Biomass and Bioenergy 109 (2018) 231–238

Contents lists available at ScienceDirect

Biomass and Bioenergy

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

Research paper

Design of a separation section in an ethanol-to-butanol process T

∗
Wesley Michaels, Hanyu Zhang, William L. Luyben, Jonas Baltrusaitis
Department of Chemical and Biomolecular Engineering, Lehigh University, 111 Research Drive, Bethlehem, PA 18015, USA

A B S T R A C T

A complete separation scheme has been designed for the effluent of a high-pressure ethanol-to-butanol catalytic
reactor, producing 250,000 tonnes of n-butanol per year. The effluent contains water, hydrogen and a diverse
range of C2-C4 oxygenates: unconverted ethanol, n-butanol, acetaldehyde, ethyl acetate, and acetal.
Fundamental phase equilibrium relationships suggested use of conventional, extractive, and heterogeneous
azeotropic distillation units to perform the separations. All reactor effluent species exit the separation process at
mole purities of at least 99%. Separation costs are estimated to range from 9.0 to 10.6 MJ/kg n-butanol, which is
comparable with the separation costs of n-butanol obtained from established acetone-butanol-ethanol (ABE)
separation process.

1. Introduction ethanol (26.7 MJ/l vs. 21.3 MJ/L, respectively, lower heating value
(LHV) basis) [7,8] and decreased water solubility, making it compatible
Ethanol production has been steadily increasing over the last several with existing fuel infrastructure and pipelines [6,10]. Therefore, n-bu-
decades [1,2]. More than 58.34 hm3 of ethanol are being produced tanol can be both an additive to fuel and a stand-alone fuel source [9].
annually in the United States, with ethanol plants concentrated in the Importantly, unlike other alcohols, the Environmental Energy Company
Midwest because of their proximity to corn production [3]. Most (US) confirmed that n-butanol can be used as a total replacement for oil
ethanol in the U.S. is produced from starch-based crops by dry- or wet- derived gasoline without any modifications to car engines [11]. Outside
mill processing, but cellulosic feedstocks, such as grass, wood and crop of fuels, n-butanol is widely used as a direct solvent for paints, coatings,
residues, are also being increasingly utilized at 15 existing and 9 pro- varnishes, resins, dyes, camphor, vegetable oils, fats, waxes, shellac,
posed cellulosic ethanol plants [4]. This biomass-derived ethanol is rubbers, and alkaloids.
almost exclusively used as a fuel. However, ethanol as a stand-alone The currently used n-butanol synthesis Oxo (hydroformylation)
fuel has lower energy density when compared to gasoline, possesses route reacts synthesis gas with propylene to form an aldehyde, which is
corrosiveness towards engine and fuel pipelines, and reacts with water subsequently hydrogenated to yield n-butanol [12,13]. The catalytic
resulting in a separation from fuel blends over time. Ethanol use in reactor operates at very high pressure (∼30 MPa) and uses two dif-
conventional internal combustion engine gasoline has been limited to ferent catalysts (Co or Rh in the first step, and Ni in the second step).
blends of volume fraction of ethanol, φEtOH, of 10% (denoted E10). Feed stocks are derived from non-renewable fossil sources. Moreover,
Recent USDA Agricultural Projections to 2025 indicate that gasoline reliance on a homogeneous catalyst results in a difficult separation,
(mostly E10) consumption in the US is declining [2,5]. Infrastructural costly catalyst preparation, and related environmental issues.
and other constraints on growth for E15 gasoline (φEtOH = 15%) and In contrast, production of n-butanol from ethanol has been de-
the small size of the market for E85 gasoline (φEtOH = 85%) [2,5] monstrated on numerous heterogeneous catalysts, making it a potential
suggest that other ethanol monetization methods are needed and may alternative source of n-butanol for chemicals and fuels. Various alkali
benefit US agriculture. exchanged zeolites (Rb-LiX, Rb-NaX and Rb-KX)13, non-stoichiometric
Higher-order alcohols, such as n-butanol, are more structurally si- Ca10(PO4)6(OH)2 with various Ca/P ratios14−16, and Cu-Mg-Al mixed
milar to gasoline and can potentially overcome many of the above- oxide catalysts [14–16] have been tried with low to moderate ethanol
mentioned problems and serve as sustainable fuels. N-butanol can po- conversion. Very few instances have been reported where the overall
tentially be used as a fuel and holds many advantages over currently reaction kinetics were measured and reported for the elementary re-
used ethanol, which has a low energy density and is hygroscopic [6]. N- action steps [17]. The resulting product mixtures are comprised of
Butanol has about 1.25 times more volumetric energy content than variety of oxygenates and hydrocarbons, including but not limited to

∗
Corresponding author. Tel.: +1 610 758 3636.
E-mail address: [email protected] (J. Baltrusaitis).

https://doi.org/10.1016/j.biombioe.2017.12.031
Received 17 July 2017; Received in revised form 27 December 2017; Accepted 29 December 2017
Available online 12 January 2018
0961-9534/ © 2018 Elsevier Ltd. All rights reserved.
W. Michaels et al. Biomass and Bioenergy 109 (2018) 231–238

C2, C4 and C6 aldehydes, alcohols, paraffins and olefins [18]. Two types 2. Process flowsheet development
of reactors have been used: vapor-phase reactors at temperatures from
300 to 450 °C were reported at atmospheric pressures using bifunctional The reactor effluent is comprised of many components as shown in
catalysts [1], and liquid phase high-pressure reactors of up to 20 MPa Figs. 1 and 2 at 3568 kmol/h, 220 °C and 7 MPa pressure with un-
were also explored at 250 °C on Cu or Ni/γ-Al2O3. Ethanol conversion converted ethanol (concentration of ethanol, xEtOH, of 8.7%) and n-
from 10 to 30% and an n-butanol selectivity of 70% are reported butanol (xBuOH = 11.8%). The other major components are hydrogen
[17,19–21]. The choice of liquid-phase operation over the vapor-phase (xH2 = 41.2%), water (xH2O = 14.3%), acetaldehyde (xAA = 4.6%),
is typically due to the higher production capacity, easier separation of ethyl acetate (xEtAc = 17%) and acetal (xAcetal = 2.5%). The normal
products, and better selectivity [17]. For these liquid-phase reactions, boiling points of these components are 20.85 °C for acetaldehyde,
ethanol of mole purity 100% was reacted in a high-pressure slurry type 71.06 °C for ethyl acetate, 78.29 °C for ethanol, 100 °C for water,
autoclave with a catalyst of mass fraction of roughly 3% [21]. Riittonen 103.5 °C for acetal, and 117.7 °C for n-butanol.
et al. investigated the kinetics of a liquid-phase catalytic packed bed The mixture exhibits complex phase equilibrium relationships
upflow reactor with residence times of up to 462 min and temperatures among the many components, and the choice of the VLLE physical
from 200 to 240 °C [17]. The reaction products were water, n-butanol, property package is critical since it strongly impacts the feasible se-
acetaldehyde, ethyl acetate, diethyl ether and 1,1-diethoxyethane paration alternatives. Table 2 demonstrates that the UNIQUAC physical
(acetal), and gaseous hydrogen. The lab-scale reactor was L = 21 cm properties (using vapor-liquid-liquid equilibria) predict the occurrence
and D = 1.6 cm. Conversion was evaluated using LHSV of 4.3 h−1 and of seven azeotropes at atmospheric pressure: six binary and one ternary.
ethanol specific velocity of 8.3 μm/s [19]. Compared with literature data, UNIQUAC accurately predicts the
Complex reaction networks and low selectivity have made such binary homogeneous azeotropic VLE in both composition and tem-
processes challenging to engineer [22]; the majority of catalysts ex- perature. Though it deviates slightly in product composition for binary
plored to produce 1,3-butadiene, for example, to date show yields of and ternary heterogeneous mixtures, it does predict temperature of
20–40%, with rare instances reaching yields of 60% [23]. N-butanol these azeotropic points very well.
catalytic production from ethanol appears to have even lower yields, The two key separations in the process are ethanol/water and ethyl
suggesting that upstream catalytic conversion will not provide high acetate/ethanol. As further justification for selecting UNIQUAC phy-
purity n-butanol without separation. Low to moderate ethanol con- sical properties, comparisons of experimental VLLE data with
version, in addition to azeotropes formed by oxygenates in reactor ef- UNIQUAC predictions are given in Fig. 3. The individual points are
fluents, frustrates the design of a downstream separation scheme for experimental data reported by Kato et al. [32] and Kurihara et al. [33],
these state-of-the-art processes. Additionally, use of pure ethanol sug- respectively. The solid lines are predicted values. The choice of UN-
gests purification of an ethanol-water mixture beyond its minimum- IQUAC appears to be appropriate for this system.
boiling azeotropic point (ethanol of mass fraction 95%) [24] via ex- The final process design configuration for this separation process
tractive, azeotropic distillation [25,26] or adsorption [27]. Thus, al- involves distillation separation techniques of several types: conven-
though implementing a cost-effective separation strategy is integral to tional, extractive and heterogeneous azeotropic. Figs. 1 and 2 give the
the economic feasibility of ethanol-to-butanol heterocatalytic processes, flowsheet of the process developed to successfully separate the reactor
the energetics of such a separation have not yet been extensively stu- effluent stream. This design encompasses a reasonable separation
died. We propose a viable product separation scheme downstream of an scheme with reasonable values of process parameters in terms of op-
ethanol-to-butanol reactor, and assess its economic feasibility, via en- erating pressures and the trade-off between reflux ratios and numbers
ergy balance, in comparison with ABE n-butanol production processes. of stages that separates this complex mixture with multiple azeotropes
The data in Table 1 highlight the energy required for ABE separation by into high purity streams.
a variety of previously reported techniques, with the optimal reported The proposed separation sequence uses extractive distillation and
values ranging from 3.8 to 8.2 MJ/kg n-butanol [28–31]. heterogeneous azeotropic distillation to separate the azeotropic mix-
The design basis for the separation section of the process was a tures. Pressure-swing distillation was also extensively studied as an
reactor effluent stream to obtain 250,000 t/year (8000 operating hours) alternative. However, the sensitivities of the azeotropic compositions to
of n-butanol. Aspen Plus simulations (version. 8.8) used the UNIQUAC pressure were modest, resulting in columns with many stages and high
physical property package to describe phase equilibria. The Aspen stage energy requirements.
numbering convention is used in which stages are numbered from the
top down (reflux drum is stage 1). We assume that all the components 2.1. Hydrogen removal
need to be purified to a composition of at least 99%. The hydrogen and
all the organic components have commercial value and are not simply The first design objective is to produce hydrogen with a purity
burned for fuel. The high water purity is set to avoid product loss and xH2 = 99% using a flash tank sequence. The non-condensible hydrogen
environmental problems. is removed first but requires refrigeration to minimize product loss.

Table 1
Comparative assessment of n-butanol separation costs of proposed process with the conventional acetone-butanol-ethanol (ABE) process.

Separation technique from ABE mixture Energy Requirement [MJ/kg n-butanol] Literature reference

Distillation 79.5 Matsumura et al. [29]

Pervaporation + distillation 7.4 Matsumura et al. [29]
Stripping + vapor permeation membrane. Distillation 3.8–14.5 Vane et al. [28]
Distillation 19.4 Kraemer et al. [30]
Extraction 5.7 Kraemer et al. [30]
Adsorption-distillation 8.2 Qureshi et al. [31]
Steam stripping or gas stripping distillation 22–24 Qureshi et al. [31]
Pervaporation 14 Qureshi et al. [31]
This work 10.6 No credit for LP steam
9.0 Credit for LP steam
7.9 Ethanol/water mixture (xEtOH = 60%) after R1 recycled back to the
reactor

232
W. Michaels et al. Biomass and Bioenergy 109 (2018) 231–238

Reactor Effluent
3568 kmol/h
H2 Refrig 220 oC, 7 MPa
1470 kmol/h CW LP Steam
6.61 MW 0.412 H2
0.997 H2 14.6 MW 13.4 MW Feed to
0.115 MW 0.046 AA
0.001 AA 20 oC Extractive
50 oC 180 oC 0.170 EtAc
0.001 EtAc 6.9 MPa
0.087 EtOH Columns
0.001 W 1124 kmol/h
0.143 W
0.0001 AA
20 oC 0.118 BuOH
0.5381 EtAc
200 kPa 0.025 Acetal 0.2759 EtOH
0.1858 W
AA 0.0001 BuOH
163.9 kmol/h
2098 kmol/h
0.0042 H2
0.0002 H2
0.9903 AA
0.0005 EtAc Benzene OR
0.0050 W 556.8 kmol/h
0.0206 W
57 oC, 20 kPa
76 oC, 110 kPa 0.0006 Acetal
8.78 MW
8.93 MW 0.8559 B
52 oC, 50 kPa 0.1229 BuOH
21 oC, 100 kPa 21.3 MW
5.02 MW
70 oC
Splitter

BuOH
AA

HAD
5
10 23
RR = 4.21 30 Water
ID = 3.17 m RR = 0.9 304.3 kmol/h RR = 8.56 Acetal
ID = 4.37 m ID = 2.83 m 0.9901 W ID = 4.41 m
69 90.49 kmol/h
39 19 0.0001 Acetal 0.0099 W
59 0.0007 B 0.9900 Acetal
0.0091 BuOH 90oC
79 oC 0.0001 BuOH
83oC 119oC 7.54 MW
12.5 MW
20.0 MW 9.72 MW

BuOH
420.2 kmol/h
1934 kmol/h 810.1 kmol/h 510 kmol/h 0.001 Acetal
0.0001 AA 0.2 mg/kg EtOH 0.0018 W 0.999 BuOH
0.3704 W 0.1765 Acetal
0.1101 Acetal 0.8217 BuOH
0.5195 BuOH

Fig. 1. Upstream flowsheet.

Further, incomplete H2 removal would significantly reduce condenser distillation column has a composition of only xH2 = 0.02%, so it has
temperature in the acetaldehyde column (Section 2) and incur even little effect on the vapor liquid equilibrium in the further separation
greater refrigeration costs. For these reasons, a series of heat exchangers process of liquid oxygenates.
that generate low pressure (LP) steam (160 °C), followed by cooling
water (CW) and refrigerant are used.
The reactor effluent is at high pressure (7 MPa) and high tempera- 2.2. Acetaldehyde distillation column (AA column)
ture (220 °C). It is cooled in three heat exchangers operating in series.
Low-pressure steam (160 °C with a saturated vapor pressure of 600 kPa) The liquid stream from the second flash drum separator is fed to a
is generated in the first heat exchanger. Using a 20 °C differential 40-stage distillation column (AA column) operating at 100 kPa. Though
temperature on the cold end of that heat exchanger yields a duty of a liquid distillate could feasibly be recycled into a reactor proceeding
13.4 MW and cools the process steam to 180 °C. The second exchanger via an aldol condensation mechanism [11], condensing the distillate
uses cooling water to reduce the process temperature to 50 °C. Heat would reduce the required reflux-drum temperature, necessitating
duty is 14.6 MW. The last exchanger uses chilled water (5 °C with a lower temperature refrigerant at higher cost. For this process, the dis-
return temperature of 15 °C) to cool the stream to 20 °C. Refrigeration tillate is therefore removed as a vapor stream from the top of the reflux
load is 6.61 MW. The stream is then fed to a flash tank separator at drum. The design objectives for this column are to remove the acet-
6.9 MPa. The vapor stream is 1444 kmol/h with xH2 = 99.83%. The aldehyde at xAA = 99% and to achieve a very low concentration of
liquid stream is flashed one more time to 200 kPa and fed to a second acetaldehyde in the bottoms stream (xAA = 0.01%) so that this light
flash tank separator whose vapor stream is 26.44 kmol/h with component does not affect the downstream separation process.
xH2 = 93.76%. This small stream is compressed back up to 6.9 MPa At 100 kPa pressure the reflux-drum temperature is 21 °C, so chilled
(compressor work of 0.115 MW) and combined with first flash tank water must be used in the condenser (heat duty of 5.02 MW). The de-
separator vapor to produce the final hydrogen product steam. sign objectives could not be achieved at pressures greater than 100 kPa
The temperature of the cooled process from the refrigerated heat despite the fact that the acetaldehyde/ethyl acetate binary separation
exchanger and pressure in the second flash drum are selected to achieve shows little pressure dependence. However, the ethyl acetate/ethanol
two objectives. The first is to attain the desired hydrogen purity, and binary separation is quite pressure dependent as shown in Fig. 4. Higher
the second is to send only a very small amount of hydrogen in the liquid pressure shifts the azeotrope to higher ethanol compositions from 0.55
stream fed to the downstream process. Any hydrogen that enters the to 0.30 vapor mole fraction of ethyl acetate, which tends to drive more
first distillation column will result in a more complex separation by ethanol overhead and may explain why high-purity acetaldehyde
reducing the required reflux-drum temperature. At 20 °C and 200 kPa, cannot be achieved at higher pressures.
the liquid stream from the second flash tank separator fed into the first The reboiler duty in the AA column is 12.5 MW using low-pressure
steam. The column diameter is 3.17 m, and the required reflux ratio is

233
W. Michaels et al. Biomass and Bioenergy 109 (2018) 231–238

0.01 kmol/h
Makeup
987 kmol/h, 0.0001 W, 0.9990 DMSO

51 oC, 40 kPa EtAc

9.14 MW 605.4 kmol/h 51 oC, 30 kPa
0.0002 AA 7.524 MW
Feed to 5
0.9990 EtAc

R1
Extractive 0.0007 EtOH

E1
Columns 23 mg/kg DMSO
7
1124 kmol/h
27
0.0001 AA RR = 0.6 505.5 kmol/h
RR = 0.245
0.5381 EtAc ID = 3.52 m 29 mg/kg EtAc
ID = 3.39 m
0.2759 EtOH 0.5987 EtOH
0.1858 W 39 1505 kmol/h 0.4011 W
0.0001 BuOH 20 mg/kg EtAc 19 10 mg/kg DMSO
0.2057 EtOH
102 oC
0.1387 W
11.0 MW 152 oC
75 mg/kg BuOH 9.23 MW
0.6555 DMSO

707.6 kmol/h
EtGly
EtOH 54 oC, 15 kPa
56 oC, 40 kPa 309.9 kmol/h 2.98 MW
3.85 MW 49 mg/kg EtAc
7 0.9990 EtOH EtGly
0.0009 W 707.6 kmol/h

R2
7 mg/kg DMSO
E2

23 mg/kg BuOH
1 mg/kg EtGly 0.9999 EtGly
7
32
RR = 0.1 RR = 0.2 Water
ID = 1.98 m ID = 2.18 m 202.7 kmol/h
39 916.2 kmol/h 44 mg/kg EtOH
10 mg/kg EtOH 19 0.9990 W
0.2275 W 0.0005 BuOH
120 oC 0.0001 BuOH 148 oC 0.0005 EtGly
6.34 MW 0.7724 EtGly 3.66 MW
6 mg/kg DMSO

Fig. 2. Downstream flowsheet.

Table 2 4.21. The optimum feed stage that minimizes both refrigeration duty in
Literature and UNIQUAC predicted azeotrope compositions (mole basis) and tempera- the condenser and reboiler duty is stage 5.
tures at 1 atm pressure.a.

Azeotrope Composition Temperature (oC) Type 2.3. Splitter distillation column

Ethyl acetate- 0.5530 ethyl 71.8 Binary
ethanol acetate The bottoms stream from the AA column is fed to a 60-stage splitter
(0.5410) (71.7) [32] column on stage 30. The design objective is to remove almost all the
Ethanol-water 0.8999 ethanol 78.2 Binary lower boiling ethyl acetate and ethanol components overhead. Some
(0.9000) (78.1) [34]
water consequentially will also be removed in the overhead distillate
Ethanol-acetal 0.9145 ethanol 78.0 Binary
(0.8904) (77.9) [35]
because of the azeotropes forming among ethyl acetate, ethanol and
Water-acetal 0.8383 97.9 Binary water. Fig. 5 represents a 3-dimensional cross-section of a 4-dimen-
(N/A) (N/A)b sional plot since water is also present in the feed stream in significant
Ethyl acetate- 0.6132 ethyl 71.0 Binary amounts. The VLE for this complex system requires more downstream
water acetate heterogeneous
columns where the ethyl acetate/ethanol and ethanol/water azeotropes
(0.7010) (70.4) [32]
Water-n-butanol 0.8383 water 97.9 Binary need to be broken. Because these columns and the HAD column both
(0.754) (92.3) [33] heterogeneous remove water, the split fraction of water in this column was found to be
Ethyl acetate- 0.5537 ethyl 70.6 Ternary a key design variable with a large impact on process economics.
ethanol-water acetate heterogeneous The remaining water and almost all of the two remaining heavier
(0.6005) (70.3) [36]
0.1691 ethanol
components (acetal and n-butanol) are removed out the bottom of the
(0.1242) splitter column. One of the separation objective for this column is to
produce a distillate that contains very little acetal and n-butanol
a
Values obtained from literature references are shown in parenthesis. (xAA = xBuOH = 0.01%). The bottoms should contain very little ethyl
b
N/A – not available in the literature. acetate and ethanol (combined mass fraction of 0.2 mg/kg). The pres-
sure is set at 50 kPa, which gives a reflux-drum temperature of 52 °C, so
cooling water can be used in the condenser. The distillate is 1124 kmol/
h with, xEtAc = 53.81%, xEtOH = 27.59%, xH2O = 18.58%,
xBuOH = 0.01%, and xAA = 0.01%. This stream is fed to a four-column

234
W. Michaels et al. Biomass and Bioenergy 109 (2018) 231–238

A B

Fig. 3. AEthanol/water; experimental and UNIQUAC prediction.

3BEthyl acetate/ethanol; experimental and UNIQUAC prediction.

Fig. 5. Splitter column; Ethyl acetate/ethanol/butanol.

agent. Two liquid phases are formed in the decanter reflux drum, and
Fig. 4. Ethyl acetate/ethanol separation. an aqueous phase with xH2O = 99% can be decanted. The ternary dia-
gram for this distillation column design is shown in Fig. 6 where the
extractive distillation separation section discussed below. two heavy components (butanol and acetal) are lumped together. The
The heavy bottoms stream is 810.1 kmol/h with xH2O = 37.04%, feed (F) is considered as a binary water/n-butanol mixture with
xAcetal = 11.01% and xBuOH = 51.95%. The trace amount of ethanol xH2O = 35%, while the organic reflux lies at the phase separation
(wEtOH = 0.2 mg/kg) is set by adjusting the reflux ratio (RR = 0.9). This boundary with xBenzene = 85%. Combining the feed and the reflux gives
stream is fed to a two-column separation section that uses hetero- the indicated mix point M. The distillation boundary connecting the
geneous azeotropic distillation to remove the water. The splitter benzene-water and n-butanol-water azeotropes has little effect on the
column reboiler duty is the largest of all the duties in the process separation with the resulting distillate (DHAD) residing in two-liquid
(20.0 MW), but the low temperature (83 °C) means a low-temperature phase region. The straight overall component line from DHAD leads
steam can be used. The column diameter is 4.37 m. through the mix point (M) to high-purity bottoms stream (n-butanol
with acetal).
The organic reflux stream coming from a decanter is fed to the top
2.4. Heterogeneous azeotropic distillation (HAD) column
tray. The flowrate is 556.8 kmol/h and the composition is mostly
benzene (xBenzene = 85.59%) with small amounts of other components
The bottoms stream from the splitter column is fed to stage 10 of a
(xH2O = 2.06%, xAcetal = 0.06%, and xBuOH = 12.29%). The organic
20-stage column, whose purpose is to remove the water. Heterogeneous
reflux is adjusted to achieve the specified small impurity of water in the
azeotropic distillation is employed to achieve this design objective
bottom stream (xH2O = 0.18%). The aqueous phase from the decanter
using benzene (cyclohexane is an alternative) as an organic separation

235
W. Michaels et al. Biomass and Bioenergy 109 (2018) 231–238

Fig. 8. E1 Column; DMSO/Ethyl Acetate/Water (40 kPa).

Fig. 6. Had column; benzene/water/butanol.

with xBuOH = 99.9%. The diameter of this vacuum column is large

contains water at xH2O = 99.0%, with a flowrate of 304.3 kmol/h.
(4.41 m). Reboiler duty is 7.54 MW.
Slight impurities of xAcetal = 0.01%, xBenzene = 0.07% and
(xH2O = 0.91%) also leave in this aqueous stream.
The decanter operates at 110 kPa and 70 °C. Overhead condenser 2.6. Extractive and recovery columns E1 and R1
duty is 8.93 MW using cooling water. Reboiler duty is 9.72 MW using
low-pressure steam. Column diameter is 2.83 m. The bottoms stream is The mixture of ethyl acetate, ethanol and water coming from the top
essentially a binary mixture of acetal and n-butanol, which is fed to the of the splitter column is fed to an extractive distillation column (E1)
downstream column for further separation. that uses dimethyl sulfoxide (DMSO) as the solvent to drive the ethyl
acetate overhead, as shown in Fig. 2. The bottoms are fed to a solvent
2.5. N-butanol distillation column recovery column (R1) to take the solvent out the bottom for recycle
back to the extractive column. The distillate from R1 is essentially a
The acetal/n-butanol separation is difficult at atmospheric pressure, binary mixture with xEtOH = 59.87% and xH2O = 40.11% that is fed
but relative volatility increases as pressure is lowered. Fig. 7 gives the downstream to a second extractive/recovery column separation unit
xy plots at pressures of 20 and 100 kPa, and suggests an increase in discussed in the next section.
volatility as pressure decreases. The n-butanol column is therefore op- The ternary diagram for E1 is shown in Fig. 8. The ethyl acetate and
erated at 20 kPa, a low practically attainable pressure which gives a ethanol have been lumped together and are shown as the F point. The
reflux-drum temperature high enough (57 °C) for the use of cooling DMSO solvent is shown as the S point. The mix point lies in a region
water. where high-purity ethyl acetate can be taken overhead. The bottoms
The separation requires many trays (70 stages) and a fairly high stream BE1 is a mixture of components, with xDMSO = 65.5%, to be
reflux ratio (RR = 4.41) to achieve the desired purities of the two recovered in R1. Fig. 9 gives the ternary diagram for R1 in which sol-
product streams: a distillate with xAcetal = 99% and a bottoms stream vent is recovered in the bottoms and a binary mixture of ethanol and

Fig. 7. Acetal/butanol separation. Fig. 9. R1 Column at 30 kPa.

236
W. Michaels et al. Biomass and Bioenergy 109 (2018) 231–238

water goes overhead. A pressure of 30 kPa permits the use of cooling

water in the condenser. The straight component-balance line con-
necting the feed (F) and pure DMSO bottoms (BR1) leads to the distillate
DR1 below azeotropic ethanol-water point.
The extractive column E1 has 40 stages with feed introduced on
Stage 27 and the DMSO solvent fed on Stage 5. Solvent flowrate is
987 kmol/h as set to achieve a very small amount of ethyl acetate in the
bottoms (wEtAc = 20 mg/kg). Operating pressure is 40 kPa with cooling
water used in the condenser. The distillate exits with xEtAc = 99.9%.
Reflux ratio is adjusted to keep a very small impurity of DMSO in the
distillate (wDMSO = 23 mg/kg). Reboiler duty is 11.0 MW. Column dia-
meter is 3.52 m.
The recovery column has 20 stages, operates at 30 kPa, has a reflux
ratio of 0.245 and a reboiler duty of 9.23 MW. The bottoms stream is
recycled back to E1 after cooling to 50 °C and adding a small makeup
stream to account for the tiny loss of solvent. The distillate is a mixture
of water and ethanol (505.5 kmol/h) that is fed to another extractive
distillation system.

2.7. Extractive and recovery columns E2 and R2

The mixture of ethanol and water can be handled in several ways. If

it is to be recycled back to the reaction section with xEtOH = 85%,
Fig. 11. R2 Column at 15 and 100 kPa.
which is lower than the azeotrope, a single column can be used. We
assume that high-purity ethanol is desired, so we design an extractive
distillation section to overcome the azeotrope using ethylene glycol as 202.7 kmol/h with xH2O = 99.9%. Reboiler duty is 3.66 MW. Column
the solvent. The ternary diagram for this separation at 40 kPa is shown diameter is 2.16 m. Reflux ratio is 0.2.
in Fig. 10. The binary feed contains ethanol at a mole fraction of 60%.
The flowrate of the ethylene glycol (denoted EtGly) solvent is 3. Separation energy balance
707.6 kmol/h. The component-balance line through the mix point
yields a bottoms BE2 with xEtGly = 77% and xH2O = 23%. The binary The 8-column separation scheme described above achieves the de-
bottoms stream is separated in R2, which operates at 15 kPa. As shown sign objective of separating the complex reactor effluent mixture into
in Fig. 11, relative volatilities of the components increase as pressure high-purity components. It uses 10.6 MJ/kg n-butanol in operation.
decreases. The operating pressure of 15 kPa pressure was selected so Two potential modifications have been identified to improve the en-
that cooling water can be used in the condenser. ergetics of the proposed scheme. First, LP steam generated in the first
The extractive column E2 has 40 stages with solvent fed on stage 7 reactor effluent heat exchanger may be heat-integrated with other unit
and operates at 40 kPa using cooling water in the condenser. The dis- operations, acting as an energy credit and reducing the overall energy
tillate contains ethanol at a purity of 99.9%. Solvent flowrate is ad- consumption from 10.6 to 9.0 MJ/kg n-butanol. Second, recycling the
justed to achieve the small ethanol composition in the bottoms mixed ethanol/water effluent, with xEtOH = 59.87%, from column R1 to
(wEtOH = 10 mg/kg). Reboiler duty is 6.34 MW. Column diameter is the reactor would obviate the need for columns E2 and R2, eliminating
1.98 m. A small reflux ratio keeps the solvent impurity in the distillate 10 MW of power consumption and reducing overall energy consump-
low (wEtGly = 1 mg/kg). tion to 7.9 MJ/kg n-butanol. This latter improvement would require
The solvent recovery column has 20 stages and produces effluent at more information about catalyst performance during the ethanol cou-
pling to n-butanol in the presence of large amounts of water. The
splitter column would need to handle larger amounts of water in this
case, increasing its reboiler duty.

4. Conclusion

The proposed distillation scheme in this work, with energy input

ranging from 7.9 to 10.6 MJ/kg n-butanol, is competitive with the ABE
separation methods reported and significantly more energy efficient
than distillation schemes cited in the same works. Further, it confers
concrete advantages over the ABE methods justifying its consideration
as a viable n-butanol separation scheme. It is optimized to purify the
reactor effluent at their actual output concentrations; the energetic
figures reported are therefore expected to be representative of true
energetic costs. In contrast to the existing ABE separation methods
which purify a model fermentation product (for example, containing
only acetone, butanol, and ethanol), the proposed scheme purifies all
components in the true reactor effluent; this in part explains the larger
energetic cost incurred. Similar to the reported ABE separation
methods, improvements in state-of-the-art ethanol conversion and n-
butanol selectivity are expected to lower butanol recovery energy,
making the proposed scheme more competitive with the lowest values
Fig. 10. E2 Column; Ethylene Glycol/Ethanol/Water (40 kPa).
reported in the literature. Importantly, conventional distillation

237
W. Michaels et al. Biomass and Bioenergy 109 (2018) 231–238

processes are an established technology and can be performed routinely [17] T. Riittonen, T. Salmi, J.-P. Mikkola, J. Wärnå, Direct synthesis of 1-butanol from
in continuous processes, facilitating implementation of the proposed ethanol in a plug flow reactor: reactor and reaction kinetics modeling, Top. Catal.
57 (2014) 1425–1429, http://dx.doi.org/10.1007/s11244-014-0314-4.
scheme at both the bench and industrial scales. Thus, while hetero- [18] T. Tsuchida, S. Sakuma, T. Takeguchi, W. Ueda, Direct synthesis of n-butanol from
catalytic and fermentative routes of n-butanol production are still de- ethanol over nonstoichiometric hydroxyapatite, Ind. Eng. Chem. Res. 45 (2006)
veloping technologies, this work substantiates that separation of a 8634–8642, http://dx.doi.org/10.1021/ie0606082.
[19] T. Riittonen, K. Eränen, P. Mäki-Arvela, A. Shchukarev, A.-R. Rautio, K. Kordas,
heterocatalytic reactor effluent with distillation is energetically com- et al., Continuous liquid-phase valorization of bio-ethanol towards bio-butanol over
petitive with other proposed methods in the literature. metal modified alumina, Renew. Energy 74 (2015) 369–378 https://doi.org/10.
1016/j.renene.2014.08.052.
[20] T.L. Jordison, C.T. Lira, D.J. Miller, Condensed-phase ethanol conversion to higher
Acknowledgments alcohols, Ind. Eng. Chem. Res. 54 (2015) 10991–11000, http://dx.doi.org/10.
1021/acs.iecr.5b02409.
JB acknowledges Lehigh University startup funds. [21] T. Riittonen, E. Toukoniitty, D.K. Madnani, A.-R. Leino, K. Kordas, M. Szabo, et al.,
One-pot liquid-phase catalytic conversion of ethanol to 1-butanol over aluminium
oxide—the effect of the active metal on the selectivity, Catalysts 2 (2012) 68–84.
References [22] W.E. Taifan, T. Bucko, J. Baltrusaitis, Catalytic conversion of ethanol to 1,3-buta-
diene on MgO: a comprehensive mechanism elucidation using DFT calculations, J.
[1] C. Angelici, B.M. Weckhuysen, P.C.A. Bruijnincx, Chemocatalytic conversion of Catal. 346 (2017) 78–91, http://dx.doi.org/10.1016/j.jcat.2016.11.042.
ethanol into butadiene and other bulk chemicals, ChemSusChem 6 (2013) [23] G. Pomalaza, M. Capron, V. Ordomsky, F. Dumeignil, Recent breakthroughs in the
1595–1614, http://dx.doi.org/10.1002/cssc.201300214. conversion of ethanol to butadiene, Catalyst 6 (2016), http://dx.doi.org/10.3390/
[2] P. Westcott, J. Hansen, USDA Agricultural Projections to 2025. Office of the Chief catal6120203.
Economist, World Agricultural Outlook Board, U.S. Department of Agriculture, [24] S.M. Walas, Phase Equilibria in Chemical Engineering, (1985) Butterworth.
2016 Prepared by the Interagency Agricultural Projections Committee. Long-term [25] A. Meirelles, S. Weiss, H. Herfurth, Ethanol dehydration by extractive distillation, J.
Projections Report OCE-2016-1. Chem. Technol. Biotechnol. 53 (1992) 181–188, http://dx.doi.org/10.1002/jctb.
[3] September 2017 EIA monthly energy review, table 10.3 from the U.S. Energy in- 280530213.
formation administration, (n.d.). [26] A. Chianese, F. Zinnamosca, Ethanol dehydration by azeotropic distillation with a
[4] http://www.ethanolproducer.com/plants/listplants/US/Proposed/Cellulosic, (n. mixed-solvent entrainer, Chem. Eng. J. 43 (1990) 59–65 https://doi.org/10.1016/
d.). 0300-9467(90)80001-S.
[5] http://www.ers.usda.gov/data-products/chart-gallery/detail.aspx?chartId= [27] W.-C. Chen, C.-T. Sheng, Y.-C. Liu, W.-J. Chen, W.-L. Huang, S.-H. Chang, et al.,
56921, (n.d.). Optimizing the efficiency of anhydrous ethanol purification via regenerable mole-
[6] E. Steen, R. Chan, N. Prasad, S. Myers, C. Petzold, A. Redding, et al., Metabolic cular sieve, Appl. Energy 135 (2014) 483–489 https://doi.org/10.1016/j.apenergy.
engineering of Saccharomyces cerevisiae for the production of n-butanol, Microb. 2014.08.112.
Cell Factories 7 (2008) 36 http://www.microbialcellfactories.com/content/7/1/36. [28] L.M. Vane, F.R. Alvarez, L. Rosenblum, S. Govindaswamy, Hybrid vapor strip-
[7] http://www.iea-amf.org/content/fuel_information/fuel_info_home/ethanol/e10/ ping–vapor permeation process for recovery and dehydration of 1-butanol and
ethanol_properties, (n.d.). acetone/butanol/ethanol from dilute aqueous solutions. Part 2. Experimental va-
[8] http://www.iea-amf.org/content/fuel_information/butanol/properties, (n.d.). lidation with simple mixtures and actual fermentation broth, J. Chem. Technol.
[9] S.K. Lee, H. Chou, T.S. Ham, T.S. Lee, J.D. Keasling, Metabolic engineering of mi- Biotechnol. 88 (2013) 1448–1458, http://dx.doi.org/10.1002/jctb.4086.
croorganisms for biofuels production: from bugs to synthetic biology to fuels, Curr. [29] M. Matsumura, H. Kataoka, M. Sueki, K. Araki, Energy saving effect of perva-
Opin. Biotechnol. 19 (2008) 556–563 https://doi.org/10.1016/j.copbio.2008.10. poration using oleyl alcohol liquid membrane in butanol purification, Bioprocess
014. Eng 3 (1988) 93–100, http://dx.doi.org/10.1007/BF00369334.
[10] C.R. Fischer, D. Klein-Marcuschamer, G. Stephanopoulos, Selection and optimiza- [30] K. Kraemer, A. Harwardt, R. Bronneberg, W. Marquardt, Separation of butanol from
tion of microbial hosts for biofuels production, Metab. Eng. 10 (2008) 295–304 acetone–butanol–ethanol fermentation by a hybrid extraction–distillation process,
https://doi.org/10.1016/j.ymben.2008.06.009. Comput. Chem. Eng. 35 (2011) 949–963 https://doi.org/10.1016/j.compchemeng.
[11] B. Ndaba, I. Chiyanzu, S. Marx, n-Butanol derived from biochemical and chemical 2011.01.028.
routes: a review, Biotechnol. Reports 8 (2015) 1–9 https://doi.org/10.1016/j.btre. [31] N. Qureshi, S. Hughes, I.S. Maddox, M.A. Cotta, Energy-efficient recovery of bu-
2015.08.001. tanol from model solutions and fermentation broth by adsorption, Bioproc. Biosyst.
[12] C.E. O'Rourke, P.R. Kavasmanek, R.E. Uhl, Manufacture of n-butanol and 2-ethyl- Eng. 27 (2005) 215–222, http://dx.doi.org/10.1007/s00449-005-0402-8.
hexanol by the rhodium Oxo process and applications of the alcohols, Monohydric [32] M. Kato, H. Konishi, M. Hirata, New apparatus for isobaric dew and bubble point
Alcohols, American Chemical Society, 1981, pp. 6–71, , http://dx.doi.org/10.1021/ method. Methanol-water, ethyl acetate-ethanol, water-1-butanol, and ethyl acetate-
bk-1981-0159.ch006. water systems, J. Chem. Eng. Data 15 (1970) 435–439, http://dx.doi.org/10.1021/
[13] R.D. Mullineaux, L.H. Slaugh, Hydroformylation of olefins, US 3239569 A, US je60046a021.
3239569 A, https://www.google.com/patents/US3239569, (1966). [33] K. Kurihara, M. Nakamichi, K. Kojima, Isobaric vapor-liquid equilibria for methanol
[14] M. León, E. Díaz, S. Ordóñez, Ethanol catalytic condensation over Mg–Al mixed + ethanol + water and the three constituent binary systems, J. Chem. Eng. Data 38
oxides derived from hydrotalcites, Catal. Today 164 (2011) 436–442 https://doi. (1993) 446–449, http://dx.doi.org/10.1021/je00011a031.
org/10.1016/j.cattod.2010.10.003. [34] J. Gmehling, Azeotropic Data, VCH, 1994.
[15] I.-C. Marcu, D. Tichit, F. Fajula, N. Tanchoux, Catalytic valorization of bioethanol [35] M. Lecat, Nouveaux azéotropes binaires: seconde liste, Recl. Des Trav. Chim. Des
over Cu-Mg-Al mixed oxide catalysts, Catal. Today 147 (2009) 231–238 https://doi. Pays-Bas 45 (1926) 620–627, http://dx.doi.org/10.1002/recl.19260450902.
org/10.1016/j.cattod.2009.04.004. [36] J. Wade, CLXII.-The influence of water and alcohols on the boiling point of esters. I.
[16] D.L. Carvalho, R.R. de Avillez, M.T. Rodrigues, L.E.P. Borges, L.G. Appel, Mg and Al A modification of Markownikoff's method of preparation, J. Chem. Soc. Trans. 87
mixed oxides and the synthesis of n-butanol from ethanol, Appl. Catal. Gen. (1905) 1656–1668, http://dx.doi.org/10.1039/CT9058701656.
415–416 (2012) 96–100 https://doi.org/10.1016/j.apcata.2011.12.009.

238