Sie sind auf Seite 1von 8

Separation and Purication Technology 124 (2014) 1825

Contents lists available at ScienceDirect

Separation and Purication Technology


journal homepage: www.elsevier.com/locate/seppur

Novel dual extraction process for acetonebutanolethanol fermentation


Antti Kurkijrvi a,, Juha Lehtonen a, Juha Linnekoski b
a
Aalto University, School of Chemical Technology, Department of Biotechnology and Chemical Technology, P.O. Box 16100, 00076 Aalto, Finland
b
VTT Technical Research Centre of Finland, Chemical Processes and Models, Espoo FI-02044 VTT, Finland

a r t i c l e i n f o a b s t r a c t

Article history: One reason why butanol fermentation is not economically feasible has to do with the energy intensity of
Received 2 July 2013 the product separation from the broth. In this study, a new approach for liquidliquid extraction is pre-
Received in revised form 19 December 2013 sented. Several solvents, previously ignored because of their non-biocompatibility, were tested and a con-
Accepted 1 January 2014
tinuous process utilizing a novel, dual extraction with solvent regeneration is proposed. The optimal
Available online 15 January 2014
solvents for this process are from nine to eleven carbon alcohols used in conjunction with alkanes of
approximately the same size. The ability to use non-biocompatible, but quite effective solvents offers a
Keywords:
signicant advantage, which is why a very low energy consumption of 3.76 MJ kg 1 butanol is achieved
Downstream processing
Extraction
with this new process.
Distillation 2014 Elsevier B.V. All rights reserved.
Simulation
Energy

1. Introduction conducted their research with the easiest possible setup, thus con-
tributing very little in terms of actual process enhancement. For
1-Butanol can be produced from biomass via acetone, butanol, example, when using extraction, methods where separation occurs
and ethanol (ABE) fermentation. During this process, solventogenic simultaneously in the reactor have received much attention in the
Clostridium spp. bacteria produce solvents, which are traditionally existing literature despite criticism from other authors, who point
separated from the broth by distillation. This is an effective, robust, out that this is unpractical for large-scale production [14]. Another
and well-known technology, but at the same time it is energy issue that is not often considered in the literature has to do with
intensive [1]. Since the production of butanol by fermentation separating ABE products from the extraction solvent. Researchers
has been economically uncompetitive compared to the synthetic often suggest using very high boiling extractants as the extraction
production route in the last decades, it is very important to mini- solvent for ABE fermentation products. Such solvents include ionic
mize the operational costs of the fermentation route by using novel liquids [8], vegetable oils [15], C-20 guerbet alcohol [16], and oleyl
technologies [2,3]. Many alternative, cheaper separation methods alcohol [1]. Using these kinds of solvents is not very economical, as
for replacing the direct distillation have been reported in the exist- high temperatures and low pressures are needed in distillation,
ing literature, but so far none of them has been adopted on an which is energy intensive. And other methods, such as ash, offers
industrial scale [3]. Some suggested methods for this purpose, only one separation stage which is not enough to get to the desired
along with the corresponding energy requirements, are as follows: product purity [13]. Therefore smaller, more polar extractants,
24.2 MJ kg 1 for steam stripping [4], 18.4 MJ kg 1 for traditional di- which are generally considered to be toxic to microbes, would be
rect distillation [5], 13.3 MJ kg 1 for extraction with oleyl alcohol a much more energy efcient option for extraction solvents when
[5], 13.8 MJ kg 1 for gas stripping [4], 8.2 MJ kg 1 for adsorption the whole process is considered. The purpose of this study is to re-
desorption [4], and 4.8 MJ kg 1 for extraction with mesitylene port results using a method that shows how effective, non-biocom-
[5]. Other methods have also been suggested: These include per- patible solvents can be used to extract ABE fermentation products
vaporation [6], perstraction [7], critical uid extraction [8], adsorb- in a continuous process.
tion [4], hollow-ber reactors [9], reverse osmosis [10], liquid
membranes [11], salt-induced phase separation [12], and continu- 2. Materials and methods
ous ash distillation [13].
Even though ABE fermentation has received a great deal of aca- 2.1. Verication of the used liquidliquid equilibrium (LLE) methods
demic attention in recent years, a large number of authors have
Experiments to verify extraction simulations were performed.
Corresponding author. Tel.: +358 505300094. An aqueous mixture containing 1.2, 0.6, and 0.2 mass% of butanol,
E-mail address: antti.kurkijarvi@aalto. (A. Kurkijrvi). acetone, and ethanol, respectively, was mixed at a mass ratio of 4:1

1383-5866/$ - see front matter 2014 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.seppur.2014.01.007
A. Kurkijrvi et al. / Separation and Purication Technology 124 (2014) 1825 19

with the solvents tested. The experiments were kept at 37 C with


constant mixing for 24 h, after which both of the phases were ana-
lyzed using a gas chromatograph with a ame ionizing detector
(GC-FID). The organic phase was further analyzed using a GC with
mass spectrometer as a detector (GCMS) to analyze the water
content. The amount of water in the aqueous phase was calculated
from material balance. The solvents selected for verifying the LLE
methods were 1-heptanol, 1-octanol, and 1-decanol. These sol-
vents were selected because they present different polarities with-
in the scope of this work. The partition coefcients were calculated
as the ratios of the mass fractions of the components in organic and
aqueous phases.

2.2. Process simulations

Simulations were performed using Aspen plus software. The


UNIFAC-LL activity coefcient model was used to model the Fig. 1. Novel dual extraction method.
liquidliquid equilibria (LLE), while the NRTL model, with the pure
component and binary interaction parameters taken from Aspen
plus, was used for the vaporliquid equilibria (VLE). During the of ABE products and water. Since the ABE products are mainly sep-
simulations, a continuous broth ow rate of 1000 kg h 1 was arated during the rst extraction, the amount of solvent used there
assumed. Initial butanol concentration was varied from 4 to primarily determines the extraction capacity, whereas the ef-
12 g l 1. The solvent mass ratio of the broth was 3:6:1 for acetone, ciency of the second extraction primarily determines the toxicity
butanol, and ethanol, respectively. The extraction columns were of the recycled fermentation broth. This kind of dual extraction
simulated using four ideal stages, and the extractions were per- system enables the use of extraction agents with distribution coef-
formed adiabatically at approximately 37 C. To make sure that cients that are several times higher than has previously been re-
the recycled broth was biocompatible, the amount of toxic solvent ported. This means that signicantly smaller amounts of solvents
in the broth after the extractions was xed at a very low value of are needed to achieve the desired extraction efciency. The ability
20 ppm. It can be assumed that this low solvent concentration does to use these very powerful solvents could signicantly improve the
not have any effect on the microbes [5]. In the heat integration, the economics of the downstream separations of an ABE process.
bottom streams of the distillation columns were used to heat up Using two extraction solvents to enhance the ABE fermentation
the feeds. The heat exchangers were assumed to operate with a is not an original or new idea. But usually the two solvents are
3 C temperature difference between the incoming cold stream mixed together in order to lower the toxicity of a high-capacity sol-
and the outgoing hot stream. The distillation columns consisted vent [17]. A method where a non-biocompatible solvent could be
of 40 stages and the pressure prole inside the columns was from used without fear of inhibiting the growth or even of killing the mi-
1.3 to 1.0 bar. The ow rate of the rst toxic solvent was varied and crobes has so far not been reported in the existing literature. The
the energy needed per kilogram of separated butanol was calcu- problem in using this kind of dual extraction process is that the
lated. In this way, the energetically optimal operating point could product mixture consists of not one, but two additional compo-
be found. nents that need to be separated from the mixture. If distillation
is the choice for the downstream purication method, this means
that additional distillation columns will be needed.
2.2.1. Dual extraction method
Many thorough articles have been published on the solvent
selection in ABE product extraction. However, extractants that 2.2.2. Solvent selection
are slightly soluble in water have been almost completely ignored. Prior studies have concluded that the optimal extraction solvent
In traditional extraction systems, these kinds of solvents would be- is non-toxic for the microorganisms used, immiscible in water,
come expensive, since much of the solvent would be lost during non-emulsion forming, non-azeotrope forming, inexpensive, read-
the extraction. Another important reason is that these solvents ily available and that it should have a high afnity towards butanol
would most likely be poisonous to the microbes. However, these [1,18]. The biocompatibility issue has been the most severe con-
problems can be overcome if the toxic solvent dissolved in the fer- straint in solvent selection [17,19]. On the other hand, the water
mentation broth is extracted in another extraction unit, as shown solubility of a solvent goes usually hand in hand with its biocom-
in Fig. 1. The aqueous, butanol-rich reactor output stream, BROTH, patibility: for the most part, the more water soluble a solvent,
is extracted in the Extraction1 using a non-biocompatible solvent, the higher its toxicity to the microbes. This also means that bio-
which has a high distribution coefcient for butanol. The solvent compatibility is inversely linked to the selectivity and capacity of
containing the extracted ABE products and some water continues the extraction process. In other words, more polar, non-biocompat-
as stream ORG1. The bottom ow of Extraction1, AQ1, is the aque- ible solvents can dissolve larger amounts of ABE products and
ous fermentation broth leaving the extraction column and it con- water than less polar, non-toxic solvents [20]. Ezeji et al. [1] claims
tains the ABE products, which were not extracted with traces of that water in the product stream will affect the product separation
the toxic extraction solvent. To remove this non-biocompatible sol- costs. For this reason, its minimization is essential.
vent from the broth, a second extraction is performed. The non- To nd the best solvent for the rst extraction, a number of
toxic extractant, SOLV2, removes traces of the toxic solvent from studies were reviewed [5,15,17,18,2126]. The chemical groups
the broth so that the fermentation broth can be safely recycled considered in these articles were alkanes, alkenes, aldehydes, ke-
back to the reactor. After both extractions the aqueous fermenta- tones, aromatic components, triglycerides, acids, alcohols, esters,
tion broth, AQ2, contains the non-extracted ABE products, and ethers, halogen components, and chemical mixtures that are dif-
small amounts of the extraction solvents. The solvent stream of cult to categorize, such as gasoline and kerosene. Easily evaporable
the second extraction, ORG2, contains the solvent used in the solvents such as diethyl ether were ignored because the conditions
second extraction, the extracted toxic solvent and trace amounts would need to be rather harsh before they could be utilized in a
20 A. Kurkijrvi et al. / Separation and Purication Technology 124 (2014) 1825

liquidliquid extraction. Furthermore, Rofer et al. [27] stated that tained solvents from both extractions, a further distillation column
esters have a tendency to form emulsions with aqueous liquids un- was needed to separate them. A partial condenser was utilized in
der agitation. As a result, all esters were ignored since the emulsion Distillation2. This means that the distillate was not condensed to
formation could render the extraction useless. To only investigate minimize the energy consumption in the following distillation.
the most efcient extractants, the solvents with a distribution Furthermore a condenser with a decanter was utilized in Distilla-
coefcient above nine were selected for further study. As a refer- tion1, which enabled a liquidliquid phase separation in the con-
ence, the solvent which has been stated to be the standard for denser. This was necessary because a fraction of water was
ABE liquidliquid extraction, oleyl alcohol, has a distribution coef- extracted in the rst extraction column. In the second option, pre-
cient value of only slightly above three [15,23,25]. According to sented in Fig. 2b, the organic streams from both of the extraction
existing literature, alcohols are very effective solvents for ABE units were mixed in a decanter before distillation. Similarly to
product extraction. This is the reason why all primary C6C11 alco- the rst conguration, a partial condenser was utilized in Distilla-
hols were included for further study, even though only n-hexanol tion1. As can be seen from Fig. 2, the rst method requires three
[17,23], n-heptanol [23,25], and n-octanol [23,25] have a distribu- distillation columns and the latter option only two. Fig. 2 does
tion coefcient above nine. Isodecanol was also added to see how not show heat integrations or the extraction columns.
branching of the carbon chain effects solvents performance. Other In the simulations the mass ow of the broth was 1000 kg h 1
solvents that also fulll this criterion are phenol [25], tert-amyl and the extractant ows were 200 and 30 kg h 1 for Solvent1
alcohol [22], and pentanal [22]. Of these selected solvents, phenol and Solvent2, respectively. In the distillations, a maximum of
was discarded because of its toxicity to humans. Other extractants 0.005 kg h 1 of the extraction solvents were allowed into the
that were discarded included tert-amyl alcohol, pentanal and product ow. The solvent pairs used for these simulations were
alcohols with seven carbons or less. They were discarded because octanolnonane and decanoldecane. These solvent pairs were se-
they indicated severe azeotropic behaviors with water or ABE, lected, because they represent a large variety of different solvent
resulting in a complicated distillation conguration according to pairs.
our simulations.
The second extraction stage of this process should leave the
2.2.5. Direct distillation
aqueous broth practically free of the toxic solvent from the rst A traditional separation method for ABE products, direct distil-
extraction step. The removal of this non-biocompatible solvent is lation, was simulated to generate a comparison point for energy
the most important aspect of the second extraction process, while consumption. In an industrial direct distillation setup, the rst col-
the ABE product capacity is only of secondary importance. Also, the umn removes the majority of water from the ABE products [5]. The
solvent for this second extraction process should have low solubil- reason why not all of the water can be removed has to do with the
ity in water to make it non-toxic for the microbes. Alkanes as a azeotrope between water and butanol. This means that the product
chemical group fulll these criteria. Furthermore, to facilitate the ow from the direct distillation setup contains a minimum mole
solvent regeneration by distillation, the second solvent should fraction of 0.763 of water at ambient pressure [29]. This water
not demonstrate severe azeotropic behavior with water or ABE makes it more difcult to purify the different ABE components
products. This ruled out the alkanes with nine carbons or less. So and in most congurations at least one product will be aqueous
the alkanes considered need to be at least ten carbons long. On mixture. A process where the ABE product mixture would contain
the other hand, this extractant should not have too high a boiling less water would be both desirable and more economical. Purifying
point so as to keep the distillation costs at a minimum. For this rea- the different ABE products via distillation is a well-known technol-
son, the solvents considered here are between ten and twelve ogy, which has already been discussed in the existing literature [5].
carbons long. One other extractant were added to this study: Therefore, it is not presented here in greater detail. Most of the
mesitylene. It was suggested as an extraction solvent for ABE prod- processes presented in the existing literature separate the ABE
uct purication by Kraemer et al. [5]. Naturally, other solvents products into pure components. To make it possible to compare
could also be used. the energy consumption of the process proposed here with other
processes presented in the existing literature, the energy needed
2.2.3. Growth inhibition and extraction to separate the ABE components from each other was added to
Of all the ABE components, butanol is the most growth inhibit- the total energy consumption. This information was obtained from
ing. For this reason, the concentration of butanol in the recycled the study by Kraemer et al. [5] and the additional energy consump-
rafnate after the extraction steps should be kept as low as possi- tion was 0.57 MJ kg 1 butanol.
ble to ensure the highest possible fermentation rate. A threshold If the separation efciency of direct distillation is reduced,
value for butanol inhibition has been reported to be between 4 either more water will be transferred to the product stream or
and 4.8 g l 1, below which the concentration does not affect the some of the products will remain in the aqueous fermentation
growth. The increasing concentration of butanol enhances the broth. In other words, the separation would take less energy, but
inhibitory effect and decreases the growth and productivity rates the separation efciency would be diminished. These cases were
until a level of 1216 g l 1 is achieved, where butanol totally inhib- studied to see if it was more energy efcient to operate the direct
its the growth. The other main ABE fermentation products, acetone distillation with reduced efciency.
and ethanol, also inhibit growth, but at levels which are much
higher. A 50% inhibition in the growth rate is reported to occur
when either component is at approximately 40 g l 1. Total inhibi- 3. Results and discussion
tion occurs when acetone is at 70 g l 1 and ethanol is between
50 and 60 g l 1 [15,28]. 3.1. Verication of the LLE model

2.2.4. Solvent regeneration method In Table 1, the measured and experimental distribution coef-
Two possible distillation column congurations were tested for cients for ABE products are reported at 37 C when using heptanol,
the solvent regeneration. A decanter was utilized in both cases to octanol, and decanol as solvents. The LLE values predicted using
remove excess water. In the rst method, presented in Fig. 2a, both the UNIFAC-LL model are in good agreement with the experimental
solvent streams from the extraction units were distilled separately. values. However, a certain degree of error exists: The simulations
Because the organic stream from the second extraction unit con- underestimate the distribution coefcient of butanol and
A. Kurkijrvi et al. / Separation and Purication Technology 124 (2014) 1825 21

Fig. 2. Possible solvent regeneration methods: (a) presents the three column method and (b) the two column method.

overestimate the distribution coefcients for acetone and ethanol. the two and three column systems, respectively. For the decanol
This means that the extraction of butanol is not as effective in the decane system, the corresponding values were 5.61 and
simulations as in reality, whereas the extraction of acetone and eth- 4.30 MJ kg 1 butanol. Because of 23.431.5% lower energy con-
anol are more effective in the simulations than in reality. The accu- sumption was obtained with the three column system, it was se-
racy of the simulations is best for butanol, which is quite important lected as the solvent regeneration method.
because butanol is the key component in the process simulations. Be-
cause of this, the error in the process simulations is small enough to 3.4. Process description
consider the simulations reliable. The solubility of water in the sol-
vent and the solubility of the solvent in water are also reported in Ta- The process proposed in this article is presented in Fig. 3. The
ble 1. The values estimated using the UNIFAC-LL model are in good only stream leaving the process is the product ow, which contains
agreement with the experimental values. It can be assumed that ABE products, water, and trace amounts of extraction solvents. The
the accuracy is on a similar level for the solvents that were not in- production of the ABE components is taken into account by adjust-
cluded in Table 1 and that the values obtained during the simulations ing the ow rate of the stream ABE, which contains the fermenta-
are reliable and in good agreement with experimental values. tion products in the ratio produced by the microbes. This mass ow
was adjusted so that the ow leaving the fermenter contained from
4 to 12 g l 1 of butanol, to cover the concentrations commonly
3.2. Direct distillation
achieved in the existing literature [18]. The extraction efciency
for the different solvents in the fermentation broth is buta-
The simulations indicate that lowering the separation efciency
nol > ethanol > acetone > water. Since acetone and ethanol are not
of direct distillation does not improve the economics of the pro-
extracted as well as butanol, their concentrations are slightly ele-
cess. When more water was allowed to exit with the ABE products,
vated in the recycled fermentation broth. The maximum levels of
then the energy saved by the lower reux ratio was lost because of
these solvents in the broth would most likely inhibit the growth
the need to vaporize more water into the product stream. This re-
of the microbes, but luckily the most reasonable operating condi-
sulted in a situation where approximately the same amount of en-
tions for this process would be with relatively high ow rates of
ergy was used despite the reux ratio. Similarly, in the case where
the Solvent1, where these concentrations are low and no inhibition
some of the ABE products were allowed to stay in the broth, less
is to be expected [15,28]. During and after the rst extraction, the
ABE products were recovered when less energy was used. The en-
fermentation broth is saturated with the toxic solvent. Therefore,
ergy that was consumed for every kilogram of butanol that was
the concentration of this toxic solvent is practically constant in
recovered varied between 19.3 and 20.6 MJ kg 1. This means that
the broth leaving the rst extraction. This concentration is not
separating the ABE components from the broth using the tradi-
dependent upon the amount of solvent used in the rst extraction.
tional method requires approximately 54% of the combustion value
So changes made in the operational parameters of the rst extrac-
of butanol when the butanol concentration of the broth is 12 g l 1.
tion have little effect on the second extraction and subsequent
distillations. According to our simulations the temperature of the
3.3. Selecting the solvent regeneration method recycled broth is 36.8 C. The energy needed to increase its
temperature by 0.2 C is approximately 0.9 MJ h 1, which
According to the simulations the energy consumptions of the corresponds to 0.15 MJ kg 1 butanol, when initial butanol concen-
octanolnonane system were 8.71 and 5.97 MJ kg 1 butanol for tration is 6 kg h 1.

Table 1
The estimated and experimental values for distribution coefcients and the water solubilities for heptanol, octanol, and decanol (1 bar, 37 C).

1-Heptanol 1-Octanol 1-Decanol


Experimental Simulated Experimental Simulated Experimental Simulated
Dacetone 1.10 1.75 1.01 1.52 0.61 1.21
Dbutanol 11.26 10.30 9.95 9.12 7.17 7.09
Dethanol 1.10 1.30 1.02 1.13 0.56 0.86
Solubility of water in solvent (m-%) 6.93 7.49 6.65 5.58 3.93 3.49
Solubility of solvent in water (m-%) 0.73 0.40 0.08 0.15 0.03 0.02
22 A. Kurkijrvi et al. / Separation and Purication Technology 124 (2014) 1825

1
Fig. 3. Mass ows, temperatures, and compositions of process ows. Initial butanol concentration 6 kg h with decanol-mesitylene solvent pair. Amount of butanol in
recycle was set to 0.5 kg h 1.

The ow, DECANT, is the aqueous ow from a decanter. This 3.5.2. Selecting the non-biocompatible solvent
ow is not recycled to the reactor; rather, it is recycled to the rst Table 2 shows the solvent ows necessary for achieving a
extraction unit because the concentration of ABE products in the 95 mass% recovery of butanol from the broth. Only the ow of
ow is higher than in the reactor. This means that the ABE products the rst solvent is stated, because the extraction efciency was
in the ow do not cause growth inhibition in the reactor and will practically independent of the choice of the second solvent. The
still be recycled. Fig. 3 shows that the heat from the distillate amount of solvent needed increases as a function of the carbon
streams, DIST1, DIST2, DIST3, and PRODUCTS, is not utilized in chain length. The carbon chain length does not have a similar cor-
the heat integration. This is to reduce the separation cost of the relation with the energy consumption per kilogram of separated
ABE products into pure components in the downstream units, butanol. Instead, decanol consumes the least amount of energy;
which are not included here. therefore it is the most economical choice. Decreasing the carbon
chain length brings the boiling points closer to the ABE solvents,
3.5. Solvent selection and therefore, more energy is needed to separate them. Also these
alcohols have the following azeotropic molar fractions with water
3.5.1. Selecting the second solvent 0.32, 0.17 and 0.08 for octanol, nonanol and decanol, respectively.
The extraction of ABE components for the most part takes place This is the main reason, why decanol has the lowest energy con-
in the rst extraction column. The fraction of all ABE solvents sumption. Moreover, increasing the chain length decreases the
recovered from the second extraction unit ranges from 0.02 mass% polarity and increases the boiling point, both of which increase
when using undecane to 0.13 mass% when using mesitylene. Even the energy consumption. This is because the reduced polarity de-
though mesitylene appears to be the most efcient solvent out of creases the extraction capacity of the solvents, and the increased
all the solvents tested, the effect is insignicant when the whole boiling point makes the reboiler temperature higher in the distilla-
process is considered. Since the amount of energy needed to regen- tion. Branching of the carbon chain appears to have very little ef-
erate the second solvent was less than 1% of the total energy con- fect on the energy consumption, as can be seen when comparing
sumption for all of the solvents tested, it can be concluded that the the energy consumptions of 1-decanol and isodecanol. The proba-
selection of the second solvent is not a signicant factor on the ble reason is that the effect of one methyl group in a long carbon
operational costs of this process. chain is not enough to promote a signicant effect.

Table 2
Mass ows and energy consumptions with different solvents in the conditions where 95 mass% of butanol is extracted from the broth with initial butanol concentration of
12 g l 1. Second solvent: decane.

1-Octanol 1-Nonanol 1-Decanol 1-Undecanol Isodecanol


kg h 1 125.14 195.21 231.25 254.27 231.25
MJ h 1(Col1) 49.82 51.51 52.02 55.44 54.80
MJ h 1(Col2) 14.36 1.90 0.18 0.10 0.11
MJ h 1(Col3) 1.26 0.012 0.007 0.010 0.025
MJ kg 1(BuOH) 5.74 4.68 4.57 4.87 4.81
A. Kurkijrvi et al. / Separation and Purication Technology 124 (2014) 1825 23

3.6. Process simulations 20 ppm criteria. If this process was built in industrial scale this
ow ratio would, with all probability, be higher. One factor that
Table 3 presents the effect of initial butanol concentration on has a relatively signicant effect on this process is the decantive
the energy consumption when the amount of butanol in the re- condenser in Distillation1. Without this decanter, the energy con-
cycle stream is xed to approximately 0.5 kg h 1. It can be seen sumption of Distillation1 would be lower because the concentra-
that when initial concentration of butanol is reduced, also less tions of the ABE components in the process would be smaller.
Solvent1 is needed in the extraction. However, at the same time However, without the decanter almost half of the product stream
energy consumption increases slightly: a decrease in butanol con- would consist of water. This would make this process very much
centration from 12 to 4 g l 1 increases energy consumption by like the direct distillation process with respect to product purity.
approximately 0.86 MJ kg 1. As was stated before, the concentra- Hence, the removal of water is essential because it gives another
tions of acetone and ethanol are elevated after the reactor, be- advantage to this process compared with the traditional distilla-
cause they are not extracted as readily as butanol. Fig. 4 tion process.
presents the amount of ABE solvents in the broth and their corre- Fig. 5 demonstrates the energy consumption of ABE product
sponding ratios with different initial butanol concentrations and separation with different extractants. Every solvent is presented
extraction solvent to broth ratios. If the extractions were equally as a function of initial butanol concentration and butanol recovery
efcient for acetone, butanol and ethanol, the total ABE concen- percentage. As can be seen the energy consumption increases
trations would be 20, 16.67, 13.33, 10 and 6.67 g l 1, with initial when more butanol is extracted from the broth. However, it should
butanol concentrations of 12, 10, 8, 6 and 4 g l 1, respectively. be noted that it is not necessarily industrially feasible to only opti-
Also the mass-ratio of acetone, butanol and ethanol would be mize the energy consumption. A certain balance should be found
3:6:1. As can be seen from Fig. 4, the increase in acetone and eth- between energy consumption and productivity. The smallest en-
anol causes the total ABE concentrations to be increased by a fac- ergy consumption for this process is 3.19 MJ kg 1 butanol. It was
tor of 1.882.87, when using decanol-mesitylene solvent pair. achieved with decanol when the initial butanol concentration
Naturally this increase could be reduced by making the extraction was 12 g l 1, but in conditions where only approximately 20 mass%
solvent to broth ratio larger, but this might not be economical in of butanol was extracted from the broth. This means reducing the
all cases. The usage of other solvent pairs gives rise to very similar amount of butanol from 12 g l 1 to 9.6 g l 1, which is not enough
results. Fig. 3 presents the process mass ows, temperatures, and from the industrial or microbial point of views. However, the en-
mass fractions in a case where the solvent pair is decanol-mesit- ergy consumption increases only slightly up to a butanol recovery
ylene and the ow of Solven1 is adjusted so that 0.5 kg h 1 buta- of 70 mass%. After this, the energy consumption starts to increase.
nol is recycled back to the reactor. The elevated concentrations of Decreasing the initial butanol concentration also increases energy
acetone and ethanol can be seen clearly in stream BROTH. How- consumption: approximately 0.33 and 0.86 MJ kg 1 additional en-
ever, the concentrations are low enough not to affect the microbes ergy is needed if the butanol concentration is lowered from 12 to 8
[15,28]. The behavior of the two extraction units can also be ob- and 4 g l 1.
served: The rst extraction is effective for the ABE components, So far, the lowest energy consumption reported in the literature
while the second extraction unit mainly extracts the solvent orig- has been achieved with a mesitylene extraction. Kramer et al. [5]
inating from the previous extraction. In Fig. 3 the ratio of organic reported that it required 4.8 MJ kg 1; however, they used ten ideal
to aqueous ow is very small in the second extraction. This is be- steps for the extraction and ideal VLE behavior for the distillation.
cause it is the amount of extraction solvent that fullls the It is unlikely that this many ideal separation steps would be

Table 3
Effect of ABE concentration on energy consumption when using decanoldecane solvent pair with butanol mass fraction of 0.0005 in recycle stream.
1 1 1 1 1
4 kg h (BuOH) 6 kg h (BuOH) 8 kg h (BuOH) 10 kg h (BuOH) 12 kg h (BuOH)
1
kg h (Decanol) 177.02 192.12 212.13 228.05 241.39
MJ h 1(Col1) 22.45 28.81 37.23 44.99 53.06
MJ h 1(Col2) 0.18 0.18 0.18 0.18 0.18
MJ h 1(Col3) 0.007 0.007 0.007 0.007 0.007
MJ kg 1(BuOH) 5.49 5.27 4.99 4.76 4.63

0.06

0.05
Fracon of ABE in broth

0.04
Ethanol
0.03 Acetone

Butanol
0.02

0.01

0
6 g/l
6 g/l

4 g/l
6 g/l
12 g/l

10 g/l
8 g/l
8 g/l

8 g/l
10 g/l

4 g/l

12 g/l

8 g/l
12 g/l

10 g/l

6 g/l
4 g/l

10 g/l

12 g/l
4 g/l

Extracon solvent : Broth rao


1:4 1:5 1:6,667 1:10

Fig. 4. ABE solvents to broth ratios as a function of initial butanol concentration and extraction solvent to broth ratios. Solvent pair: decanol-mesitylene.
24 A. Kurkijrvi et al. / Separation and Purication Technology 124 (2014) 1825

4 g/l
7
(BuOH)
6 6 g/l
(BuOH)
5
MJ kg-1 8 g/l
(BuOH) 4
(BuOH)
10 g/l
3 (BuOH)
12 g/l
2
(BuOH)
1

40 %
40 %

20 %
97 %

40 %
90 %

97 %
70 %
20 %

20 %

95 %
97 %

40 %
70 %
95 %

97 %

70 %

95 %
40 %

90 %

97 %
95 %

20 %
70 %
90 %

20 %

90 %
90 %
95 %
70 %
1-Octanol 1-Nonanol 1-Decanol 1-Isodecanol 1-Undecanol

Fig. 5. Energy consumption with ve different extractants, six butanol recovery percentages and ve initial butanol concentrations.

achieved on an industrial scale [30]. For that reason, the process than 4 MJ kg 1 of butanol, which is the lowest energy consumption
proposed here only utilized 4 ideal stages. When ten ideal separa- rate reported in the existing literature.
tion steps was tested with decanol to broth ratio of 0.20, over
99.4 mass% butanol was recovered instead of only 92.4 mass%, Acknowledgements
when using four stages. Moreover, the ideal VLE does not take into
account the azeotropes and other un-idealities of the polar compo- The nancial support by Neste Oil and Stora Enso is gratefully
nents in the system. In other words, when using ideal VLE the sep- acknowledged. We also greatly appreciate the effort by these
arations are easier and the energy consumptions are lower. companies to patent the idea presented in this article.
When the separation of ABE components from each other is
taken into consideration, the optimal point for this process from
References
an energy point of view would require only 3.76 MJ kg 1 butanol.
This is lower than any energy consumption reported in the litera- [1] C.T. Ezeji, N. Qureshi, H.P. Blaschek, Butanol fermentation research: upstream
ture so far for ABE product separation. Even when operating at a and downstream manipulations, TCR 4 (2004) 305314.
point where 80 mass% of the butanol is extracted, requires only [2] P. Drre, New insights and novel developments in clostridial acetone/butanol/
isopropanol fermentation, Appl. Microbiol. Biotechnol. 49 (1998) 639648.
4.0 MJ kg 1 butanol, which is less than 11% of the combustion va- [3] E.M. Green, Fermentative production of butanol the industrial perspective,
lue of butanol. This is already very close to the 10% energy content Curr. Opin. Biotechnol. 22 (2011) 17.
of butanol, which is the acknowledged target for energy efciency [4] N. Qureshi, S. Hughes, I.S. Maddox, M.A. Cotta, Energy-efcient recovery of
butanol from model solutions and fermentation broth by adsorption,
of ABE product separation [5]. With lowered initial butanol con-
Bioprocess. Biosyst. Eng. 27 (2005) 215222.
centrations of 8 and 6 g l 1 the energy consumptions are 4.49 [5] K. Kraemer, A. Harwardt, R. Bronneberg, W. Marquardt, Separation of butanol
and 4.54 MJ kg 1, when 80 mass% butanol is recovered. These cor- from acetonebutanolethanol fermentation by a hybrid extraction distillation
respond to 12.4% and 12.6% of the combustion value of butanol. process, Comput. Chem. Eng. 35 (2011) 949963.
[6] F. Liu, L. Liu, X. Feng, Separation of acetonebutanolethanol (ABE) from dilute
The advantage of using solvents with extremely high distribu- aqueous solutions by pervaporation, Sep. Purif. Technol. 42 (2005) 273282.
tion coefcients is clear: If the solvents previously suggested in [7] N.G. Grobben, G. Eggink, F.P. Cuperus, H.J. Huizing, Production of acetone,
the literature are used, then several times more solvent is needed butanol and ethanol (ABE) from potato wastes: fermentation with integrated
membrane extraction, Appl. Microbiol. Biotechnol. 39 (1993) 494498.
to match the extraction capacity of the toxic solvents suggested in [8] L.D. Simoni, A. Chapeaux, J.F. Brennecke, M.A. Stadherr, Extraction of biofuels
our work. Naturally, this kind of increase in the amount of solvent and biofeedstocks from aqueous solutions using ionic liquids, Comput. Chem.
would mean remarkably higher operational costs. For this reason, Eng. 34 (2010) 14061412.
[9] R. Shukla, W. Kang, K.K. Sirkar, Acetonebutanolethanol (ABE) production in a
none of the published extraction-based processes can compete in novel hollow ber fermentor-extractor, Biotechnol. Bioeng. 34 (1989) 1158
terms of energy consumption with the process proposed here. On 1166.
the other hand, the process presented here requires more initial [10] A. Garcia, E.L. Lannotti, J.L. Fischer, Butanol fermentation liquor production and
separation by reverse osmosis, Biotechnol. Bioeng. 28 (1986) 785791.
investment in the equipment. It remains to be seen, whether the [11] M. Matsumura, H. Kataoka, M. Sueki, K. Araki, Energy saving effect of
reduced operational costs will promote this dual extraction process pervaporation using oleyl alcohol liquid membrane in butanol purication,
enough to be used in industrial scale someday or not. Bioprocess. Eng. 3 (1988) 93100.
[12] F.S. Santos, S.G. DAvila, M. Aznar, Salt effect on liquidliquid equilibrium of
water + 1-butanol + acetone system: experimental determination and
4. Conclusions thermodynamic modeling, Fluid Phase Equilib. 187188 (2001) 265274.
[13] A.P. Mariano, C.B.B. Costa, D.F. de Angelis, F.M. Filho, D.I.P. Atala, M.R.W.
Marciel, R.M. Filho, Dynamics of a continuous ash fermentation for butanol
The ability to use continuous fermentation coupled with production, Chem. Eng. Trans. 20 (2010) 285290.
efcient extraction solvents have a great potential to reduce the [14] A. Oudshoorn, L.A.M. van der Wielen, A.J.J. Straathof, Assessment of options for
selective 1-butanol recovery from aqueous solution, Ind. Eng. Chem. Res. 48
production costs of bio-butanol. It is possible to use these solvents
(2009) 73257336.
if a second extraction is applied to reduce the toxic effect. We [15] W.E. Barton, A.J. Daugulis, Evaluation of solvents for extractive butanol,
found that decanol was the best solvent for this process when used fermentation with CIostridium acetobutylicum and the use of poly(propylene
in conjunction with an alkane or similar component of approxi- glycol) 1200, Appl. Microbiol. Biotechnol. 36 (1992) 632639.
[16] S. Ishii, M. Taya, T. Kobayashi, Production of butanol by Clostridium
mately the same size. With this method, the energy consumption acetobutylicum in extractive fermentation system, J. Chem. Eng. Jpn. 18
of the ABE fermentation product recovery can be lowered to less (1985) 125130.
A. Kurkijrvi et al. / Separation and Purication Technology 124 (2014) 1825 25

[17] P.J. Evans, H.Y. Wang, Enhancement of butanol formation by Clostridium [24] Q. Li, H. Cai, B. Hao, C. Zhang, Z. Yu, S. Zhou, L. Chenjuan, Enhancing clostridial
acetobutylicum in the presence of decanol-oleyl alcohol mixed extractants, acetonebutanolethanol (ABE) production and improving fuel properties of
Appl. Environ. Microbiol. 54 (1988) 16621667. ABE-enriched biodiesel by extractive fermentation with biodiesel, Appl.
[18] L. Adhami, B. Griggs, P. Himebrook, K. Taconi, Liquidliquid extraction of Biochem. Biotechnol. 162 (2010) 23812386.
butanol from dilute aqueous solutions using soybean-derived biodiesel, Am. [25] A.M. Dadgar, G.L. Foutch, Evaluation of solvents for the recovery of Clostridium
Oil. Chem. Soc. 86 (2009) 11231128. fermentation products by liquidliquid extraction, Biotechnol. Bioeng. Symp.
[19] S.R. Rofer, H.W. Blanch, C.R. Wilke, Extractive fermentation of acetone and butanol Set. 15 (1985) 611620.
process design and economic evaluation, Biotechnol. Prog. 3 (1987) 131140. [26] F. Kollerup, A.J. Daugulis, Ethanol production by extractive fermentation
[20] A. Oudshoorn, C. van den Berg, C.P.M. Roelands, A.J.J. Straathof, L.A.M. van der solvent identication and prototype development, Can. J. Chem. Eng. 64 (1986)
Wielen, Short-cut calculations for integrated product recovery options in 598605.
fermentative production of bio-bulk chemicals, Process Biochem. 45 (2010) [27] S.R. Rofer, H.W. Blanch, C.R. Wilke, In situ extractive fermentation of acetone
16051615. and butanol, Biotechnol. Bioeng. 31 (1987) 135143.
[21] A. Ishizaki, S. Michiwaki, E. Crabbe, G. Kobayashi, K. Sonomoto, S. Yoshino, [28] D.T. Jones, D.R. Woods, Acetonebutanol fermentation revisited, Microbiol.
Extractive acetonebutanolethanol fermentation using methylated crude palm Rev. 50 (1986) 484524.
oil as extractant in batch culture of Clostridium saccharoperbutylacetonicum Nl-4 [29] W.L. Luyben, Control of the heterogeneous azeotropic n-butanol/water
(ATCC 13564), J. Biosci. Bioeng. 87 (1999) 352356. distillation system, Energy Fuels 22 (2008) 42494258.
[22] J.J. Malinowski, A.J. Daugulis, SaIt effects in extraction of ethanol, 1-butanol [30] P.C. Wankat, LiquidLiquid Extraction, in: P.C. Wankat (Ed.), Separation
and acetone from aqueous solutions, AIChE J. 40 (1994) 14591465. Process Engineering: Includes Mass Transfer Analysis, 3rd ed., Prentice Hall,
[23] W.J. Groot, H.S. Soedjak, P.B. Donck, R.G.J.M. van der Lans, K.Ch.A.M. Luyben, Upper Saddle River, 2011, pp. 499936.
J.M.K. Timmer, Butanol recovery from fermentations by liquidliquid extraction
and membrane solvent extraction, Bioprocess. Eng. 5 (1990) 203216.

Das könnte Ihnen auch gefallen