Bioresource Technology60 (1997) 143-151
1997 Elsevier Science Limited
ELSEVIER
PIl:S0960-8524(97)O0011-4
RECIRCULATION OF PROCESS WATER IN THE
PRODUCTION OF ETHANOL FROM SOFTWOOD
Mats Larsson, a Mats G a l b e a & G u i d o Z a c c h i a*
aDepartment of Chemical Engineering 1, Lund University, Lund, Sweden
(Received 4 June 1996; revised version received 13 December 1996; accepted 20 December 1996)
Abstract
In the production of ethanol from lignoceUulosics,
large quantities of water are needed in various process
steps. In industrial processes, recycling of process
streams is necessary to minimise fresh water require-
ments and decrease the amount of wastewater
produced. This leads to increases in various sub-
stances, both volatile and non-volatile, which inhibit
fermentation. In the present study ethanol was pro-
duced from softwood in a bench-scale process
development unit. The stillage stream from the distilla-
tion step was fractionated by evaporation and various
evaporation fractions characterised and their inhibitory
effect on fermentation investigated. It was found that
the volatile fractions have no negative effect on the
fermentation, while the non-volatile fraction inhibited
fermentation and resulted in a decreased yield. Simula-
tion of a six-effect evaporation unit based on the
experimental data shows that the intermediate evapora-
tion condensate fractions are most suited for disposal
The chemical and biological oxygen demands are only
about 10% of the value in the stiUage stream. Evap-
oration has been shown to be an effective way to
drastically diminish the build-up of inhibitory com-
pounds when process streams are recirculated, but the
energy demand is high. Various process configurations
to reduce the energy demand are suggested. 1997
Elsevier Science Ltd,
Key words: Ethanol, recirculation, inhibitory, bench-
scale process development unit, evaporation.
INTRODUCTION
Ethanol from renewable sources has gained in
interest during the last 20 years. One reason for this
is that provided no fossil fuel is used in the produc-
tion, ethanol produced from biomass does not
contribute to the greenhouse effect since the carbon
dioxide formed in the production and the combus-
tion of ethanol is taken up by photosynthesis during
the growth of new biomass. Another benefit with
*Author to whom correspondence should be addressed.
143
All rights reserved. Printed in Great Britain
0960-8524/97 $17.00
using ethanol instead of petrol is the removal of
aromatic hydrocarbons from the emissions. Ethanol
from renewable sources is usually produced from
starch, and the production cost is very dependent on
the market for cattle feed, since the main byproduct
is a protein-rich residue. It is expected that in the
future, cheaper raw materials, such as lignocellu-
losics, will be used.
Lignocellulose as a raw material for ethanol pro-
duction costs less than starch and larger quantities
are available, allowing for large-scale production
(Lou, 1994). The dependence on the market for
byproducts is less than for starch-based production,
since the main byproduct, lignin, is a valuable solid
fuel with very low ash content. Lignocellulosic
materials are consequently preferable to starch as a
raw material for the production of fuel ethanol (The
Swedish National Board for Industrial and Technical
Development, 1994).
Several processes based on enzymatic hydrolysis
on lignocellulosic materials have been suggested in
the literature (Jurasek, 1979; Saddler et al., 1982;
Grethlein et al., 1984; Mes-Hartree et al., 1988; Lynd
et al., 1991). Most of these have a similar process
configuration consisting of prehydrolysis of hemicel-
lulose, enzymatic hydrolysis of cellulose and
fermentation of the liberated sugars to ethanol.
Techniques involved in the various production steps,
such as prehydrolysis by steam pretreatment (Clark
and Mackie, 1987), enzymatic hydrolysis (Esterbauer
et al., 1983) and simultaneous saccharification and
fermentation (SFF) (Wright et al., 1988), have been
improved over the years. For ethanol to be competi-
tive with petrol, the utilisation of the raw material
must be high, since the raw material contributes
more than 30% of the production cost (von Sivers
and Zacchi, 1996; Busche et al., 1992; US Depart-
ment of Energy, 1993; Hinman et al., 1992). This
requires high yields in the saccharification steps.
Furthermore, micro-organisms which ferment
glucose to ethanol with high yields and which can
withstand potential inhibitors present in the raw
material and formed in the preceding steps are
required. The yeast Saccharomyces cerevisiae has
144 M. Larsson, M. Galbe, G. Zacchi
proved to be very resistant to the inhibitors present
in lignocellulosic hydrolysates and it efficiently fer-
ments hexoses (Palmqvist et al., 1997).
Industrial-scale processes differ from small-scale
laboratory investigations. In the latter, process para-
meters can be carefully controlled and usually pure
water or buffer solutions are employed to optimise
the conditions. In industrial processes, recycling of
process streams is necessary to minimise fresh water
requirements and decrease the amount of waste-
water produced. This recycling leads to increases in
various substances, both volatile and non-volatile, in
the hydrolysis and fermentation steps (Galbe and
Zacchi, 1993). A previous investigation (Galbe and
Zacchi, 1991) has shown that more than 10-fold
higher concentrations of non-volatile substances can
occur at different locations in the process. The
inhibitory effects on enzymatic hydrolysis and fer-
mentation of the volatile and non-volatile
compounds released from willow during steam pre-
treatment have been studied by Palmqvist et al.
(1996a). The non-volatile fraction has been found to
inhibit fermentation greatly and hydrolysis modera-
tely. It is thus of considerable importance to
determine process streams that are suited for recir-
culation in terms of fermentability and
environmental impact. The waste streams from the
process must be taken care of in the best possible
way. For a totally closed process, the minimum
requirement is that an amount of water equivalent
to that present in the raw material has to be dis-
posed of.
In the present study, softwood was converted to
ethanol in a bench-scale process development unit
(Palmqvist et al., 1996b). The stillage stream from
the distillation was fractionated by evaporation and
the inhibitory effects of the volatile and the non-
volatile compounds assayed by fermentation.
Environmental impact was estimated by measuring
the chemical oxygen demands (COD) and the bio-
logical oxygen demands (BODT) of various fractions.
Wood chips
Steam1bar '~b'e
SO2 , atrnent
Steam 20-28
bar
Cydone ~_1
Ethanol
Condensate~ ~ Distillation
[..r_~ Steam
Residue _ i L~
Evaporation [
Stiflage
Fig. 1. Schematic flowsheet for the bench-scale unit.
The inhibitory effects and the characteristics of the
waste stream for softwood as raw material are com-
pared with those of willow.
METHODS
Figure 1 shows a schematic flowsheet of the process
configuration studied, using a bench-scale process
development unit for converting softwood to
ethanol. The bench-scale unit consists of a steam
pretreatment unit, a reactor for enzymatic hydro-
lysis, a fermentor, a distillation column, and an
evaporation unit. The different units are not con-
nected physically so the material is transferred
manually from one step to the next. By using the
same material throughout the entire process, a real
process can be experimentally simulated. The equip-
ment is described in more detail elsewhere
(Palmqvist et al., 1996).
Substrate
A mixture of pine and spruce from a Swedish pulp
and paper mill was used as substrate. The composi-
tion of the mixture is shown in Table 1 (STORA
Teknik AB, Sweden). The dry matter (DM) content
was 48wt%. The material was chopped and
screened and fractions from 2 to 5 mm selected for
pretreatment.
For similar substrates, the amount of extractives
(not measured) is reported in the literature to be in
the range 2 - 6 w t % and the ash content around
0.5 wt% (Ladisch, !983).
Pretreatment
Here 2454 g of softwood (DM) was presteamed for
30 min in a small pressure vessel, using steam of
2 bar. Immediately after presteaming, the material
was transferred to a plastic bag and impregnated
with gaseous sulphur dioxide (SO2). The amount of
SO2 added was approximately 5 wt% of the DM.
PS:~u~~~._~
3 hydrolysis
E.nzym.ati.c
Washing
, Water
L~nln
~-- Yeast
J , ~ Fermentation
 Recirculating process water in ethanol production
Table 1. Composition of the raw material
wt% of dry
wood
Galactan 1.9
Glucan 40.5
Mannan 11.8
Arabinan 1.2
Xylan 5.1
Lignin 27.6
The plastic bag with its contents was stored over-
night at 4C. The SO2 impregnated material was
then subjected to steam pretreatment at 210C for
5 min. For spruce, these conditions have been found
to be optimal in terms of the sugar yield in
enzymatic hydrolysis (Saddler et al., 1989).
Enzymatic hydrolysis
Hydrolysis was performed on the fibrous material
and the filtrate at a dry-matter content of fibrous
material of 5 wt%, supplemented with 0.2 g Cellu-
clast 2L (Novo AS, Denmark) and 0.05 g Novozym
188 (Novo AS, Denmark) per gram substrate (DM),
corresponding to 12 FPU/g substrate. The hydrolysis
was performed for 96 h at 40 + 1C, pH 4.8, as con-
trolled by titration with 10wt% NaOH solution.
Samples were withdrawn after 0, 1, 3.5, 6, 16, 24, 40,
48, 72, and 96 h. After hydrolysis, the hydrolysate
was filtered in a filter press and the solid residue
removed.
Fermentation
The softwood hydrolysate was fermented in non-
sterile conditions at 30C in an intermittently stirred
301 tank containing 25 1 medium. Prior to fermenta-
tion, the pH was adjusted to 5.5 with Ca(OH)2 and
the hydrolysate supplemented with 1.0g/l yeast
extract, 0.25 g/1 (NH4)EPO4, 0.025 gfl MgSO4-7H20,
and 12 g/l NaHzPO4. The medium was inoculated
with compressed baker's yeast, Saccharomyces cere-
visiae (J/istbolaget, Rotebro, Sweden), to a final cell
concentration (DM) of 7.5 g/l. Samples were taken
at 0, 1, 2, 3, 4, 5 and finally 8 h, at which time all
glucose had been consumed.
Distillation and evaporation
Since the distillation was operated continuously the
fermentation broth was stripped in the evaporation
unit prior to distillation, to avoid problems associ-
ated with suspended particles in the packed
distillation column. The evaporated fraction was
then distilled. After distillation, the non-volatile
fraction from the stripping stage was mixed with the
stillage stream from the distillation stage, in order to
maintain the process configuration in Fig. 1. The
mixture was evaporated and the condensate col-
lected in five fractions. The COD, BOD7 and
fermentability of these fractions, as well as those of
145
the stillage stream and evaporation residue, were
determined.
Assessment of the inhibitory effects of evaporation
condensates
The inhibitory effects of evaporation condensates, to
be recycled in the process, were assessed by fer-
mentation using S. cerevisiae. Fermentation of both
total and fractionated stillage was carried out in
100 ml glass flasks sealed with cotton stoppers. The
volume of the fermentation broth was 75 ml and the
pH was adjusted to 5.5 before inoculation. Glucose
at a concentration of 50 g/l was added to samples of
the original stillage, the five evaporation condensate
fractions and the evaporation residue, which were
then inoculated to a final cell concentration (DM) of
10 g/1. The nutrients used in the fermentation of the
hydrolysate were also added. Fermentation was per-
formed in a shake-incubator for 5 h at 30C. A
reference fermentation containing only 50 g/l
glucose and nutrients was run for comparison.
Samples were withdrawn at the start and then every
30 min for 5 h.
Analysis
All samples from the hydrolysis, fermentation and
distillation/evaporation steps were filtered through
0.2mm filters (MFS-13, Micro Filtration Systems,
USA) and analysed on an HPLC unit (Shimadzu,
Japan) equipped with a refractive index detector
(Shimadzu, Japan). Sugars were separated using an
Aminex HPX-87P column (Biorad, USA) at 80C
with water, at a flow rate of 0.5 ml/min as eluent.
The ethanol, glycerol, and acetic acid were sepa-
rated on an Aminex HPX-87H column (Biorad,
USA) at 65C using 5 mM H2SO4, at a flow rate of
0.5 ml/min as eluent.
COD and BOD7 were determined at the certified
laboratory Cenox AB (Lund, Sweden) using the
methods of Dr Lange, LCK Cuvettes and Swedish
Standard 02 81 43 (SIS, 1990).
Yields
All yields are based on the theoretically available
hexosan and pentosan contents of the raw material
and the ethanol yield is based on the theoretically
available amounts of fermentable sugars, glucose
and mannose.
Specific heat and boiling point elevation
The physical data necessary for computer simulation
of the evaporation of the distillation stillage, the
specific heat and the boiling point elevation were
determined for various dry-matter contents in the
evaporation residue. The specific heat at 25C and
atmospheric pressure was determined at the Depart-
ment of Thermochemistry, Lund University
(Suurkuusk and Wads6, 1974). The boiling point for
the evaporation residue at atmospheric pressure was
146 M. Larsson, M. Galbe, G. Zacchi
determined relative to that of pure water using a
reflux boiler.
RESULTS
Yields
During pretreatment 4.6 wt% of the material was
lost. This was partially due to the formation of vola-
tile compounds which, to some extent, leave the
cyclone together with the flash vapour. Sixty-four
wt% of the material ended up as fibrous material
(cellulose and lignin) and the rest, 31.4wt%, con-
sisted of dissolved solids, mainly hydrolysed
hemicellulose and degraded lignin. The yields of
glucose and mannose in the liquid after pretreat-
ment were 21 and 71 wt%, respectively, based on the
theoretical amount available in the raw material.
After enzymatic hydrolysis, 65 wt% of the theo-
retical glucose yield based on raw material was
achieved. The mannose yield is not affected by
enzymatic hydrolysis since most of the hemicellulose
is hydrolysed during the pretreatment step. The con-
centrations of glucose and other sugars in the
hydrolysis as a function of time are shown in Fig. 2.
The hydrolysate contained 26.2g/1 glucose and
7.1 g/1 mannose prior to fermentation. Based on
total fermentable sugars in the hydrolysate, the
ethanol yield was 0.41 g/g, corresponding to 80% of
the theoretical value and resulting in an overall yield
of 15.6 g ethanol/100 g of original dry matter. This
corresponds to 53% of the theoretical yield based on
the amount of fermentable sugars available in the
raw material. The concentrations of sugars and
ethanol produced in fermentation are shown in Fig.
3. The maximum productivity in fermentation was
7.0 g ethanol/(l h) with a final productivity of 3.8 g
3t
25
o - - - . . . . .
Time [h]
Fig. 2. Sugar concentrations versus time during enzymatic hydrolysis.
ethanol/(l h) when all the fermentable sugars were
consumed. Pentoses and galactose do not contribute
to the yield since the strain of S. cerevisiae used
could not ferment these sugars.
After distillation, approximately 98.5 wt% of the
ethanol present in the fermentation broth was
obtained in the distillate. The missing ethanol was
present in the two first condensate fractions from
the evaporation. The loss of ethanol in the distilla-
tion step can be lowered with improved control.
Characterisation of recycling and waste streams
The amounts of COD and the BOD7 in the stillage
stream, the five condensate fractions from the evap-
oration and the evaporation residue were
determined in order to estimate the environmental
impact of discharging these streams. Figure 4 shows
COD and BOD7 for the volatile fraction from the
evaporation step as a function of the relative
amount evaporated. Data for willow obtained by
Palmqvist et al. (1996) are included for comparison.
Both COD and BOD7 decrease rapidly. For soft-
wood, the first condensate has a COD around
9000 mg/1, which drops to below 3000 mg/l when
20% of the stillage has been evaporated. The BOD7
for softwood and willow is 65-75% of the COD.
The slight increases in COD and BOD7 for willow at
high relative amounts of evaporated liquid are
explained by entrainment of non-volatile com-
pounds, e.g. glycerol, in the condensate. In the latter
stage of the concentration of the liquid in the evap-
orator, the viscosity of the liquid increased, which
resulted in deposition of the solutes on the walls of
the separator, thereby increasing the loss by evap-
oration. The problem is more noticeable when small
amounts of liquid are to be concentrated, since the
loss will be relatively greater. In this study the loss
~ Glucose
............... /
 Recirculating process water in ethanol production 147

25.00 & ~ |
o Arabhlos, I

20.00
-~ , , , . ~
. Bhsnol
II

l 1
Time [hi

Fig. 3. Concentration of sugars and ethanol as functions of time during fermentation.

was 6.3 wt%. The problem would be avoided in a The values of COD and BOD7 for the stillage
continuous commercial evaporation unit, e.g. a fall- stream and the evaporation residue for softwood
ing-film evaporator. and willow are shown in Table 2.

, # CO0-Softwood I
- BOD-Soflw ood I

...+.. coo.o,, I
~6000 .. -x... BOD-WIIow l
5000
4000 '\,
~3000
0
2000 . . . . -_- . . . . -&
l- ....-X
1000 .o.O..o.

0 I l l I I I I I
0 10 20 30 40 50 60 70 80 90
Rnlatlve m o u n t of evlporlted liquid [%]

Fig. 4. Amounts of COD and BOD7 in the evaporation condensates for softwood (this study) and willow (Palmqvist et al.,
1996).

2~able 2. The COD and BOD7 for the stillage stream and the evaporation residue for sol~ood and willow
Stream COD BOD7 Percentage of
(mg/I) (mg/l) stillage
Stillage, softwood 26500 12800 --
Residue, softwood 182000 86900 17.5
Stillage, willow 33 300 19 800 --
Residue, willow 208 000 95 400 10.7
148 M. Larsson, M. Galbe, G. Zacchi

Due to the problem with the evaporation The specific heat at various dry-matter contents in
described earlier, the percentage of stillage for the the evaporation residue at 25C and atmospheric
evaporation residue is defined as the difference pressure are shown in Fig. 5. Data were fitted to a
between the amount of stillage and the total amount polynomial using a least-squares fit, yielding the fol-
of evaporation condensate. The higher values of lowing expression:
COD and BOD7 for willow compared with softwood
can be accounted for by a higher content of xylan in Cp = 4.180-4.675x+5.722x3-3.419x 4 (1)
the raw material. During pretreatment, xylan is
hydrolysed into xylose, which cannot be converted (x: kg DM/kg solution; Cp: kJ/kgC). The elevations
into ethanol with the strain of S. cerevisiae used in of the boiling point (BPE) of two evaporation resi-
the fermentation. dues with different dry-matter contents are shown in
The inhibitory effects of stillage, the five evapora- Fig. 6. Data were interpolated using the expression
tion condensate fractions and the evaporation
residue were measured by fermentation. Ethanol BPE = 3.3798x+ 13.10x 2 (2)
yields and final productivities are listed in Table 3.
Due to difficulties in the glucose analysis of the (x: kg DM/kg solution; BPE: C).
evaporation residue, the ethanol yield for the resi-
due is based on a calculated initial glucose
concentration. Simulation
The fermentability of the residue was, unfortuna- The experimental data were used in the simulation
tely, performed with a residue which was diluted five of a six-effect evaporation unit. The configuration of
times, resulting in a non-volatile concentration only the unit, shown in Fig. 7, is based on the unit
1.5 times that of the stillage stream. This was actually used for the evaporation of black liquor by
repeated 3 months later, with a non-diluted residue, Nym611a AB, a pulp and paper mill in the south of
which had been stored at 4C and resulted in a yield Sweden. The feed enters effect number 4 and the
of 0.41 g ethanol/g glucose and a productivity of liquid passes co-currently to effect number 6. From
3.0 g ethanol/l h. effect 6 the liquid is taken to effect 3 and then
counter-currently to effect 1. The temperature of the
cooling water is assumed to be 55C which, with a
Table 3. Ethanol productivity and yield in the fermenta- temperature difference of 5C in the condenser,
tion of fractions from the evaporation stage results in a pressure of 0.20 bar in effect number 6.
Percentage of Productivity Yield Saturated live steam of 3 bar (133.5C) has been
stillage (g EtOH/I h) (g EtOH/g chosen as the heating medium in effect 1. The calcu-
glucose)
lations are based on the assumption that all effects
Fraction 1 5.0 4.5 0.41 have the same area. The overall heat transfer coeffi-
Fraction 2 8.9 5.2 0.40 cients were assumed to vary between 3.2 and
Fraction 3 11.5 5.2 0.39
Fraction 4 10.1 5.1 0.38 2.0 kW/m2C, with the highest value for effect 1 and
Fraction 5 47.0 5.7 0.40 the lowest for effect 6.
Stillage -- 6.7 0.34 The feed to the evaporation unit has a flow rate
Residue 17.5 6.6 0.35 of 100 metric ton/h, corresponding to a wood intake
Reference -- 4.2 0.37 (DM) of 100000 metric ton/year and an ethanol pro-

4.5

3.5

~2.5

12
~
V)
1.5
1
0.5
I ~ I I I
0.1 0.2 0.3 0.4 0.5 0.6

Dry m alter content [kg/Izg]

Fig. 5. Specific heat for the evaporation residue as function of dry-matter content.
 Recirculating process water in ethanol production
os
==
3 2
1
0 0.1
Fig. 6. Boiling point elevations of evaporation residues at atmospheric pressure.
duction of 18400 metric tons/year. The temperature
and dry-matter content of the feed are 40C and
4wt%, respectively. The product flow rate is 6
metric tons/h with a dry-matter content of 65 wt%.
The enthalpy for the feed is calculated using the
specific heat at 25(2 and assuming that the variation
in specific heat with temperature is negligible. The
boiling point elevation is assumed to be dependent
only on the dry-matter content in the liquid. In the
energy calculations the condensate streams are
treated as pure water.
The calculated values of flow rates, temperatures
and COD for the various condensate streams leaving
each effect are shown in Table 4. The live steam
consumption is 23.5 metric ton/h, resulting in a
steam economy of 4 kg evaporated vapour/kg live
steam. For these operating conditions, the heating
areas were found to be 380 m 2 for each effect.
DISCUSSION
The overall ethanol yield obtained in this study, 53%
of theoretical, must be considered as low. A higher
yield is required in order to produce ethanol to a
competitive cost. Enhanced ethanol yield is expected
Live steam
Fig. 7. Schematic flowsheet for the evaporation unit.
149
0.2 0.3 0.4 0.5 0.6
Dry matter content [kg/kg]
when the hydrolysis will be performed in a fed-batch
mode and the fermentation in a continuous mode.
The main object with this study was, however, not to
present optimal conditions to obtain a high ethanol
yield, but to investigate the inhibitory effect on fer-
mentation when process streams are recycled.
The first fermentation test with diluted evapora-
tion residue showed that the ethanol yields for the
fractions containing non-volatile compounds, stillage
and residue was lower than for the volatile fractions
and the reference. The results were similar to those
obtained by Palmqvist et al. (1996) for willow. The
second test with evaporation residue that has been
stored at 4C for 3 months showed a decrease in
productivity but an increase in ethanol yield. The
yield was of the same magnitude as for the volatile
fractions. This shows that something had happened
with the residue during storage and that the inhibi-
tors might have been degraded. In an industrial
process, 3 months of storage is not an economical
alternative, so fresh residue, with its negative impact
on fermentation, must be excluded from the process.
Further research is necessary to identify the com-
pounds that inhibit fermentation. We have found
that the non-volatile fraction of the distillation stil-
lage contains compounds inhibitory to fermentation.
Cooling
water
150 M. Larsson, M. Galbe, G. Zacchi
Table 4. Results from simulation of the evaporation unit
Condensate Flow Temperature COD
from rate (C) (mg/l)
effect (metric tons/h)
2 22140 114.4 3000
3 20664 101.7 3000
4 16 992 89.1 3000
5 10260 77.4 11000
6 11304 70.0 5200
Condenser 12 492 60.0 3400
To avoid accumulation of non-volatile compounds
in the process, evaporation is an alternative when
process streams are recirculated. If the evaporated
fractions are characterised in terms of COD, BOD7,
sugar content and fermentability, process streams
suitable for recirculation can be chosen.
The benefits using evaporation are two-fold:
potential inhibitors are removed and the amounts of
COD and BOD7 in the disposal stream are lowered
to less than one-tenth of the values in the stillage
stream. The removal of inhibitors is necessary to
utilise the raw material efficiently and the reduction
of COD and BOD7 in the disposal stream decreases
the cost of further wastewater treatment. Results
from the simulation of a six-effect evaporator unit
show that the condensate streams most suitable for
disposal are streams leaving effect 2, 3, 4 and the
condenser. The values of COD and BOD7 for these
streams are similar, about 10% of stillage value. The
non-volatile evaporation residue can be burned and
used as fuel for the generation of process steam.
The drawback with evaporation is the live steam
consumption, 10.2 metric ton/metric ton of ethanol,
which makes evaporation very energy demanding.
This should be compared with the live steam con-
sumption in distillation, 2.8 metric ton/metric ton of
ethanol. The steam consumption in the evaporation
unit must therefore be lowered. Various energy-
saving techniques, such as preheating of internal
liquid streams and the feed to the evaporation unit,
as well as alternative configurations of the evapora-
tion unit, are available. Total optimisation of the
evaporation unit is, however, at best expected to
improve the steam economy from 4.0 to 5.5 kg evap-
orated liquid/kg live steam for a six-effect
evaporation unit (Jernqvist, personal communica-
tion). Further reduction in steam consumption can
be achieved by increasing the dry-matter content in
the feed, by recirculating process streams or hydrol-
ysing at higher dry-matter content. Combinations of
these might also be an alternative. Depending on
how recirculation is performed, the ethanol concen-
tration in the fermentation broth may increase
(Galbe and Zacchi, 1993), which decreases distilla-
tion costs.
The integration of a stripper in the evaporation
unit is an alternative which will further reduce total
steam consumption. Another alternative is precon-
centration of the distillation stillage with reverse
osmosis, which reduces energy consumption in the
evaporation step.
All of these energy-saving alternatives require fur-
ther investigation to make evaporation a
cost-effective alternative for closing the water
balance in the process.
ACKNOWLEDGEMENTS
The authors wish to thank the Swedish Ethanol
Development Foundation (SSEU) and The Swedish
National Board for Industrial and Technical
Development (NUTEK) for their financial support.
REFERENCES
Busche, R. M., Scott, C. D., Davison, B. H. & Lynd, L. R.
(1992). Ethanol, the uimate feedstock. A technoeco-
nomic evaluation of ethanol manufacture in fluidized
bed reactors operating with immobilized cells. Appl. Bio-
chem. Biotechnol., 34/35, 395-417.
Clark, T. A. & Mackie, K. L. (1987). Steam explosion of
the softwood Pinus radiata with sulphur dioxide addi-
tion. I. Process optimisation. J. Wood Chem. Technol.,
7(3), 373-403.
Esterbauer, H., Hayn, M., Jungschaffer, G., Tafratzhofer,
E. & Schurz, J. (1983). Enzymatic conversion of lig-
nocellulosic materials to sugars. J. Wood Chem. TechnoL,
3(3), 261.
Galbe, M. & Zacchi, G. (1991). Simulation of ethanol
production processes based on enzymatic hydrolysis of
lignocellulosic materials using ASPEN PLUS. Appl. Bio-
chem. Biotechnol. , 34/35, 93-104.
Galbe, M. & Zacchi, G. (1993). Simulation of ethanol
production based on enzymatic hydrolysis of lignocellu-
losic material using ASPEN PLUS. Appl. Biochem.
Biotechnol., 34/35, 639-649.
Grethlein, H. E., Allen, D. C. & Converse, A. O. (1984).
A comparative study of the enzymatic hydrolysis of the
acid-pretreated white pine and mixed hardwood. Bio-
technol. Bioengng, 26(12), 1498.
Hinman, N. D., Schell, D. J., Riley, C. J., Bergeron, P. W.
& Waer, P. J. (1992). Preliminary estimate of the cost of
ethanol production for SSF technology. Appl. Biochem.
Biotechnol., 34/35, 639-649.
Jernqvist, A., Dept of Chemical Eng., Lund University.
Personal communication.
Jurasek, L. (1979). Enzymatic hydrolysis of pretreated
aspen wood. Develop. Ind. Microbiol., 20, 177.
Ladisch, M. R. (1983). Enzymatic hydrolysis of biomass.
Enzyme Microb. Technol. 5, March.
Lou, Y. (1994). The engineering and economic evaluation
for chemical utilisation of biomass. Renewable Energy,
5(2), 866-874.
Lynd, L. R., Cushman, J. H., Nichols, R. J. & Wyman, C.
E. (1991). Fuel ethanol from cellulosic biomass. Science,
251, 1318-1323.
Mes-Hartree, M., Dale, B. E. & Craig, W. K. (1988).
Comparison of steam and ammonia pretreatment for
enzymatic hydrolysis of cellulose. Appl. Microbiol. Bio-
technol., 29(5), 462-468.
Palmqvist, E., Hahn-H~gerdal, B., Galbe, M. and Zacchi,
G. (1996a). The effect of water soluble inhibitors from
steam-pretreated willow on enzymatic hydrolysis and
ethanol fermentation. Enzyme and Microbial Technol.,
19, 470-476.
 Recirculating process water in ethanol production
Palmqvist, E., Hanh-H~igerdal, B., Galbe, M., Larsson, M.,
Stenberg, K., Szengyel, Z., Tengborg, C. and Zacchi, G.
(1996b). Design and operation of a bench-scale process
development unit for the production of ethanol from
lignocellulosics. Biores. Technol., 58, 171-179.
Palmqvist, E., Meinander, N. and Hahn-H[igerdahl, B.
(1997). Modelling the effect of acetic-acid, furfural, and
p-hydroxybenzioc acid on ethanol fermentation by
industrial yeast's, using a 3-factorial experimental
design. Biotech. Bioeng. (submitted).
Saddler, J. N., Schwald, W., Smaridge, T., Chan, M. and
Breil, C. (1989). The influence of SO2 impregnation and
fraction of product recovery and enzymatic hydrolysis of
steam-treated sprucewood. Commission of the European
Communities, Enzyme System, p. 321.
Saddler, J. N., Brownell, H. H., Clermont, L. P. & Levitin,
N. (1982). Enzymatic hydrolysis of cellulose and various
pretreated wood fractions. Biotechnol. Bioengng, 24,
1389.
SIS (The Swedish Standardisation Committee). (1990).
Determination of Biochemical Oxygen Demand, BOD7, of
151
Waste Water - - Dilution Method. SIS < Stockholm, Swe-
den.
Suurkuusk, J. & Wads6, I. (1974). Design and testing of a
improved precise drop calorimeter for the measurement
of the heat capacity of small samples. J. Chem. Thermo-
dyn., 6, 667.
The Swedish National Board for Industrial and Technical
Development (NUTEK) (1994). Energy Crops-94, pp.
1102-2574 (in Swedish).
US Department of Energy (1993). Assessment of costs
and benefits of flexible and alternative fuel use in the
US transportation sector. Technical Report 11: evalu-
ation of a potential wood-to-ethanol process (Report
No. DOE/EP-0004). US DoE, Washington, DC.
von Sivers, M. & Zacchi, G. (1996). Ethanol from lig-
nocellulosics - - a review of the economy. Biores.
Technol., 56, 131-140.
Wright, J. D., Wyman, C. E. & Grohman, K. (1988).
Simultaneous saccharification and fermentation of lig-
nocellulose: process evaluation. Appl. Biochem.
Biotechnol., 18, 75.