–1. Ethanolcontent was measured by gas chromatography,using an HP apparatus equipped with a flameionisation detector and a Carbowax column operat-ing at 85 °C. The 20M injector and detector temper-atures were maintained at 150 °C.
Statistical analysis
Statistical evaluation was carried out usingStatistica 8.0 software (Statsoft Inc., Tulsa, OK).Independent t-tests of glucan and xylan conversionsin EH and yields in SSF experiments were per-formed for some pairs of conditions to ascertainwhether they differed significantly, where the sig-nificance level was � 0.05. The two-tailed t-testprobability is denoted here by p.
Results and discussion
Compositions
Results of the compositional analysis are pre-sented in Table 1. The most abundant polysaccha-rides in the raw hurds were glucan and xylan, to-gether comprising w � 59 % of the dry matter. Thehemp-hurd composition reported by Moxley et al.17
is in line with that obtained here, except that theirraw material consisted of 4.7 % arabinan. The com-position of hemp hurds was compared to that ofother lignocellulosic raw materials such as wheatstraw,26,27 corn stover,28,29 softwoods (spruce,30
pine26) and hardwoods (salix,11 poplar,31 aspen,birch26). The glucan content of hemp hurds wasslightly higher than that of wheat and barley straw,and it was in the range of corn stover, softwoodsand hardwoods. Regarding the xylan, mannan andlignin contents, major differences were obtained be-tween softwoods and hemp hurds. The hemicellu-lose of softwoods consisted of more mannan, butless xylan, and the lignin content of softwoods wasalso higher. Galactan and arabinan are minor con-stituents of each raw material, and none of theircontents varied significantly among the materialssurveyed.
The glucan contents of the pretreated solidfraction varied between 57–63 %. The increase inglucan and lignin contents, compared to the originalmaterial, was due to solubilisation of the hemi-cellulose fraction through pretreatment. Indeed,mannan and galactan present in the raw hurds dis-appeared entirely from the solid fraction, except forthe case of the 200 °C treatment; a significant dropin xylan content was observed as well. Vignonet al.19 reported the composition of two kinds of
Z. BARTA et al., Refining Hemp Hurds into Fermentable Sugars or Ethanol, Chem. Biochem. Eng. Q. 24 (3) 331–339 (2010) 333
steam-pretreated hemp hurds (220 and 240 °C,3 min, 0.1 % H2SO4). A major difference was foundin the glucan contents (48 and 53 %, respectively)compared to those in Table 1, however, the authorssuggested that their glucan contents might be un-derestimated.
Steam pretreatment
In steam pretreatment, as the steam heats thematerial, it simultaneously condenses. Dependingon the heat flux between the environment and thereactor, i.e., on the heat losses to ambient, furthercondensation can occur. Increased steam condensa-tion results in a more diluted slurry. Table 2 showsthat higher solid contents were obtained in the 10 Lreactor than in the smaller one, which could be dueto different equipment designs, and concomitantdifferences in heat loss. As acid concentrations inthe liquid fraction obtained in the large reactor wereonly twice that in the small one, the pH values didnot differ significantly.
Fig. 1 shows the dry matter and component re-coveries after pretreatment as % of original. Similartrends were observed for galactan and mannan aswell (data not shown). The recovered portion de-creased when the pretreatment severity was en-hanced, i.e., by elevating the temperature. Dry mat-ter loss can be referred to as the volatile fraction,since this represents the dry mass difference be-tween the initial material and the non-volatile frac-tion (slurry).
At least 82 % of the initial glucan was recov-ered in the slurry, mainly in the water-insolublefraction. Lost glucan was the portion hydrolysed toglucose, however, later it was further degraded toHMF. The latter compound can decompose tolevulinic acid and formic acid, or condense with thelignin.10 Recoveries of galactan and mannan fluc-tuated between 17–92 % and 17–98 % of original,respectively. These compounds were recoveredalmost exclusively in the liquid phase; however,at 200 °C a remarkable amount of mannan (54 %
334 Z. BARTA et al., Refining Hemp Hurds into Fermentable Sugars or Ethanol, Chem. Biochem. Eng. Q. 24 (3) 331–339 (2010)
T a b l e 1– Composition of raw hemp hurds and pretreated separated, washed solid fractions expressed as % of dry matter. Stan-
dard deviations were calculated from triplicates.
Raw hemp hurds2 L 10 L
220 °C 230 °C 200 °C 210 °C 220 °C
glucan 40.1 ± 1.1 62.8 ± 0.5 60.3 ± 1.8 57.4 ± 0.8 57.5 ± 1.1 58.6 ± 3.0
mannan 0.9 ± 0.1 0.0 ± 0.0 0.0 ± 0.0 0.7 ± 0.0 0.0 ± 0.0 0.0 ± 0.0
galactan 0.3 ± 0.0 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0
xylan 18.4 ± 0.3 2.6 ± 0.2 2.1 ± 0.1 6.9 ± 0.9 2.9 ± 0.0 3.1 ± 0.1
arabinan 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0
acid-insoluble lignin 21.7 ± 0.2 36.2 ± 0.3 38.8 ± 0.2 34.7 ± 1.1 37.9 ± 0.1 40.9 ± 1.4
water extractives 3.3 ± 0.2 – – – – – – – – – –
ethanol extractives 1.5 ± 0.4 – – – – – – – – – –
total ash 2.3 ± 0.0 1.2 ± 0.1 1.5 ± 0.0 1.4 ± 0.1 1.3 ± 0.0 1.3 ± 0.0
structural inorganics 1.2 ± 0.0 – – – – – – – – – –
acid-insoluble ash 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0
T a b l e 2– Solid contents of the pretreated slurry and acidity, and acid concentrations in liquid fractions. Standard deviations
were calculated from duplicates.
2 L 10 L
220 °C 230 °C 200 °C 210 °C 220 °C
total solids, w/% 5.9 ± 0.1 5.5 ± 0.1 21.2 ± 0.8 13.0 ± 0.5 13.3 ± 0.1
water-insoluble solids, w/% 4.4 ± 0.1 4.2 ± 0.1 15.3 ± 0.4 10.7 ± 0.3 10.3 ± 0.1
pH 3.6 3.6 3.7 3.7 3.6
acetic acid, �/g L–1 2.6 2.4 4.7 5.2 5.5
formic acid, �/g L–1 0.8 0.8 1.9 1.8 1.8
of original) remained in the fibre fraction. Lossesof galactan and mannan were similar to xylanat a given pretreatment condition, except at 200 °C,where they were significantly lower. Galactose andmannose can also be degraded to HMF.
Xylan recovery varied between 18–66 % oforiginal. The conditions investigated were more se-vere than the optimum for xylan, hence xylan re-covery decreased drastically by increasing the tem-perature. The water-soluble xylan made up a signif-icant share of the recovered part. These poor recov-eries might be due to the rather long residence timein addition to the moisture content of the raw hurdsfed to the reactor, which was low as well (around10 %). These conditions have been reported to haveunfavourable effects on xylan recoveries.11 Re-leased pentoses can degrade into furfural that canthen decompose to formic acid or condense with thelignin.10 Since no arabinan was detected in the rawhemp hurds, the furfural present must have ori-ginated from xylan. Concerning the stoichiometricreaction of xylose into furfural and the estimatedtotal amount of furfural, see Table 3, the % of xylanthat were degraded to furfural were calculatedas follows: 26 and 25 % for the 2 L reactor at220 and 230 °C, respectively, and 6, 14, and14 % for the 10 L reactor at 200, 210, and 220 °C,respectively. By subtracting these values fromthe xylan losses, it can be concluded that a signifi-
cant amount of the original xylan was lost to theformation of formic acid or pseudolignin detailedbelow.
The aim of using two different reactors was toinvestigate whether the same temperature and resi-dence time resulted in the same severity. It wasobserved that, at the same temperature (220 °C),larger losses were obtained in case of the 10 L reac-tor. Dry matter, glucan, or xylan losses obtained inthe large reactor at 210 and 220 °C were similar tothose obtained in the small one at 220 and 230 °C,respectively. These indicate that a given treatmenttemperature resulted in more severe pretreatmentin the 10 L reactor. This might be due to differencesin the heating and cooling profiles of the two reac-tors. The 2 L reactor takes ~1 min to reach the pre-treatment temperature, whereas the larger one re-quires ~3 min. Opening the slurry receiver alsotakes more time with the larger one, due to differingequipment designs. Since the pH values werenearly the same in both reactors at each tempera-ture, the severity was therefore not much affectedby the acidity.
Acid-soluble lignin recoveries were measuredfrom the solid fraction as the residue after atwo-step acid hydrolysis. All the lignin recoverieswere around 100 %, i.e., the extent of ligninsolubilisation was low; additionally, in the 10 L re-actor, the values were higher than 100 %. This ap-
Z. BARTA et al., Refining Hemp Hurds into Fermentable Sugars or Ethanol, Chem. Biochem. Eng. Q. 24 (3) 331–339 (2010) 335
F i g . 1 – (a) Dry matter, (b) acid-insoluble lignin, (c) glucan and (d) xylan recoveries after pretreatment as % of original
parent contradiction can be resolved by taking intoaccount the fact that during pretreatment thesolubilised lignin partially undergoes repolymerisa-tion, and reactions between lignin and sugars orsugar degradation products also occur, formingpseudolignin, which contributes to the amount ofacid-insoluble lignin.32
Enzymatic hydrolysis
Fig. 2 shows the conversions of glucan andxylan in enzymatic hydrolysis. Glucan conversionafter pretreatment in the 10 L reactor at 210 °C was83 % of theoretical, which was significantly higherthan that in case of the 200 °C (p � 10–5) and220 °C (p � 0.0038) pretreatments. In the 2 L reac-tor, there was no significant difference between thetwo temperatures (p � 0.3876).
The xylan content of the pretreated solid frac-tions was low (see Table 1), therefore the accuracyof xylan conversion was low as well. Consideringthe two reactor sizes separately, it can be seen thatxylan conversion decreased monotonically with theelevation of temperature. In other studies, the sametrend has been observed, i.e., xylan conversionsdecreased when pretreatment severity was en-hanced.11,33 Xyloglucan, one of the major hemi-cellulosic polysaccharides, can be present in threemacromolecular domains in lignocellulosics.34
These domains differ in resistance and accessibility.More severe pretreatment solubilises the less recal-citrant xyloglucan to a greater extent; hence theproportion of resistant, less-accessible regions in-creases in the pretreated solid fraction, which mightinterpret the declining xylan conversions in EH.Pretreating hemp hurds in the small reactor at220 °C resulted in significantly higher glucanand xylan conversion than in the large one at thesame temperature (p � 0.0343 and 0.0008, respec-tively).
Simultaneous saccharificationand fermentation
Fig. 3 shows the SSF yields based on glu-can and mannan contents of the substrate. The high-est ethanol yield from washed fibres, 70 % of theo-retical, was obtained from the pretreatment at210 °C. This was significantly higher than thatat 200 °C (p � 0.0003); nevertheless, the differen-ce was not significant compared to that at220 °C (p � 0.4925). No significant differenceswere obtained either in the 2 L reactor betweenthe two investigated temperatures (p � 0.2081)or between the two reactor sizes at 220 °C(p � 0.2360).
Overall product yields
Table 3 shows the product masses obtainedfrom 100 g of dry hemp hurds. After pretreatment,solubilised sugars were present mainly in theiroligomeric forms. The results of the Aspen Plusflash calculations for furan derivatives and acidsshowed that, for a given volatile compound, elevat-ing the pretreatment temperature resulted in a de-creased mass recovery (amount in slurry/estimatedtotal amount produced). At a given condition massrecoveries increased in the following order: furfural< formic acid < acetic acid < HMF, the last remain-ing entirely in the liquid phase. Generally, in the2 L reactor, more furfural and HMF were obtainedin the slurry than in the 10 L reactor, hence the totalamounts of these compounds were also higher inthe small reactor. At 220 °C, less sugar was lost inthe 2 L reactor compared to the 10 L one; however,more furfural and HMF were obtained. Conse-quently, more furan derivatives were further de-graded or formed pseudolignin at 220 °C in the10 L reactor.
Overall sugar yields were calculated from Ta-ble 3 values by the following equation:
336 Z. BARTA et al., Refining Hemp Hurds into Fermentable Sugars or Ethanol, Chem. Biochem. Eng. Q. 24 (3) 331–339 (2010)
F i g . 2 – Conversions of glucan and xylan in enzymatichydrolysis (EH) as % of theoretical
F i g . 3 – Ethanol yields in simultaneous saccharificationand fermentation (SSF) expressed as % of theoretical, based onglucan and mannan contents of the substrate
yield sugar �
��mass in liquidfraction mass obtained inEH
ma
sugar sugar
ss in raw materialsugar
where sugar denotes either glucose or xylose. Theoligomeric sugars are also included in the liquidfraction term, expressed as their monomer equiva-lents. Overall glucose yields varied between 63–75 %of theoretical and they showed the same patternas the EH glucan conversions, since the amountsof monomeric and oligomeric glucose in the liquidfraction after pretreatment were rather small. Over-all xylose yields ranged between 13 and 63 %of theoretical and decreased monotonically withhigher temperatures in both reactors. Beside thexylose released in EH, however, xylose solubil-ised during pretreatment made up a significantshare.
In terms of overall glucose yields, the best pre-treatment condition was that of 210 °C, where336 g glucose was obtained from 1 kg of dry rawmaterial, while with regard to total sugar yield, i.e.,glucose � xylose overall yields, it was shifted to200 °C. At this temperature, 414 g sugar was recov-ered from 1 kg of dry raw material. In regard to thehighest overall ethanol yield, with the pretreatmentat 210 °C, 141 g ethanol could be produced from1 kg of dry raw material.
Ordinary baker’s yeast, the robust full-scaleethanol fermenting microorganism, is unable toconvert pentoses to ethanol, hence in a process ap-plying this organism, these sugars can formco-products such as solid fuel or are burnt on-site.35
Furthermore, they can also be utilized as carbonsource of on-site cellulase enzyme fermentation.36
A lot of effort has been made to developpentose-fermenting yeast strains, however the
Z. BARTA et al., Refining Hemp Hurds into Fermentable Sugars or Ethanol, Chem. Biochem. Eng. Q. 24 (3) 331–339 (2010) 337
T a b l e 3– Product masses expressed in g obtained from 100 g hemp hurds dry matter. Oligo-sugars are given as monomer equivalents.
hemp hurds dry matter 100.0
glucose potential 44.6
xylose potential 20.9
2 L 10 L
220 °C 230 °C 200 °C 210 °C 220 °C
Steam pretreatment
oligo-glucose 0.9 0.9 0.9 0.7 0.4
glucose 0.2 0.1 0.0 0.1 0.1
oligo-xylose 3.3 2.8 7.4 3.3 1.5
xylose 0.7 0.3 0.5 0.7 0.2
furfural in slurry 0.9 0.8 0.2 0.5 0.5
furfural in flash vapoura 2.6 2.6 0.6 1.3 1.4
HMFb 0.2 0.2 0.0 0.1 0.1
HMFb in flash vapoura t.a.c t.a.c t.a.c t.a.c t.a.c
acetic acid 3.3 3.0 1.8 2.7 2.6
acetic acid in flash vapoura 0.7 0.8 0.3 0.5 0.6
formic acid 1.0 1.0 0.7 0.9 0.9
formic acid in flash vapoura 0.4 0.4 0.2 0.3 0.3
Enzymatic hydrolysis
glucose 32.3 31.7 27.3 32.8 27.5
xylose 1.1 0.8 5.2 1.7 0.9
SSFd
ethanol 13.3 12.7 8.5 14.1 12.5
aestimated with Aspen Plusbhydroxymethylfurfuralctrace amountsdsimultaneous saccharification and fermentation
co-fermentation of hexoses and pentoses has notbeen solved on an industrial scale yet.37
If a pentose-fermenting yeast were used, assum-ing the same ethanol yield from xylose as from glu-cose, an addition of 0.8 g ethanol (per 1 kg of dry rawmaterial) could be produced from the xylan of thesolid fraction obtained at 210 °C. Raw hemp hurdscontain a remarkable amount of xylan; however, itsrecovery through steam pretreatment was poor at thistemperature. If the ultimate aim is sugar or ethanolproduction, xylan recovery must be improved. Atwo-step steam pretreatment is a possible alternativefor preventing the less-recalcitrant hemicellulose frac-tion from further degradation, thus improving its re-covery.30 Nevertheless, the additional pretreatmentstep increases both the capital investment and opera-tional costs. Sassner et al.11 reported that impregnat-ing the raw material with acid before steam pretreat-ment resulted in closer-to-optimum conditions forglucan and xylan recoveries. As demonstrated in thisstudy, the optimum temperatures for glucan and xylanrecoveries differed greatly, therefore this alternativeshould be considered as well.
Conclusions
Hemp hurds, being an agro-industrial by-prod-uct with high carbohydrate content, are a potentialcandidate for second-generation fuel-ethanol produc-tion. Steam pretreatment of non-impregnated hemphurds was investigated for sugar and ethanol produc-tion by enzymatic hydrolysis and SSF, respectively.Dry matter and sugar recoveries throughout thepretreatments decreased monotonically with increas-ing temperature. Regarding enzymatic digestibilityof the fibre fraction and ethanol production, no sig-nificant difference was observed between the two in-vestigated temperatures in the 2 L reactor. In the 10L reactor, the most favourable condition was at 210°C in terms of overall glucose and ethanol yield;however, the maximum sugar yield was obtained at200 °C. These yields were in the same size order asthe highest yields reported in other studies,11,27,30,33
where a similar experimental procedure (steam pre-treatment, EH and SSF) was applied on variouslignocellulosic raw materials. To improve xylan re-covery, further investigation of impregnation andtwo-step steam pretreatment is required.
ACKNOWLEDGEMENTS
The Hungarian-Spanish IntergovernmentalS&T cooperation program and the Hungarian Na-tional Research Fund (OTKA – K 72710) are grate-fully acknowledged for their financial support. En-zymes were kindly donated by Novozymes A/S(Bagsværd, Denmark).
L i s t o f s y m b o l s
c � concentration, mol dm–3
p � corresponding level of significance
Q � volume flow rate, mL min–1
w � mass fraction, %
X � conversion, %
Y � yield, %, g kg–1
� � mass concentration, g L–1
� � volume ratio
A b b r e v i a t i o n s
ASE � accelerated solvent extraction
DM � dry matter
EH � enzymatic hydrolysis
FPU � filter paper unit
HMF� hydroxymethylfurfural
HP � Hewlett Packard
HPLC � high-performance liquid chromatography
NREL � National Renewable Energy Laboratory
SSF � simultaneous saccharification and fermentation
WIS � water-insoluble solids
R e f e r e n c e s
1. Balat, M., Balat, H., Appl. Energ. 86 (11) (2009) 2273.
2. Saunders, J., Rosentrater, K., Bioresour. Technol. 100 (13)(2009) 3277.
3. Taherzadeh, M. J., Karimi, K., BioRes. 2 (3) (2007) 472.
4. Briens, C., Piskorz, J., Berruti, F., Int. J. Chem. React.Eng. 6 (2008) R2.
5. Kumar, S., Singh, S. P., Mishra, I. M., Adhikari, D. K.,Chem. Eng. Technol. 32 (4) (2009) 517.
6. Tomas-Pejo, E., Oliva, J. M., Ballesteros, M., J. Sci. Ind.Res. India 67 (11) (2008) 874.
7. KoonOng, L., Planter 80 (941) (2004) 517.
8. Hendriks, A. T. W. M., Zeeman, G., Bioresour. Technol.100 (1) (2009) 10.
9. Carvalheiro, F., Duarte, L. C., Girio, F. M., J. Sci. Ind.Res. India 67 (11) (2008) 849.
10. Fengel, D., Wegener, G., Wood: chemistry, ultrastructure,reactions. Walter De Gruyter, Berlin, New York, 1989.
11. Sassner, P., Galbe, M., Zacchi, G., Appl. Biochem. Bio-technol. 121 (2005) 1101.
12. Duff, S. J. B., Murray, W. D., Bioresour. Technol. 55 (1)(1996) 1.
13. Saddler, J. N., Ramos, L. P., Breuil, C., Biotech. Agric.Ser. 9 (1993) 73.
14. Sun, Y., Cheng, J. Y., Bioresour. Technol. 83 (1) (2002) 1.
15. Garcia, C., Jaldon, Dupeyre, D., Vignon, M. R., BiomassBioenerg. 14 (3) (1998) 251.
16. Dang, V., Nguyen, K. L., Bioresour. Technol. 97 (12)(2006) 1353.
17. Moxley, G., Zhu, Z., Zhang, Y.-H. P., J. Agric. Food Chem.56 (17) (2008) 7885.
18. Datwyler, S. L., Weiblen, G. D., J. Forensic Sci. 51 (2)(2006) 371.
19. Vignon, M., Garcia-Jaldon, C., Dupeyre, D., Int. J. Biol.Macromol. 17 (6) (1995) 395.
338 Z. BARTA et al., Refining Hemp Hurds into Fermentable Sugars or Ethanol, Chem. Biochem. Eng. Q. 24 (3) 331–339 (2010)
20. Ballesteros, M., Oliva, J. M., Negro, M. J., Manzanares,P., Ballesteros, I., Process Biochem. 39 (12) (2004) 1843.
21. Hames, B., Ruiz, R., Scarlata, C., Sluiter, A., Sluiter,J., Templeton, D., Laboratory Analytical ProcedureNREL/TP-510-42620, National Renewable Energy Labo-ratory, Golden, CO, 2008.
22. Sluiter, A., Ruiz, R., Scarlata, C., Sluiter, J., Templeton, D.,Laboratory Analytical Procedure NREL/TP-510-42619,National Renewable Energy Laboratory, Golden, CO,2008.
23. Sluiter, A., Hames, B., Ruiz, R., Scarlata, C., Sluiter, J.,Templeton, D., Crocker, D., Laboratory Analytical Proce-dure NREL/TP-510-42618, National Renewable EnergyLaboratory, Golden, CO, 2008.
24. Sluiter, A., Hames, B., Ruiz, R., Scarlata, C., Sluiter,J., Templeton, D., Laboratory Analytical ProcedureNREL/TP-510-42622, National Renewable Energy Labo-ratory, Golden, CO, 2008.
25. Sluiter, A., Hames, B., Ruiz, R., Scarlata, C., Sluiter,J., Templeton, D., Laboratory Analytical ProcedureNREL/TP-510-42623, National Renewable Energy Labo-ratory, Golden, CO, 2008.
26. Saddler, J. N., Bioconversion of forest and agriculturalplant residues. C A B International, Wallingford, 1993.
27. Linde, M., Jakobsson, E. L., Galbe, M., Zacchi, G., Bio-mass Bioenerg. 32 (4) (2008) 326.
28. Kalman, G., Varga, E., Reczey, K., Chem. Biochem. Eng.Quart. 16 (4) (2002) 151.
29. Ohgren, K., Bura, R., Saddler, J., Zacchi, G., Bioresour.Technol. 98 (13) (2007) 2503.
30. Soderstrom, J., Pilcher, L., Galbe, M., Zacchi, G., BiomassBioenerg. 24 (6) (2003) 475.
31. Esteghlalian, A., Hashimoto, A. G., Fenske, J. J., Penner,M. H., Bioresour. Technol. 59 (2–3) (1997) 129.
32. Heitz, M., Capek-Ménard, E., Koeberle, P. G., Gagné, J.,Chornet, E., Overend, R. P., Taylor, J. D., Yu, E., Bio-resour. Technol. 35 (1) (1991) 23.
33. Ohgren, K., Galbe, M., Zacchi, G., Appl. Biochem. Bio-technol. 124 (2005) 1055.
34. Benko, Z., Siika-aho, M., Viikari, L., Reczey, K., EnzymeMicrob. Technol. 43 (2) (2008) 109.
35. Sassner, P., Zacchi, G., Biotechnol. Biofuels 1 (4) (2008)15 April.
36. Palmqvist, E., Hahn-Hagerdal, B., Szengyel, Z., Zacchi, G.,Reczey, K., Enzyme Microb. Technol. 20 (4) (1997) 286.
37. Hahn-Hagerdal, B., Karhumaa, K., Fonseca, C., Spen-cer-Martins, I., Gorwa-Grauslund, M. F., Appl. Microbiol.Biotechnol. 74 (5) (2007) 937.
Z. BARTA et al., Refining Hemp Hurds into Fermentable Sugars or Ethanol, Chem. Biochem. Eng. Q. 24 (3) 331–339 (2010) 339
II
Erdei et al. Biotechnology for Biofuels 2010, 3:16http://www.biotechnologyforbiofuels.com/content/3/1/16
Open AccessR E S E A R C H
© 2010 Erdei et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative CommonsAttribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction inany medium, provided the original work is properly cited.
ResearchEthanol production from mixtures of wheat straw and wheat mealBorbála Erdei*1, Zsolt Barta2, Bálint Sipos2, Kati Réczey2, Mats Galbe1 and Guido Zacchi1
AbstractBackground: Bioethanol can be produced from sugar-rich, starch-rich (first generation; 1G) or lignocellulosic (second generation; 2G) raw materials. Integration of 2G ethanol with 1G could facilitate the introduction of the 2G technology. The capital cost per ton of fuel produced would be diminished and better utilization of the biomass can be achieved. It would, furthermore, decrease the energy demand of 2G ethanol production and also provide both 1G and 2G plants with heat and electricity. In the current study, steam-pretreated wheat straw (SPWS) was mixed with presaccharified wheat meal (PWM) and converted to ethanol in simultaneous saccharification and fermentation (SSF).
Results: Both the ethanol concentration and the ethanol yield increased with increasing amounts of PWM in mixtures with SPWS. The maximum ethanol yield (99% of the theoretical yield, based on the available C6 sugars) was obtained with a mixture of SPWS containing 2.5% water-insoluble solids (WIS) and PWM containing 2.5% WIS, resulting in an ethanol concentration of 56.5 g/L. This yield was higher than those obtained with SSF of either SPWS (68%) or PWM alone (91%).
Conclusions: Mixing wheat straw with wheat meal would be beneficial for both 1G and 2G ethanol production. However, increasing the proportion of WIS as wheat straw and the possibility of consuming the xylose fraction with a pentose-fermenting yeast should be further investigated.
BackgroundThe use of bioethanol can reduce our dependence on fos-sil fuels, while at the same time decreasing net emissionsof carbon dioxide, the main greenhouse gas [1,2]. How-ever, large-scale production of bioethanol is beingincreasingly criticized for its use of food sources as rawmaterial. Brazil's bioethanol production consumes largequantities of sugar cane, while in the USA, corn is used[3]. Other starch-rich grains, such as wheat and barley,are mostly used in Europe [4]. The use of such sugar-richfeedstock causes the escalation of food prices, owing tocompetition on the market [5,6]. Therefore, future expan-sion of biofuel production must be increasingly based onbioethanol from lignocellulosic materials, such as agricul-tural byproducts, forest residues, industrial wastestreams or energy crops [7,8]. These feedstocks, whichare being used in second-generation (2G) bioethanol pro-duction, are abundant, and their cost is lower than that of
food crops [9]. In Europe, wheat straw has the greatestpotential of all agricultural residues because of its wideavailability and low cost [10].
To efficiently utilize lignocellulosic products, pretreat-ment is required to hydrolyse the hemicelluloses to makethe celluloses more accessible to the enzymes. One of themost suitable kinds of pretreatment for lignocellulosicmaterial is steam explosion [11]. Combining steam explo-sion with acid catalysts is considered one of the mostpromising techniques for the commercialization of theprocess [12]. Several studies have shown that impregna-tion of wheat straw with small amounts of H2SO4 beforesteam pretreatment results in improved sugar yields[13,14].
During pretreatment, several sugar degradation prod-ucts such as 5-hydroxymethyl-furfural (HMF) andfurfural (degradation products of hexoses and pentoses,respectively), weak organic acids and phenolic com-pounds from lignin degradation are released into thehydrolysate, and have been shown to inhibit both yeast[15,16] and enzymes [17]; however, these compoundsaffect cell growth more than ethanol formation. It has
* Correspondence: [email protected] Lund University, Department of Chemical Engineering, P.O. Box 124, SE-221 00 Lund, SwedenFull list of author information is available at the end of the article
Erdei et al. Biotechnology for Biofuels 2010, 3:16http://www.biotechnologyforbiofuels.com/content/3/1/16
Page 2 of 9
also been shown by Larsson et al. that the ethanol yield inthe presence of several inhibitors decreased only slightlycompared with the reference fermentation [18]. Further-more, the addition of weak acids has an intense inhibitoryeffect on growth of Saccharomyces cerevisiae, but leads toincreased ethanol yield at low concentrations [19,20].Therefore, we hypothesized that mixing starch hydro-lysate with the lignocellulosic stream would dilute theinhibitor concentration in the cellulose hydrolysate andprobably improve the fermentation, and at the same time,the presence of inhibitors might also improve the ethanolyield from the starch fraction.
To obtain efficient ethanol fermentation with S. cerevi-siae, numerous nutrients, including trace metals and vita-mins, are required during the process. Chemicalscontribute significantly to the cost of large-scale produc-tion [21]; although it was not in the scope of this study toinvestigate this, their use should thus be minimized.Wheat hydrolysate, which is relatively cheap comparedwith chemicals, has been proven to be a potential supple-ment for lignocellulosic hydrolysate, because it is not onlya sugar-containing material, but is also a complex nutri-ent source [22,23].
The production cost of ethanol is not only dependenton the yield but also on the concentration of ethanol inthe fermentation broth, because of the high energydemand in the distillation step. In this step, the ethanolconcentration in the broth after fermentation is increasedto 94% using two stripper columns and a rectification col-umn, which are heat-integrated by operating at differentpressures. A significant increase in energy demand isobserved at an ethanol concentration below 4% [24]. Ahigher ethanol concentration can be achieved in thebroth by adding starch-rich material to the lignocellulosicprocess, leading to a lower energy demand in distillation,thus reducing the production cost.
The aim of this study was to evaluate the simultaneoussaccharification and fermentation (SSF) of mixtures ofcellulosic material (steam pretreated wheat straw; SPWSand presaccharified wheat meal (PWM). The effect onethanol concentration and ethanol yield of varying theproportions of starch and cellulose fraction in SSF wasinvestigated and compared with the pure starch and purecellulose alternatives.
MethodsThe experimental procedure is illustrated in Figure 1. Thewheat meal was subjected to liquefaction and supple-mented with α-amylase, followed by 2 hours of presac-charification with amyloglucosidase. The entire wort(PWM) was used in SSF. The wheat straw was impreg-nated with a weak sulfuric acid solution and thensqueezed using a hydraulic press. The pressed materialwas pretreated in a steam pretreatment unit and the
whole slurry was then mixed with the PWM. The twomaterials were mixed in different proportions to investi-gate the effects on the ethanol yield and the ethanol con-centration.
Raw materialsWheat straw was kindly provided by Lunds Civila Ryttar-förening (Lund, Sweden). It was chopped in a hammermill, sieved to obtain pieces of 2-10 mm, and then storedat room temperature before pretreatment. Wheat meal(dry-milled grain) with an average particle size of 2.5 mmwas kindly provided by Sileco (Halland, Sweden) andstored at 5°C before use.
Enzymesα-Amylase (Termamyl® SC; Novozymes A/S, Bagsværd,Denmark) and amyloglucosidase (Spirizyme® Fuel;Novozymes) amylolytic enzymes were used for starch liq-uefaction and saccharification, respectively. Theamylolytic activity of these enzymes were not measured,because they were loaded based on their weight, as it isrecommended by the manufacturer [25]. In the SSFexperiments, cellulase (Celluclast 1.5 L) and β-glucosi-dase (Novozym 188) enzyme preparations (bothNovozymes) were used. Celluclast 1.5 L had an activity of65 filter paper units (FPU)/g, measured using the IUPACprotocol [26], and 33 IU/g β-glucosidase activity accord-
Figure 1 Experimental procedure used to assess the effects of mixtures of wheat straw and wheat meal on simultaneous sac-charification and fermentation (SSF).
Wheat StrawWheat Meal
Liquefaction
Termamyl SCT=85 °C pH=5.5 t=3 h
Presaccharification
Spirizyme FuelT=60 °C pH=4.2 t=2 h
Impregnation
0.2% H2SO420 g liquid/g dry straw
t=1 h T=20 °C
SteamPretreatment
T=190 °Ct=10 min
SteamPretreatment
T=190 °Ct=10 min
Simultaneous Saccharificationand Fermentation
S. Cerevisiae 5 g/LSubstrate: 5% WIS
T=36.5 °C pH=4.8 t=72, 96 hCelluclast 1.5L+ Novozym 188
Simultaneous Saccharificationand Fermentation
S. Cerevisiae 5 g/LSubstrate: 5% WIS
T=36.5 °C pH=4.8 t=72, 96 hCelluclast 1.5L+ Novozym 188
SlurrySlurry
Press
P=270 bar DM=43%
Press
P=270 bar DM=43%
Excessliquid
Erdei et al. Biotechnology for Biofuels 2010, 3:16http://www.biotechnologyforbiofuels.com/content/3/1/16
Page 3 of 9
ing to the method of Berghem and Petterson [27].Novozym 188 had a β-glucosidase activity of 350 IU/g.
Compositional analysisThe carbohydrate and lignin contents of the raw wheatstraw, the starch-free fibre, and the solid fraction of thepretreated wheat straw were determined according to thestandard National Renewable Energy Laboratory (NREL)method [7,28]. Finely ground samples were treated with72% H2SO4 for 1 h at 30°C, then diluted to 4% H2SO4 andautoclaved for 1 hour at 121°C. Sugar contents were anal-ysed with high performance liquid chromatography(HPLC) (LC-10AD; Shimadzu, Kyoto, Japan), acid-insol-uble lignin was measured by weighing after overnightdrying at 105°C, and acid-soluble lignin was determinedby spectrophotometry using a wavelength of 240 nm.Each sample was analysed in duplicate.
The liquid fraction of the SPWS and the supernatantafter fermentation were analysed for total sugar contentaccording to an NREL procedure [29]. In this method, thesample is treated with 4% H2SO4 at 121°C for 1 h, andthen analysed by HPLC.
The fraction of acid-insoluble ash was determined afterthe two-step acid hydrolysis described above, and againon the ash of the residue. Both samples were heated at550°C until the sample weight remained constant. Totalash refers to the inorganic part of raw material or solidfraction after pretreatment.
To determine the starch content, the wheat meal wassubjected to a two-step enzymatic hydrolysis consistingof liquefaction and saccharification. All batches werehydrolysed using a 7 L evaporator (Rotavapor® R-153;Büchi Labortechnik AG, Flawil, Switzerland). The drymatter content was set to 35%. In the first step, wheatmeal slurry supplemented with 0.5 g/kg dry matter (DM)and Termamyl® SC was liquefied at 85°C, pH 5.5, for 3 h.In the second step, Spirizyme® Fuel was added at a ratio of0.5 mL/kg DM at pH 4.2, and the slurry was treated at60°C for 24 h to ensure total starch hydrolysis. The wortwas filtered and the glucose content of the supernatantwas measured using HPLC. The washed solid residue isreferred to as the starch-free residue (SFR).
PWMWheat meal was presaccharified as described above,except that the duration of saccharification was 2 hinstead of 24 h. PWM was then used in SSF.
SPWSThe wheat straw was immersed in an aqueous solution of0.2% H2SO4 at a liquid:dry straw weight ratio of 20. It wasstored in sealed buckets for 1 h, and was then squeezed ina manual 3 L press (Fisher Maschinenfabrik Gmbh, Burg-kunstadt, Germany) to an average dry matter content of
43%. Steam pretreatment was performed in a unit(described previously [30]) comprising a 10 L pressurizedvessel, with a flash cyclone in which the pretreated mate-rial was released and collected. Previously optimized con-ditions for wheat straw [14] were used; that is, thetemperature was maintained at 190°C for 10 min usingsaturated steam. Each batch that was fed into the reactorwas 600 g wet weight. The steam-pretreated wheat straw(SPWS) was then subjected to SSF.
SSFSSF experiments were performed in 2 L laboratory fer-mentors (Infors AG, Bottmingen, Switzerland) with afinal working weight of 1.4 kg. PWM, SPWS slurry andvarious mixtures of these (denoted as mixtures A, B, Cand D) were used as substrates in SSF with a total water-insoluble solids (WIS) content of 5%. The PWM:SPWSWIS ratios for the mixtures investigated were 0.8:4.2,1.5:3.5, 2.0:3.0 and 2.5:2.5, respectively. When SSF wasperformed on pure PWM, the WIS content was set to2.8% to restrict the ethanol concentration to 60 g/L.Details of the substrates can be found in Table 1. SSFexperiments were performed using Celluclast 1.5 L andNovozym 188, at dosages of 15 FPU/g glucan and 17 β-glucosidase IU/g glucan, respectively. SSF of pure PWMwas not supplemented with Celluclast 1.5 L or Novozym188. S. cerevisiae (ordinary baker's yeast; Jästbolaget AB,Stockholm, Sweden) was suspended in sterilized waterand added to the fermentor at a concentration of 5 g DM/L. As nutrients, (NH4)2HPO4, MgSO4·7 H2O and yeastextract were used at concentrations of 0.5, 0.025 and 1.0g/L, respectively. The fermentor was loaded with theSPWS and the nutrients, which were sterilized separatelyat 121°C for 20 minutes, but the PWM was not sterilizedbecause the starch and the enzymes already added wouldhave been damaged at these conditions. SSF was per-formed at 36.5°C, and the pH was maintained at 5 ± 0.2by addition of 10% NaOH solution. The experimentswere run for 72 h in the case of pure PWM, and for 96 hin the case of SPWS or mixtures of the substrates. Allsamples withdrawn were filtered through 0.2 μm filtersbefore being analysed by HPLC.
Analysis of sugars, ethanol and byproductsThe content of reducing sugars was measured colorimet-rically using dinitrosalicylic acid, according to Miller'smethod [31]. The liquid fractions from pretreatment,samples from acid hydrolysis and the supernatants of SSFbroth were analysed by HPLC, in a chromatographequipped with a refractive index detector. Cellobiose, glu-cose, mannose, xylose, galactose and arabinose were sep-arated on an ion-exchange column (Aminex HPX-87P;Bio-Rad Laboratories, Hercules, CA, USA) at 85°C. Ultra-pure water was used as eluent at a flow rate of 0.6 mL/
Erdei et al. Biotechnology for Biofuels 2010, 3:16http://www.biotechnologyforbiofuels.com/content/3/1/16
Page 4 of 9
min. Lactic acid, glycerol, acetic acid, ethanol, HMF andfurfural were separated (Aminex HPX-87H column; Bio-Rad Laboratories) at 65°C. The eluent was 0.005 MH2SO4 at a flow rate of 0.5 mL/min.
Results and DiscussionThe ethanol yields were calculated as a percentage of themaximal theoretical yield for glucose (0.51 g/g) that couldhave been produced if all the glucose present in the slurryand the PWM, including both monomers and oligomersin the liquid and glucan fibres in the WIS, had been con-verted to ethanol. The theoretical amount of glucosereleased during the hydrolysis was calculated by multiply-ing the amount of glucan by 1.11.
Material composition and pretreatmentThe composition of the raw materials is shown in Table 2.The DM content of wheat meal consists of 72.7% starchand 24.3% SFR, showing that this part of the crop couldalso be an important source of lignocellulose. Severalstudies have investigated ways in which the ethanol yieldcould be enhanced by utilizing this part of the crop[32,33]. However, pretreatment is required to facilitatethe enzymatic hydrolysis of SFR fibres, which was notperformed in the present study. The compounds deter-mined in SFR constituted about 63% of the DM.
The total solids recovery by the steam pretreatmentwas 79%. The slurry obtained after pretreatment had atotal solids content of 11.1% and a WIS content of 7.6%.Almost 100% of the glucose was recovered, based on thecontent in the raw material, of which about 95% wasrecovered in the solid fraction (Figure 2). The recovery ofxylose, which is the main sugar in hemicellulose, was67%, the major fraction of which was obtained in the liq-uid phase (prehydrolysate).
The proportions of glucan and lignin were increased bypretreatment, from 38.8 and 16.1% (Table 2) of the DM inthe raw material to 67.6 and 23.1% (Table 3; sugars arepresented as monomers) of the DM in the SPWS, respec-tively, indicating a high degree of hemicellulose hydroly-sis. This was confirmed by the xylose concentration of24.9 g/L in the prehydrolysate, corresponding to 65% ofthe theoretical yield available in the raw material (Figure2). Other sugars were also present, mostly in monomerform. Some inhibitors, such as acetic acid, HMF andfurfural were also detected after pretreatment (Table 3).The concentrations of acetic acid, HMF and furfural cor-responded to yields of 1.2, 0.3 and 1.0 g/100 g dry straw,respectively, which is somewhat higher than the valuesreported by Linde et al. for wheat straw pretreated underthe same conditions [14]. The concentration of furfuralwas higher than that of HMF, because the hemicelluloseconsisted mostly of pentoses, which is typical of herba-ceous crops [34], and the cellulose fraction was barelyhydrolysed in steam pretreatment.
Enzymatic hydrolysis of wheat mealIt has been shown previously that the amyloglucosidasedosage can be reduced by 5-10% when saccharification iscarried out before fermentation [25]. However, a highglucose concentration at the beginning of SSF with ligno-cellulosics should be avoided to prevent end-productinhibition of the enzymes and osmotic stress to the yeastcells. β-glucosidase activity is reduced by 80% in the pres-ence of only 10 g/L glucose when p-nitrophenyl-β-D-gli-copyronoside is used as substrate, and less significantlywith cellobiose [35]. In that study also, a high degree ofinhibition of cellulase activity was observed at a glucoseconcentration range of 0 to 100 g/L. Osmotic stressaffects the yeast cell when the glucose in the solution is >150 g/L [25,36]. Therefore, instead of completely saccha-
Table 1: Details of the substrates used in the SSF experiments
Substrate WIS Glucose equivalents
Total PWM SPWS PWM SPWS
WIS Liquid WIS Liquid
SPWS 5 -- 5.0 -- -- 3.7 0.2
Mixture A 5 0.8 4.2 0.1 2.5 3.2 0.2
Mixture B 5 1.5 3.5 0.3 5.0 2.6 0.1
Mixture C 5 2.0 3.0 0.4 6.6 2.2 0.1
Mixture D 5 2.5 2.5 0.5 8.3 1.9 0.1
PWM 2.8 2.8 -- 0.5 9.4 -- --
PWM = presaccharified wheat meal; SPWS = team-pretreated wheat straw; WIS = water-insoluble solids.Data are presented as percentage of working weight.
Erdei et al. Biotechnology for Biofuels 2010, 3:16http://www.biotechnologyforbiofuels.com/content/3/1/16
Page 5 of 9
rifying the wheat meal, we chose to perform partial sac-charification (presaccharification). Optimum presacch-arification in a starch-based material is about 50-70 dex-trose equivalents (DE), which is an indication of the totalamount of reducing sugars, expressed as D-glucose, pres-ent in the solution [25,37].
In the present study, we achieved 68 DE during starchhydrolysis after a total reaction time of 5 h (liquefactionand subsequent saccharification). The glucose monomer
fraction after presaccharification was 53% of the amountof glucose measured at the end of the reaction (Figure 3).The glucose concentration in the PWM after presacchari-fication was 164 g/L. The PWM was diluted before SSF(Table 1).
Effect of PWM on the ethanol concentrationSignificant differences, in terms of the initial rate of etha-nol formation, were observed between SSF on pureSPWS and SSF on mixtures containing different propor-tions of PWM. During the first 2 hours of SSF, the ethanolproductivity was 1.6 g/L/hour in the case of pure SPWS,whereas it was ≥4.7 g/L/hour for PWM alone or PWMmixed with SPWS. This could be due to the high water-soluble sugar content of PWM (Table 1) present at thebeginning of fermentation, mainly as glucose, which wasconsumed rapidly (data not shown), resulting in anincreased rate of ethanol formation. In pure SPWS, themajor part of the glucose is in polymeric form bound inthe solid phase, and this had to be hydrolysed before fer-mentation. However, glucose was measured in the solu-tion during the initial 8 hours, which means thathydrolysis is not the rate-limiting step in this reaction.However, furfural and HMF may cause a lag-phase in eth-anol fermentation [38], because ethanol production isinhibited by the degradation of these compounds tofurfuryl alcohol and HMF alcohol, respectively. The mostrapid ethanol formation (6.7 g/L/hour) was obtained withpure PWM.
Ethanol concentrations increased with increased glu-cose content in the SSF experiments (Figure 4). The high-est concentration obtained was 56.5 g/L (at 96 h) whenSSF was carried out on mixture D. Increasing the ethanolconcentration is beneficial to the energy demand of distil-lation [24].
Effect of PWM on the yieldThe ethanol yield is usually reported as g EtOH/g DM ofthe raw material. However, this means of expressing theyield was not appropriate for this study because mixturesof materials were used. Therefore, the yields areexpressed as a percentage of the theoretical maximum,considering only the glucose available in the substrates, asgalactose and the pentoses are not usually fermented toethanol by S. cerevisiae [30]. These sugars were not con-sumed in any of the SSF experiments, which validates thisassumption (data not shown).
Figure 5 shows the ethanol yields for the SSF of variousmixtures of substrate. The yield from pure SPWS was68%, which is in agreement with the results obtained byLinde et al. [14]. The yield observed for pure PWM wasabout 91%, which is also typical for SSF of starch-basedmaterials [25]. In mixtures of the two substrates, the yieldincreased as the proportion of PWM was increased. The
Table 2: Composition of raw wheat straw and wheat meal, in % of DM, including breakdown of the starch-free residue.
Component Percentage of DM, mean ± SD
Wheat meal Raw wheat straw
Starch 72.7 --
SFRa 24.3 --
% of SFRa
Glucan 17.5 ± 0.1 38.8 ± 0.5
Mannan BDL 1.7 ± 0.2
Xylan 14.4 ± 0.0 22.2 ± 0.3
Galactan 1.6 ± 0.0 2.7 ± 0.1
Arabinan 8.5 ± 0.0 4.7 ± 0.1
ASLb 3.1 ± 0.0 2.4 ± 0.0
AILc 15.1 ± 3.0 16.1 ± 0.1
Total ash 2.3 ± 0.3 5.8 ± 0.1
AIAd BDL 2.4 ± 0.4
Where standard deviations are presented, these were calculated from duplicate measurements on the same sample.aStarch-free residue.bAcid-soluble lignin.cAcid-insoluble lignin.dAcid-insoluble ash.BDL = below detection limit; DM = dry matter.
Figure 2 Recovery of solids, sugars and water-insoluble lignin. Percent of the theoretical, after steam pretreatment of wheat straw at 190°C for 10 minutes with 0.2% H2SO4.
Erdei et al. Biotechnology for Biofuels 2010, 3:16http://www.biotechnologyforbiofuels.com/content/3/1/16
Page 6 of 9
highest ethanol yield, 99%, was obtained for SSF of mix-ture D. This yield is rather high and could be due to someerrors in the raw materials analysis. However, this wouldaffect all trials in the same way, thus the most importantresult is the difference in yield between the experimentalpoints, which shows a clear trend of increasing yield asthe ratio of the PWM increased, and most importantly,resulting in higher yield than that from SSF of purePWM. This is very favourable in terms of process econ-
omy and can be explained by the presence of inhibitors atlow concentrations (see below).
Small amounts of lactic acid were produced after 40 hof fermentation in all cases. Lactic acid can be formedfrom both hexoses and pentoses as a microbial metabolicproduct. The lactic acid yield in this process was 1 g/gconsumed sugar regardless of the type of monomeric
Figure 4 Variation in ethanol concentration with time during si-multaneous saccharification and fermentation (SSF) of steam-pretreated wheat straw (SPWS), presaccharified wheat meal (PWM) and mixtures of the two (A to D). PWM to SPWS ratios in the mixtures A to D were 0.8:4.2; 1.5:3.5; 2.0:3.0 and 2.5:2.5, respectively.
Table 3: Composition of WIS and liquid (prehydrolysate) fractions in steam-pretreated wheat straw slurry.
Steam-pretreated wheat straw
WIS Prehydrolysate
Components Percentage of DM Components Oligosaccharides Monosaccharides
Glucan 67.6 ± 0.5 Sugars, g/L
Mannan BDL Glucose 0.9 ± 0.1 2.3
Xylan 0.7 ± 0.0 Mannose 4.1 ± 0.1 1.1
Galactan BDL Xylose 2.9 ± 0.7 22.0
Arabinan 0.4 ± 0.0 Galactose BDL 1.4
ASLa 5.1 ± 0.2 Arabinose BDL 2.9
AILb 23.1 ± 0.2 Inhibitors, g/L
Total ash 1.0 ± 0.1 Acetic acid 1.7
AIAc 0.4 ± 0.2 HMF 0.4
Furfural 1.5
Total ash, % of prehydrated material
0.3 ± 0.2
All the sugars in the prehydrolysate are reported as monomer sugars. Where standard deviations are presented, it was calculated from duplicate measurements on the same sample.aAcid-soluble lignin.bAcid-insoluble lignin.cAcid-insoluble ash.BDL = below detection limit; DM = dry matter; WIS = water-insoluble solids.
Figure 3 Starch hydrolysis profiles. (1) Liquefaction, (2a) presaccha-rification and (2) saccharification.
Erdei et al. Biotechnology for Biofuels 2010, 3:16http://www.biotechnologyforbiofuels.com/content/3/1/16
Page 7 of 9
sugar. The amount of additional ethanol that could havebeen produced from hexoses if there had not been anylactic acid formation can be estimated as follows:
0.46 g/g is 90% of the maximum theoretical ethanolyield for hexose sugars. After applying this correction forlactic acid, the yield slightly exceeds the theoretical maxi-mum for mixture D (Figure 5).
Several fermentation inhibitors were present in theslurry during SSF when SPWS was also added to the mix-tures. The most significant inhibitors, because of theirconcentrations, were acetic acid and furfural. Acetic acidis one of the most important weak acids of the fermenta-tion inhibitors, as its pKa is close to the pH of fermenta-tion, and therefore a significant amount is in theundissociated form. The undissociated form can diffusethrough the plasma membrane, and may dissociate insidethe cell, depending on the pH. To avoid the drop in intra-cellular pH, the cell must pump out protons by the actionof the plasma membrane ATPase [39,40]. This means thatmore ATP has to be generated to maintain the intracellu-lar pH, which is achieved in anaerobic conditions by pro-ducing more ethanol, resulting in increased ethanol yield[19,38,41]. Increased ethanol yield has also been noted inthe presence of furfural at low concentrations [42]. How-ever, at higher concentrations of acetic acid (> 10 g/L)[18], the ethanol yield is decreased. Similarly, furfural atconcentrations of > 3 g/L reduces the ethanol productiv-ity to a great extent [43]. The ethanol yield as a functionof the acetic acid concentration in SSF is illustrated inFigure 6. The values given are the average of the concen-trations at 0 and 24 h, because acetic acid was releasedfrom the hemicellulose fraction at the beginning of SSF.The maximum yield was 99% (obtained at an acetic acidconcentration of 1.0 g/L), which is even higher than the
yield from PWM fermentation when no acetic acid waspresent. These results are in good accordance with a pre-vious study reported by Larsson et al.; however, thoseauthors expressed the ethanol yield as a function of thetotal acid concentration [18]. That study also showed thatthe presence of acetic acid (5 g/L), formic acid (10 g/L),levulinic acid (23 g/L), furfural (1.2 g/L) and HMF (1.3 g/L) only slightly decreased the ethanol yield comparedwith the reference fermentation. The combination of ace-tic acid and furfural has been shown to have a negativeeffect on growth [44]; however, at the low concentrationsproduced in the present study, they might have a positiveeffect on ethanol production yield.
The specific raw material demands and the straw tomeal ratios for the process scenarios are listed in Table 4.In the pure wheat straw-based process, 5.4 kg dry rawmaterial is needed to produce 1 L of ethanol, whereas thecorresponding amount of wheat meal is only 2.4 kg. Thisis simply because of the difference in the proportions ofcarbohydrates in the raw materials. The raw mate-rial:meal ratios are in the range of 0.5-2.6 in the mixtures.
The residue:crop ratio for wheat is typically about1.3:1.0 (w/w) [45]. Approximately 30-40% of the straw isleft on the field for soil protection, leaving the sameamount of residue for biomass utilization when the cropis harvested (that is, the straw:wheat ratio is 1.0:1.0). Inthe case of mixture D, which gave the highest yield, theproportion of wheat meal was double that of the wheatstraw. However, the yield was still rather high (87% of the-oretical yield) when straw and meal were used in the pro-portions at harvest, as in mixture B. By comparison, thetotal yield obtained if the fermentations are carried outseparately is 78%.
ConclusionsIn this study, we investigated the effects on ethanol yieldof mixing different proportions of PWM and SPWSbefore SSF. The highest yield was obtained when equalamounts of PWM and SPWS (based on WIS) were used.
mass 0.46 (mass massadditional ethanol lactic acid pentoses= ⋅ − consumed)
Figure 5 Ethanol yield (shaded grey areas) based on glucose only in simultaneous saccharification and fermentation (SSF) of pure substrates and their mixtures (A to D). Potential ethanol yield after correction for the lactic acid produced from hexoses (white areas).
Figure 6 Ethanol yield as a function of acetic acid concentration during simultaneous saccharification and fermentation (SSF) (av-erage of acetic acid concentrations at 0 and 24 h).
Erdei et al. Biotechnology for Biofuels 2010, 3:16http://www.biotechnologyforbiofuels.com/content/3/1/16
Page 8 of 9
Thus, a mixed substrate is favourable in terms of finalethanol yield, probably due to the stress on S. cerevisiaecaused by weak acids present in SPWS. At the same time,it is also easier to reach a high ethanol concentrationusing such as mixture than when using wheat straw onlyas a raw material.
Increasing the proportion of WIS of the lignocellulosicmaterial should be studied further in an attempt toimprove the ethanol production from mainly lignocellu-losics. Bearing in mind the significant proportion ofhemicelluloses in wheat straw, a pentose-fermentingyeast should also be considered as a potential alternative.Assuming 70% ethanol yield from pentoses, the final eth-anol concentration in the fermentation broth could befurther improved by 3-5 g/L ethanol. To decrease the costof chemicals, decreasing the amount of added nutrients isan option to consider and further investigate when wheathydrolysate is used as a supplement to SSF with lignocel-lulosic substrate.
Competing interestsThe authors declare that they have no competing interests.
Authors' contributionsBE, MG and GZ designed and coordinated the study. ZB and BS carried out theexperiments. BE, ZB and BS analysed the results. BE and ZB wrote the paper,and KR, MG and GZ reviewed the paper. All authors read and approved thefinal manuscript.
AcknowledgementsWe gratefully acknowledge the Swedish Energy Agency for its financial sup-port, and the European Community Vocational Training Programme (Leonardo da Vinci) for mobility support.
Author Details1Lund University, Department of Chemical Engineering, P.O. Box 124, SE-221 00 Lund, Sweden and 2Budapest University of Technology and Economics, Department of Applied Biotechnology and Food Science, 1111 Budapest, Szt. Gellért tér 4, Hungary
References1. Wyman CE: Ethanol production from lignocellulosic biomass: overview.
In Handbook on Bioethanol: Production and Utilization Edited by: Wyman CE. Washington: Taylor & Francis; 1996:1-18.
2. McMillan JD: Bioethanol production: status and prospects. Renewable Energy 1997, 2-3:295-302.
3. Wheals AE, Basso LC, Alves DMG, Amorim HV: Fuel ethanol after 25 years. Trends Biotechnol 1999, 12:482-487.
4. Galbe M, Zacchi G: A review of the production of ethanol from softwood. Appl Microbiol Biotechnol 2002, 6:618-628.
5. Catic I, Rujnic-Sokele M: Agriculture products - food for living being or for machinery. Polimeri 2008, 4:243-243.
6. Knocke C, Vogt J: Biofuels - challenges & chances: How biofuel development can benefit from advanced process technology. Eng Life Sci 2009, 2:96-99.
7. Claassen PAM, van Lier JB, Contreras AML, van Niel EWJ, Sijtsma L, Stams AJM, de Vries SS, Weusthuis RA: Utilisation of biomass for the supply of energy carriers. Appl Microbiol Biotechnol 1999, 6:741-755.
8. Solomon BD, Barnes JR, Halvorsen KE: Grain and cellulosic ethanol: History, economics, and energy policy. Biomass Bioenergy 2007, 6:416-425.
9. Tan KT, Lee KT, Mohamed AR: Role of energy policy in renewable energy accomplishment: The case of second-generation bioethanol. Energy Policy 2008, 9:3360-3365.
10. Kim S, Dale BE: Global potential bioethanol production from wasted crops and crop residues. Biomass Bioenergy 2004, 4:361-375.
11. McMillan JD: Pretreatment of lignocellulosic biomass. In Enzymatic Conversion of Biomass for Fuels Production. ACS Symposium Series 566 Edited by: Himmel ME, Baker JO, Overend RP. Washington: American Chemical Society; 1994:292-324.
12. Hu G, Heitmann JA, Rojas OJ: Feedstock pretreatment strategies for producing ethanol from wood, bark, and forest residues. Bioresources 2008, 1:270-294.
13. Ballesteros I, Negro MJ, Oliva JM, Cabanas A, Manzanares P, Ballesteros M: Ethanol production from steam-explosion pretreated wheat straw. Appl Biochem Biotechnol 2006, 12:496-508.
14. Linde M, Jakobsson EL, Galbe M, Zacchi G: Steam pretreatment of dilute H2SO4-impregnated wheat straw and SSF with low yeast and enzyme loadings for bioethanol production. Biomass Bioenergy 2008, 4:326-332.
15. Sanchez B, Bautista J: Effects of furfural and 5-hydroxymethylfurfural on the fermentation of Saccharomyces cerevisiae and biomass production from Candida guilliermondii. Enzyme Microb Technol 1988, 5:315-318.
16. Taherzadeh MJ, Gustafsson L, Niklasson C, Liden G: Physiological effects of 5-hydroxymethylfurfural on Saccharomyces cerevisiae. Appl Microbiol Biotechnol 2000, 6:701-708.
17. Tengborg C, Galbe M, Zacchi G: Reduced inhibition of enzymatic hydrolysis of steam-pretreated softwood. Enzyme Microb Technol 2001, 9-10:835-844.
18. Larsson S, Palmqvist E, Hahn-Hagerdal B, Tengborg C, Stenberg K, Zacchi G, Nilvebrant NO: The generation of fermentation inhibitors during dilute acid hydrolysis of softwood. Enzyme Microb Technol 1998, 3-4:151-159.
19. Pampulha ME, Loureiro-Dias MC: Combined effect of acetic-acid, ph and ethanol on intracellular ph of fermenting yeast. Appl Microbiol Biotechnol 1989, 5-6:547-550.
Received: 11 February 2010 Accepted: 2 July 2010 Published: 2 July 2010This article is available from: http://www.biotechnologyforbiofuels.com/content/3/1/16© 2010 Erdei et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.Biotechnology for Biofuels 2010, 3:16
Table 4: Specific raw material demand
Substrate Raw material required (kg total DM per L EtOH) Straw:meal
Wheat straw only 5.4 -
Mixture A 4.0 2.6
Mixture B 3.1 1.1
Mixture C 2.9 0.7
Mixture D 2.5 0.5
Wheat meal only 2.4 0
DM = dry matter.
Erdei et al. Biotechnology for Biofuels 2010, 3:16http://www.biotechnologyforbiofuels.com/content/3/1/16
Page 9 of 9
20. Taherzadeh MJ, Niklasson C, Liden G: Acetic acid - friend or foe in anaerobic batch conversion of glucose to ethanol by Saccharomyces cerevisiae? Chem Engg Sci 1997, 15:2653-2659.
21. Sassner P, Galbe M, Zacchi G: Techno-economic evaluation of bioethanol production from three different lignocellulosic materials. Biomass Bioenergy 2008, 5:422-430.
22. Jones AM, Ingledew WM: Fuel alcohol production: Assessment of selected commercial proteases for very high gravity wheat mash fermentation. Enzyme Microb Technol 1994, 8:683-687.
23. Brandberg T, Karimi K, Taherzadeh MJ, Franzen CJ, Gustafsson L: Continuous fermentation of wheat-supplemented lignocellulose hydrolysate with different types of cell retention. Biotechnol Bioeng 2007, 1:80-90.
24. Galbe M, Sassner P, Wingren A, Zacchi G: Process engineering economics of bioethanol production. In Biofuels Edited by: Olsson, L. Springer Berlin/Heidelberg; 2007:303-327.
25. Destexhe A, Peckous L, Picart L: Using enzymes in ethanol production. Novozymes A/S 2004.
26. Ghose TK: Measurement of cellulase activities. Pure Appl Chem 1987, 2:257-268.
27. Berghem LER, Petterson LG: Mechanism of enzymatic cellulose degradation - isolation and some properties of a beta-glucosidase from Trichoderma viride. Eur J Biochem 1974, 2:295-305.
28. Sluiter A, Hames B, Ruiz R, Scarlata C, Sluiter J, Templeton D: Determination of Structural Carbohydrates and Lignin in Biomass Golden, CO: National Renewable Energy Laboratory; 2004.
29. Sluiter A, Hames B, Ruiz R, Scarlata C, Sluiter J, Templeton D: Determination of Sugars, Byproducts, and Degradation Products in Liquid Fraction Process Samples Golden, CO: National Renewable Energy Laboratory; 2004.
30. Palmqvist E, Hahn-Hägerdal B, Galbe M, Larsson S, Stenberg K, Szengyel Z, Tengborg C, Zacchi G: Design and operation of a bench-scale process development unit for the production of ethanol from lignocellulosics. Bioresour Technol 1996, 2:171-179.
31. Miller GL: Use of dinitrosalicylic acid reagent for determination of reducing sugar. Anal Chem 1959, 3:426-428.
32. Linde M, Galbe M, Zacchi G: Bioethanol production from non-starch carbohydrate residues in process streams from a dry-mill ethanol plant. Bioresour Technol 2008, 14:6505-6511.
33. Palmarola-Adrados B, Galbe M, Zacchi G: Combined steam pretreatment and enzymatic hydrolysis of starch-free wheat fibers. Appl Biochem Biotechnol 2004, 1:989-1002.
34. Tomas-Pejo E, Oliva JM, Ballesteros M, Olsson L: Comparison of SHF and SSF processes from steam-exploded wheat straw for ethanol production by xylose-fermenting and robust glucose-fermenting Saccharomyces cerevisiae strains. Biotechnol Bioeng 2008, 6:1122-1131.
35. Xiao Z, Zhang X, Gregg D, Saddler J: Effects of sugar inhibition on cellulases and beta-Glucosidase during enzymatic hydrolysis of softwood substrates. Appl Biochem Biotechnol 2004, 1/3:1115-1126.
36. Nikolic S, Mojovic L, Rakin M, Pejin D, Nedovic V: Effect of different fermentation parameters on bioethanol production from corn meal hydrolyzates by free and immobilized cells of Saccharomyces cerevisiae var. ellipsoideus. J Chem Technol Biotechnol 2009, 4:497-503.
37. Montesinos T, Navarro JM: Production of alcohol from raw wheat flour by amyloglucosidase and Saccharomyces cerevisiae. Enzyme Microb Technol 2000, 6:362-370.
38. Palmqvist E, Hahn-Hägerdal B: Fermentation of lignocellulosic hydrolysates, II: inhibitors and mechanisms of inhibition. Bioresour Technol 2000, 1:25-33.
39. Russell JB: Another explanation for the toxicity of fermentation acids at low pH - anion accumulation versus uncoupling. Journal of Applied Bacteriology 1992, 5:363-370.
40. Verduyn C, Postma E, Scheffers WA, van Dijken JP: Effect of benzoic acid on metabolic fluxes in yeasts: a continuous-culture study on the regulation of respiration and alcoholic fermentation. Yeast 1992, 7:501-517.
41. Taherzadeh MJ, Niklasson C, Liden G: Acetic acid - friend or foe in anaerobic batch conversion of glucose to ethanol by Saccharomyces cerevisiae? Chem Eng Sci 1997, 15:2653-2659.
42. Palmqvist E, Hahn-Hägerdal B, Galbe M, Zacchi G: The effect of water-soluble inhibitors from steam-pretreated willow on enzymatic hydrolysis and ethanol fermentation. Enzyme and Microbial Technology 1996, 6:470-476.
43. Navarro AR: Effects of furfural on ethanol fermentation by Saccharomyces cerevisiae - mathematical models. Curr Microbiol 1994, 2:87-90.
44. Palmqvist E, Grage H, Meinander NQ, Hahn-Hagerdal B: Main and interaction effects of acetic acid, furfural, and p-hydroxybenzoic acid on growth and ethanol productivity of yeasts. Biotechnol Bioeng 1999, 1:46-55.
45. Milbrandt A: A Geographic Perspective on the Current Biomass Resource Availability in the United States Golden, CO: National Renewable Energy Laboratory; 2004.
doi: 10.1186/1754-6834-3-16Cite this article as: Erdei et al., Ethanol production from mixtures of wheat straw and wheat meal Biotechnology for Biofuels 2010, 3:16
III
HUNGARIAN JOURNAL
OF INDUSTRIAL CHEMISTRY VESZPRÉM
Vol. 36(1-2) pp. 5-9 (2008)
TECHNO-ECONOMIC ASPECTS OF ON-SITE CELLULASE PRODUCTION
ZS. BARTA1�, P. SASSNER
2, G. ZACCHI2, K. RÉCZEY
1
1Budapest University of Technology and Economics, Department of Applied Biotechnology and Food Science, H-1111 Budapest Gellért tér 4, HUNGARY
�E-mail: [email protected] 2Lund University, Department of Chemical Engineering, P. O. Box 124, S 221 00 Lund, SWEDEN
On-site cellulase production for lignocellulosic ethanol production based on SO2-impregnated steam pretreatment followed by simultaneous saccharification and fermentation was investigated from a techno-economic aspect using Aspen Plus and Aspen Icarus softwares. The enzyme fermentation was assumed to operate batch-wise with a cycle time of 100 hours. The base case included sixteen 343-m3 aerated fermentors arranged in four lines operating according to a merry-go-round pattern. Besides the base case, three cases, with improved productivities, were investigated. The cost of the on-site enzyme production was estimated to range between 6.7-16.5 Eurocent/L ethanol. The cost of carbon source was not included in the total production cost, since the pretreated material was produced in the process.
Keywords: Process simulation, Cellulase fermentation, On-site, Ethanol production, Economics
Introduction
The second generation fuel-ethanol production has not been demonstrated on full-scale so far, although some pilot plants already exist in Europe and North-America. The lignocellulosic ethanol production is the most complex technology compared to sugar- and starch-based processes, which are already well-known and mature. Due to their complex structure the lignocellulosic feedstock require pretreatment prior to the cellulose hydrolysis and ethanol fermentation, that adds one more step to the process.
One alternative of cellulose hydrolysis is the enzymatic way. Although substantial improvements have been made in the last decades, the cost of enzyme is still a major problem in the enzymatic process. In this study on-site cellulase fermentation was modeled and economic evaluation for the enzyme production was conducted.
Materials and Methods
The simulation software
The process was modeled by Aspen Plus flow-sheeting software (Aspen Tech Inc, Cambridge, MA, USA) capable to solve mass and energy balances. It is a powerful tool in comparing different process configurations in terms of efficiency, energy demand or –coupled with Aspen Icarus Process Evaluator (Aspen
Tech Inc, Cambridge, MA, USA)– production cost. The later software is able to evaluate the process economics, nevertheless in our case it was used for sizing and estimating the capital investment. The built-in databases of Aspen Plus did not contain all the chemical components e.g. the ones of wood such as cellulose, lignin etc. They were obtained from the biomass databank of NREL.
Economic evaluation
Before performing economic evaluation the process equipments had to be sized. Most of them were sized manually on Excel worksheets except the heat exchanger which was sized by Icarus using the report file from Aspen Plus containing the results of material and energy balances. The manual sizing was also based on Aspen Plus simulation data.
The fixed capital investment –both the direct and indirect costs– was estimated by Icarus, where equipments not present in the Aspen Plus flowsheet, such as pumps, compressors and additional vessels were also included. The built-in database of Icarus was used for cost estimation of all the process components except the fermentors where modifications were made introducing factors to obtain the costs given by a Swedish supplier. The fermentors, however, were cost-estimated as stainless-steel (SS 304) storage tanks and their agitators as well as cooling coils were added separately. The annual fixed capital investment was calculated by use of an annuity factor of 0.110, corresponding to 15-year life of the plant, 7% interest
6
rate, linear deprecation and zero scrap-value. The reference year was 2008 and 8000 working hours per year were assumed.
The working capital investment was calculated according to the recommendation of Peters and Timmerhaus [1]. To obtain its annual representation the working capital was multiplied by the interest rate.
Table 1 summarizes the specific costs employed in the operating cost estimation.
Table 1: Cost used in the evaluation
Chemicals, nutrients Soy-meal (48% protein) 0,16 €/kg (NH4)2SO4 0,10 €/kg KH2PO4 0,10 €/kg FeSO4*7H2O 0,11 €/kg NH3 (25%) 0,22 €/kg cc. H2SO4 0,05 €/kg Defoamer 2,15 €/kg Utilities Electricity 48,4 €/MWh Cooling water 0,02 €/m3 Other costs Insurance 1% of fixed capital Maintenance 2% of fixed capital By-product credit CO2 3,2 €/t
Base case description
The enzyme production step was based on literature data [2,3], and the process step was implemented in an Aspen Plus model including all major process steps
shown in Fig. 1 described in detail in a previous study [4]. The ethanol plant was assumed to be located in Sweden, with the capacity to process 200 000 dry tons of spruce annually.
The pre-treated material stream was divided into a major stream fed directly to the simultaneous saccharification and fermentation (SSF) step and a minor stream (7.5% in the base case) led to the Trichoderma fermentation where the enzyme amount required by SSF assumed to be 15 FPU/g WIS (filter paper unit/g water insoluble solid) was obtained. The whole broth could be added to SSF since it was carried out at 37 °C and above 35 °C the growth of mycelia is entirely inhibited. Using the whole culture had several advantages: i) no additional separation was needed, which decreased the cost; ii) the enzymes adsorbed on the surface of the lignin and the cells as well as the ones trapped in the cytoplasm could also be utilized.
All the sugars present in the fermentation medium were taken into account in anhydro equivalent i.e. the polymer and monomer sugars in the pretreated material and the carbohydrate content of the soy-meal (26%) were assumed to be consumed entirely. The yields were the same for the hexosans and pentosans (Table 2). It must be mentioned that the fermentation whose results were used in the model was carried out on sulphite-pulp [2]. In order to apply these data key-assumptions had to be made: the lignin content did not affect the enzyme production, which was concluded in the same article, furthermore the monomer sugars present in the medium did not result in catabolite repression.
The base case included 16 aerated agitated fermentors, each 343 m3 in volume, arranged in four lines. The working volume was 72% of the total one.
Feedstockhandling
Steam-pretreatment Simultaneous saccharificationand fermentation (SSF)
Enzyme production
Yeast cultivation
Distillation
SeparationEvaporation Drying
Pellet productionComb. heat&powerWastewater treatment
Ethanol
Stillage
Liquid Solid
SO2 Steam Water
Condensate
Molasses
Liquid
Syrup
Steam Solid fuel
Methane
Spruce
Electricity
a.)
b.)
c.)
Fig.1: Boundary conditions of the modelled wood-to-ethanol process.
(a. steam pre-treated spruce slurry, b. condensate recycled to enzyme fermentation, c. fermentation broth)
7
Table 2: The features of T. reesei MCG-77 fermentation
Temperature 30 °C [2] pH 6 [2] Fermentation time 90 h [2] Cycle time 100 h [5] Aeration rate 0.5 VVM 1 [2] Power to the broth 0.5 kW/m3 [5] Mycelium yield 0.27 g/g CH [3] Soluble protein yield 0.26 g/g CH [3] Activity yield 185 FPU/g CH [2] Specific activity 0.71 FPU/mg protein * CH concentration 2 2 % [2] Productivity 61 FPU/(l*h) [2]
*calculated 1 Air volume/working volume/minute 2 Carbohydrate concentration given in anhydro equivalent
Cooling was performed by use of cooling coils. The fermentors operated in atmospheric pressure, and were not pressure-rated for steam sterilization. The pre-treated material and the makeup water coming entirely from the evaporation step were considered sterile, hence only cleaning-in-place was applied in the tanks. The cost of nutrient sterilization was assumed to be negligible.
Compressed air(2,7 bar)
Fermentation broth
Medium
1.
2.
3.
4.
Fig.2: Schematic flowsheet of Kornuta process
(1 line – 4 fermentors).
The four fermentors in a given line followed the same schedule, however they started being shifted in 25-hour intervals. At 25 hour 10% of the culture in the first vessel was transferred to the second one and used as inoculum (the second one gave inoculum to the third one etc.). The culture was at its peak growth and the cellulase concentration was low enough, hence fast sugar formation i.e. catabolite repression in the second vessel could be avoided. The fermentation lasted for 90 hours and was followed by a 10-hour harvesting, cleaning, charging period giving a 100-hour cycle time. After 100 hours from the start of the first fermentation the fourth vessel was ready to transfer inoculum to the first one closing the line to a loop (Fig. 2). This operation pattern was referred as “Kornuta merry-go-round” [5].
In case of contamination inoculum could be transferred from a vessel in another line and both lines could continue uninterrupted. The air supply was provided by compressors, one for each line. The four lines had a common medium preparation vessel that received the pretreated material, the makeup water and the nutrients whose concentrations were the following: 0.5% soy-meal, 0.15% (NH4)2SO4, 0.07% KH2PO4, 0.001% FeSO4 7H2O. The outlet stream before being fed to the fermentors was cooled down to 30 °C in a heat exchanger. The system contained 16 inlet and 16 outlet pumps.
Other investigated cases
Besides the base case (A) three hypothetical cases with improved productivities were investigated (Table 3). In case B the activity yield was enhanced by 50%, which also connoted 1.5-fold productivity. In case C the carbohydrate content (CH) was increased to 4% and the same yield with doubled productivity was assumed. In case D both parameters was enhanced, which resulted in tripled productivity. Table 3: Modified parameters in the various cases
Base case (A)
Enhanced yield (B)
Enhanced CH conc. (C)
Enhanced yield, CH (D)
Activity y., FPU/g CH
185 278 (1,5x)
185 278 (1,5x)
CH conc. 2% 2% 4% (2x)
4% (2x)
Prod., FPU/(l*h)
61 92 (1,5x)
122 (2x)
183 (3x)
Results and Discussion
While according to the model the Trichoderma fermentation consumed all the sugars being fed, it did not alter the amount of other substances (lignin, inhibitors etc.). The water consumption declined monotonous from A to D, whereas the other components had two levels.
Table 4: Component flows entering and departing
the enzyme fermentation
Flow, kg/h A B C D
In
Hexosans 782 535 782 535
Pentosans 15 11 15 11
Hexoses 352 241 352 241
Pentoses 63 43 63 43
Lignin 439 300 439 300
Water 55814 38167 27031 18485
Produced
Enzyme 304 208 304 208
Mycelium 317 217 317 217
CO2 712 487 712 487
8
Table 5: Total capital investment and the annual costs in M€
A B C D
Total capital investment, M€ 34 25 19 16 Costs, M€/year
Capital 3.72 (41%) 2.74 (43%) 2.06 (41%) 1.75 (46%) Chemicals, nutrients 0.49 (5%) 0.34 (5%) 0.44 (9%) 0.30 (8%) Utilities 3.89 (43%) 2.62 (41%) 1.96 (39%) 1.28 (34%) Other costs 1.02 (11%) 0.75 (11%) 0.56 (11%) 0.48 (12%) By-product credit, M€/year CO2 -0.02 -0.01 -0.02 -0.01 Total, M€/year 9.09 6.44 5.00 3.81
They were higher in scenario A/C and lower at B/D (Table 4). It can be due to the two activity yields applied which determined the carbon source demand as well as the product formation.
The total capital investment, i.e. the sum of fixed and working capitals varied in a range between 16 and 34 M€ which multiplied by the annuity factor gave the annual capital cost of 1.8-3.7 M€/year (Table 5).
Besides the capital the utilities namely the electricity used by agitators, compressors, pumps was found the other largest contributor in production cost. The cost of cooling water was negligible. The carbon-dioxide credits were two order smaller than the costs. The sum of chemicals, nutrients and other costs were estimated not being more than 20% of the total. It must be pointed out, that the cost of carbon source was not included in either the annual or the specific enzyme production cost, since the pre-treated material was produced in the process.
The on-site cellulase production reduced the produced ethanol amount providing the same feedstock utilization, since the carbohydrates were consumed partially by the enzyme fermentation. The ethanol plant using commercial enzyme produced 59 563 m3 ethanol per year, whereas the base case (A) and case C merely 55 000 m3. The cases B/D with enhanced activity yield produced more ethanol (56 441 m3/year), since less pre-treated material was needed for the cellulase fermentation.
y = 464.032x-0.815
R2 = 0.997
0
2
4
6
8
10
12
14
16
18
50 100 150 200
Produktivitás, FPU/(l*h)
cent/L EtOH
Fig. 3: Specific enzyme cost as a function of
productivity.
By increasing the productivity the specific enzyme cost reduced monotonously. The fitted curve was close to hyperbola having the index of 0.8 (Fig. 3). The breakdown of specific enzyme cost also shows that the main contributors were the capital and the utilities (Fig. 4). In the base case 16.5 Eurocent/L EtOH was found. At tripled productivity (D) the specific enzyme cost was 6.7 Eurocent/L (41% of the base case).
11.4
6.7
9.1
16.5
0
2
4
6
8
10
12
14
16
18
A B C D
cent/L EtOHTotal
Capital
Chemicals
Utilities
Other
Fig. 4: Breakdown of specific enzyme cost (produced ethanol: 55 000 m3/year at A/C, 56 441 m3/year at B/D).
Summary
The total cost of on-site cellulase production (diminished with cost of carbon source) was estimated to range between 6.7-16.5 Eurocent/L ethanol for the four investigated scenarios. Capital investment and electricity were found the main contributors.
Acknowledgement
The 6th Framework Programme of the European Commission is gratefully acknowledged for its financial support (NILE-project, contract no. 019882).
9
REFERENCES
1. PETERS, M. S., TIMMERHAUS, K. D., Plant Design and Economics for Chemical Engineers, McGraw-Hill, New York (1991)
2. DOPPELBAUER, R., ESTERBAUER, H., STEINER, W., LAFFERTY, R. M., STEINMÜLLER, H., Applied Microbiology and Biotechnology 26 (1987) 485-494
3. ESTERBAUER, H., STEINER, W., LABUDOVA, I., HERMANN, A., HAYN, M., Bioresource Technology 36 (1991) 51-65
4. SASSNER, P., GALBE, M., ZACCHI, G., Biomass and Bioenergy 32 (2008) 422-430
5. NYSTROM, J.M., ALLEN, A. L., Biotechnology and Bioengineering Symposium 6 (1976) 57-74
IV
SAGE-Hindawi Access to ResearchEnzyme ResearchVolume 2010, Article ID 734182, 8 pagesdoi:10.4061/2010/734182
Research Article
Process Design and Economics of On-Site Cellulase Production onVarious Carbon Sources in a Softwood-Based Ethanol Plant
Zsolt Barta,1 Krisztina Kovacs,1, 2 Kati Reczey,1 and Guido Zacchi2
1 Department of Applied Biotechnology and Food Science, Budapest University of Technology and Economics,Szt. Gellert ter 4, 1111 Budapest, Hungary
2 Department of Chemical Engineering, Lund University, P.O. Box 124, 221 00 Lund, Sweden
Correspondence should be addressed to Zsolt Barta, zsolt [email protected]
Received 15 January 2010; Accepted 29 March 2010
Academic Editor: Robert D. Tanner
Copyright © 2010 Zsolt Barta et al. This is an open access article distributed under the Creative Commons Attribution License,which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
On-site cellulase enzyme fermentation in a softwood-to-ethanol process, based on SO2-catalysed steam pretreatment followed bysimultaneous saccharification and fermentation, was investigated from a techno-economic aspect using Aspen Plus and AspenIcarus Process Evaluator softwares. The effect of varying the carbon source of enzyme fermentation, at constant protein andmycelium yields, was monitored through the whole process. Enzyme production step decreased the overall ethanol yield (270L/dry tonne of raw material in the case of purchased enzymes) by 5–16 L/tonne. Capital cost was found to be the main costcontributor to enzyme fermentation, constituting to 60–78% of the enzyme production cost, which was in the range of 0.42–0.53 SEK/L ethanol. The lowest minimum ethanol selling prices (4.71 and 4.82 SEK/L) were obtained in those scenarios, wherepretreated liquid fraction supplemented with molasses was used as carbon source. In some scenarios, on-site enzyme fermentationwas found to be a feasible alternative.
1. Introduction
Production of ethanol from lignocellulosic materials is a verycomplex process which consists of various interdependentsteps, such as pretreatment of the raw material, enzymatichydrolysis of the polysaccharides into sugar monomers,fermentation of the sugars to ethanol, and purification ofethanol. Since the process has not yet been demonstratedon a commercial scale, only a limited number of studiesare available on its techno-economic aspects, and largevariations in the estimated overall ethanol production costs(from about 0.93 to 5.49 SEK/L ethanol) can be seen, due todifferences in the process design and in the assumptions usedin the studies [1–10]. One of the most important parametersthat influences the production costs is the annual capacityof the plants, that is, lower production costs are usuallyobtained for plants that process above 650 000 tons of dryraw material per year. Other differences are found in theconversion technologies, the types of raw materials used, theoverall ethanol yields assumed, the investment parameters,and whether utilities such as process steam and electricity areincluded as costs in the assessment [3].
According to recent techno-economic evaluations, themain contributors to the overall costs of producing ethanolfrom biomass are the raw material (30%–40%) and thecapital investment (30%–45%), followed by the cellulaseenzymes (10%–20%) [6, 7, 11–13]. The cost of cellulases notonly represents a significant part in the overall productioncosts but is also one of the most uncertain parameters inthe evaluations [3]. Most authors assume that cellulases arepurchased from enzyme manufacturers and calculate withan estimated future enzyme price, which varies from about0.2 to 0.7 SEK/L ethanol in the investigations reviewed in[1, 6, 7, 11–15]. However, some other studies presume thaton-site or near-site production on cheap lignocellulosic rawmaterials will be desirable to meet the targeted enzymecosts of <0.5 SEK/L [8, 16–20]. In any case, improvement ofcellulolytic microorganisms, enhancement of the hydrolyticcapacity of cellulases, and optimization of the technology ofenzyme production are essential today in order to furtherreduce the enzyme costs for the biomass-to-bioethanolprocess.
Spruce is the most abundant wood in Sweden, and itwas shown to be a suitable raw material for bioethanol
2 Enzyme Research
production in several studies [21–24]. Hypercellulolyticmutants of Trichoderma reesei, the most widely used fungusfor cellulase production, were reported to grow well, andsecrete high amounts of cellulolytic enzymes on steam-pretreated spruce [25, 26]. The most economical way ofemploying the enzymes produced would be the directuse of whole crude fermentation broths, containing fungalcells and substrate residues, in order to avoid expensivecell removal, enzyme concentration, and purification steps.Previous investigations showed that due to the effect ofmycelium-bound enzymes, application of the whole brothof T. reesei could not only lead to cost reduction but also toimproved saccharification and enhanced ethanol yields [27–30]. These suggest that on-site enzyme production with T.reesei could be a possible alternative to purchasing cellulasesfor a bioethanol plant using spruce as raw material.
In the present study, on-site cellulase production in a full-scale bioethanol plant was modelled together with the wholeethanol production process, and the economic impact of theenzyme fermentation step on the ethanol production costwas assessed. Cellulases were assumed to be produced using amutant of T. reesei, employing the whole crude fermentationbroth of the fungus in the ethanol production step. Theeffect of varying the carbon source of enzyme fermentation,at constant protein and mycelium yields, was investigatedthrough the whole process. Different mixtures of pretreatedliquid fraction, slurry, and molasses were evaluated as feedfor enzyme production.
2. Materials and Methods
2.1. Raw Material. The dry spruce chips contain 37.9%glucan, 9.9% mannan, 1.8% galactan, 4.3% xylan, 1.3%arabinan, and 28.0% lignin. These values were derivedfrom compositional analyses performed in EU-project NILE(contract no. 019882) according to the standardized methodof National Renewable Energy Laboratory (NREL, Golden,CO) [31]. The remaining part is made up of acetyl groups,extractives, and other compounds which were estimatedfrom a previous study [6]. The dry matter (DM) contentwas assumed to be 50%. Theoretically, 356 L of ethanol couldbe produced from the hexose sugars per dry tonne of rawmaterial.
2.2. Overall Process Description. The proposed ethanol plantis assumed to be located in Sweden and process 200 000 drytonne spruce chips annually. It is run by 28 employees, and isassumed to be in operation for 8000 hours per year.
The process scheme is shown in Figure 1. Each step,except cellulase enzyme fermentation (CEF), has beendescribed in detail elsewhere [6] and will only be discussedhere briefly, focusing mainly on the minor modifications.Live steam was assumed to be available at 20 and 4 bar,and secondary steam is used to replace live steam wheneverpossible.
2.3. Reference Case. The conversion of carbohydrates iscarried out in steam pretreatment and in simultaneous
saccharification and fermentation (SSF) (Figure 1). Processdata for steam pretreatment (210◦C, 2.5% SO2) and SSF werebased on results recently obtained from experimental workperformed at the Department of Chemical Engineering,Lund University, Sweden.
Water needed to adjust the dry matter in the SSFstep to 10% water-insoluble solids (WIS) is added beforepressing the pretreated slurry. The diluting stream consistsof fresh water and part of the evaporation condensate. Italso contains ammonia to neutralize the slurry. The pressedliquid supplemented with molasses containing 50% sucroseis used in yeast cultivation (YC) without adding extra freshwater, hence the inhibitor concentrations in YC and SSFare approximately the same. Yeast seed train consisting ofthree stages provides SSF with 7.5% inoculum. Only thefirst and second stages are designed to be sterile, that is,those vessels are pressure-rated for steam sterilization. In SSF,the concentration of ordinary baker’s yeast and the enzymedosage are 3 g DM/L and 10 FPU (filter paper unit)/g WIS,respectively. The SSF takes place in twelve agitated nonsterilefermentors with a total volume of 920 m3 each. An SSFcycle including filling, fermentation, draining, and cleaninglasts for 60 hours. The number of the YC fermentors wascalculated from the cycle time, which was assumed to be 15hours for all YC stages.
According to the model calculations the ethanol contentof the SSF broth is 3.8 wt-% which corresponds to aconcentration of 40.4 g/L in the liquid phase. Distillationand molecular sieve adsorption are used to produce pure(99.8 wt-%) ethanol. The distillation step consists of twostripper columns and a rectifier, which are heat integratedby operating at different pressures. The remaining water inthe overhead vapour leaving the rectifier is removed in thedehydration columns that are regenerated with pure ethanolvapour. The regenerate is returned to the rectifier.
The stillage of the stripper columns is separated in afilter press resulting in a solid fraction with a WIS contentof 40%. The liquid fraction of the stillage is concentratedto 60% DM in an evaporation system which contains fiveeffects in a forward-feed arrangement, that is, only thefirst effect is heated by live steam; the subsequent onesutilize the vapour from the previous effect, operating athigher pressure. Boiling point elevation was accounted for[32], and overall heat transfer coefficients were estimatedto vary between 500 and 2000 W/m2◦C, depending on thetemperature and concentration of the liquid. Based on thework of Olsson and Zacchi [33], it was assumed that byapplying a stripper column after evaporation, recycling ofpart of the evaporation condensate to dilute the wholeslurry was possible. The rest of the condensate is sentto the wastewater treatment facility, where together withthe condensed flash streams mainly originating from thepretreatment, it is treated by anaerobic digestion followed byan aerobic step [6]. It was assumed that 50% of the chemicaloxygen demand (COD) was converted with a yield of 0.35 m3
methane/kg COD consumed.Steam and electricity are generated by burning the
biogas, the concentrated liquid fraction and part of the solidfraction of the stillage. The generated steam is allowed to
Enzyme Research 3
Feedstockhandling
Steampretreatment
YC, CEFSSF
Distillation,dehydration
Wastewatertreatment
Evaporation Filtration
Heat and powerproduction
Drying
Spruce chips SO2 Steam Ethanol
Stillage
Liquidfraction
Solidfraction
SteamElectricity
Condensate
Biogas
Solid fuel
Syrup
Diluting liquid stream
Figure 1: Overall process scheme for the proposed ethanol plant. In the reference case, there was no enzyme production; the enzymes werepurchased. CEF: cellulase enzyme fermentation, YC: yeast cultivation and SSF: simultaneous saccharification and fermentation.
expand to 4 bar through the turbine system, however, partof the steam is withdrawn at 20 bar for pretreatment anddrying. The heat from flue gas condensation could be utilizedby integrating a district heating system with the heat andpower producing facility, however, this was not included inthe model. The excess solid residue, that is, the solid fractionnot required for steam generation, is dried in a superheatedsteam dryer to 88% DM. The secondary steam generated bydrying is utilized in the process.
2.4. Description of Enzyme Fermentation. Process data forCEF were obtained from the literature [34, 35], how-ever, some key assumptions were also made. The appliedTrichoderma strain was assumed to be able to producecellulase enzymes in the presence of monosaccharides, thatis, it was not catabolite-repressed (e.g., T. reesei RUT C30).The mycelium, soluble protein, and activity yields were0.27 g, 0.26 g, and 185 FPU per g carbohydrate in anhydroequivalent [35], respectively, which resulted in a specificactivity of 710 FPU/g protein. After complete hydrolysisof polysaccharides, all the monosaccharides are consumedentirely in CEF, while other compounds are not involved inany reaction. Based on the work of Szengyel and Zacchi [36],it was assumed that inhibition due to compounds present inthe pretreated material, such as furan derivatives and organicacids, did not occur.
The 5% inoculum is received from the second stageof a two-stage seed train. Both stages operate with 5%inoculum at a cycle time of 30 hours. The first stage receivesinoculum from a stock culture, while the second is inoculatedwith the broth of the first. Concerning the composition,the seed stages are assumed to be run on the same feedas the production stage, where 120 hours cycle time ispresumed. As this time is double of the cycle time of SSF,the number of vessels are 24 in the enzyme productionstage. Considering the ratio of the cycle times of seed andproduction stages, the number of seed vessels is 6 in bothstages. In all scenarios these numbers were kept constant,hence the total vessel volume varied in a range of 37–121 m3
in the production stage. The fermentors of the seed train
are pressure-rated and can be sterilized at 120◦C, however,it was assumed at the production stage that sterilizationwas not necessary. Cleaning-in-place is sufficient, since theevaporation condensate and the pretreated material wereconsidered to be sterile and the fresh water added beforepressing the pretreated slurry is sterile-filtered beforehand.Furthermore, the nutrients (soy-meal 0.5%, (NH4)2SO4
0.15%, KH2PO4 0.07%, and FeSO4·7H2O 0.001%) and themolasses were assumed not to cause any contamination,so the seed vessels can be sterilized empty. At all stages,30◦C and pH 5 are kept. The feed is cooled down in aheat exchanger and the heat released during fermentationis removed by cooling water that circulates in jackets at thefirst stage and in coils at later stages. The cooling jacket isfavourable in terms of cleaning, however, it is not sufficientat larger volume. The pH is controlled using ammonia.Aeration of 0.5 VVM was assumed to ensure sufficientagitation when the solid content was less then 1% WIS. Thewhole broth containing mycelia and enzymes is added to SSF.This can be done, since SSF is carried out at 37◦C, and above35◦C the growth of mycelia is completely inhibited [35].
2.5. Enzyme Fermentation Configurations. Three configura-tions, denoted with A-C, were investigated in the model ofenzyme fermentation (Figure 2). They differed in the carbonsource. In scenario A, part of the liquid fraction of the dilutedslurry was used, while in scenario B, the liquid fraction wassupplemented with molasses to increase the sugar content. Inscenario C, a mixture of the liquid fraction and the pressedslurry was prepared, and used as feed for CEF. Scenario Cwas divided into three subcases, C1–3 depending on theWIS content of the mixture, that is, 1%, 2%, and 3%.For scenarios A and B, sensitivity analysis was performed,denoted with +. The specific activity of the soluble proteinswas enhanced 1.5-fold, resulting in an increase of 50% in theproductivity in terms of enzyme activity, while protein andmycelium yields remained the same.
The liquid fraction also contained water-insoluble par-ticles, as a WIS retention of 99% was assumed in thefiltration of the slurry. In the CEF feed the total carbohydrate
4 Enzyme Research
YC
CEF
SSFWholeslurry
Diluting liquid stream
Filter press Pressedslurry
Liquid fraction
Broth
Molasses
Nutrients
(a)
YC
CEF
SSFWholeslurry
Diluting liquid stream
Filter press Pressedslurry
Liquid fraction
Broth
Molasses
Molasses
(b)
YC
CEF
SSFWholeslurry
Diluting liquid stream
Filter pressPressedslurry
Liquid fraction
Broth
Molasses
Nutrients
(c)
Figure 2: Layout of cellulase enzyme fermentation (CEF), yeast cul-tivation (YC) and simultaneous saccharification and fermentation(SSF) at scenarios A to C (a–c, respectively).
0 . 4 2
0 . 5 3
0 . 4 3
0 . 5 2 0 .5 0 . 4 9 0 . 5
0
0 . 1
0 . 2
0 . 3
0 . 4
0 . 5
0 . 6
A A + B B + C 1 C 2 C 3
T o t a l C a p i t a l C h e m i c a l s
U t i l i t i e s O t h e r
E n
z y m
e p r
o d u
c t io
n c
o s t
( S E
K / L
)
Figure 3: Cost contributors of enzyme production in SEK/Lethanol. Molasses was added to chemicals. Other refers to mainte-nance and insurance. No extra labour was accounted for the enzymefermentation area. Carbon source A: pretreated liquid fraction, B:pretreated liquid fraction and molasses and C: pretreated liquidfraction and pressed pretreated slurry with a total WIS content of1%, 2% and 3%; +: 1.5-fold specific activity.
4.88
4.96
4.83 4.82
4.71
4.99 4.97 4.96
4.5
4.6
4.7
4.8
4.9
5
5.1
Ref A A+ B B+ C1 C2 C3
Min
imu
met
han
olse
llin
gpr
ice
(SE
K/L
)
Figure 4: Minimum ethanol selling price. Carbon source A:pretreated liquid fraction, B: pretreated liquid fraction and molassesand C: pretreated liquid fraction and pressed pretreated slurry witha total WIS content of 1%, 2% and 3%; +: 1.5-fold specific activity;Ref: reference case with purchased enzyme preparation.
content expressed in monomer equivalent (ME) and WISconcentration, in parentheses, were the following: A: 4.6%(0.5%), A+: 4.7% (0.7%), B: 10% (0.8%), B+: 10% (0.9%),C1: 4.9% (1%), C2: 5.5% (2%), and C3: 6.0% (3%). Incases C1–3, aeration was assumed to be insufficient forensuring homogeneity, therefore agitators were built in, andpower-to-broth value was set to 40 W/m3. In scenario B,molasses served as a complex nutrient source, hence nutrientsupplementation was omitted.
2.6. Analysis Methods. Mass and energy balances weresolved using the commercial flowsheeting program AspenPlus 2006.5 (Aspen Technology, Inc., Cambridge, MA).Physical property data for biomass components such aspolysaccharides and lignin were derived from the NRELdatabase [37]. Fixed capital investment (FCI) costs wereestimated either with Aspen Icarus Process Evaluator 2006.5(Aspen Technology, Inc.) or from vendor quotation. Theconstruction material was assumed to be 304 stainless steelfor all process vessels. To obtain the annual FCI, an annuityfactor of 0.110 was used, corresponding to a depreciationperiod of 15 years and an interest rate of 7%. Workingcapital investment (WCI) was calculated according to therecommendations in literature [38]. Annual WCI is theproduct of WCI and interest rate.
All costs are presented in Swedish kronor (SEK, 1 US$ ≈7.3 SEK, 1 C ≈ 10.5 SEK). In the reference case the purchaseprice of enzyme was assumed to be 28.5 SEK per millionFPU. Cost of nutrients (in SEK/kg) applied in CEF werethe following: soy-meal 1.5, (NH4)2SO4 0.9, KH2PO4 1.0,FeSO4·7H2O 1.0. Cost of raw material, chemicals, utilities,labour, insurance, maintenance, and revenues of ethanol,coproducts, and electricity are reported in recent studies[6, 13]. Minimum ethanol selling price (MESP) refers to theethanol price at break-even point, that is, at this price, annualcost and income are equal.
Enzyme Research 5
3. Results and Discussion
3.1. Process Design: The Effects of On-Site Enzyme Fermenta-tion. After steam pretreatment, a slurry with a WIS contentof 27% was obtained, which was much higher than thatin experimental work. This was due to lower heat lossesapplied in the model (10% of the adiabatic heat demand).The whole slurry was diluted to a WIS content of 10.1%-10.2%, depending on the scenario, before pressing.
Components fed into CEF and SSF on a yearly basis,other features of CEF and overall ethanol yield are shownin Table 1. On-site enzyme production has the advantagethat pentoses, not fermentable by ordinary baker’s yeast,can also be utilized. Nevertheless, in regard to hexoses,CEF competes with ethanol fermentation, since the moresugar is used for CEF, the less sugar is available for ethanolproduction. The highest mass flow of C5 sugars (pentosansand pentoses in ME) into CEF was obtained in scenarioA, which corresponded to 13% of C5 sugars recoveredafter pretreatment, hence even at the best utilization, 87%was burnt or dried. Regarding the composition of sugarsin the CEF feed, scenarios A and A+ gave the highestC5 proportion (21%). This would increase considerably, ifthe raw material of the ethanol plant consisted of morepentosans, for example, if agricultural byproducts were usedinstead of softwood. In scenarios A, A+, and C1–3, the C5
ratio decreased monotonously with increased WIS content inthe feed, since the solid fraction contained relatively less C5
sugars compared to the sugar composition of liquid fraction.In the cases supplemented with molasses (B and B+), the C5
ratio was only 9%, due to the high ratio of sugars of molasses(56%-57%). The productivity of CEF varied in a range of70–230 FPU/(L·h) (Table 1). As the fermentation time wasmaintained at 108 hours in each scenario, the productivityonly depended on the carbohydrate concentration andactivity yield.
Although CEF decreased the C6 flow (hexosans and hex-oses) in SSF feed, the majority of C6 sugars were fermentedto ethanol (Table 1). Even in the worst scenario (C3), thedecrement of C6 sugars (in ME) was only 5850 tonne/year,which corresponded to 6% of the C6 sugars (in ME) fedto SSF in the reference case. The yeast amount required bySSF slightly decreased with increasing WIS contents in CEF(Table 1), since by feeding more WIS into CEF, the WISflow fed to SSF decreased, which resulted in smaller totalfermentation volume in SSF at constant WIS concentration.Excluding the cases of 1.5-fold specific activity, the producedenzyme protein varied little (Table 1). The produced amountdepended on the WIS flow fed into SSF, hence the highestamount was obtained for scenario B due to the lowestWIS consumption in CEF. Mycelium had the same trend asenzyme protein, since their ratio was fixed by their yields.
The ethanol concentration in the SSF broth variedbetween 3.6 and 3.7 wt-%, which corresponded to a rangeof 38.1 to 39.7 g/L in the liquid phase. They were slightlylower than in the reference case, due to drop of the contentsof hexoses and hexosans in the liquid and solid phases,respectively. The overall ethanol yield of the process in CEFcases also decreased compared to the reference (Table 1). The
decrements were lowest in scenarios B and B+ owing to thelarge proportion of sugars coming from molasses. The yieldsin scenarios A and C1–3 were equal, which implied that themodel was indifferent to the WIS content of CEF in the rangeinvestigated.
3.2. Economics: Specific Enzyme Cost, Minimum EthanolSelling Price and Annual Cash Flows. Cost elements ofenzyme production expressed in SEK/L ethanol are shownin Figure 3. Capital and chemical costs were found to bethe main contributors in each scenario. Utilities refer to theelectricity consumption of the compressor and the agitators(scenarios C1–3) and to the demand for cooling water. Thecost for steam used for sterilization of the empty seed vesselswas assumed to be negligible. The lowest capital costs wereobtained in scenarios B and B+, due to high carbohydrateconcentration in CEF, which resulted in small fermentorvolume. However, the cost of chemicals was the highest inthese scenarios, owing to the extra cost of molasses. In casesC1–3, increasing the WIS concentration in CEF resulted ina reduced fermentation volume, and in consequence, lowercapital and chemical costs.
Comparing the base scenarios (A, B, and C1–3), case Bproved to have the highest total cost of enzyme production,due to the additional cost of molasses. On the otherhand, in regard to MESP, case B was the most favourable,furthermore, this was the only base scenario with on-siteenzyme production in which the MESP was lower than thatin the reference case, with purchased enzymes (Figure 4).In spite of the extra expenses, molasses could improve theprocess economics considerably, since CEF supplementedwith molasses reduced the overall ethanol yield, the mostimportant parameter in the production cost of ethanol [39],to a smaller extent. The 1.5-fold activity yield resulted ina decrease of 16 and 19% in total enzyme production costcompared to the base scenarios A and B, respectively, whichcorresponded to a decrement of 2.6 and 2.3% in MESP.
Annual cash flows are presented in Table 2, calculated fora selling price of ethanol of 5.5 SEK/L. The CEF increasedthe capital costs significantly (10%–14%) compared to thereference case. The second largest cost contributor after thecapital cost was the raw material cost, which did not changedue to constant annual capacity. Also these costs have beenproved to be the main contributors to the production costof lignocellulosic ethanol in previous, similar studies [3, 6, 7,13]. Chemical expenses increased by 12%–34% compared tothe reference case. The utility costs were the lowest amongthe cost elements in each scenario, since only process andcooling water had to be purchased, as steam and electricitywere generated on-site.
Ethanol, the main product, gave 83%-84% of the annualrevenues. Co-products refer to solid fuel, that is, the driedexcess solid residue, and the carbon-dioxide produced inCEF, YC, and SSF, which was also assumed to be marketable.Solid fuel contributed to 97% of the co-products incomein each scenario. While steam generation met the steamrequirement of the process, produced electricity was con-sumed on-site only partially. The excess electricity varied
6 Enzyme Research
Table 1: Features of enzyme and ethanol productions. Both CEF and SSF are carried out in batch operation. Carbon source A: pretreatedliquid fraction, B: pretreated liquid fraction and molasses, and C: pretreated liquid fraction and pressed pretreated slurry with a total WIScontent of 1%, 2%, and 3%; +: 1.5-fold specific activity.
Reference A A+ B B+ C1 C2 C3
Components into CEF, tonne/year
Hexosans — 430 368 298 236 776 1374 1861
Pentosans — 7 6 5 4 13 23 32
Hexoses — 5133 3352 6242 4120 4812 4256 3805
Pentoses — 1487 971 622 413 1394 1233 1102
WISa — 806 690 558 442 1453 2575 3486
C5 ratio in CEFb, % — 21 21 9 9 20 18 16
WIS content in CEF, % — 0.5 0.7 0.8 0.9 1.0 2.0 3.0
Productivity of CEFc, FPU/(L·h) — 70 107 154 230 75 84 93
Components into SSF, tonne/year
Hexosans 62980 62550 62612 62683 62745 62205 61606 61120
Hexoses 32634 27521 29304 30499 31228 27841 28397 28848
Yeast 3168 3160 3160 3160 3168 3144 3128 3112
Enzyme protein 1816e 1656 1104 1682 1121 1651 1642 1635
Mycelium — 1727 1152 1754 1170 1722 1713 1705
WISa, ktonne/year 117.98 117.55 117.61 117.68 117.74 117.19 116.59 116.09
Overall EtOH yieldd, L/dry tonne 270 254 260 263 266 254 254 254aIncludes hexosans and pentosans.b(pentosans in ME + pentoses)/total sugar in ME.cCorresponds to an enzyme fermentation time of 108 h.dIncludes yeast and enzyme productions and ethanol losses in the process.eThe purchased enzyme preparation was assumed to contain 10% protein.
Table 2: Annual costs, revenues, and profit of the proposed ethanol plant in MSEK. Carbon source A: pretreated liquid fraction, B: pretreatedliquid fraction and molasses, and C: pretreated liquid fraction and pressed pretreated slurry with a total WIS content of 1%, 2%, and 3%;+: 1.5-fold specific activity.
Reference A A+ B B+ C1 C2 C3
Annual cost (MSEK)
Raw material 99.5 99.5 99.5 99.5 99.5 99.5 99.5 99.5
Capital 135.5 152.7 151.6 151.0 149.1 154.2 153.8 153.5
Chemicals 28.3 32.4 31.7 38.0 35.4 32.2 31.9 31.6
Enzymes 33.6 — — — — — — —
Utilities 2.3 2.5 2.4 2.5 2.4 2.5 2.4 2.4
Other 20.8 21.3 21.2 21.2 21.2 21.3 21.3 21.3
Total 320.0 308.3 306.4 312.2 307.6 309.7 309.0 308.3
Annual income (MSEK)
Ethanola 297.4 279.8 285.7 289.7 292.2 279.8 279.7 279.7
Co-products 45.1 45.9 45.3 48.5 46.8 46.1 46.5 46.8
Electricity 11.1 9.8 10.2 10.0 10.3 9.5 9.4 9.4
Total 353.5 335.6 341.2 348.2 349.3 335.4 335.7 335.8
Annual profit (MSEK) 33.6 27.2 34.8 36.0 41.7 25.7 26.7 27.5aEthanol price was assumed to be 5.5 SEK/L.
between 35% and 41% of the amount produced, the lowestbeing in cases C1–3, due to the consumption of agitators,whereas the highest being in the reference case. The highestprofit was achieved in scenarios B and B+, where the MESPswere the lowest. Table 2 clearly shows the importance of co-product and electricity revenues, since the income of ethanoldoes not exceed the expenses.
4. Conclusions
By means of the developed model of on-site enzyme produc-tion, embedded in a softwood-based ethanol process, variousstreams were studied as carbon sources in enzyme fermen-tation at constant soluble protein and mycelium yields.The majority of sugars consumed in enzyme fermentation
Enzyme Research 7
were C6, especially when molasses was present. The overallethanol yields were lower in those scenarios, where enzymefermentation was included, than in the reference case, wherea purchased cellulase preparation was applied. This was dueto the fact that enzyme production decreased the amountof carbohydrates available for the yeast to produce ethanol.When molasses was used as additional carbon source theminimum ethanol selling price was the lowest among thescenarios, resulting in the highest annual profit, althoughthe specific enzyme production cost was found to be thehighest.
On-site enzyme fermentation contributed to 9%–11%of the ethanol production cost. The feasibility of includingenzyme production in the lignocellulosic ethanol processhighly depends on the full-scale price of commercial cellulaseenzyme preparation, which is still very uncertain. At thepremises of the study, some scenarios proved to be morefeasible than that with purchased enzymes, which impliesthat on-site enzyme production can be an alternative. Toachieve further improvement in the economics of a processintegrating cellulase fermentation, the enzyme demandof SSF has to be decreased, whereas the activity yieldand productivity, the two most important parameters ofenzyme fermentation in terms of cost reduction, have to beincreased.
Similarly to previous evaluations [6, 7], the present studydemonstrated the importance of the overall ethanol yield andthe co-product revenues in regard to the process economics.On-site enzyme production is the most feasible, when theleast C6 sugars are consumed, hence it decreases the overallethanol yield to a smallest extent. A plant using a rawmaterial with higher C5 proportion, such as agricultural by-products, could become relatively more viable by integratingenzyme fermentation. However, further investigation isrequired to prove this statement.
Abbreviations
CEF: Cellulase enzyme productionCOD: Chemical oxygen demandDM: Dry matterFCI: Fixed capital investmentFPU: Filter paper unitME: Monomer equivalentMESP: Minimum ethanol selling priceNREL: National renewable energy laboratorySEK: Swedish kronorSSF: Simultaneous saccharification and
fermentationWCI: Working capital investmentWIS: Water-insoluble solidsYC: Yeast cultivation
Acknowledgment
The Hungarian National Research Fund (OTKA – K 72710)is gratefully acknowledged for its financial support.
References
[1] A. Aden, M. Ruth, K. Ibsen, et al., “Lignocellulosic biomassto ethanol process design and economics utilizing co-currentdilute acid prehydrolysis and enzymatic hydrolysis for cornstover,” Tech. Rep. NREL/TP-510-32438, National RenewableEnergy Laboratory, 2002.
[2] G. M. Bohlmann, “Process economic considerations forproduction of ethanol from biomass feedstocks,” IndustrialBiotechnology, vol. 2, no. 1, pp. 14–20, 2006.
[3] M. Galbe, P. Sassner, A. Wingren, and G. Zacchi, “Processengineering economics of bioethanol production,” in Biofuels,vol. 108 of Advances in Biochemical Engineering/Biotechnology,pp. 303–327, Springer, Berlin, Germany, 2007.
[4] C. N. Hamelinck, G. van Hooijdonk, and A. P. C. Faaij,“Ethanol from lignocellulosic biomass: techno-economic per-formance in short-, middle- and long-term,” Biomass andBioenergy, vol. 28, no. 4, pp. 384–410, 2005.
[5] L. R. Lynd, R. T. Elander, and C. E. Wyman, “Likely featuresand costs of mature biomass ethanol technology,” AppliedBiochemistry and Biotechnology A, vol. 57-58, pp. 741–761,1996.
[6] P. Sassner, M. Galbe, and G. Zacchi, “Techno-economicevaluation of bioethanol production from three differentlignocellulosic materials,” Biomass and Bioenergy, vol. 32, no.5, pp. 422–430, 2008.
[7] A. Wingren, M. Galbe, and G. Zacchi, “Techno-economicevaluation of producing ethanol from softwood: comparisonof SSF and SHF and identification of bottlenecks,” Biotechnol-ogy Progress, vol. 19, no. 4, pp. 1109–1117, 2003.
[8] R. Wooley, M. Ruth, J. Sheehan, K. Ibsen, H. Majdeski, andA. Galvez, “Lignocellulosic biomass to ethanol process designand economics utilizing co-current dilute acid prehydrolysisand enzymatic hydrolysis—current and futuristic scenarios,”Tech. Rep. NREL/TP-580-26157, National Renewable EnergyLaboratory, 1999.
[9] K. S. So and R. C. Brown, “Economic analysis of selectedlignocellulose-to-ethanol conversion technologies,” AppliedBiochemistry and Biotechnology A, vol. 77–79, pp. 633–640,1999.
[10] T. Eggeman and R. T. Elander, “Process and economic analysisof pretreatment technologies,” Bioresource Technology, vol. 96,no. 18, pp. 2019–2025, 2005.
[11] A. Wingren, M. Galbe, C. Roslander, A. Rudolf, and G.Zacchi, “Effect of reduction in yeast and enzyme concentra-tions in a simultaneous-saccharification-and-fermentation-based bioethanol process: technical and economic evaluation,”Applied Biochemistry and Biotechnology A, vol. 122, no. 1–3,pp. 485–499, 2005.
[12] A. Wingren, M. Galbe, and G. Zacchi, “Energy considerationsfor a SSF-based softwood ethanol plant,” Bioresource Technol-ogy, vol. 99, no. 7, pp. 2121–2131, 2008.
[13] P. Sassner and G. Zacchi, “Integration options for high energyefficiency and improved economics in a wood-to-ethanolprocess,” Biotechnology for Biofuels, vol. 1, article 4, 2008.
[14] D. Gregg and J. N. Saddler, “Bioconversion of lignocellulosicresidue to ethanol: process flowsheet development,” Biomassand Bioenergy, vol. 9, no. 1–5, pp. 287–302, 1995.
[15] D. J. Gregg, A. Boussaid, and J. N. Saddler, “Techno-economicevaluations of a generic wood-to-ethanol process: effect ofincreased cellulose yields and enzyme recycle,” BioresourceTechnology, vol. 63, no. 1, pp. 7–12, 1998.
8 Enzyme Research
[16] Z. Barta, P. Sassner, G. Zacchi, and K. Reczey, “Techno-economic aspects of on-site cellulase production,” HungarianJournal of Industrial Chemistry, vol. 36, pp. 5–9, 2008.
[17] S. T. Merino and J. Cherry, “Progress and challenges in enzymedevelopment for biomass utilization,” in Biofuels, vol. 108 ofAdvances in Biochemical Engineering/Biotechnology, pp. 95–120, Springer, Berlin, Germany, 2007.
[18] R. Wooley, M. Ruth, D. Glassner, and J. Sheehan, “Processdesign and costing of bioethanol technology: a tool for deter-mining the status and direction of research and development,”Biotechnology Progress, vol. 15, no. 5, pp. 794–803, 1999.
[19] W. J. Frederick, S. J. Lien, C. E. Courchene, N. A. DeMartini,A. J. Ragauskas, and K. Iisa, “Production of ethanol fromcarbohydrates from loblolly pine: a technical and economicassessment,” Bioresource Technology, vol. 99, no. 11, pp. 5051–5057, 2008.
[20] Q. A. Nguyen and J. N. Saddler, “An integrated model for thetechnical and economic evaluation of an enzymatic biomassconversion process,” Bioresource Technology, vol. 35, no. 3, pp.275–282, 1991.
[21] M. Alkasrawi, A. Rudolf, G. Liden, and G. Zacchi, “Influenceof strain and cultivation procedure on the performanceof simultaneous saccharification and fermentation of steampretreated spruce,” Enzyme and Microbial Technology, vol. 38,no. 1-2, pp. 279–286, 2006.
[22] M. Galbe and G. Zacchi, “A review of the production ofethanol from softwood,” Applied Microbiology and Biotechnol-ogy, vol. 59, no. 6, pp. 618–628, 2002.
[23] A. Rudolf, M. Alkasrawi, G. Zacchi, and G. Liden, “Acomparison between batch and fed-batch simultaneous sac-charification and fermentation of steam pretreated spruce,”Enzyme and Microbial Technology, vol. 37, no. 2, pp. 195–204,2005.
[24] K. Stenberg, M. Bollok, K. Reczey, M. Galbe, and G.Zacchi, “Effect of substrate and cellulase concentration onsimultaneous saccharification and fermentation of steam-pretreated softwood for ethanol production,” Biotechnologyand Bioengineering, vol. 68, no. 2, pp. 204–210, 2000.
[25] T. Juhasz, Z. Szengyel, K. Reczey, M. Siika-Aho, and L. Viikari,“Characterization of cellulases and hemicellulases producedby Trichoderma reesei on various carbon sources,” ProcessBiochemistry, vol. 40, no. 11, pp. 3519–3525, 2005.
[26] Z. Szengyel, G. Zacchi, A. Varga, and K. Reczey, “Cellulaseproduction of Trichoderma reesei Rut C 30 using steam- pre-treated spruce. Hydrolytic potential of cellulases on differentsubstrates,” Applied Biochemistry and Biotechnology A, vol. 84–86, pp. 679–691, 2000.
[27] K. Kovacs, G. Szakacs, and G. Zacchi, “Comparative enzy-matic hydrolysis of pretreated spruce by supernatants, wholefermentation broths and washed mycelia of Trichoderma reeseiand Trichoderma atroviride,” Bioresource Technology, vol. 100,no. 3, pp. 1350–1357, 2009.
[28] D. J. Schell, N. D. Hinman, C. E. Wyman, and P. J. Werdene,“Whole broth cellulase production for use in simultaneoussaccharification and fermentation,” Applied Biochemistry andBiotechnology, vol. 24-25, pp. 287–297, 1990.
[29] X. Liming and S. Xueliang, “High-yield cellulase productionby Trichoderma reesei ZU-02 on corn cob residue,” BioresourceTechnology, vol. 91, no. 3, pp. 259–262, 2004.
[30] K. Kovacs, G. Szakacs, and G. Zacchi, “Enzymatic hydrolysisand simultaneous saccharification and fermentation of steam-pretreated spruce using crude Trichoderma reesei and Tricho-derma atroviride enzymes,” Process Biochemistry, vol. 44, no.12, pp. 1323–1329, 2009.
[31] A. Sluiter, B. Hames, R. Ruiz, C. Scarlata, J. Sluiter, andD. Templeton, Determination of structural carbohydrates andlignin in biomass, Laboratory Analytical Procedure, NationalRenewable Energy Laboratory, Golden, Colo, USA, 2004.
[32] M. Larsson, M. Galbe, and G. Zacchi, “Recirculation ofprocess water in the production of ethanol from softwood,”Bioresource Technology, vol. 60, no. 2, pp. 143–151, 1997.
[33] J. Olsson and G. Zacchi, “Simulation of the condensatetreatment process in a kraft pulp mill,” Chemical Engineeringand Technology, vol. 24, no. 2, pp. 195–203, 2001.
[34] R. Doppelbauer, H. Esterbauer, W. Steiner, R. M. Lafferty, andH. Steinmuller, “The use of lignocellulosic wastes for produc-tion of cellulase by Trichoderma reesei,” Applied Microbiologyand Biotechnology, vol. 26, no. 5, pp. 485–494, 1987.
[35] H. Esterbauer, W. Steiner, I. Labudova, A. Hermann, and M.Hayn, “Production of Trichoderma cellulase in laboratory andpilot scale,” Bioresource Technology, vol. 36, no. 1, pp. 51–65,1991.
[36] Z. Szengyel and G. Zacchi, “Effect of acetic acid and furfural oncellulase production of Trichoderma reesei RUT C30,” AppliedBiochemistry and Biotechnology A, vol. 89, no. 1, pp. 31–42,2000.
[37] R. Wooley and V. Putsche, “Development of an ASPENPLUS physical property database for biofuel components,”Tech. Rep. NREL/MP-425-20685, National Renewable EnergyLaboratory, 1996.
[38] M. S. Peters and K. D. Timmerhaus, Plant Design andEconomics for Chemical Engineers, McGraw-Hill, New York,NY, USA, 1980.
[39] J. D. Wright, “Ethanol from biomass by enzymatic hydrolysis,”Chemical Engineering Progress, vol. 84, no. 8, pp. 62–74, 1988.
V
RESEARCH Open Access
Techno-economic evaluation of stillagetreatment with anaerobic digestion in asoftwood-to-ethanol processZsolt Barta1, Kati Reczey1, Guido Zacchi2*
Abstract
Background: Replacing the energy-intensive evaporation of stillage by anaerobic digestion is one way ofdecreasing the energy demand of the lignocellulosic biomass to the ethanol process. The biogas can be upgradedand sold as transportation fuel, injected directly into the gas grid or be incinerated on-site for combined heat andpower generation. A techno-economic evaluation of the spruce-to-ethanol process, based on SO2-catalysed steampretreatment followed by simultaneous saccharification and fermentation, has been performed using thecommercial flow-sheeting program Aspen Plus™. Various process configurations of anaerobic digestion of thestillage, with different combinations of co-products, have been evaluated in terms of energy efficiency and ethanolproduction cost versus the reference case of evaporation.
Results: Anaerobic digestion of the stillage showed a significantly higher overall energy efficiency (87-92%), basedon the lower heating values, than the reference case (81%). Although the amount of ethanol produced was thesame in all scenarios, the production cost varied between 4.00 and 5.27 Swedish kronor per litre (0.38-0.50 euro/L),including the reference case.
Conclusions: Higher energy efficiency options did not necessarily result in lower ethanol production costs.Anaerobic digestion of the stillage with biogas upgrading was demonstrated to be a favourable option for bothenergy efficiency and ethanol production cost. The difference in the production cost of ethanol between using thewhole stillage or only the liquid fraction in anaerobic digestion was negligible for the combination of co-productsincluding upgraded biogas, electricity and district heat.
BackgroundEthanol produced from sugar, starch and lignocellulosicbiomass is a liquid biofuel with the potential to replacesome of the liquid fossil fuels used in transportationtoday. Currently, bio-ethanol is produced from sugar-and starch-containing materials [1]. However, it is clearthat the large-scale use of ethanol as fuel will requirelignocellulosic biomass to be used as raw material [2].The conversion of lignocellulosic material to ethanol ismore complex than ethanol production from sugar orstarch. Although pilot-scale and pre-commercial demon-stration plants have been brought into operation
recently [3-5], the process concept has not yet beendemonstrated on an industrial scale.Many process alternatives have been proposed for the
production of ethanol from lignocellulosic materials. Themain difference between them is the way in which cellu-lose and hemicellulose are hydrolysed to fermentablesugars [6-8]. A process based on enzymatic hydrolysisand fermentation is considered to be a promising alterna-tive for the conversion of lignocellulosic carbohydrates toethanol [6,9]. Compared with separate enzymatic hydro-lysis and fermentation, simultaneous saccharification andfermentation (SSF) has been shown to be less capitalintensive and to result in higher overall ethanol yields[10-12]. In order to obtain a high conversion of cellulosein enzymatic hydrolysis the raw material must be pre-treated [13]. One of the most thoroughly investigated
* Correspondence: [email protected] of Chemical Engineering, Lund University, PO Box 124,S 221 00 Lund, SwedenFull list of author information is available at the end of the article
Barta et al. Biotechnology for Biofuels 2010, 3:21http://www.biotechnologyforbiofuels.com/content/3/1/21
© 2010 Barta et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative CommonsAttribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction inany medium, provided the original work is properly cited.
methods is steam pretreatment, with or without a catalyst[14-17].The economics of the lignocellulosic ethanol process
is highly dependent on the income from co-products[18,19]. During the downstream process the lignin-richsolid residue can be separated, dried and pelletized. Pel-lets are sold as solid fuel on the residential pellet market[20]. Alternatively, steam and electricity can be gener-ated by burning the solid residue together with the con-centrated liquid fraction of the stillage. When thewastewater streams are treated with anaerobic digestion(AD) and aerobic treatment steps, the biogas producedand the sludge formed can also be incinerated on-site ina combined heat and power (CHP) facility [21,22].Above a certain level of biogas production, upgrading ofthe biogas can also be an option and the upgraded gascan then be sold as a transportation fuel or injecteddirectly to the gas grid [23,24]. As the overall energydemand (for both heat and electricity) of the lignocellu-losic ethanol process decreases the energy output in theform of co-products [20], it must be reduced as muchas possible and a high degree of heat integration istherefore required. Replacing energy-intensive processsteps by less energy-demanding ones, such as replacingevaporation of the stillage stream by AD, can furtherdecrease the heat demand of the process [21].In this study, the techno-economic aspects of a
spruce-to-ethanol process have been investigated using aprocess concept based on SO2-catalysed steam pretreat-ment followed by SSF. In Sweden, spruce is consideredto be the main alternative as raw material in the conver-sion of lignocellulosic biomass to ethanol, due to itsabundance and the relatively high content of carbohy-drates [25,26]. This evaluation was focused on investi-gating alternative options for stillage treatment, such asAD of the liquid fraction of the stillage or AD of thewhole stillage stream. Furthermore, the ways in whichvarious process configurations, with different combina-tions of co-products, affect the overall process in termsof energy efficiency and production cost has also beenstudied. Sensitivity analyses were performed with regardto the variations in the price of electricity, upgraded bio-gas, pellets and district heat in order to obtain a widerview of the relation between the scenarios from an eco-nomic perspective. The primary aim was not to deter-mine an absolute ethanol production cost but todevelop a useful modelling tool for comparing differentprocess scenarios, for identifying possible process bottle-necks and for the identification of future research activ-ities that have the greatest potential to lower the cost ofbio-ethanol production. Comparisons of the costobtained in this evaluation with those reported in simi-lar studies applying other assumptions should be per-formed with great care. Differences in technological (for
example, capacity, recoveries, yields) and/or in financialparameters (for example, interest rate, depreciation per-iod) can render such comparisons invalid.
MethodsProcess description - the reference caseThe process scheme for the reference scenario is shownin Figure 1. Each step has been previously described indetail [18] and will only briefly be discussed here, focus-ing mainly on minor modifications. The proposed etha-nol plant is assumed to be located in Sweden and toconvert 200 000 dry tonnes of spruce chips into49 416 m3 of ethanol annually. It is run by 28 employ-ees and is assumed to be in operation for 8000 h/year.Live steam is assumed to be available at 20 and 4 barand secondary steam is used to replace live steam when-ever possible.The dry matter (DM) of spruce contains 43.5% glucan,
12.8% mannan, 2.1% galactan, 5.1% xylan, 1.5% arabinanand 29.4% lignin, all determined by compositional analy-sis performed in the EU project NILE (contract No.019882), according to the standardized method of theNational Renewable Energy Laboratory (NREL, CO,USA) [27]. The remainder is made up of acetyl groups,extractives and other compounds, which were estimatedusing the data of a previous study [19]. The DM contentis assumed to be 50%. Theoretically, 420 L of ethanolcan be produced from the hexose sugars available in atonne of dry raw material.The conversion of carbohydrates is carried out in
steam pretreatment and in SSF. The conditions usedfor steam pretreatment (210°C, 2.5% SO2, conversionfactors for some reactions: glucan to glucose 0.161,glucan to hydroxymethylfurfural 0.013, xylan to xylose0.674, xylan to furfural 0.220, water-insoluble lignin towater-soluble lignin 0.082) and SSF (conversion factorsfor some reactions: glucan to glucose 0.756, xylan toxylose 0, glucose to ethanol 0.9, glucose to glycerol0.005) were based on results recently obtained fromexperimental work performed at the Department ofChemical Engineering, Lund University, Sweden(unpublished). Part of the evaporation condensate andammonia are added before pressing the pretreatedslurry, in order to adjust the dry matter in the SSFstep to 10% water-insoluble solids (WIS) and to neu-tralize the slurry, respectively. Simultaneous saccharifi-cation and fermentation is performed at 37°C withordinary baker’s yeast at a concentration of 3 g DM/Land an enzyme dosage of 10 FPU (filter paper unit)/gWIS. It takes place in 12 agitated non-sterile fermen-tors each with a volume of 920 m3. Yeast is cultivatedon the pressed liquid fraction of the diluted pretreatedslurry, supplemented with molasses, while the enzymesare purchased.
Barta et al. Biotechnology for Biofuels 2010, 3:21http://www.biotechnologyforbiofuels.com/content/3/1/21
Page 2 of 11
The ethanol concentration obtained after SSF is3.5 wt-%. Distillation and molecular sieve adsorption areused to produce pure (99.8 wt-%) ethanol. The distilla-tion step consists of two stripper columns and a recti-fier, which are heat integrated by operating at differentpressures. Ethanol recovery is assumed to be 99.5% ineach column.The stillage of the stripper columns is separated in a
filter press resulting in a solid fraction with a WIS con-tent of 40%. In any scenario where pellets are produced,washing is included in the stillage separation step inorder to decrease the sulphur content in the solid frac-tion. It is assumed that washing - with a soluble solidrecovery of 90% in the liquid stream - is sufficient tomeet the requirement regarding the sulphur content ofpellets.The liquid fraction of the stillage is concentrated to
60% DM in an evaporation system containing five effectsin a forward-feed arrangement. Boiling point elevationwas taken into account [28] and overall heat transfercoefficients were estimated to vary between 400 and2000 W/m2°C. Based on the work of Olsson et al. [29],it is assumed that, by applying a stripper column afterevaporation to remove volatile compounds, part of theevaporation condensate could be recycled to dilute thewhole slurry without affecting SSF. The rest of the con-densate, together with the condensed flash streams ori-ginating from pretreatment and drying, is treated by ADfollowed by an aerobic treatment step. These steps aredescribed below.Steam and electricity are generated by burning the
concentrated liquid fraction, part of the solid fraction ofthe stillage, the biogas and the sludge. The generatedsteam is allowed to expand to 4 bar through a high-pressure turbine system. However, part of the steam iswithdrawn at 20 bar for pretreatment and drying.
District heating was not included in the reference case.The excess solid residue (87% of the total) - the solidfraction not required for steam generation, is dried in asuperheated steam dryer to 88% DM and then pelle-tized. The secondary steam generated by drying isutilized in the process.
Alternative stillage treatment scenariosIn the case of alternative stillage treatment the heat-demanding evaporation plant was excluded from theethanol process. These scenarios are summarized inFigure 2 and Table 1. In scenarios A1 to A4 the stillageis separated in a filter press (described above) and theliquid stream is treated by AD in order to producebiogas. The effluent from AD is subjected to an aerobictreatment step. In scenario B the whole stillage is trea-ted directly by AD. However, the effluent of AD is sepa-rated and only the liquid fraction is fed to the aerobictreatment step. The same type of filter press as thatused for stillage separation is used and it is assumedthat the WIS content of the solid fraction is 35%. Thissolid fraction, comprising lignocellulosic residue andanaerobic sludge, is assumed to be burnt in the CHPplant. In all alternative treatment scenarios sludgeseparation after the aerobic treatment is performed witha belt filter, which increases the sludge DM to 30%. Thepressed sludge is incinerated on-site.Anaerobic digestion is performed under mesophilic
conditions and, hence, the inlet flow is cooled down to37°C before being fed to the first digester. The assumeddegradation factors during AD are: (i) 90% for solublesugars, organic acids, ethanol, glycerol, enzyme andyeast; (ii) 50% for polysaccharides, extractives, degrada-tion products and water-soluble lignin; and (iii) 0% forwater-insoluble lignin. The methane and AD sludgeyields are assumed to be 0.35 Nm3/kg COD (chemical
Figure 1 Overall process scheme for the proposed ethanol plant in the reference case. Part of the evaporation condensate, together withthe condensed flash streams originating from pretreatment and drying, is anaerobically digested followed by an aerobic treatment step. Mat:material; SSF: simultaneous saccharification and fermentation.
Barta et al. Biotechnology for Biofuels 2010, 3:21http://www.biotechnologyforbiofuels.com/content/3/1/21
Page 3 of 11
oxygen demand) removed and 0.03 kg DM/kg COD fed,respectively. In order to obtain the appropriate levels ofN and P, solutions of NH3 (25%) and H3PO4 (50%) areadded in dosages of 18 g/kg and 4 g/kg COD, respec-tively. An organic loading rate of 10 kg COD/m3/day isapplied and, hence, the hydraulic retention time variesbetween 6.2 and 9.9 days. The specific power requiredfor stirring depends on the scenario and is assumed tobe 20 W/m3 for scenarios A1 to A4 and 40 W/m3 forscenario B. The digesters, arranged in series, are contin-uous stirred tank reactors with a volume of around 3500m3 each and their number varies between 2 and 12,including the reference scenario. As the water-insoluble
lignin was considered to be inert in terms of biogas pro-duction, its COD was not included in the COD used tocalculate the reactor volume and the demand of chemi-cals. The biogas is assumed to consist of 50 wt-%methane, 46% CO2 and 4% water. The design data andcost for AD were obtained from a supplier of waste-water treatment plants with experience in the treatmentof wastewater from the pulp industry (PURAC AB,Lund, Sweden), based on calculated flows and estimatedCOD content. Biogas upgrading is performed in somescenarios where the biogas is sold as a co-product sup-plied to the gas grid. In these scenarios all the biogasproduced is upgraded using pressure swing adsorption
Figure 2 Alternative stillage treatment scenarios. Either the liquid fraction of the stillage is anaerobically digested (A), or the whole stillage isfed directly to anaerobic digestion (B). The dotted parts are optional, see Table 1 for details. The wastewater streams, such as the condensedflash streams from pretreatment and drying, are also sent to anaerobic digestion but they are not shown here. BF: belt filter; FP: filter press.
Table 1 Differences in stillage processing in the scenarios investigated
Scenario Washing anddrying
Pellet production Biogasupgrading
Turbinesystem
Burnt in CHP (besidessludge)
Co-products*
DH
Reference Yes Part of sfrac of stillage(dried)
No HP Part of sfrac of stillage, syrup,bg
Pellets No
A1 Yes Part of SF Yes HP Part of SF, tail gas Up bg,pellets
No
A2 Yes TA of SF No HP-LP Bg Pellets Yes
A3 No - Yes HP-LP Sfrac of stillage, tail gas Up bg Yes
A4 No - No HP-LP Sfrac of stillage, bg - Yes
B - - Yes HP-LP Sfrac of eff, tail gas Up bg Yes
*The electricity generated is not included in co-products. Although it is produced in all scenarios, in some cases electricity consumption exceeds production.
Bg: biogas; CHP: combined heat and power; DH: district heating; eff: effluent of anaerobic digestion; HP: high-pressure; LP: low-pressure; SF: solid fuel; Sfrac: solidfraction; TA: total amount; Up: upgraded.
Barta et al. Biotechnology for Biofuels 2010, 3:21http://www.biotechnologyforbiofuels.com/content/3/1/21
Page 4 of 11
technology. However, 5% of the methane is assumed tobe retained in the tail gas that is burnt on-site.The whole effluent from AD (scenarios A1 to A4), or
the separated liquid stream from the AD effluent (sce-nario B), is first treated aerobically. The organic matteris removed almost entirely and sludge is produced at ayield of 0.3 kg sludge DM/kg organic matter. The laststep is ozone treatment, which degrades some phenoliccompounds that have not been broken down in theaerobic treatment. The effluent from the ozone treat-ment is regarded as clean water and is recycled to theprocess partially or entirely depending on the scenario.The recycled water stream is sterile-filtered before beingused to dilute the pretreated slurry. However, thecost of this filter was not included in the economicevaluation.Similarly to the reference case, in scenario A1 the
CHP plant only covers the heat demand of the process,while in scenarios A2 to A4 and B excess heat is pro-duced that is used to generate electricity through anadditional low-pressure turbine with a discharge pres-sure of 0.75 bar. The condensation heat of the outletsteam from the low-pressure turbine and the heatobtained in flue gas condensation are used for districtheating. Detailed descriptions of the turbine system andflue gas condensation can be found elsewhere [20]. Thereturn water of 6 bar from the district heating system isheated up from 45°C to 90°C by passing through theflue gas condenser and the turbine condenser.
AnalysisMass and energy balances were solved using the com-mercial flow sheeting program Aspen Plus™, version2006.5 (Aspen Technology, Inc, MA, USA). Data on thephysical properties of biomass components such as poly-saccharides and lignin were taken from the NREL data-base [30]. Aspen HX-Net, version 2006.5 (AspenTechnology) was used to design a near-optimal heatexchanger network, which was implemented in the pro-cess model in Aspen Plus. The energy efficiency, basedon the lower heating values, is defined as the energyoutput in the products (ethanol, pellets, biogas, electri-city and district heat) divided by the energy input com-prising raw material, molasses, enzymes and the fuelequivalent of the electric power requirement, which wascalculated using an electricity-to-fuel ratio of 0.4.The fixed capital investment cost was estimated either
with Aspen Icarus Process Evaluator, version 2006.5(Aspen Technology) setting 2009 as costing year or fromvendor quotations. The construction material for all pro-cess vessels is assumed to be 304 stainless steel. Workingcapital was calculated using the recommendation ofPeters and Timmerhaus [31] with a slight modification.The annualized fixed capital cost was determined by
multiplying the fixed capital investment by an annualiza-tion factor of 0.110, corresponding to a 15-year deprecia-tion period and an interest rate of 7% [32]. Theannualized working capital is the product of workingcapital investment and interest rate.All costs are presented in Swedish kronor (SEK,
1 euro ≈ 10.5 SEK; US$ 1 ≈ 7.3 SEK). The purchaseprice of enzymes is assumed to be 28.5 SEK/millionFPU, which was obtained by multiplying an old estimateof cellulase price [33] by 1.5. The costs of raw material,chemicals, utilities, labour, insurance and maintenance,and the revenues from co-products have been reportedin recent studies [19,20]. The total cost of aerobic andozone treatment, the cost of biogas upgrading and theselling price of upgraded biogas are assumed to be0.5 SEK/kg COD, 100 and 600 SEK/MWh upgradedbiogas, respectively. Other costs comprise labour, insur-ance and maintenance.
Results and discussionProcess design and energy efficiencyImportant process details of the scenarios are given inTable 2. The largest biogas production was obtained inscenario B, due to the increased flow of polysacchar-ides, soluble sugars and other degradable components.In the reference case the organic matter fed to ADconsisted of only volatile substances originating fromthe pretreatment, evaporation and drying steps. Theoverall degradation factor - the COD removed dividedby the COD fed - including or excluding the water-insoluble lignin, varied between 28% and 60% orbetween 58% and 66%, respectively. The slight differ-ence in the biogas production in scenarios A1 to A4was due to the drying step, since the condensed outletsteam from the dryer increased the COD flow ofothers for scenarios A1 and A2.Removing the evaporation step decreased the heat
duty of the process considerably, from 27.8 MW to17.1-17.9 MW. The ethanol production was the same inall scenarios (6177 L/h); hence the specific heatdemands were reduced from 16.2 to 9.9-10.4 MJ/L etha-nol. The minor increase in the overall heat dutyobtained by removing the drying step from the process(scenarios A3 and A4 compared with scenarios A1 andA2) is not inconsistent. On the one hand, it is due tothe different heat-exchanger networks with or withoutdrying. On the other hand, the superheated steam dry-ing produces almost as much heat as it requires (datanot shown) and the secondary steam leaving the dryingstep can be utilized at other stages in the process.The electricity produced on-site varied considerably,
unlike the total power requirement of the process. Thehighest electricity production was obtained for scenarioA4, where both the biogas and the solid residue are
Barta et al. Biotechnology for Biofuels 2010, 3:21http://www.biotechnologyforbiofuels.com/content/3/1/21
Page 5 of 11
burnt in the CHP plant. In scenarios A1 and A2 thepower produced was lower than the power demand inthe process, thus requiring the purchase of electricity,while in all other scenarios a surplus of electricity wasproduced that could be sold. The largest pellet produc-tion was obtained for scenario A2, where, unlike otherscenarios, the entire solid residue was pelletized. In thecase of district heating, the more heat that can be deliv-ered, the more difficult it is to find an appropriate plantlocation. There are currently less than 10 district heat-ing systems in Sweden [20] that match the district heatduty for scenario A4 (48.9 MW). Scenarios with districtheat production of around 30 MW may be easier toimplement as there are more systems operating on thisscale.The reference case had significantly lower overall
energy efficiency (81%) than the alternative stillage treat-ment scenarios (87-92%), see Figure 3. This was due tothe difference in the overall heat demand, since thereduction in the heat demand of the ethanol processincreased the energy output in the co-products. Theenergy efficiency obtained for ethanol varied between33% and 34%, which emphasizes the importance of co-products in an energy-efficient process. If the upgradedbiogas was used as a fuel as well as the ethanol, 51%-58% of the energy input would be recovered as trans-portation fuel.
All the scenarios including district heating (A2 to A4,B) had a higher overall efficiency than scenarios withoutdistrict heating (reference case and A1). Concordantresults were obtained in a similar previous study, inwhich various process configurations for ethanol pro-duction from spruce, based on evaporation of the liquidfraction of stillage, were investigated [20]. The overallenergy efficiency of the scenarios including district heat-ing varied only slightly (89%-92%), whereas the energyefficiency would differ significantly (46%-85%) if the dis-trict heating was excluded from these scenarios (forexample, during the summer, when the released heathas to be removed by cooling water).
Capital investment and ethanol production costsThe lowest and highest total capital investment costswere obtained for scenarios A1 (1199 million SEK) andA4 (1360 million SEK), respectively (Table 3). The maincontributors to the total direct cost were the pretreat-ment unit, the fermentation stage (yeast cultivation andSSF) and the CHP plant. The highest cost for CHP wasobtained in scenario A4, where it constituted 44% of thetotal direct cost. Anaerobic digestion had the highestdirect cost in scenario B, where the whole stillagestream was anaerobically digested. When the COD flowof lignin was included in the COD flow used to designAD, the direct cost of AD increased to 225 million SEK
Table 2 Process details of the various scenarios
Reference A1 A2 A3 A4 B
Organic matter to anaerobic digestion t COD/h 1.9 9.7 9.7 9.5 9.5 30.9
Polysaccharides t COD/h - 0.1 0.1 0.1 0.1 2.4
Soluble sugars t COD/h - 2.2 2.2 2.2 2.2 2.4
WIL t COD/h - 0.9 0.9 0.9 0.9 17.8
Others* t COD/h 1.9 6.5 6.5 6.3 6.3 8.2
Organic matter removed t COD/h 1.1 5.8 5.8 5.7 5.7 8.5
COD removed/COD fed incl. WIL % 58 60 60 60 60 28
COD removed/COD fed excl. WIL % 58 66 66 66 66 65
Energy flow† of raw biogas MW 3.8 19.9 19.9 19.5 19.5 29.4
Energy flow† of upgraded biogas MW - 18.9 - 18.5 - 27.9
Overall heat duty of the process MW 27.8 17.6 17.6 17.9 17.9 17.1
Total power demand of the process MW 4.1 4.4 4.5 3.8 4.0 4.5
Electricity produced in CHP MW 5.9 3.0 4.0 11.7 17.1 9.8
District heat MW - - 4.7 34.1 48.9 31.1
Energy flow† of pellets‡ MW 48.5 39.3 56.2 - - -
The following input energy flows (in MW, based on lower heating values) were the same in all the scenarios: raw material 105.1; molasses 1.7; and enzymes 0.7.The positive and negative differences between power demand and electricity produced indicate purchasing or selling of electricity, respectively. A summary ofthe scenarios is given in Table 1.
* Includes organic acids, ethanol, glycerol, enzyme, yeast, extractives, degradation products and water-soluble lignin, depending on the scenario.† Based on lower heating values.‡Only the fraction of the solid fuel sold is given here, i.e. the fraction incinerated on-site is excluded.
CHP: combined heat and power production; COD: chemical oxygen demand; WIL: water-insoluble lignin.
Barta et al. Biotechnology for Biofuels 2010, 3:21http://www.biotechnologyforbiofuels.com/content/3/1/21
Page 6 of 11
in scenario B. The ratio of working capital to fixed capi-tal varied between 2.4% and 6.4% in the scenariosinvestigated.The major contributors to ethanol production cost
were the cost of capital, raw material and chemicals, indescending order (Figure 4), whereas the utilities andother costs made minor contributions. The alternativestillage treatment increased the cost of capital, chemicalsand other costs compared with the reference case. Thehighest cost of utilities was obtained for scenariosA1 and A2, where electricity had to be purchased.The favourable combinations for the co-products ofthe alternative stillage treatment scenarios were:(i) upgraded biogas and pellets (scenario A1); and (ii)upgraded biogas, electricity and district heat (scenariosA3 and B). However, the combination of (a) pellets anddistrict heat (scenario A2) and (b) electricity and districtheat (scenario A4) resulted in lower co-product income.With respect to the production cost of ethanol, the alter-
native stillage treatment configurations proved to be goodoptions in some cases. Scenarios A1, A3 and B resulted inlow ethanol production costs (4.00-4.24 SEK/L), whilescenarios A2 and A4 were less economical (5.27 and 5.02SEK/L, respectively). Scenario A2 had an even higher etha-nol production cost than the reference case (5.14 SEK/L).Scenarios A1 and A2 had almost the same costs. However,
Figure 3 Overall energy efficiency, based on lower heating values (LHV), expressed as percentage of the input. A summary of thescenarios is given in Table 1. REF: reference case.
Table 3 Breakdown of the total capital investment costin million Swedish Kronor (SEK)
Reference A1 A2 A3 A4 B
Raw material handling 10 10 10 10 10 10
Pretreatment 115 115 115 115 115 115
Yeast cultivation and SSF 119 119 119 119 119 119
Distillation 47 47 47 47 47 47
Separation1 26 26 26 26 26 25
Evaporation 47 - - - - -
Drying and pelletproduction
42 45 47 - - -
CHP2 155 110 136 266 343 248
Storage 33 28 28 28 28 29
Heat exchanger network 18 11 11 11 11 12
AD 15 75 75 74 74 111
Total direct cost 628 586 613 696 773 716
Total indirect cost 578 551 557 539 556 557
Fixed capital investment 1206 1137 1170 1236 1329 1273
Working capital 69 62 75 32 32 32
Total capital investment 1275 1199 1246 1268 1360 13051 Refers to stillage separation in the reference case and in scenarios A1-A4and to separation of the effluent of AD in scenario B.2 Includes the flue gas condenser for scenarios with district heating (A2-A4and B).
A summary of the scenarios is given in Table 1.
AD: anaerobic digestion; CHP: combined heat and power production; SSF:simultaneous saccharification and fermentation.
Barta et al. Biotechnology for Biofuels 2010, 3:21http://www.biotechnologyforbiofuels.com/content/3/1/21
Page 7 of 11
they differed greatly in their co-product incomes. Produ-cing 4.7 MW district heat at a price of 280 SEK/MWh and56.2 MW pellets at 195 SEK/MWh (scenario A2) resultedin much less income than producing 18.9 MW upgradedbiogas with a selling price of 600 SEK/MWh and anupgrading cost of 100 SEK/MWh together with 39.3 MWpellets (scenario A1; Table 2).The performance of the AD step is very uncertain due
to a lack of experimental data on the continuous treat-ment of stillage from a wood-based ethanol process.Therefore two sub-scenarios were developed fromscenario B and the ethanol production cost was deter-mined. In one sub-scenario the COD flow of water-insoluble lignin was included in the COD flow used todesign the anaerobic digesters and for calculating thedemand of chemicals in AD (however, the conversion ofwater-insoluble lignin was maintained at 0%) and theethanol production cost increased to 4.71 SEK/L, com-pared with 4.00 SEK/L in scenario B. In the other sub-scenario it was assumed that the polysaccharides in thesolid fraction were not converted to biogas - theirdegradation factors were set to 0%. Since the COD flowinto AD was the same as that in scenario B, the designof the AD step remained unaltered. Although the biogasproduction decreased by 4%, the electricity generated
and the district heat produced increased by 14% and12%, respectively, which resulted in the same overallenergy efficiency as for scenario B and an ethanol pro-duction cost of 4.28 SEK/L. Although the ethanol pro-duction cost in the two sub-scenarios increasedcompared with scenario B, it did not exceed the ethanolproduction cost in the reference case (5.14 SEK/L).
Sensitivity to co-product pricesThe sensitivity of ethanol production cost to changes inthe prices of the co-products was monitored by chan-ging the price of one co-product at a time from -40% to+40%. The results are shown in Figure 5. The effect ofchanging price of the co-products depends on the shareof the given co-product with regard to the total income:the higher the share the greater the effect. The electri-city price had the highest impact on scenario A4, fol-lowed by scenarios A3 and B. However, scenarios A3and B had the lowest ethanol production cost in almostthe whole range investigated. Without selling electricitycertificates (200 SEK/MWh) scenario A1 would be themost profitable if the spot price of electricity decreasedbelow 260 SEK/MWh. Scenario B was that most affectedby the price of biogas. Even with a 40% lower price ofbiogas the scenarios including biogas upgrading (A1,
Figure 4 Breakdown of ethanol production cost in SEK/L. Chemicals include enzymes. ‘Others’ refers to the cost of labour, insurance andmaintenance. The cost of aerobic and ozone treatment was added to the capital cost: the cost of upgrading was taken into account byreducing the income of upgraded biogas. A summary of the scenarios is given in Table 1. REF: reference case; SEK: Swedish kronor; Upgr:upgraded.
Barta et al. Biotechnology for Biofuels 2010, 3:21http://www.biotechnologyforbiofuels.com/content/3/1/21
Page 8 of 11
A3, B) resulted in lower ethanol production costs thanthe scenarios including biogas burning (reference case,A2, A4), although the difference was much smaller. Theincome from biogas could also decrease due to a lowerproduction of biogas than that assumed. However, aslong as the product of the amount of biogas producedand the price does not decrease by more than 40%, thescenarios with biogas upgrading are more favourablefrom an economic point of view. However, there isnothing to suggest that the raw biogas productionassumed in scenarios A1 to A4 is overestimated. In sce-nario B, it has also been shown that assuming zerodegradation of the polysaccharides in AD resulted in aslight decrease in biogas production.The pellet price had the highest impact on scenario
A2. For a pellet price higher than 235 SEK/MWh sce-nario A1 would result in a lower ethanol productioncost than any of the other scenarios. Finally, the districtheat price had a considerable influence on scenarios A4,
A3 and B, in descending order. Scenarios A3 and B hadthe lowest ethanol production cost when the price ofdistrict heat was higher than 195 SEK/MWh, whereasbelow this value scenario A1 became the most econom-ical. Hence, scenario A1 would have the lowest ethanolproduction cost, in the case where there was no incomefrom district heating - for example, the plant locationwould not be appropriate for district heating.
ConclusionsA techno-economic model of anaerobic digestion of thestillage stream has been developed based on the indivi-dual degradation of each component and was used tostudy different process configurations with various co-product combinations in a spruce-to-ethanol process. Inconcordance with the results obtained by Wingren et al.[21], the heat demand of the scenarios with alternativestillage treatment (with anaerobic digestion of the stil-lage) was significantly lower than that for the reference
Figure 5 Ethanol production cost (SEK/L) as a function of co-product prices. Co-products: (A) electricity, (B) upgraded biogas, (C) pelletsand (D) district heat. A summary of the scenarios is given in Table 1. REF: reference case; SEK: Swedish kronor.
Barta et al. Biotechnology for Biofuels 2010, 3:21http://www.biotechnologyforbiofuels.com/content/3/1/21
Page 9 of 11
case, which included evaporation of the liquid fractionof the stillage. Due to the reduced process heat demand,the overall energy efficiency was improved in the alter-native stillage treatment scenarios compared with thereference case. The highest energy efficiencies wereobtained in the scenarios including district heating.Although the energy efficiencies of these scenarios var-ied slightly, their ethanol production cost differed signif-icantly. Hence, it can be concluded that high energyefficiency does not necessarily result in improved eco-nomics. The implementation of district heating enableshigh energy efficiency by utilizing the heat available atlow temperatures. However, it restricts the location ofthe plant as there must be a demand for the heatavailable.The prices of co-products may vary considerably and,
therefore, sensitivity analyses were performed in orderto monitor the change in ethanol production cost. Twoof the scenarios, which had the same co-product combi-nation (upgraded biogas, electricity and district heat),but different feeds to the AD (only the liquid fraction ofthe stillage or the whole stillage stream), resulted in thelowest ethanol production cost over a wide range of co-product prices. These scenarios responded very similarlyto changes in the co-product prices, resulting in almostidentical ethanol production costs - with this combina-tion of co-products the feed to AD does not influencethe process economics significantly. If the price of pel-lets is increased to a certain level, or district heating cannot be implemented, the scenario including the produc-tion of pellets and upgraded biogas becomes the mostprofitable. Hence, it can be concluded that alternativestillage treatment with biogas upgrading proved to be afavourable option in many respects.There is still a need for the further development of
the model regarding the wastewater treatment stage inboth AD and the following aerobic step. It would beuseful to implement the management of the inorganiccompounds that originate from the raw material orwhich are added for pH adjustment. This is essentialfor the recirculation of the process water in order toavoid the accumulation of these compounds in theprocess.
AbbreviationsAD: anaerobic digestion; CHP: combined heat and power; COD: chemicaloxygen demand; DM: dry matter: FPU: filter paper unit; SEK: Swedish kronor;SSF: simultaneous saccharification and fermentation; WIS: water-insolublesolid.
AcknowledgementsThe authors gratefully acknowledge the 6th Framework Programme of theEuropean Commission (NILE project, contract No. 019882) and the NationalResearch Fund of Hungary (OTKA K72710) for their financial support.
Author details1Department of Applied Biotechnology and Food Science, BudapestUniversity of Technology and Economics, Szt Gellért tér 4, H 1111 Budapest,Hungary. 2Department of Chemical Engineering, Lund University, PO Box124, S 221 00 Lund, Sweden.
Authors’ contributionsZB carried out the process simulations and economic evaluation and wrotethe paper. ZB and GZ designed the study and analysed the results. GZ andKR reviewed the manuscript. All authors read and approved the finalmanuscript.
Competing interestsThe authors declare that they have no competing interests.
Received: 13 June 2010 Accepted: 15 September 2010Published: 15 September 2010
References1. Gray KA, Zhao LS, Emptage M: Bioethanol. Curr Opin Chem Biol 2006,
2:141-146.2. Farrell AE, Plevin RJ, Turner BT: Ethanol can contribute to energy and
environmental goals. Science 2006, 5760:506-509.3. Lignocellulosic ethanol pilot plant in Salamanca, Spain (Abengoa
Bioenergy Inc). 2010 [http://www.abengoabioenergy.com/corp/web/en/acerca_de/oficinas_e_instalaciones/bioetanol/europa/biocarburantes_cast_leon/index.html].
4. Lignocellulosic ethanol pilot plant in Ottawa, Canada (Iogen Inc). 2010[http://www.iogen.ca/company/facilities/index.html].
5. Lignocellulosic ethanol pilot plant in Örnsköldsvik, Sweden (SEKAB E-Technology AB). 2010 [http://www.sekab.com/default.asp?id=2018&refid=2030].
6. Galbe M, Zacchi G: A review of the production of ethanol from softwood.Appl Microbiol Biotechnol 2002, 6:618-628.
7. Lenihan P, Orozco A, O’Neill E, Ahmad MNM, Rooney DW, Walker GM:Dilute acid hydrolysis of lignocellulosic biomass. Chem Eng J 2010,2:395-403.
8. von Sivers M, Zacchi G: A techno-economical comparison of threeprocesses for the production of ethanol from pine. Bioresour Technol1995, 1:43-52.
9. Wooley R, Ruth M, Glassner D, Sheehan J: Process design and costing ofbioethanol technology: a tool for determining the status and directionof research and development. Biotechnol Prog 1999, 5:794-803.
10. Soderstrom J, Galbe M, Zacchi G: Separate versus simultaneoussaccharification and fermentation of two-step steam pretreatedsoftwood for ethanol production. J Wood Chem Technol 2005, 3:187-202.
11. Wingren A, Galbe M, Zacchi G: Techno-economic evaluation of producingethanol from softwood: Comparison of SSF and SHF and identificationof bottlenecks. Biotechnol Prog 2003, 4:1109-1117.
12. Tomas-Pejo E, Oliva JM, Ballesteros M: Realistic approach for full-scalebioethanol production from lignocellulose: a review. J Sci Ind Res 2008,11:874-884.
13. Alvira P, Tomas-Pejo E, Ballesteros M, Negro MJ: Pretreatment technologiesfor an efficient bioethanol production process based on enzymatichydrolysis: a review. Bioresour Technol 2010, 13:4851-4861.
14. Ohgren K, Zacchi G, Galbe M: Optimization of steam pretreatment ofSO2-impregnated corn stover for fuel ethanol production. Appl BiochemBiotechnol 2005, 121:1055-1067.
15. Cara C, Ruiz E, Ballesteros M, Manzanares P, Negro MJ, Castro E: Productionof fuel ethanol from steam-explosion pretreated olive tree pruning. Fuel2008, 6:692-700.
16. Sipos B, Reczey J, Somorai Z, Kadar Z, Dienes D, Reczey K: Sweet sorghumas feedstock for ethanol production: enzymatic hydrolysis of steam-pretreated bagasse. Appl Biochem Biotechnol 2009, 1:151-162.
17. Barta Z, Oliva JM, Ballesteros I, Dienes D, Ballesteros M, Reczey K: Refininghemp hurds into fermentable sugars or ethanol. Chem Biochem Eng Q2010, 3:301-309.
18. Barta Z, Kovacs K, Reczey K, Zacchi G: Process design and economics ofon-site cellulase production on various carbon sources in a softwood-based ethanol plant. Enzyme Res 2010.
Barta et al. Biotechnology for Biofuels 2010, 3:21http://www.biotechnologyforbiofuels.com/content/3/1/21
Page 10 of 11
19. Sassner P, Galbe M, Zacchi G: Techno-economic evaluation of bioethanolproduction from three different lignocellulosic materials. BiomassBioenergy 2008, 5:422-430.
20. Sassner P, Zacchi G: Integration options for high energy efficiency andimproved economics in a wood-to-ethanol process. Biotechnol Biofuels2008, 1:4-4.
21. Wingren A, Galbe M, Zacchi G: Energy considerations for a SSF-basedsoftwood ethanol plant. Bioresour Technol 2008, 7:2121-2131.
22. Wilkie AC, Riedesel KJ, Owens JM: Stillage characterization and anaerobictreatment of ethanol stillage from conventional and cellulosicfeedstocks. Biomass Bioenerg 2000, 2:63-102.
23. Abatzoglou N, Boivin S: A review of biogas purification processes. BiofuelBioprod Bior 2009, 1:42-71.
24. Greer D: Fundamentals of biogas conditioning and upgrading. BioCycle2010, 2:27-30.
25. Tengborg C, Stenberg K, Galbe M, Zacchi G, Larsson S, Palmqvist E, Hahn-Hagerdal B: Comparison of SO2 and H2SO4 impregnation of softwoodprior to steam pretreatment on ethanol production. Appl BiochemBiotechnol 1998, 70-72:3-15.
26. Soderstrom J, Pilcher L, Galbe M, Zacchi G: Combined use of H2SO4 andSO2 impregnation for steam pretreatment of spruce in ethanolproduction. Appl Biochem Biotechnol 2003, 105:127-140.
27. Sluiter A, Hames B, Ruiz R, Scarlata C, Sluiter J, Templeton D: Determinationof structural carbohydrates and lignin in biomass. Laboratory AnalyticalProcedure Colorado: National Renewable Energy Laboratory 2004.
28. Larsson M, Galbe M, Zacchi G: Recirculation of process water in theproduction of ethanol from softwood. Bioresour Technol 1997, 2:143-151.
29. Olsson J, Zacchi G: Simulation of the condensate treatment process in akraft pulp mill. Chem Eng Technol 2001, 2:195-203.
30. Wooley R, Putsche V: Development of an ASPEN PLUS physical propertydatabase for biofuel components. Technical Report Colorado: Nationalrenewable Energy Laboratory 1996.
31. Peters MS, Timmerhaus KD: Plant Design and Economics for ChemicalEngineers. New York: McGraw-Hill 1980.
32. Levin HM, McEvan PJ: Cost-effectiveness Analysis: Methods andApplications. CA: Sage 2000.
33. von Sivers M, Zacchi G: Ethanol from wood - A techno - Economicevaluation of the enzymatic hydrolysis process. LUTKDH/(TKKA-7006) 1993,1-27.
doi:10.1186/1754-6834-3-21Cite this article as: Barta et al.: Techno-economic evaluation of stillagetreatment with anaerobic digestion in a softwood-to-ethanol process.Biotechnology for Biofuels 2010 3:21.
Submit your next manuscript to BioMed Centraland take full advantage of:
• Convenient online submission
• Thorough peer review
• No space constraints or color figure charges
• Immediate publication on acceptance
• Inclusion in PubMed, CAS, Scopus and Google Scholar
• Research which is freely available for redistribution
Submit your manuscript at www.biomedcentral.com/submit
Barta et al. Biotechnology for Biofuels 2010, 3:21http://www.biotechnologyforbiofuels.com/content/3/1/21
Page 11 of 11
