Biotechnol. Prog.

2004, 20, 14211429

1421

Process Considerations and Economic Evaluation of Two-Step

Steam Pretreatment for Production of Fuel Ethanol from Softwood
Anders Wingren, Johanna So1 derstro1 m, Mats Galbe, and Guido Zacchi*
Department of Chemical Engineering, Lund University, P.O. Box 124, SE-221 00 Lund, Sweden

To increase the overall ethanol yield from softwood, the steam pretreatment stage
can be carried out in two steps. The two-step pretreatment process was evaluated
from a techno-economic standpoint and compared with the one-step pretreatment
process. The production plants considered were designed to utilize spruce as raw
material and have a capacity of 200 000 tons/year. The two-step process resulted in a
higher ethanol yield and a lower requirement for enzymes. However, the two-step
process is more capital-intensive and has a higher energy requirement. The estimated
ethanol production cost was the same, 4.13 SEK/L (55.1 /L) for both alternatives.
For the two-step process different energy-saving options were considered, such as a
higher concentration of water-insoluble solids in the filter cake before the second step,
and the possibility of excluding the pressure reduction between the steps. The most
optimistic configuration, with 50% water-insoluble solids in the filter cake in the feed
to the second pretreatment step, no pressure reduction between the pretreatment steps,
and 77% overall ethanol yield (0.25 kg EtOH/kg dry wood), resulted in a production
cost of 3.90 SEK/L (52.0 /L). This shows the potential for the two-step pretreatment
process, which, however, remains to be verified in pilot trials.

Introduction
The enzymatic process for the production of fuel
ethanol from lignocellulosic materials has been recognized as the most promising option in terms of ethanol
yield and low production cost (1-3). In this process, the
pretreatment step is of vital importance because cellulose
in its native form cannot be hydrolyzed to any great
extent by enzymes. In softwood the cellulose is the
primary sugar source for ethanol production, and mannan, the main constituent of the hemicelluloses, is the
second. It has been shown that cellulose and mannan
have different optima in terms of sugar recovery in the
pretreatment step (4, 5). Cellulose requires a higher
temperature than hemicellulose for optimal sugar recovery. During steam pretreatment, the sugars formed may
be degraded. Glucose, which is liberated from the cellulose, is further degraded to 5-hydroxymethylfurfural
(HMF), levulinic acid, and formic acid. The hydrolysis of
hemicellulose generates pentoses, which are converted
to furfural and formic acid. These degradation products,
together with lignin degradation products and released
organic acids, act as inhibitors in the fermentation step
(6-9) and in the enzymatic hydrolysis step (10). More
severe pretreatment conditions will cause greater degradation of hemicellulosic sugars with loss of yield and
possible inhibition as a consequence (7, 9, 11). However,
a rather high severity is required to enhance the enzymatic digestibility of the cellulose (8, 12). Steam pretreatment is improved by using an acid catalyst such as
H2SO4 or SO2. The acid increases the recovery of hemicellulosic sugars and improves the enzymatic hydrolysis
of the solid residue (13-17).
* 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/bp049931v CCC: $27.50

The difference in optimal pretreatment conditions

between cellulose and hemicellulose has led to the
proposal of a two-step pretreatment process in which the
first step is optimized for high recovery of hemicellulose,
(i.e., mannose) and the second step is optimized for
glucose recovery (14, 18, 19). The first step should be
performed at low severity to hydrolyze the hemicellulose.
The remaining solid material from the first step may then
be washed to recover hydrolyzed hemicellulosic sugars
and to avoid further degradation of sugars to inhibiting
substances. The solid material is then treated again in
the second pretreatment step to soften the structure of
the cellulose in order to increase the enzymatic digestibility in the subsequent process steps. The optimal
conditions for two-step steam pretreatment of softwood
have been investigated by Boussaid et al., Nguyen et al.,
and Soderstrom et al. (5, 12, 20-22). Impregnation with
SO2 has been shown to result in higher ethanol yields
than when H2SO4 is used (23). Sulfur dioxide is thus the
acid catalyst used in this evaluation.
The purpose of this study was to investigate the
economical feasibility of two-step pretreatment and
compare it to a one-step process. Since the outcome of
the pretreatment step affects the whole process, it is
important not to restrict the evaluation to the pretreatment step only; therefore the subsequent process steps
were also included. Process data for the two processes
considered was based on experimental work carried out
at the Department of Chemical Engineering, Lund
University, Sweden (5, 23, 24). The data were implemented in the flowsheeting program Aspen Plus (25) to
solve the mass and energy balances in the process.

Base Case Process Description

The process evaluated, based on steam pretreatment
and simultaneous saccharification and fermentation

2004 American Chemical Society and American Institute of Chemical Engineers

Published on Web 08/26/2004

Biotechnol. Prog., 2004, Vol. 20, No. 5

1422

Figure 1. Simplified flowsheet for the process considered in this study.

Figure 2. One-step and two-step pretreatment.

(SSF), is illustrated in Figure 1. The raw material

consists of fresh spruce with a water content of 50%.
Eight areas within the production plant are considered:
feedstock handling, pretreatment, SSF, distillation, dewatering, evaporation, steam generation, and other
facilities. The latter consists of off-site facilities such as
storage facilities for chemicals and products and administration buildings.
Pretreatment. A flowsheet for the pretreatment steps
can be seen in Figure 2. Fresh chips of spruce are
transported from the raw material handling area to the
pretreatment step where they are impregnated with
sulfur dioxide before being fed to the pretreatment
reactor. High-pressure steam, around 22 bar, is used to
treat the wood chips at the desired temperature for the
residence time required. The steam consumption in the
pretreatment units was set 10% higher than for a system
working adiabatically to account for heat losses and
steam leakage in the steam distribution system.
After pretreatment the material is flashed in two steps.
After the first flashing to 4 bar the vapor formed is
condensed at 138 C in heat exchanger 1 (HX1) and then
further flashed to atmospheric pressure and condensed
at 95 C (in HX2). In the third heat exchanger (HX3) the
vapor at 1 bar, from the second flashing of the pretreated
material, is condensed at 95 C. The purpose of reducing
the pressure in two steps is to recover some of the latent
heat at a higher temperature. The second pretreatment
step in the two-step process is similar to the first.
Dewatering and washing steps are included after the first

pretreatment step utilizing recycled evaporation condensates available at 60 C as washing water. The content
of water-insoluble solids (WIS) in the filter cake is
assumed to be 30% in the base case process, but this
parameter was varied to study the impact on the ethanol
production cost. In the simulations it was assumed that
filtration is carried out in a counter-current fashion in
two stages, and the relation between the addition of
washing water and washing efficiency was based on
studies of the removal of lignin from pulp (26). The
amount of washing water was adjusted to remove 95%
of the sugars, and sugars fed to the second step were
assumed to be unaffected, i.e., not degraded to byproducts. The whole filtrate from the washing step together
with the slurry leaving the second step is transferred to
the SSF step.
An alternative to the two-step process described above
is to exclude the flashing after the first pretreatment step
and to dewater and wash the material under pressure.
The filtrate in this process configuration is flashed to
atmospheric pressure, and the latent heat in the vapor
is recovered at 95 C (HX8 in Figure 2). Because the
pressure, and thus the temperature, of the slurry is
maintained, the amount of steam needed in the second
step is decreased. The addition of washing water also
results in a decrease in the temperature of the slurry
being fed to the second step. It has been shown, however,
that sugars not washed out, e.g., those fed to the second
step, are degraded to a lower extent than might be
expected. In a study by Soderstrom et al. (27) two-step

Biotechnol. Prog., 2004, Vol. 20, No. 5

1423

Figure 3. Composition of raw material and yields after pretreatment, expressed as g/100 g dry raw material. For the two-step case
yields are given as overall yields from both the first and the second step.

pretreatment with and without dewatering and washing

between the pretreatment steps were compared using
SSF for the evaluation of the overall ethanol yield.
Although the mannose recovery after pretreatment was
higher with dewatering and washing between the steps,
the overall ethanol yield was about the same for both
cases. Thus, a case in which the washing was excluded
while assuming the overall ethanol yield to be the same
as in the two-step base case was also evaluated in this
study.
Pretreatment data for the one-step base case is based
on experimental data presented by Galbe et al. (24) The
pretreatment temperature was 215 C, residence time
was 5 min, and SO2 was added at 2% of the water content
in the wood. These conditions correspond well to the
optimum in terms of overall sugar yield, as well as
ethanol yield, as determined by Stenberg et al. (18).
Data for two-step pretreatment originates from studies
carried out by Soderstrom et al. (5, 23). Various conditions were studied in terms of overall ethanol yield, and
the optimal conditions were found to correspond to a first
step at 190 C for 2 min and a second step at 210 C for
5 min. Both pretreatment steps were preceded by impregnation with 2% sulfur dioxide based on the liquid
content in the wood.
The pretreatment was modeled as a reactor working
with fixed fractional conversion of the components in the
wood. The reactions considered were hydrolyses of hexosans and pentosans to monomers, their degradation to
HMF and furfural, acetic acid production from acetyl
groups in the wood, and lignin degradation. Two lumped
components, one volatile and one nonvolatile, were
defined to account for further degradation products such

as formic acid, levulinic acid, and unknown products. The

composition of the raw material and the yields in the two
pretreatment alternatives are given in Figure 3. The
main difference between the two process configurations
is the higher fractional conversion of mannan to mannose
in two-step pretreatment compared to the one-step
process. Furthermore, a larger proportion of the cellulose
is converted to glucose in the two-step pretreatment
process than in the one-step process. However, in the twostep process slightly more of the cellulose is degraded to
byproducts, which cannot be used for ethanol production.
Of the further degraded components 85% is volatiles,
most of which end up in the flash steam together with
50% of the degraded lignin. For the two-step process the
amount of volatile lignin was set to 25% of the degraded
lignin, expressed as weight, which is the same amount
as for the one-step process.
Simultaneous Saccharification and Fermentation. It was assumed that SSF was performed in a fedbatch mode and process data was based on the same
experimental series as for the one-step pretreatment (24).
In these experiments, commercial bakers yeast, Saccharomyces cerevisiae, was added to a concentration of 5 g/L.
In a commercial plant a more realistic option is to use
some of the sugars available in the hydrolysate for inhouse yeast production. In this study it was assumed that
yeast must be produced in-house to a concentration of 2
g/L in SSF using fermentable sugars available in the
hydrolysate from the pretreatment step. The amount of
sugar needed will thus be dependent on the volumetric
flow rate to the SSF step. As a result of this in-house
yeast production the calculated ethanol yield will be lower
than the yield obtained in the experiments. At the end

Biotechnol. Prog., 2004, Vol. 20, No. 5

1424

of SSF the total amount of freshwater added is the same

as if the process had been carried out in conventional
batch mode with an initial load of WIS of 8.4%. The load
of cellulases is 15 FPU/g WIS, and 95% of the glucan is
hydrolyzed to glucose. Ninety-two percent of the fermentable sugars (glucose and mannose) that remain after
yeast has been produced is fermented to ethanol, and the
remainder (8%) is converted to byproducts, e.g., glycerol,
or remain unfermented. In the experiments that this
study is based on, commercial bakers yeast was used as
the fermentation organism. Because this yeast cannot
ferment pentoses to ethanol, it was assumed that these
sugars will remain unfermented. The total reaction time
in the SSF step is 72 h, and the total cycle time, including
draining and cleaning, is assumed to be 84 h.
Fed-batch SSF studies, similar to those described above
for the one-step process, have not been carried out with
two-step pretreated material. However, to be able to
compare the processes it was assumed that the results
in terms of yields in the SSF step would be the same for
both processes. This assumption is justified by the fact
that the yields obtained in batch SSF, with a substrate
load of 5%, of materials pretreated in one step and two
steps were both around 90% based on the amounts of
glucan, glucose, and mannose fed to the SSF step (23,
24).
Distillation and Evaporation. The downstream
processing begins with a distillation step, which consists
of two separate strippers and a rectifier operating at
different pressures. The overhead vapor from the first
stripper is used as heating medium in the reboiler of the
second stripper, and the overhead vapor from the second
stripper is used as heating medium in the rectifier. The
product is ethanol at 94 wt %. No further ethanol
purification was considered in the analysis. The stillage
from the strippers is filtered, and the solid fraction is sent
to the drying step. The liquid fraction, with a dry matter
content of around 3.5%, is concentrated to a dry matter
content of 50% in a multiple-effect evaporator consisting
of five stages with a forward feed arrangement.
Drying, Steam Generation, and Pellet Production. The solid residue from the filtration step and the
evaporation residue are mixed and dried in a steam dryer
to a dry matter content of 85%. Part of this dried solid is
incinerated to produce the steam needed in the pretreatment, the distillation, the evaporation, and the drying
steps. The excess material is pelletized and sold as a solid
fuel.

Methods
The methodology used in this work is similar to that
described previously (28); however, a brief description,
including the updates to the model, is provided here.
Aspen Plus from Aspen Technology (25) was used to
simulate the process. Rigorous material and energy
balance calculations using detailed equipment models
were carried out to determine flow rates, composition,
and energy flow for all streams in the process. Sizing of
the equipment was carried out utilizing the Icarus
Process Evaluator (IPE) (29) and by using rules of thumb
(30). Equipment cost was estimated using the IPE,
various reports, and vendor quotations. The cost of the
pretreatment reactors was based on a quotation from
Stake Technology (31). To evaluate the consistency of the
equipment cost provided by the IPE, vendor quotations
were obtained from a Swedish engineering company. The
process equipment compared was columns, pumps, heat
exchangers, and vessels including fermentors. The cost

Table 1. Costs Used in the Evaluation.

type

cost

raw material
wood
chemicals
sulfur dioxide
lime
defoamer
sodium hydroxide
(NH4)2HPO4
MgSO4
enzymes
byproduct income
solid fuel
carbon dioxide
utilities
electricity
cooling water
process water
other costs
labor (per employee)
insurance

unit

0.42
1.5
0.70
20.0
2.00
1.50
4.41
19.0
0.79
0.03
250
0.14
1.40
500 000
1

maintenance

SEK/kg DM
SEK/kg
SEK/kg
SEK/kg
SEK/kg
SEK/kg
SEK/kg
SEK/106 FPU
SEK/kg DM
SEK/kg
SEK/MWh
SEK/m3
SEK/m3
SEK/year
% of fixed capital
investment
% of fixed capital
investment

estimated with the IPE deviated by (20% from the

vendor costs, with the largest deviation for the smallest
equipment. For this kind of a comparative study this
deviation is acceptable. For quoted equipment, i.e., the
pretreatment reactor and pellet machine, and for equipment costs obtained from the literature, i.e., the dryer
and steam boiler, the sixth-tenths rule was used to
calculate the equipment cost, C2, at a capacity, A2, from
data at a specified capacity, A1, with a cost C1:

()

C2 ) C1

A2
A1

0.6

The IPE was also used to estimate the cost of auxiliary

equipment, such as piping and instrumentation, to yield
the direct cost. Indirect capital, which consists of costs
not directly related to the process equipment itself, such
as engineering, freight, start-up cost, and contingency
costs, was also estimated with the IPE. The fixed capital
investment consists of the sum of the direct and indirect
costs. Working capital was estimated as suggested by
Peters and Timmerhaus (32).
The annual cost of the fixed capital was obtained by
multiplying the fixed capital investment by an annualization factor of 0.104, corresponding to an interest rate
of 6% and 15 years pay-off time. Zero salvage value was
assumed. Raw material cost, 90 SEK/MWh, was based
on the actual cost of forest residue in Sweden, whereas
the assumed income from the solid fuel coproduct, 140
SEK/MWh, is slightly lower than the current price for
pellets and briquettes. The cost of enzymes, 19 SEK/
million FPU, was based on an estimate for producing the
enzymes at the plant (33). All costs are summarized in
Table 1.
The plant was assumed to be a grass-roots plant
designed to process 200 000 tons of dry raw material
yearly. The on-line time was set to 8000 hours per year.
The product was assumed to be 94% ethanol, but all costs
and energy consumptions reported are based on a pure
ethanol product. Costs reported in US$ (or US) are
based on an exchange rate of 7.5 SEK/US$. The aim of
the study was to compare different pretreatment configurations. The calculated costs should therefore not be
regarded as absolute, and care must be taken when

Biotechnol. Prog., 2004, Vol. 20, No. 5

1425

Table 2. Composition of Streams in the Base Cases of One- and Two-Step Pretreatmenta
one-step
stream
liquid phase
water
hexoses
pentoses
others

25.0

18.2

0.9

solid phase
hexosans
pentosans
lignin
acetyl groups

15.0
1.8
7.0
0.2

total flow

50.0

temp (C)
pressure (bar)

25
1

7.6

3.6

0.0
1.4

0.0
0.4

two-step
5

31.1
4.8
1.4
1.4

14.9

4.9

3.6

0.0
0.8

0.0
0.3

8.6
0.1
6.6
18.2
217
22

8.9
144
4

4.0
100
1

53.9

14.9

100
1

217
22

5.6
144
4

3.9
100
1

30.2
4.6
1.7
1.0

38.7
0.2
0.1
0.1

47.8
4.4
1.6
1.0

19.5

10.0
0.2
6.6

10.0
0.2
6.6

54.3

55.9

54.7

70
1

82
1

100
1

10

9.8

4.7

0.0
0.3

0.0
0.1

11
90.5
6.1
1.8
1.7
7.8
6.2

19.5
217
22

10.1
144
4

4.7
100
1

114.0
91
1

Flow rates in ton/h. Stream numbers refer to Figure 2.

Table 3. Energy Duties (MW) in the Pretreatment Step

steam 1a
steam 2a
HX1
HX2 + HX3
HX4 + HX8
HX5
HX6 + HX7
net energy demand
in pretreatment
steam requirement
(kg/kg dry wood)

one-step

two-step

pressure (bar)

14.1

11.6
15.2
-2.8
-2.5
0.0
-5.9
-3.5
12.0

22
22
4
1

-4.3
-2.6

7.2
0.73

4
1

1.37

a Calculated as the heat needed to produce the steam from a

boiler fed water at 25 C.

comparing the costs obtained in this study with results

from similar studies based on other assumptions.

Results and Discussion

Base Cases. Stream data from the simulations for the
two base cases is shown in Table 2. The yield of WIS (kg
WIS/kg raw material) after pretreatment is 0.61 and 0.56
for the one-step and two-step process, respectively. For
both process configurations the concentration of WIS
after flashing is higher in the simulations than in the
experiments, and the difference is mainly due to the
higher heat loss in the small laboratory equipment
compared to that in the simulated full-scale, continuously
working reactor. A high heat loss results in condensation
of the flashed vapor and thus dilution of the slurry.
For the one-step base case the energy requirement in
the pretreatment stage is 14.1 MW, which corresponds
to 0.73 kg steam/kg dry wood, see Table 3. The energy
duty associated with the condensation of flash vapor is
in total 6.9 MW, which results in a net energy requirement for the pretreatment step of 7.2 MW. The total
energy requirement for the whole plant is 38.3 MW.
Sixty-one percent of the solid residue from the dryer must
thus be incinerated to produce the required amount of
steam.
Two-step pretreatment requires 26.7 MW or 1.37 kg
steam/kg dry wood, and the recoverable energy from the
flash vapor is 14.8 MW, which results in a net energy
requirement of 12.0 MW. The total process energy
requirement is slightly higher than for the one-step
process; 40.6 MW and 64% of the solid residue must be
incinerated. The difference between the two processes in
terms of total process energy requirement, 2.3 MW, is
lower than would be expected from the difference in the

pretreatment step only, which is 4.8 MW. However, the

energy requirement in the downstream processing steps
is lower for the two-step process than for the one-step
process as a result of less freshwater being needed in
SSF. This is a result of the lower yield of WIS in the twostep process than in the one-step process.
Utilization of the secondary steam requires low-temperature process steps for the steam to have a value. If
this is not possible, the energy in the vapor formed must
be removed by condensation using cooling water, and the
net energy requirement for pretreatment will be higher
than that calculated here. In the processes considered
in this study, the downstream processing steps require
substantial amounts of steam. For instance, the flashed
vapor can be used to preheat the feed to the distillation
step or in the reboiler of the stripper.
The flow of ethanol after SSF in the one-step base case
is 5860 kg/h (0.23 kg EtOH/kg dry wood), which is 71.8%
of the theoretical maximum that could be produced from
the fermentable hexoses (glucose and mannose) available
in the raw material. If no yeast had been produced, the
ethanol yield would have been the same as the yield
obtained in the experimental study, i.e., 76% (24). The
ethanol production in the two-step base case is 6090 kg/h
(0.24 kg EtOH/kg dry wood), corresponding to a yield of
74.6% with yeast production and 78.1% without yeast
production. Two-step pretreatment followed by batch SSF
at 5% WIS has resulted in an experimental overall
ethanol yield of 81% based on the glucan and mannan
available in the raw material (23).
It was assumed that the sugars fed to the second
pretreatment reactor were not degraded. In the base case
two-step process 95% of the sugars was removed before
the second pretreatment step. If 20% of the sugars fed
to the second pretreatment reactor were degraded, the
overall ethanol yield would be reduced by only 0.2% age
points, which justifies the assumption of no degradation.
Table 4 summarizes the costs associated with the two
base cases evaluated, showing an overall ethanol production cost of 4.13 SEK/L (55.1 /L) for both processes. The
major advantages of the two-step process are its higher
ethanol yield and its lower enzyme requirement. The
amount of enzymes added is based on the amount of WIS
fed to SSF. As more material is hydrolyzed in two-step
pretreatment, less WIS is present in SSF and thus less
enzyme solution is needed. The lower amount of WIS
after pretreatment in two steps also reduces the need for
freshwater to the SSF step, thereby reducing the capital
cost and energy consumption in downstream processing.
The cost of the enzymes was in this study estimated to

Biotechnol. Prog., 2004, Vol. 20, No. 5

1426
Table 4. Costs in the Base Case Processes
one-step
yearly cost
MSEK

M$

two-step

cost/L EtOH
SEK

yearly cost

cents MSEK

raw matl
83.0 11.1
1.41 18.8
chemicals
50.4
6.7
0.86 11.4
byproducts -31.3 -4.2 -0.53 -7.1
utilities
11.5
1.5
0.20
2.6
other costs
37.9
5.1
0.64
8.6
capital
91.5 12.2
1.55 20.7
total
243
32.4 4.13 55.1

M$

cost/L EtOH
SEK

cents

83.0 11.1
1.36 18.1
51.5
6.9
0.84 11.2
-29.5 -3.9 -0.48 -6.4
12.1
1.6
0.20
2.6
39.3
5.2
0.64
8.6
96.4 12.9
1.58 21.0
253
33.7 4.13 55.1

be 0.6 SEK/L EtOH (8.0 /L EtOH) for the one-step base

case. Himmel et al. (34) have also estimated the cost for
enzymes produced from hydrolysate slurry. For a plant
capacity of 2000 dry tons per day of feedstock the cost
was 9 /L EtOH. National renewable energy laboratory
(NREL) expects a cost of 2.6 /L EtOH (35). However, in
these two studies a lower dosage of enzyme was used
than in the present study.
The fixed capital investment was estimated to be 847
and 893 million Swedish kronor (MSEK) for the one-step
and the two-step processes, respectively. The main reason
for the two-step process being more expensive is the extra
pretreatment reactor and the additional flash vessels and
the heat exchangers. The dewatering and washing step,
however, was not included in the economic evaluation.
The higher cost of pretreatment is somewhat offset by
the lower capital cost of downstream processing. In
addition to the higher capital cost, the two-step process
suffers from a higher consumption of sulfur dioxide and
a lower income from the solid fuel coproduct. The latter
is due to the higher energy consumption.
Process Alternatives for the Two-Step Process.
In the base case two-step process the slurry is flashed,
dewatered to 30% WIS, and washed with recycled condensate from the evaporation unit to recover 95% of the
sugars. For these conditions the net energy duty in the
pretreatment step is 40% higher in the two-step base case
than in the one-step base case. In an attempt to reduce
the energy consumption in the two-step case, the dependency of the concentration on WIS in the filter cake (a)
and the impregnation temperature (b) on the energy duty
was studied. In addition, the consequence of excluding
the flashing (c) and the washing (d) between the steps
are discussed.
(a) Higher Concentration of WIS in the Filter
Cake. By increasing the fraction of WIS in the filter cake
it is possible to reduce the energy consumption, since less
liquid needs to be heated in the second step. More
efficient dewatering also reduces the amount of catalyst
needed in the impregnation preceding the second step,
as the amount is based on the water content of the
material. Figure 4 shows the net energy duty as a
function of concentration of WIS in the filter cake for the
two-step pretreatment (Case a1). If the concentration can
be increased, from the base case value of 30% to 50%,
the net energy duty will be reduced from 12.0 to 8.9 MW,
a reduction of 26%. A WIS concentration of 70% would
reduce the net duty by 37% to 7.6 MW. However,
dewatering and filtration of the material from the first
pretreatment step have not been technically optimized,
a WIS concentration of 37% being reached on lab scale
(21). A higher concentration is expected for an optimized,
large-scale system. Fifty percent WIS was therefore taken
as the currently most realistic value.

Figure 4. Net energy duty in the pretreatment step as a

function of WIS for various alternatives. Dashed lines represent
the two-step base case and the solid horizontal line the onestep base case. Case a1: flashing to 1 bar after first step,
washing water at 60 C, 95% recovery of sugars. Case c1: no
flashing after first step, washing water at 60 C, 95% recovery
of sugars. Case c2: no flashing after first step, washing water
at 180 C, 95% recovery of sugars. Case d1: flashing to 1 bar,
no washing. Case d2: no flashing after first step, no washing.

(b) Impregnation Temperature. After the first step

the slurry has a temperature of 100 C, but as a result
of the addition of washing water, the temperature
decreases. A higher concentration of WIS in the filter
cake results in a higher temperature of the feed to the
second reactor due to the lower addition of cold washing
water. A WIS concentration of 50% yields a feed of 85
C, whereas a WIS concentration of 30% corresponds to
70 C. The temperature, however, affects the catalyst
impregnation of the material. For a liquid catalyst such
as sulfuric acid impregnation will probably be better at
a high temperature as a result of a higher diffusion rate.
However, for a gaseous catalyst, such as sulfur dioxide,
the temperature has to be low for the gas to dissolve in
the liquid. From a process point of view this means that
the slurry, after being washed, may have to be cooled
further before impregnation can take place. Thus, the
effect of having to cool the slurry before the second
impregnation stage was investigated. Two WIS concentrations in the filter cake, 30% and 50% (Case b1 and
Case b2), were studied. The results are shown in Figure
5. The worst case is when the filter cake cannot be
dewatered to more than 30% WIS and has to be cooled
to 20 C before impregnation. For these conditions the
net energy duty increases to almost 16 MW.
(c) No Flashing after the First Step. An alternative
to the process configurations discussed above is to exclude
the flashing after the first pretreatment step. The dewatering, washing, and impregnation steps are carried
out under pressure, after which the slurry is fed directly
to the second pretreatment step. It was assumed, as in
the cases discussed above, that recycled evaporation
condensates at 60 C can be used as washing water for
95% sugar recovery. The option of heating the washing
stream to 180 C (HX4) was also studied. Figure 4 shows
the results where the concentration of WIS is also
included as a parameter. The difference, in terms of net
energy demand, is minor between flashing to 1 bar (Case

Biotechnol. Prog., 2004, Vol. 20, No. 5

Figure 5. Net energy duty in the pretreatment step as a

function of impregnation temperature before the second step
for 30% WIS (Case b1) and 50% WIS (Case b2) in the filter cake.
Dashed lines represent the two-step base case.

a1) and without flashing and washing water with a

temperature of 60 C (Case c1). However, if the washing
water is heated to 180 C (Case c2), close to the temperature of the slurry from the first step, the recovery of
energy from the filtrate increases, resulting in a net
energy reduction.
(d) No Washing of the Filter Cake. The different
options discussed above assume that 95% of the sugars
released from the first pretreatment step are washed out
of the filter cake before the second step. The option of
excluding the washing step was studied with and without
flashing after the first step, Case d1 and Case d2,
respectively. Although no washing water is added, sugars
are recovered as a result of the dewatering step alone.
For instance, at 50% WIS in the filter cake and no
flashing, 65% of the sugars is removed from the filter
cake. These two cases are also illustrated in Figure 4.
When no flashing is carried out after the first step, a very
low net energy consumption is achieved, even at low
concentrations of WIS in the filter cake. At 40% WIS,
the net duty is even lower than for the one-step base case
(<7.2 MW).
Clearly, the choice of pretreatment conditions and the
demand for dewatering of the slurry before the second
step affect the heat duty in the pretreatment stage and
thus the ethanol production cost. The ethanol cost as a
function of the net heat duty in the pretreatment step is
shown in Figure 6. The highest energy duty, 16 MW,
corresponds to Case b1, in which the slurry has to be
cooled to 20 C before impregnation, and there is 30%
WIS in the filter cake. For this energy duty the cost of
the two-step processes increases from the two-step base
case value of 4.13 to 4.23 SEK/L; see Figure 6. However,
if the process can be run as Case d2, with 50% WIS, the
cost decreases to 4.03 SEK/L.
Potential Yield for a Two-Step Process. Recovering
more of the fermentable sugars available in the raw
material can increase the ethanol yield. Figure 7 shows
the ethanol yield for the two-step process as a function
of the conversion of mannan to mannose in the pretreatment step for various degrees of cellulose conversion in
the SSF step. In previous experimental studies high

1427

Figure 6. Ethanol cost as a function of net heat duty in

pretreatment for the two-step process. Dashed line represents
the two-step base case, and the solid line represents Case d2
with 50% WIS in the filter cake.

Figure 7. Ethanol production and ethanol yield as a function

of mannose recovery in the pretreatment step for different
amounts of cellulose converted to glucose in the SSF step.
Dashed lines represent the two-step base case.

overall ethanol yields were obtained with a two-step

steam pretreatment process. The composition of the raw
material used in the experimental studies varied and was
different from the composition assumed in this analysis.
Thus, it is difficult to compare overall yields of ethanol.
A material containing more glucan and mannan may
result in a lower ethanol yield expressed in percent, but
in a higher ethanol yield expressed in L/ton raw material.
However, in this analysis the content of hexosans is lower
than that in the previous studies, and thus does not
overestimate the yield (5,21).
In the two-step base case an overall ethanol yield of
74.6% is obtained. This assumes a mannan-to-mannose
conversion of 85% in the pretreatment step and 95%
hydrolysis yield from cellulose. As stated above, the
simulated data includes yeast production and the yields
are thus lower than the corresponding experimental
yields obtained where bakers yeast was added. The

1428

Biotechnol. Prog., 2004, Vol. 20, No. 5

is washed out before the second pretreatment step, 4.00

SEK/L, is obtained when pretreatment is configured
according to Case c2 in Figure 4, with 50% WIS in the
filter cake (C95% in Figure 9).
A best case was identified in which the most favorable configuration and data were included: a yield of
77.0% (80.5% without yeast production) corresponding
to a mannose recovery of 85% and 100% hydrolysis of
cellulose, an energy consumption in the pretreatment
unit corresponding to the case where flashing was
excluded, no washing water added (Case d2), and 50%
WIS in the filter cake. Under these conditions the ethanol
production cost was estimated to be 3.90 SEK/L, which
is 5.6% lower than the 4.13 SEK/L in the one-step base
case process.

Conclusions

Figure 8. Ethanol production cost as a function of ethanol yield

for the two-step process. Dashed lines indicate the two-step base
case, and the solid line represents the highest yield achieved.

highest experimental overall ethanol yield achieved in a

two-step process was 81% (23). This indicates a mannanto-mannose conversion of 85% and 100% hydrolysis of
cellulose according to Figure 7, represented by 77% in
the simulations. These are realistic assumptions as a
successful SSF experiment provides an ethanol yield
above 90%, indicating the complete hydrolysis of cellulose. Figure 8 shows the ethanol production cost as a
function of the ethanol yield for the two-step process. The
energy consumption was assumed to be the same as in
the two-step base case. At the highest yield attained in
experimental work, 77.0%, the lowest ethanol production
cost is 4.04 SEK/L.
Summary of the Different Cases. Figure 9 summarizes some of the cases discussed in this study. The
two-step configuration that results in the highest ethanol
production cost, 4.23 SEK/L, is that in which the slurry
from the first step has to be cooled to 20 C before
impregnation. The lowest cost when 95% of the sugars

The two-step process results in an ethanol yield higher

than that of the one-step process, 74.6% compared to
71.8%. Another advantage is the lower requirement of
enzymes and water in the SSF step. Major drawbacks
are, however, the higher capital cost and the higher
energy consumption. The overall ethanol cost was estimated to be 4.13 SEK/L (55.1 /L) for both the one-step
and the two-step base case. The lowest cost estimated
for the two-step process, 3.90 SEK/L (52.0 /L), requires
a high ethanol yield, high concentration of WIS in the
filter cake between the steps, and that the sugars being
fed to the second step are not degraded. The higher yield
has been demonstrated experimentally, but the two other
assumptions still need to be verified.
An important parameter in this evaluation is the
overall ethanol yield, which has a significant impact on
the ethanol production cost. Therefore, it is of importance
to follow the progress made in both the one-step and twostep process. The difference in yield between the two base
cases, 2.8%, can serve as a rough guide when comparing
future experiments.
Another aspect of importance is the higher complexity
of a two-step pretreatment process compared to a onestep process. Whether it is possible to operate the twostep process in the way suggested in this paper must be
technically verified on a larger scale. Hopefully, some of

Figure 9. Summary of costs. One-step BC: one-step base case, 71.8% ethanol yield. Two-step BC: two-step base case, washing
water at 60 C, 74.6% ethanol yield. Two-step worst case: slurry from the first step cooled to 20 C, 30% WIS in filter cake, and
74.6% ethanol yield. C95%: no flashing after first step, washing water at 180 C, 50% WIS in filter cake, and 74.6% ethanol yield.
optimistic case: no flashing after first step, no washing, 50% WIS in filter cake, 77.0% ethanol yield.

Biotechnol. Prog., 2004, Vol. 20, No. 5

the questions raised in this paper can be answered when

the new pilot plant that is being constructed in the north
of Sweden is in operation (36). It will operate in a
continuous mode (2 ton dry raw material per day) with
pretreatment in one or two steps. The dewatering and
washing steps are designed to work under pressure, as
suggested in this paper.

Acknowledgment
The Swedish Energy Agency is gratefully acknowledged for its financial support.

References and Notes

(1) Galbe, M.; Zacchi, G. A review of the production of ethanol
from softwood. Appl. Microbiol. Biotechnol. 2002, 59, 618628.
(2) Wright, J. D. Ethanol from biomass by enzymatic hydrolysis.
Chem. Eng. Prog. 1988, 84 (8), 62-74.
(3) Wooley, R.; Ruth, M.; Glassner, D.; Sheehan, J. Process
design and costing of bioethanol technology: A tool for
determining the status and direction of research and development. Biotechnol. Prog. 1999, 15 (5), 794-803.
(4) Tengborg, C.; Stenberg, K.; Galbe, M.; Zacchi, G.; Larsson,
S.; Palmqvist, E.; Hahn-Hagerdal, B. Comparison of SO2 and
H2SO4 impregnation of softwood prior to steam pretreatment
on ethanol production. Appl. Biochem. Biotechnol. 1998, 7072, 3-15.
(5) Soderstrom, J.; Pilcher, L.; Galbe, M.; Zacchi, G. Two-step
steam pretreatment of softwood with SO2 impregnation for
ethanol production. Appl. Biochem. Biotechnol. 2002, 98-100,
5-21.
(6) Palmqvist, E.; Hahn-Hagerdal, B.; Szengyel, Z.; Zacchi, G.;
Reczey, K. Simultaneous detoxification and enzyme production of hemicellulose hydrolyzates obtained after steam
pretreatment. Enzyme Microb. Technol. 1997, 20 (4), 286293.
(7) Heitz, M.; Capek-Menard, E.; Koeberle, P. G.; Gagne, J.;
Chornet, E.; Overend, R. P.; Taylor, J. D.; Yu, E. Fractionation
of Populus tremuloides at the pilot plant scale: optimization
of steam pretreatment conditions using the STAKE 2 technology. Bioresour. Technol. 1991, 35 (1), 23-32.
(8) Shevchenko, S. M.; Beatson, R. P.; Saddler, J. N. The nature
of lignin from steam explosion/enzymatic hydrolysis of softwood-structural features adn possible uses. Appl. Biochem.
Biotechnol. 1999, 77-79, 867-876.
(9) Nguyen, Q. A.; Tucker, M. P.; Boynton, B. L.; Keller, F. A.;
Schell, D. J. Dilute acid pretreatment of softwoods. Appl.
Biochem. Biotechnol. 1998, 70-72, 77-87.
(10) Tengborg, C.; Galbe, M.; Zacchi, G. Reduced inhibition of
enzymatic hydrolysis of steam-pretreated softwood. Enzyme
Microb. Technol. 2001, 28, 835-544.
(11) Boussaid, A.; Robinson, J.; Cai, Y.; Gregg, D. J.; Saddler,
J. N. Fermentability of the hemicellulose-derived sugars from
steam-exploded softwood (douglas fir). Int. Conf. Biotechnol.
Pulp Paper Ind. 1999, 7th C, 284-289.
(12) Nguyen, Q. A.; Tucker, M. P.; Keller, F. A.; Eddy, C. K.
Two-stage dilute-acid pretreatment of softwoods. Appl. Biochem. Biotechnol. 2000, 84-86, 561-576.
(13) Mackie, K. L.; Brownell, H. H.; West, K. L.; Saddler, J. N.
Effect of sulphur dioxide and sulphuric acid on steam explosion of aspenwood. J. Wood Chem. Technol. 1985, 5 (3), 405425.
(14) Clark, T. A.; Mackie, K. L.; Dare, P. H.; McDonald, A. G.
Steam explosion of softwood pinus radiata with sulphure
dioxide addition. 2. Process characterization. J. Wood Chem.
Technol. 1989, 9 (2), 135-166.
(15) Schwald, W.; Breuil, C.; Brownell, H. H.; Chan, M.;
Saddler, J. N. Assessment of pretreatment conditions to
obtain fast complete hydrolysis on high substrate concentration. Appl. Biochem. Biotechnol. 1989, 20-21, 29-44.

1429
(16) Eklund, R.; Galbe, M.; Zacchi, G. The influence of SO2 and
H2SO4 impregnation of willow prior to steam pretreatment.
Bioresour. Technol. 1995, 52 (3), 225-229.
(17) Grethlein, H. E.; Allen, D. C.; Converse, A. O. A comparative study of the enzymatic hydrolysis of the acid-pretreated
white pine and mixed hardwood. Biotechnol. Bioeng. 1984,
26 (12), 1498-1505.
(18) Stenberg, K.; Tengborg, C.; Galbe, M.; Zacchi, G. Optimization of steam pretreatment of SO2-impregnated mixed softwoods for ethanol production. J. Chem. Technol Biotechnol.
1998, 71 (4), 299-308.
(19) Kim, K. H.; Tucker, M. P.; Keller, F. A.; Aden, A.; Nguyen,
Q. A. Continuous countercurrent extraction of hemicellulose
from pretreated wood residues. Appl. Biochem. Biotechnol.
2001, 91-93, 253-267.
(20) Boussaid, A.; Esteghlalian, A. R.; Gregg, D. J.; Lee, K. H.;
Saddler, J. N. Steam pretreatment of douglas-fir wood chipsCan conditions for optimum hemicellulose recovery still
provide adequate access for efficient enzymatic hydrolysis?
Appl. Biochem. Biotechnol. 2000, 84-86, 693-705.
(21) Soderstrom, J.; Pilcher, L.; Galbe, M.; Zacchi, G. Two-step
steam pretreatment of softwood by dilute H2SO4 impregnation
for ethanol production. Biomass Bioenergy 2003, 24, 475486.
(22) Soderstrom, J.; Pilcher, L.; Galbe, M.; Zacchi, G. Combined
use of H2SO4 and SO2 impregnation for steam pretreatment
of spruce in ethanol production. Appl. Biochem. Biotechnol.
2003, 105-108, 127-140.
(23) Soderstrom, J.; Galbe, M.; Zacchi, G. Separate versus
simultaneous saccharification and fermentation of two-step
steam pretreated softwood for ethanol production. J. Wood
Chem. Technol., submitted for publication.
(24) Personal communication with Mats Galbe at Department
of Chemical Engineering, Lund University, Sweden.
(25) Aspen Plus; Aspen Technology Inc.: Cambridge, MA, 2001.
(26) Grahs, L.-E. Displacement washing of packed beds of
cellulose fibres; Part 3. A comparison of the washing behaviour of sodium and lignin. Sv Papperstidn. 1976, 4, 123-128.
(27) Soderstrom, J.; Galbe, M.; Zacchi, G. Effect of washing on
yield in one- and two-step steam pretreatment of softwood
for production of ethanol. Biotechnol. Prog. 20 (3), 744-749.
(28) Wingren, A.; Galbe, M.; Zacchi, G. A techno-economic
evaluation of producing ethanol from softwood-a comparison
of SSF and SHF and identification of bottlenecks. Biotechnol.
Prog. 2003, 19 (4), 1109-1117.
(29) Icarus Process Evaluator, ver. 12.0; Aspen Technology
Inc.: Cambridge, MA, 2002.
(30) Perry, P. H.; Green, D. Perrys Chemical Engineers Handbook, 6th ed.; McGraw-Hill: New York, 1984.
(31) Stake Technology Ltd., Norval, Ontario, Canada, 2003.
(32) Peters, M. S.; Timmershaus, K. D. Plant Design and
Economics for Chemical Engineers; 3rd ed.; McGraw-Hill:
New York, 1980.
(33) von Sivers, M.; Zacchi, G. Ethanol from wood-A technoeconomic evaluation of the enzymatic hydrolysis process,
LUTKDH/(TKKA-7006), 1993.
(34) Himmel, M. E.; Ruth, M. F.; Wyman C. E. Cellulase for
commodity products from cellulosic biomass. Curr. Opin.
Biotechnol. 1999, 10 (4), 358-364.
(35) Aden, A.; Ruth, M. F.; Ibsen, K.; Jechura, J.; Neeves, K.;
Sheehan, J.; Wallace, B. Lignocellulosic biomass to ethanol
processing design and economics utilizing co-current dilute
acid prehydrolysis and enzymatic hydrolysis for corn stover.
NREL/TP-510-32438. 2002.
(36) www.etek.se; 2003.

Accepted for publication July 16, 2004.

BP049931V