Vous êtes sur la page 1sur 9

Biotechnol. Prog.

2003, 19, 1109−1117 1109

REVIEW
Techno-Economic Evaluation of Producing Ethanol from Softwood:
Comparison of SSF and SHF and Identification of Bottlenecks
Anders Wingren, Mats Galbe, and Guido Zacchi*
Department of Chemical Engineering, Lund University, P.O. Box 124, S-221 00 Lund, Sweden

The aim of the study was to evaluate, from a technical and economic standpoint,
the enzymatic processes involved in the production of fuel ethanol from softwood.
Two base case configurations, one based on simultaneous saccharification and
fermentation (SSF) and one based on separate hydrolysis and fermentation
(SHF), were evaluated and compared. The process conditions selected were based
mainly on laboratory data, and the processes were simulated by use of Aspen plus.
The capital costs were estimated using the Icarus Process Evaluator. The ethanol
production costs for the SSF and SHF base cases were 4.81 and 5.32 SEK/L or 0.57
and 0.63 USD/L (1 USD ) 8.5SEK), respectively. The main reason for SSF being
lower was that the capital cost was lower and the overall ethanol yield was higher. A
major drawback of the SSF process is the problem with recirculation of yeast follow-
ing the SSF step. Major economic improvements in both SSF and SHF could be
achieved by increasing the income from the solid fuel coproduct. This is done by
lowering the energy consumption in the process through running the enzymatic
hydrolysis or the SSF step at a higher substrate concentration and by recycling
the process streams. Running SSF with use of 8% rather than 5% nonsoluble
solid material would result in a 19% decrease in production cost. If after distillation
60% of the stillage stream was recycled back to the SSF step, the production cost
would be reduced by 14%. The cumulative effect of these various improvements
was found to result in a production cost of 3.58 SEK/L (0.42 USD/L) for the SSF
process.

Contents Introduction
Introduction 1 In Sweden there is an increasing interest in using
Materials and Methods 3 ethanol as a neat or blended fuel in the transportation
sector as a substitute for fossil fuels such as gasoline and
Process Description 3
diesel. In 2001 the first full-scale plant in Sweden for
General Process and Design Data 4 producing ethanol from a renewable raw material was
Simulation Software 4 started in the city of Norrköping. It has the capacity of
Economics 4 producing 50 000 m3 of ethanol per year to be blended
5 with gasoline up to a level of 5% ethanol. The plant uses
Results and Discussion
wheat as the major raw material. If larger quantities of
Base Cases 5 the gasoline are to be replaced, a cheap and abundant
Sensitivity Analysis 5 raw material needs to be employed. In Sweden lignocel-
Improvements in the Base Cases 6 lulosic materials such as softwood are readily available
Effect of Yield 6 in the form of logging waste and waste from the forest
industry. It has been estimated that there is sufficient
Initial Substrate Loading in SSF 6 biomass available for 26% of the gasoline or 40% of the
and EH Steps diesel used in the transportation sector to be replaced
Process Integration 7 by ethanol (1). Since such biomass is a renewable and
Effect of Residence Time in SSF 7 indigenous raw material, its use would also make Sweden
and EH Steps on Ethanol Cost less dependent on imported gasoline.
Plant Capacity 7 The reasons for using ethanol from renewable sources
Cumulative Effect of 7 as a fuel are many. In contrast with fossil oil, the net
Improvements
Conclusions 8 * To whom correspondence should be addressed. Tel: +46 (0)-
46 82 97. Fax +46 (0)46 45 26. E-mail: guido.zacchi@chemeng.lth.se.

10.1021/bp0340180 CCC: $25.00 © 2003 American Chemical Society and American Institute of Chemical Engineers
Published on Web 06/10/2003
1110 Biotechnol. Prog., 2003, Vol. 19, No. 4

Anders K. Wingren is a PhD student at the Department


of Chemical Engineering at Lund University in Sweden,
where he also received his MsChE in 1999 with focus on
Guido Zacchi obtained his PhD in Chemical Engineering process design. Current research activities involve techno-
in 1979 at Lund University. A full professor in Chemical economic evaluations of ethanol production processes from
Engineering at Lund University since 1989, he was ap- various biomass resources. Mr. Wingren also teaches un-
pointed as Dean of research for Chemistry and Chemical dergraduate courses given by the Department of Chemical
Engineering for 1997-2000 and since January 2000 has Engineering.
been a member of the Board for Lund Institute of Technol-
ogy. He is also Head of the Chemical Engineering depart-
ment at Lund University. A major part of his research has
been directed to the field of biochemical engineering with sulfuric acid and hydrochloric acid to hydrolyze the
projects in the production of ethanol from lignocellulosic cellulose and hemicellulose to sugars. A drawback of the
materials, production of polymers from renewable raw concentrated acid process, however, is the costly recovery
materials, mass transfer in biotechnical systems, math-
ematical modelling, process simulation, and separation
of the acid.
processes. Zacchi has supervised about 20 students that have Another method is steam pretreatment of the biomass,
graduated with a PhD and is at present supervisor for 9 PhD followed by enzymatic hydrolysis (EH) of the remaining
students. He has also been supervisor for more than 60 cellulose. During steam pretreatment the wood is sub-
Master students. He has published more than 100 scientific jected to high-pressure steam and to acidic conditions,
papers in international journals with peer review. which leads to the hydrolysis of the hemicellulose and
makes the cellulose more accessible to subsequent enzy-
matic attack.
Although the enzymatic process is known to give yields
of ethanol higher than those of the acid process (7), there
is a larger degree of uncertainty connected with the
former process than with the latter. The cost of the
enzymes used in the process is an example of this.
Presently, they contribute significantly to the cost of
ethanol, over 20% without recycling being taken into
account (8). Thus, the cost would need to be reduced
considerably to make the process economically viable. In
the long run the cost of the enzymes is likely to drop as
a result of both increasing efficiency in the production of
the enzymes and higher production volumes due to
Mats J. Galbe is a Senior Researcher at the Department
applications in other areas such as textiles, detergents,
of Chemical Engineering at Lund University in Sweden. He
received his MScChE and his PhD from Lund University. and pulp (8).
He has extensive experience in research on the production The two dominating process configurations that use
of ethanol from various lignocellulosic sources. He is cur- enzymes for saccharification are separate hydrolysis and
rently in charge of a national process-development unit for fermentation (SHF) and simultaneous saccharification
ethanol production situated in Lund and also acts as and fermentation (SSF). SSF has been regarded as the
supervisor for PhD students. major option here because for various substrates and
under varying pretreatment conditions it results in
higher yields and also shorter residence times (9-11).
emission of carbon dioxide is almost zero (2) since the Two drawbacks of the SSF process, however, at least
carbon dioxide emitted during ethanol combustion is when softwood is utilized as substrate, are the formation
assimilated when new biomass is produced. Also, the of lactic acid (12) and the inability to recycle yeast.
emission of other compounds, such as nitrogen oxides, The purpose of the present techno-economic evaluation
carbon monoxide, and volatile organic compounds (VOCs), was to compare the SSF and SHF process configurations
is reduced (3). and determine where bottlenecks in the two processes
Many process alternatives for the production of fuel lie. The evaluation of these two base cases, SSF and SHF,
ethanol from lignocellulosic materials have been pro- involved as much as possible use of laboratory data for
posed. The main differences between these concern the yields and other process conditions. The impact that
way in which cellulose and hemicellulose are hydrolyzed changes in different process parameters have on the
to fermentable sugars. The dilute acid process (4, 5) and production cost was also studied with the aim of sug-
the concentrated acid process (6) employ such acids as gesting possible improvements that could be made in the
Biotechnol. Prog., 2003, Vol. 19, No. 4 1111

Figure 1. The two base cases, SSF and SHF. SSF: simultaneous saccharification and fermentation. SHF: separate hydrolysis and
fermentation. EH: enzymatic hydrolysis.

future. These include the achieving of a higher overall In the pretreatment step, other components, besides
ethanol yield, carrying out SSF and EH using higher those described above, are also formed. The chemistry is
initial substrate concentrations, reducing the residence very complex, and most of the compounds produced are
times required in SSF and EH, and increasing the unknown. These unknowns were described in the simu-
capacity of ethanol plants. The approach of recycling lations in terms of three components formed from the
process streams to achieve better process integration was degradation of hexoses, pentoses, and lignin, respectively.
evaluated from a technical and economic standpoint. In Ten percent of the lignin was assumed to be degraded or
the cases considered, the streams prior to and after solubilized. For the hexosans and pentosans that did not
distillation were recycled to the hydrolysis or the SSF form monomers, HMF, or furfural, the conversion was
step, reducing the need for fresh water. A sensitivity assumed to be such that 2% remained unchanged after
analysis was also conducted for evaluating the impact of pretreatment.
the cost and/or the load of enzymes on the overall After pretreatment the concentrations of nonsoluble
production cost of ethanol. The study aimed neither at solids (solid fraction remaining after filtration and wash-
determining the absolute costs of ethanol production nor ing) are high, and water is needed to dilute the feedstock
an economic optimization in production of it, but rather into the EH or the SSF units. For each of the two base
at comparing the production costs of the two processes cases examined the water added is considered to be fresh
on an equal basis and studying the effects of different water, the concentration of nonsoluble solid material
parameters on the ethanol production costs. being set to be 5% (w/w). The use of fresh water and the
low substrate concentration in the SSF and EH steps are
Materials and Methods scarcely feasible from a techno-economic standpoint,
Process Description. Figure 1 shows the SSF and because large effluent streams would be created and the
SHF processes that were evaluated in the study. capital costs and energy consumption in the downstream
The raw material was assumed to consist of spruce processing steps would be high. However, this water and
with a water content of 50%; 45% of the dry matter is solid mixture was chosen since most of the laboratory
glucan, 13% is mannan, and 2% is galactan. Pentosans data is based on the use of fresh water and a 5%
and lignin constitute 7% and 28%, respectively. The concentration. Recycling the process streams and the use
balance is made up of ash, extractives, and acetyl groups. of a higher concentration of nonsoluble solid material are
It is assumed that the woodchips are transported to options considered in the study.
the plant by truck and that in the feedstock handling In the SHF process, pretreatment is followed by EH.
area (not shown) the chips are reduced in size if needed The data used in the study is based on the work of
and are stored in a stack before being fed to the Tengborg et al. (15). The residence time was set to 96 h,
pretreatment area. the temperature to 38 °C, and the yield of glucose from
The pretreatment reactor is assumed to work as a cellulose to 69% at an enzyme activity of 19 FPU/g
StakeTech reactor (13). The woodchips are first impreg- cellulose. The total time for filling, cleaning, and empty-
nated with sulfur dioxide and are then subjected to high- ing was assumed to be 12 h. The reactors were assumed
pressure steam. After pretreatment, the pressure is to be agitated tanks, each 1000 m3 in size and with a
reduced in a two-step procedure, being first reduced to 4 working volume of 800 m3 operated in a simple batch
bar and then to atmospheric pressure. Thus, some of the mode. Agitators with a power consumption of 20 W/m3
flash vapor is available at 144 °C and can be used as a (16) ensure good mixing.
heating medium in the distillation or evaporation steps. In the filtration and washing steps 99% of the solubi-
In the pretreatment step most of the hemicellulose is lized material are recovered. The high recovery can be
broken down to monomeric sugars and byproducts. The attributed to the fact that most of the solubilized material
data on process conditions and yields is based on the work is recovered in the filtration step prior to the addition of
of Tengborg (14). The amount of SO2 used was 2.4 kg/ water in the washing step. Filtration and washing are
100 kg dry wood. The residence time, at a temperature performed in a two-stage counter-current procedure, the
of 215 °C, was 3 min. Some 15.5% of the pentosans are washing curve being based on a washing model developed
converted to furfural, and 1.3% of the hexosans to for the recovery of lignin from pulp residue (17). The
5-hydroxy-2-methylfurfural (HMF). Sixty-five percent of equipment involved is assumed to be rotating-drum
the mannan and galactan and 70% of the xylan and vacuum filters. The solid stream leaving the filter
arabinan are converted to monomers. Fourteen percent contains 40% nonsoluble solid material.
of the cellulose is broken down to glucose. Acetic acid is In the fermentation step of the SHF process the
obtained from the acetyl groups in the wood at a hexoses are converted to ethanol, as well as to byproducts
fractional conversion of 97%. and yeast. The fractional conversion of the hexoses to
1112 Biotechnol. Prog., 2003, Vol. 19, No. 4

ethanol was set to 88%. Five percent of the glucose is Table 1. Costs Used in Evaluation
converted to glycerol, and the remaining glucose would price unit
be used for the production of cell mass. No pentose
raw material (spruce) 440 SEK/ton
fermentation was considered in either the SSF or the utilities
SHF case. The residence time was 24 h. The size of the process water 1.4 SEK/m3
fermentor and the 12 h needed for cleaning and filling cooling water 0.14 SEK/m3
where the same as for the hydrolysis step. After fermen- electricity 200 SEK/MWh
chemicals
tation, the yeast is recycled by two centrifuges connected sulfur dioxide 1200 SEK/ton
in parallel, and no additional yeast is needed. ammonia 2000 SEK/ton
The data for the SSF step were taken from studies by base (lime) 1150 SEK/ton
Stenberg (11, 12). The SSF step is assumed to be run defoamer 20000 SEK/ton
sodium hydroxide 2200 SEK/ton
with use of the same type of fermentors as for the SHF enzymes 19 SEK/1 × 106 FPU
process, with a residence time of 48 h. All reactions that yeast 3100 SEK/ton
take place in EH and fermentation in the SHF base case byproduct credit
are also assumed to take place in SSF and to involve the solid fuel 706 SEK/ton
same conversion. For SSF, however, the conversion of carbon dioxide 100 SEK/ton
other costs
glucan to glucose is set to 86%. The amount of enzymes labor 500000 SEK/(employee year)
added, 32 FPU/g cellulose, is higher than in the SHF base maintenance 2 % of fixed capital
case. The yeast load was initially 4 g/L, somewhat less insurance 1 % of fixed capital
than the concentration used by Stenberg. Since the yeast
in the slurry leaving the fermentation step is difficult to of the streams involved in the process. The physical data
separate from the other solid components, it was assumed for the components was obtained either from the Aspen
that no recycling was possible. inhouse databank or from the literature. Data on most
The highly diluted ethanol solution needs to be con- of the wood components, such as cellulose and lignin,
centrated before it can be used as a liquid fuel. The were obtained from NREL’s databank on biomass com-
distillation section consists of a stripper connected to a ponents (21). The data not directly available was esti-
rectifier. The stripper has 20 stages and the rectifier 40, mated by use of rules of thumb or from data on similar
with a stage efficiency of 50% and 70%, respectively. The components.
stripper runs at a higher pressure than the rectifier. The Economics. Most of the equipment was sized with the
vapor leaving the stripper is used as a heating medium Icarus Process Evaluator (IPE) (22). The report file from
in the reboiler of the rectifier. The feed is preheated by Aspen Plus containing the results of energy and material
the stillage from the stripper. balances was used in the sizing calculations. IPE was
The stillage from the stripper, containing nonvolatiles used mainly for the sizing of heat exchangers, tanks
such as pentoses and HMF, must be concentrated before (except for the fermenters and the hydrolysis vessels),
it can be used as a fuel in the steam boiler or be sold as filters, evaporators, and columns. Some equipment was
solid fuel. Concentration is carried out in a 5-effect sized manually for capital cost evaluation. This was done
evaporator train, with a resulting concentrate of 50% for the equipment in the feedstock handling area and for
water content. the fermentor and hydrolysis vessels. The StakeTech
The sugars from the evaporator and the lignin from reactors were sized using data provided by Stake Tech-
the filtration step are mixed and are dried to a water nology (23).
content of 11% and thereafter incinerated to supply the IPE was also used for estimating the material costs
process with steam. Because lignin is a nonfossil fuel the and installation costs for nearly all equipment used in
internal steam production does not result in a net the process. Exceptions were the Stake Tech reactor, for
emission of carbon dioxide to the atmosphere. The solid which cost estimates were based on a quotation from
residue not incinerated is made into briquettes and sold Stake Technology (23), and the dryer, boiler, and bri-
as solid fuel. quette machine, for which data was obtained from the
The enzymes employed can either be purchased or literature (24, 25). IPE was also used to estimate the
produced at the plant, using different carbon sources such material costs and installation costs for auxiliary equip-
as prehydrolysate from the pretreatment or residual ment such as piping, insulation, electrical, etc.
cellulose from the SSF or the EH steps (18). Because The total direct cost (TDC) is the total cost of major
there are still many uncertainties in the production of equipment and auxiliary equipment, together with the
the enzymes, a lumped cost of 19 SEK (Swedish Kronor)/ cost of the buildings directly associated with the process.
million FPU (19) was employed (cost originally from Other costs associated with the project were then
discussion with an enzyme company), this cost being added, such as indirect construction costs, freight, engi-
varied then to determine the effect on the overall produc- neering, construction management, overhead, and con-
tion cost of the ethanol. tract fees. The contingencies were assumed to be 10% of
General Process and Design Data. The capacity for the fixed capital investment (FCI).
each of the two base cases considered is 195,600 dry ton Working capital was calculated, in the manner recom-
(metric) of raw material per annum, the plant being mended by Peters and Timmerhaus (26), as the sum of
assumed to be situated in the northern part of Sweden. the cost of the raw material and chemicals needed for 30
The plant has access to large quantities of cooling water days production, 30 days of finished products in stock,
and operates continuously 8,000 h per year. The first accounts receivable, and 30 days of cash on hand for
choice of construction material is assumed to be SS304 monthly payments of wages, raw material, etc. The
except the steam explosion equipment, which is made of yearly capital cost was calculated by use of an annuity
SS316. factor of 0.096, corresponding to a 15-year life of the plant
Simulation Software. The process was modeled by and a 5% rate of return, a straight-line deprecation and
Aspen Plus simulation software from Aspen Tech (20). zero scrap-value being assumed.
This software was used to solve the mass and energy Table 1 summarizes the operating costs employed in
balances and to calculate the thermodynamic properties the evaluation. The cost of the raw material was assumed
Biotechnol. Prog., 2003, Vol. 19, No. 4 1113

Table 2. Stream Data for Two Base Cases; All Values in ton/h
stream 1 2 SHF 3 SHF 4 SHF 5 SHF 6 SSF 3 SSF 4 SSF 5 SSF 6
hexoses 0.0 4.4 0.5 0.0 0.6 0.6 0.3 0.0 0.3 0.3
pentoses 0.0 1.4 1.4 0.0 1.4 1.4 1.4 0.0 1.4 1.4
ethanol 0.0 0.0 5.1 5.1 0.0 0.0 5.9 5.8 0.1 0.0
water 24.4 294.0 287.7 0.3 18.8 1.9 286.9 0.4 18.7 1.8
hexosans 15.2 9.4 0.0 0.0 3.0 3.0 0.0 0.0 1.4 1.4
pentosans 1.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
lignin 6.7 6.1 0.1 0.0 6.1 6.1 0.1 0.0 6.1 6.1
yeast 0.0 0.0 0.4 0.0 0.4 0.4 0.0 0.0 1.8 1.8
other 0.7 3.3 3.8 0.0 3.2 3.2 4.0 0.0 3.4 3.4
total (ton/h) 48.9 318.6 299.0 5.4 33.4 16.5 298.5 6.2 33.2 16.2

to be 440 SEK/dry ton or 84 SEK/MWh. Since the solid


fuel has a higher heat of combustion and a lower water
content than the raw material it has a higher value. An
income of 706 SEK/ton dry material was assumed cor-
responding to 125 SEK/MWh. The current cost of biofuels
such as briquettes and pellets is in Sweden around 160
SEK/MWh (transportation included). However, the cost
is likely to decrease if many ethanol plants are built.
The carbon dioxide produced in the fermentation step
was assumed to be sold. It is questionable whether this
would be realistic if many ethanol-from-wood plants were
established because the market for carbon dioxide in
Sweden could become saturated. These matters have no
major impact on the production cost of the ethanol,
however. Figure 2. Effect of the enzyme cost on the ethanol production
The plant was assumed to be operated by 25 persons, cost.
operators and supervisors included. A lumped wage rate Table 3. Capital Costs (MSEK) for Base Cases with
of 500,000 SEK/year per person, including social security Breakdown of Direct Capital Costs
charges, was used.
% of % of
SSF5% TDC SHF5% TDC
Results and Discussion
direct costs
Base Cases. Figure 1 shows the schematic flowsheet feedstock handling 11 2 11 2
for the two base cases considered. The data for some of pretreatment 110 20 110 16
the key streams involved is listed in Table 2. For the SSF SSF/EH 128 24 212 30
base case (SSFBC), the overall production of ethanol was filtration 15 3 15 2
found to be 7900 L/h or 63000 m3/year. For the SHF base fermentation 86 12
distillation 54 10 52 7
case (SHFBC) the production was 6900 L/h or 55000 m3/ evaporation 61 11 61 9
year. The yield of ethanol prior to distillation, expressed drying, steam generation 145 27 144 20
as the percentage of the amount that could theoretically and briquetting
be produced from the fermentable sugars (hexoses) storage 14 3 14 2
contained in the raw material, was 70% for the SSF case total direct cost (TDC) 538 705
and 62% for the SHF case. other costs 297 412
fixed capital investment (FCI) 835 1117
The power requirement of the steam boiler is 72.8 MW working capital 75 67
for SSF and 72.3 MW for SHF, the electric power needed total capital investment (TCI) 910 1184
being 5.24 MW and 6.28 MW, respectively. The high
energy consumption results in a very low production of material cost are the largest contributors to the overall
solid fuel. As already pointed out, this is due to the low production cost; see Table 4. The enzyme cost contributes
initial substrate concentration in the SSF and EH steps significantly to the overall cost for both the SSF and SHF
and to the fact that no streams are recycled. For both case. In the SSF base case, the yeast cost corresponds to
base cases, the most energy-demanding step is that of 0.5 SEK/L ethanol. Thus, yeast recycling would be very
evaporation followed by distillation. beneficial for the SSF process.
The total capital investment was estimated to be 910 The solid residue remaining after the production of
MSEK (million Swedish kronor) for the SSF case and process steam is negligible, the credit being only 0.01
1184 MSEK for the SHF case. Table 3 shows a break- SEK/L in both base cases. The production of carbon
down of the total capital investment. For the SSF case dioxide reduces the cost by 0.07 SEK/L, for both SSF and
the steam generation area, including the dryer, the SHF.
burner, and the boiler, has the largest contribution, 27%, Sensitivity Analysis. As already indicated, both the
to the total direct cost, the SSF step having the second cost of enzymes and the enzyme requirements are still
largest. For the SHF process, EH is the most expensive very uncertain. A sensitivity analysis was carried out to
process step, representing 30% of the direct cost, steam study the effect that variations in the cost of the enzymes
generation being the second most expensive. would have on the ethanol production cost. The result is
The overall ethanol production cost for the base cases shown in Figure 2. If the enzyme cost was reduced by
were estimated to be 4.81 and 5.32 SEK/L for the SSF 50%, the ethanol cost would be reduced to 4.46 and 5.08
and SHF processes, respectively. Thus, the production SEK/L for SSF and SHF, respectively. Thus, SSF is more
cost in the SHF case is 11% higher than in the SSF case. sensitive to variations in the enzyme cost due to the
For both SSF and SHF, the capital cost and the raw higher enzyme requirement that applies to the base case.
1114 Biotechnol. Prog., 2003, Vol. 19, No. 4

Table 4. Cost Breakdown for Two Base Cases and Some Improvements Suggesteda
SSFBC SHFBC SHFIY2 SSF8 SHF8 SSF560A SSF540B
wood 1.36 1.57 1.36 1.36 1.57 1.36 1.36
enzymes 0.70 0.48 0.42 0.70 0.48 0.70 0.70
yeast 0.50 0.00 0.00 0.31 0.00 0.50 0.50
byproducts -0.08 -0.08 0.01 -0.49 -0.61 -0.59 -0.52
other chemicals and utilites 0.35 0.43 0.37 0.29 0.35 0.28 0.30
labor, maintenance, insurance 0.59 0.84 0.73 0.54 0.72 0.57 0.57
capital 1.39 2.08 1.82 1.20 1.70 1.32 1.30
total (SEK/L) 4.81 5.32 4.71 3.91 4.21 4.15 4.21
a SSFBC: SSF base case. SHFBC: SHF base case. SHFIY2: SHF process with the same ethanol yield as SSFBC. SSF8 and SHF8: 8%
nonsoluble solid material in the SSF and EH steps, respectively. SSF560A: SSF process where 60% of the stream after distillation is
recycled. SSF540B: SSF process where 40% of the stream before distillation is recycled.

Improvements in Base Cases. Effect of Yield. The


SHF alternative suffers from its low overall yield of
ethanol. Accordingly, the effect of an increase in yield
on the production cost was studied for the SHF process.
The yield was increased from the base case value of 62%
to 65% and 70%, respectively. The higher yield was
assumed to be due to a higher fractional conversion of
the cellulose in the EH, increasing to 75% and 86%,
respectively. The highest yield is the same as that used
in the SSF base case. The ethanol cost was reduced from
the base case value of 5.32 to 5.09 and 4.71 SEK/L,
respectively. Table 4 presents a cost breakdown for the Figure 3. Effect of increasing the total substrate loading in
case of the highest overall yield (SHFIY2) together with the SSF and EH steps on ethanol cost.
the base cases, as well as for various of the other process
improvements discussed below. If the overall yield of
ethanol for the SHF process could be as high as the SSF
base case the production cost would be lower for SHF,
which would have a higher capital cost but a lower
enzyme cost, the cost for yeast being zero. The increase
in yield would require a higher enzyme concentration,
however, which was not taken into account.
Initial Substrate Loading in SSF and EH Steps.
In both base cases the stream going to the SSF step or
to the EH step is diluted to consist of only 5% nonsoluble
solid material. This results in the stream going to
downstream processing in the plant being large in volume
and highly diluted. This would have a considerable effect
on the capital cost of the plant since larger and thus more
Figure 4. Effect of increasing the total substrate loading in
expensive equipment will be required. Also, the energy the SSF and EH steps. Ethanol production cost vs final ethanol
consumption depends on the level of dilution. This is the concentration.
case for distillation and evaporation, in particular, since
these process steps involve liquid removal by evaporation. reduction for the SSF8 case is due to the increase in the
Stenberg et al. (11) studied the effect of increasing the production of solid fuel to 5500 kg/h, which contributes
initial concentration of nonsoluble solid material in SSF. to almost 46% of the reduction in cost. The reduction in
The concentration was varied between 2% and 10% at a capital cost and yeast cost contribute by about 20% each.
cellulase activity of 32 FPU/g cellulose. At 10%, no
ethanol production at all occurred, but at 7.5% the yield Running the process at a higher concentration of
of ethanol was about as high as when it was run at 5%, nonsoluble solid material makes the SSF and the EH
although the time required for the highest ethanol yield steps more difficult. This is due to the higher concentra-
to be reached was somewhat longer. Since there is a clear tion of inhibitory substances or to the higher concentra-
potential for improvement here, studying the effect of tion of fibrous material, which could reduce the mass
differences in concentration on the overall production cost transfer rate. For EH another problem would be end-
is of genuine interest. product inhibition of the cellulases due to a higher sugar
This effect was studied by simulating the SSF and the concentration. Using more powerful agitators might
SHF process at higher initial concentrations of nonsoluble possibly increase the mass transfer rate. Another ap-
solid material than the 5% that applied to the base cases, proach would be to run the SSF step or the EH step in a
with concentrations as high as 8% (SSF8 and SHF8). The fed-batch mode. The advantage of a fed-batch as opposed
same overall yield of ethanol was assumed as in the two to a simple batch mode is that the initial concentration
base cases. When the initial loading of substrate in- of nonsoluble solid material can be kept at a low level
creases, the production costs are reduced considerably, and this will have a positive effect on the agitation. The
the SSF8 case by nearly 19% compared with the SSF base concentration of inhibitors is also likely to be less than
case (Figure 3). Figure 4 shows how the final ethanol in the corresponding batch process because several
concentration, due to the higher substrate loading, affects compounds, e.g., furfural and HMF, are converted to less
the ethanol cost, and a cost breakdown for the SSF8 and inhibitory compounds by the yeast during fermentation
SHF8 cases is shown in Table 4. The largest cost (27). End-product inhibition of the cellulases in the EH
Biotechnol. Prog., 2003, Vol. 19, No. 4 1115

Figure 6. Effect of the residence times in the EH step for the


Figure 5. Effect of process stream recirculation on the ethanol SHF process and the SSF step for the SSF process on the
cost of the SSF process. R1: recycling before distillation. R2: ethanol cost.
recycling after distillation.
reduction. For both alternatives, the most important
step, however, will not be reduced in a fed-batch process contribution to the cost reduction was the increase in
compared to batch. byproduct credit for the solid fuel due to the reduction
Process Integration. In the two base cases the water in the energy demand of downstream processing. For the
added to the process was assumed to be fresh water. An SSF540B case this production was 5800 kg/h or 0.24 kg/
alternative would be to replace the fresh water stream kg dry raw material and for the SSF560A it was 6800
by recycled process water. Two possibilities regarding this kg/h or 0.28 kg/kg dry raw material. The reduction in
have been considered: capital also reduced the overall production cost.
(1) Recycling of the stream prior to distillation. This Effect of Residence Time in SSF and EH Steps on
alternative leads in both SSF and SHF to an increase in Ethanol Cost. Since the SSF and EH steps contribute
the concentration of ethanol as well as of the inhibitory significantly to the overall capital cost in both base cases,
components (R1 in Figure 1). the effects of a reduction in residence time on the
(2) Recycling of the stream after distillation. This would production costs was investigated. In the SSF case, two
result in an increase in the concentration of nonvolatile alternatives were evaluated: the residence time being
substances alone (R2 in Figure 1). reduced from 48 to 36 and 24 h, respectivley. In the SHF
Recycling could affect the performance of the SSF step case, hydrolysis was simulated by use of a residence time
and the EH step negatively through reducing the yield of 72, 48, and 24 h. Reducing the residence time by 50%
and/or the rate. Recycling prior to distillation reduces the resulted in an ethanol production cost of 4.65 and 4.93
size of the streams to both the distillation and the SEK/L for the SSF and the SHF cases, respectively
evaporation step, thereby decreasing the capital cost as (Figure 6). Thus, the cost reduction that is possible is
well as the energy consumption in both of these steps. greater in the SHF case.
On the other hand, recycling after distillation reduces Plant Capacity. The effect of plant capacity was
the capital cost and the energy consumption in the investigated by doubling the plant capacity in the two
evaporation step only. The two alternatives were simu- base cases. All of the costs for raw material, utilities, etc.,
lated using 40% and 60% recycling of the streams in the as well as the number of employees, were assumed to be
SSF case. Although no evaluation of the SHF case was unaffected. For SSF the overall capital cost increased to
made, the results for SHF can be expected to be similar 1570 MSEK, making this cost 73% higher than for the
to those for the SSF case. base case. The SHF case increased to 2055 MSEK, which
The possibility of recycling the process streams has is 74% higher than for the base case. The cost for ethanol
been demonstrated by Stenberg et al. (28) in performing production was reduced to 4.48 and 4.84 SEK/L for SSF
SHF experiments and by Alkasrawi et al. (29) in per- and SHF, respectively. This decrease was due to the
forming similar studies of SSF. In the latter investigation decrease in capital cost and in indirect costs such as
it was shown that a 60% reduction in the addition of labor, insurance, and maintenance. All other costs re-
freshwater could be achieved by recycling the stillage mained unchanged on a per liter of ethanol basis, since
stream, without affecting either the overall yield of they are linearly dependent upon the capacity of the
ethanol or the rate of fermentation. When 40% of the plant. However, the cost for the raw material could
freshwater was replaced by the distillation stream, the increase, since the collection area increases with the
yield was unaffected, although the initial rate was increase in capacity. Since this effect is uncertain, we
decreased somewhat. If still more water was replaced by chose not to include it in the evaluation.
recycled streams, the overall yield of ethanol was de- Cumulative Effect of Improvements. The actual
creased. production process is likely to be a combination of the
Figure 5 shows the results of the simulations and of concepts discussed above. Accordingly, it is of interest to
the economic evaluations. The overall production costs evaluate the simultaneous effect of the most promising
were reduced by as much as 18% and 14%, respectively, alternatives. As the results of screening studies per-
for 60% recycling prior to (R1) and after distillation (R2). formed in our process development unit indicate (30), the
Table 4 presents a breakdown of the costs for the case in process should be able to run at a higher initial concen-
which 40% of the stream was recycled prior to distillation tration of nonsoluble solid material, while at the same
(SSF540B) and the case in which 60% of it was recycled time a part of the stillage stream could be recycled. An
after distillation (SSF560A). A recycling of 40% prior to SSF-based process run at 8% nonsoluble solid material
distillation reduced the freshwater stream, F1, by 43%, having the same yields as the SSF base case was
whereas 60% recycling after distillation yielded a 58% simulated. Twenty-six percent of the stillage stream was
1116 Biotechnol. Prog., 2003, Vol. 19, No. 4

recycled, resulting in a fresh water addition equivalent (2) Lave, L.; MacLean, H.; Hendrickson, C.; Lankey, R. Life
to a 10% concentration of nonsoluble solid material in Cycle Analysis of Alternative Automobile Fuel/Propulsion
the SSF step, without recycling occurring. Technologies. Environ. Sci. Technol. 2000, 34, 3598-3605.
Since an adapted yeast would work much better than (3) Wyman, C. E. Ethanol Production from Lignocellulosic
Biomass: Overview. In Handbook on Bioethanol: Production
the yeast employed in the present study, a possible 50% and Utilization; Wyman, C. E., Ed.; Taylor and Francis Inc.:
reduction in yeast was also included in this scenario. Bristol, PA, 1996; pp 1-18.
The production cost was estimated at 3.73 SEK/L (4) Nagle, N.; Ibsen, K.; Jennings, E. A Process Economic
without the reduction in the amount of yeast added being Approach To Develop a Dilute-Acid Cellulose Hydrolysis
taken into account. If the concentration of yeast could Process to Produce Ethanol from Biomass. Appl. Biochem.
be reduced by 50%, the cost would drop to 3.58 SEK/L. Biotechnol. 1999, 77-79; 595-607.
The power requirement of the steam boiler was reduced (5) Fransson, G.; Lindstedt, J.; Sundlöf, L.; Wiklund, K.-G.
to 38.3 MW, which is a reduction of 47% as compared Dilute Acid Hydrolysis of Cellulose In Pilot- and Fullscale,
with the SSF base case. This produces a large excess of Report 2000-002:1, EnergiCentrum Norr; Report written in
Swedish.
solid fuel, 7200 kg/h, resulting in a net income of 0.54
(6) Barrier, J. W.; Moore, M. R.; Farina, G. E.; Broder, J. D.;
SEK/L ethanol.
Forsythe, M. L.; Lightsey, G. R. Experimental production of
ethanol from agricultural cellulosic materials using low-
Conclusions temperature acid hydrolysis. Biomass Energy Dev.; Proceed-
ings of the 3rd Southern Biomass Energy Research Confer-
According to the data used in the evaluation, use of
ence; Plenum Press: New York, 1986; pp 587-600.
the SSF process results in a lower ethanol production
(7) Galbe, M.; Zacchi, G. A Review of the Production of Ethanol
cost. The production costs were estimated to 4.81 and from Softwood. Appl. Microbiol. Biotechnol. 2002, 59, 618-
5.32 SEK/L for the SSF and SHF base cases, respectively. 628.
The main reasons for the cost of the SHF process being (8) Gregg, D. J.; Boussaid, A.; Saddler, J. N. Techno-Economic
higher are higher capital cost and lower overall ethanol Evaluation of Generic Wood-To-Ethanol Process: Effect of
yield. If the yield of ethanol were the same, the ethanol Increased Cellulose Yields and Enzyme Recycle. Bioresour.
cost would be lower for SHF than for SSF. Thus, one can Technol. 1998, 63, 7-12.
conclude that the SHF process is by no means inap- (9) Wright, J. D.; Wyman, C. E.; Grohmann, K. Simultaneous
propriate. In fact, Söderström et al. (31) has shown that Saccharafication and Fermentation of Lignocellulose: Process
for two-step pretreatment the SHF configuration gives Evaluation. Appl. Biochem. Biotechnol. 1988, 18, 75-90.
a slightly higher yield than SSF does. However, the EH (10) Wyman, C. E.; Spindler, D. D.; Grohmann, K. Simulta-
was in their study performed at a lower concentration of neous Saccharafication and Fermentation of several Ligno-
cellulosic Feedstocks to Fuel Ethanol. Biomass Bioenergy
nonsoluble solid material (2%) than was SSF (5%). 1992, 3(5), 301-307.
The income from the solid fuel is crucial to the (11) Stenberg, K.; Bollók, M.; Réczey, K.; Galbe, M.; Zacchi, G.;
economics of both the SSF and the SHF process. Since Effect of Substrate and Cellulase Concentration in Simulta-
the base cases investigated require large amount of neous Saccharification and Fermentation (SSF) of Steam-
energy, the production of solid fuel they provide is low. pretreated Softwood for Ethanol Production. Biotechnol.
Experiments have shown that it is possible to reduce the Bioeng. 2000, 68(2), 204-210.
addition of fresh water either by running the SSF or EH (12) Stenberg, K.; Galbe, M.; Zacchi, G. The Influence of Lactic
step at a higher substrate concentration or by recycling Acid Formation on the Simultaneous Saccharafication and
of the process streams both alternatives with a consider- Fermentation (SSF) of Softwood to Ethanol. Enzyme Microb.
Technol. 2000, 26(1), 71-79.
able reduction in the cost of ethanol.
(13) Stake Technology Ltd., Norval, Ontario, Canada.
A major drawback of the SSF process is the addition (14) Tengborg, C. Bioethanol Production: Pretreatment and
of yeast required. The residue slurry leaving the SSF step Enzymatic Hydrolysis of Softwood. Ph.D. Thesis, Lund Uni-
contains, along with the yeast, other solids such as lignin versity, 2000, Report LUTKDH/(TKKA-1004).
and cellulose. Recirculation is thus very difficult to (15) Tengborg, C.; Galbe, M.; Zacchi, G. Influence of Enzyme
perform, although if it could be performed it would reduce Loading and Physical Parameters on the Enzymatic Hydroly-
the production cost significantly. Another option is, of sis of Steam-Pretreated Softwood. Biotechnol. Prog. 2001,
course, to lower the concentration of yeast. This can be 17(1), 110-117.
done by using an adapted yeast. (16) Wooley, R.; Ruth, M.; Sheehan, J.; Ibsen, K. Lignocellulosic
The cumulative effect of the most promising alterna- Biomass to Ethanol Process Design and Economics Utilizing
tives was evaluated for the SSF process. The effect of a Co-Current Dilute Acid Prehydrolyze and Enzymatic Hy-
drolysis Current and Futuristic Cases. NREL/TP-580-26157,
combination of higher substrate concentration in the SSF 1999.
step, the recycling of process streams, and lower concen- (17) Grähs, L. E. Displacement Washing of Packed Beds of
tration of yeast was investigated. A realistic case would Cellulose Fibers, part 1. A Comparison of the Behaviour of
be a process involving 8% nonsoluble solid material Sodium and Lignin. Swed. Pap. J. 1976, 79(4), 123-128.
initially in the SSF step and 26% recycling of the stillage (18) Szengyel, Z.; Zacchi, G.; Varga, A.; Réczey, K. Cellulase
stream, which together with a 50% reduction in the yeast Production of Trichoderma reesei RUT C30 Using Steam
requirement would result in an ethanol production cost pretreated Spruce. Hydrolytic Potential of Cellulases on
of 3.58 SEK/L. This is a decrease of 26% compared with Different Substrates. Appl. Biochem. Biotechnol. 2000, 84-
the SSF base case. 86, 679-691.
(19) Stenberg, K. Ethanol from Softwood: Process Development
Acknowledgment based on Steam Pretreatment and SSF, Ph.D. Thesis, Lund
University, 1999, Report LUTHKDH/(TKKA-1001).
The Swedish Energy Agency is gratefully acknowl- (20) Aspen Plus, version 11.1; Aspen Technology, Inc.: Cam-
edged for its financial support. bridge, MA.
(21) Wooley, R. J.; Putsche, V. Development of an ASPEN PLUS
References and Notes Physical Property Database for Biofuels Components; NICH
Report TP-425-20685, 1996
(1) Swedish Energy Agency; Biofuels from the Forest, ISSN (22) Icarus Process Evaluator, version 11.1; Aspen Technology,
1403-1892; report written in Swedish; 2002. Inc.: Cambridge, MA.
Biotechnol. Prog., 2003, Vol. 19, No. 4 1117

(23) StakeTech System; personal communication. in Ethanol Production from Softwood Based on Enzymatic
(24) Von Sivers, M.; Zacchi, G. Ethanol from Wood: A Techno- Hydrolysis. Appl. Biochem. Biotechnol. 1998, 70-72; 697-
Economical Evaluation of the Enzymatic Hydrolysis Process, 708.
Lund University, 1993, LUTKDH/(TKKA-7005). (29) Alkasrawi, M., Galbe, M., Zacchi, G. Recirculation of
(25) Wimmerstedt, R.; Linde, B. Assessment of Technique and Process Streams in Fuel Ethanol Production from Softwood
Economy of Biofuel Drying, Värmeforsk Service AB, ISSN Based on Simultaneous Saccharification and Fermentation,
0282-3772; in Swedish, 1998. Appl. Biochem. Biotech. 2002, 98-100; 849-861.
(26) Peters, M. S.; Timmerhaus, K. D. Plant Design and (30) Personal Communication with Mats Galbe, Department of
Economics for Chemical Engineers, 4th ed.; McGraw-Hill Chemical Engineering 1, Lund University.
Inc.: New York, 1991. (31) Soderstrom, J.; Pilcher, L.; Galbe, M.; Zacchi, G.. Two-step
(27) Taherzadeh, M. J.; Eklund, R.; Gustafsson, L.; Niklasson, Steam Pretreatment of Softwood With SO2 Impregnation for
C.; Lidén, G. Characterization and Fermentation of Dilute- Ethanol Production. Appl. Biochem. Biotechnol. 2002, 98-
Acid Hydrolyzates from Wood. Ind. Chem. Res. 1997, 36, 100, 5-21.
4659-4665. Accepted for publication April 15, 2003.
(28) Stenberg, K.; Tengborg, C.; Galbe, M.; Zacchi, G.; Palm-
qvist, E.; Hahn-Hägerdahl, B. Recycling of Process Streams BP0340180

Vous aimerez peut-être aussi