2. M. Errico et al.
possible alternatives. Superstructures (Agrawal, 2003) and individual splits (Rong,
2003) are the most common techniques used to develop as completely as possible set of
alternatives. Independently of the tool selected it is of paramount importance disposing
of methods or criteria for the flowsheet generation. Extractive distillation, applied to the
bioethanol production will be considered in details, since its energy efficiency was
already proved (Meirelles, 1992). The classical separation sequence, together with the
alternative configurations proposed in the literature, are described and, by the definition
of a systematic methodology, a connection between all the sequences is defined.
The first step of the procedure includes the definition of the simple column´s subspace
and the selection of the sequence that better matches with the pre-defined criteria. Once
that one or more promising sequences are selected, the corresponding complex
sequences are considered. The relationship between the simple column configurations
and the complex ones is an important tool that helps the designer in the alternative
generation process. It was already proved that, once a simple column sequence has been
selected, there is a specific subspace of complex configuration generated from it. This
result becomes even more significant considering the connection existing from the best
simple column sequence and the corresponding complex configurations. In other words,
the best complex configuration can be identified among the complex configurations
generated from the best simple column.
2. The simple column configurations
The configurations are mainly developed for the bioethanol purification, for this reason
1694.240 kmol hr-1
of a diluted ethanol-water solution containing 5% mol of ethanolis
considered as a feed for all the cases considered. All the simulations were performed by
means of the process simulator Aspen Plus V7.3. The NRTL method was utilized to
evaluate the activity coefficients. Ethylene glycol was chosen as solvent. The ethanol
minimum purity was set to 0.999 on molar fraction basis to enable its use in oil derived
gasoline blends.
All the simple columns in the sequences were first simulated utilizing the Winn-
Underwood-Gilliland short-cut method. Then the number of stages, feed location and
reflux ratio were optimized utilizing the sensitivity analysis tool implemented in the
RadFrac rigorous method. A solvent to feed ratio of 0.87 was employed in all the
configurations simulated. The installation cost evaluation was performed with Aspen
Plus Economic Analyzer. The distillation columns were considered equipped with sieve
trays spaced 0.6 m, 2.5 m for the bottom sump height and 1.25 m for the vapor
disengagement were also taken into account. For the auxiliary heat equipment, fixed
tube condensers and floating head kettle reboilers were considered. In each sequence the
solvent recovered is cooled from about 470 K to 303 K. This heat can be easily
recovered in the process, for this reason the cost of that cooler is not considered. Finally,
the capital cost was annualized considering a mean operational time of 10 years.
The simple column configuration recently introduced by Li and Bai, 2012 was utilized
as a reference for the comparison of the alternative sequences. As claimed by the
authors the importance to define energy efficient sequences for the ethanol-water
separation is nowadays related to the bioethanol production. For this reason, as reported
in Figure 1, the pre-concentrator column was included to their original configuration.
The main limit in proposing a standalone configuration is the impossibility to claim
about its absolute convenience in terms of energy consumption or capital costs.This
3. Optimal synthesis and design of extractive distillation for bioethanol separation: from
simple to complex columns 375
limitation comes out due to the different approach used to develop the alternative
configurations.
Figure 1: Reference simple column configuration
Alternative sequences are obtained by introduction one, two or three partial condensers
in the transfer streams between the columns. Moreover the possibility to eliminate the
fourth column and directly recycle the third column´s distillate to the prefractionator is
considered.
Employing of a partial condenser for transferring the azeotropic feed from the pre-
concentrator to the extractive column, has the benefit to maintain a high solvent
concentration on the feed tray and the trays immediately below. The possibility to
reduce the number of columns was introduced considering that the distillate of the
solvent recovery column has a concentration close to the main feed of the pre-
concentrator column. For this reason is possible to by-pass the forth column recycling
back the solvent column´s distillate directly to the first one.
This arrangement has the extra benefit to produce a single water stream and to increase
the ethanol recovery.
Among all the possibilities, the configuration reported in Figure 2 with a partial
condenser for transferring the azeotropic feed from the pre-concentrator column to the
extractive one and a vapor recycle from the solvent recovery column to the pre-
concentrator was selected as the best option in terms of energy consumption, capital
cost investment and solvent consumption.
This configuration realizes 26% savings in the total condenser duty, 22% in the total
reboiler duty and 16% in the capital costs. The three column sequence represents a valid
alternative to the classical configuration considered as a reference.
C1 C2 C3 C4
Number of stages 44 28 17 19
Reflux ratio (molar) 2.420 0.185 0.390 3.035
Feed stage 30 24 6 17
Solvent feed stage --- 5 --- ---
Column diameter (m) 1.37 0.78 0.51 0.27
Design pressure (kPa) 101.00 101.00 101.00 101.00
Condenser duty (kW) 5270.538
Reboiler duty (kW) 6307.069
Capital cost (k$ yr-1
) 161.3
Water 1
Az. feed
Fermentation
Broth
Solvent Recycle
Solvent
Make up
3
4
5
6
7
Ethanol
8
9
Water 2
Distillate Recycle
1
2
4. M. Errico et al.
Figure 2: Three column configuration
3. The complex column configurations
The complex configurations presented in Figure 3 are obtained substituting the non
product streams with a thermal coupling and by column recombination following the
procedure introduced by Errico and Rong, 2012.
Figure 3: Complex configuration obtained
from the configuration on Fig. 2
C1 C2 C3
Number of stages 44 28 17
Reflux ratio (molar) 2.337 0.246 0.397
Feed stage 30 25 6
Vapor recycle stage 34 --- ---
Column diameter (m) 1.37 0.79 0.51
Design pressure (kPa) 101.0 101.0 101.0
Condenser duty (kW) 3871.042
Reboiler duty (kW) 4907.893
Capital cost (k$ yr-1
) 133.1
7
8
Az. feed
Fermentation
Broth
Water
Solvent
Make up
4
5
6
Ethanol
Distillate Recycle
1
2
3
Solvent Recycle
Solvent Recycle
7
8
Az. feed
Fermentation
Broth
Water
Solvent
Make up
Ethanol
Distillate Recycle
1
2
3
4
5
6
VTC
Fermentation
Broth
Az. feed
Distillate Recycle
Water
Solvent Recycle
4
5
6
8
Solvent
Make up
Ethanol
1
2
3
7
Az. feed
Water
Fermentation
Broth
Solvent Recycle
4
5
6
8
Solvent
Make up
Ethanol
1
2
3
Vap. side stream
(a)
(b)
(c)
5. Optimal synthesis and design of extractive distillation for bioethanol separation: from
simple to complex columns 377
The thermally coupled configuration reported in Figure 3(a) is obtained from the one in
Figure 2 by substitution of the second column reboiler with a thermal coupling, the
corresponding thermodynamically equivalent sequence is reported in Figure 3(b). The
configuration of Figure 3(c) is obtained from the one in Figure 3(b) by substitution of
the column section 7 with a vapor withdrawal. Since only two columns are utilized this
sequence is very attractive for the possible capital costs savings. Its performances are
considered in details and reported in Table 1:
C1 C2
Number of stages 44 41
Reflux ratio (molar) 2.325 0.419
Feed stage 30 25
Recycle feed stage 31 ---
Solvent feed stage --- 5
Vapor side stream stage --- 28
Column diameter (m) 1.35 0.84
Design pressure (kPa) 101.00 101.00
Condenser duty (kW) 3852.160
Reboiler duty (kW) 4902.925
Capital cost (k$ yr-1
) 107.8
Table 1: Design parameters, energy requirement and capital cost of the configuration reported in
Fig. 3(c)
It is possible to notice that compare to the sequences in Figure 2, the two column
configuration has a similar energy requirement but more significantly a reduction of
19% of the capital cost is observed.
Anyway for the configuration in Figure 3(c) it was not possible to completely recover
the solvent and the necessary solvent makeup rise up to 0.619 kmol hr-1
compare to the
value of 0.004 utilized in the three column configuration of Figure 2. At this point the
question to be answered is: Does the trade-off between the solvent cost and the saving
in the capital costs make the configuration attractive from an economical point of view?
In the case considered ethylene glycol was used as a solvent, according to the world´s
largest petrochemical market information provider (ICIS), its price is of about 84 $
kmol-1
. Considering an operational time of 6000 hr yr-1
for the required makeup stream
the annual operational cost increases of 311976 $ yr-1
. This value makes the sequence
not convenient. Anyway the generalization of the result obtained is not so direct like in
the case of ideal mixtures. For this reason another solvent case is considered in details.
3.1. Solvent case study: glycerol
A more detailed analysis of the two-column sequence is required to prove if the
reduction of the capital costs balance the increase of the solvent makeup. The case
where a by-product is used as entrainer could represent a convenient alternative. One
possibility is represented by integrated biodiesel-bioethanol plants. The glycerol
produced in the biodiesel process can be efficiently used as solvent in the extractive
distillation. A solvent to feed ratio of 0.49 was chosen as optimal value to reach the
ethanol purity target. The recovery column´s pressure was set to 0.85 bar in order to
avoid the thermal decomposition of glycerol. Table 2 reports the design parameters, the
energy utilization, and the capital cost for the configurations in Figure 2 and the
corresponding intensified version of Figure 3(c). It is possible to notice that the
configuration with two columns has a slight reduction of the energy demand, but more
significantly the capital costs were reduced by 16%. Once again the solvent recovery
6. M. Errico et al.
was not complete and 0.010 kmol hr-1
were necessary to replenish the loss. The cost of
the glycerol is estimate at 75 $ kmol-1
that correspond to an annual cost of 4500 $ yr-1
.
In this case the saving in the capital costs overcomes the additional cost of the solvent
and the two-column configuration realizes an overall saving of 13%.
Differently from the previous solvent, the increase of the glycerol makeup stream does
not penalize the economy of the whole process.
Table 2: comparison between configurations on Fig. 2 and 3(a) using glycerol as entrainer
4. Conclusions
After the best simple column sequences have been selected, the analysis of the complex
configurations completes the study for the possible energy and capital cost reduction.
Following the thermal coupling principle and the possibility of column section
recombination, different subspaces of alternatives are generated. It was found that the
presence of the thermal coupling between the extractive and the solvent recovery
column does not improve the energy performance of the configuration, but only slightly
decreases the total capital cost. More interesting are the results obtained for the two-
column configurations.
For this configuration is not possible to realize a complete solvent recovery and its final
convenience mainly depends on the solvent´s cost. It was proved that utilizing glycerol
this sequence is still competitive compare to the three-column sequence from which was
derived.
References
A. Meirelles, 1992, Ethanol dehydration by extractive distillation, J. Chem. Tech. Biotechnol., 53,
181-188
B.-G. Rong, A. Kraslawski, 2003, Partially thermally coupled distillation systems for
multicomponent separations, AIChE J., 49, 1340-1347
G. Li, P. Bai, 2012, New operation strategy for separation of ethanol-water by extractive
distillation. Ind. Eng. Chem. Res., 51, 2723-2729
M. Errico, B.-G. Rong, 2012, Synthesis of new separation processes for bioethanol production by
extractive distillation, Sep. Pur. Tech., 96, 58-67
R. Agrawal, 2003, Synthesis of multicomponent distillation column configurations, AIChE J., 49,
379-399
Figure 2 Figure 3(a)
C1 C2 C3 C1 C2
Number of stages 44 28 9 44 36
Reflux ratio (molar) 2.337 0.149 0.580 2.337 0.197
Feed stage 30 25 3 30 25
Vapor recycle feed stage 34 --- --- 34 ---
Vapor side stream stage --- --- --- --- 28
Solvent feed stage --- 3 --- --- 3
Column diameter (m) 1.37 0.74 0.65 1.37 0.79
Design pressure (kPa) 101.0 101.0 85.0 101.0 85.0
Condenser duty (kW) 3820.645 3750.638
Reboiler duty (kW) 5018.671 4934.136
Capital cost (k$ yr
-1
) 134.3 112.9
