Journal of Advanced Chemical Engineering

To achieve efficient bioethanol production, it is important to develop energy-saving dehydration processes. Thus, there are many reports about energy-saving dehydration technology from a fermentative ethanol solution [1-5]. Horizoe et al. [6-8] developed an ethanol dehydration system that involved extractive distillation with a light hydrocarbon solvent. In Figure 1 our proposed system for bioethanol dehydration through extractive distillation is showed. It consists of a distillation column, an extractive distillation column, a solvent separation column, and a water-butane separation tank. During distillation, the bioethanol solution (5-10 wt%) is condensed to around 90-96 wt%. The condensed ethanol solution is then dehydrated using the extractive distillation column with butane as the solvent. Next, the dehydrated ethanol/butane mixture is flowed from the bottom of the column to a solvent separation column, where 99.9 wt% of the ethanol is recovered. Subsequently, butane is separated with water and recycled. The heat of vapor recompression is used as the heat source for the column reboiler for distillation and extractive distillation. To evaluate the energy consumption of this process, vapor-liquid equilibrium data and their accurate model are needed (Figure 1).

Figure 1: Extractive distillation process with butane solvent.

Vapor-liquid equilibrium data for the butane+ethanol system have been previously obtained by Deak et al. [9], Soo et al. [10], Kretschmer et al. [11], Holderbaum et al. [12], and Dahlhoff et al. [13]. Vaporliquid equilibrium data for the isobutane+ethanol system have been obtained by Ouni et al. [14].

In an earlier publication, our group reported on the vapor-liquid equilibria of the butane+ethanol and isobutane+ethanol systems using process gas chromatography at 313-403 K, and determined the modified Redlich-Kwong (RK) equation of state (EOS) parameters to correlate the experimental data [15]. However, no data exists for the vapor-liquid equilibria of the mixed butane+ethanol system. To allow the use of mixed butane solvents, which are less expensive than isobutane and butane solvents, the vapor-liquid equilibria of the isobutane+butane+ethanol system were measured at 313-403 K. The measured data were compared with the calculation results of the modified RK EOS, of which parameters were determined in a previous work [16], and the predictive Soave-Redlich-Kwong (PSRK) EOS data.

Materials

In Table 1, there is lists of purities and suppliers of the materials used in this study.

Table 1: Properties and suppliers of the materials.

Apparatus and measuring procedure

The measurement apparatus and precise measurement procedures have been described in previous reports [15]. The vapor-liquid equilibrium cell has a viewing window and vapor and liquid phase are circulated with a magnetic pump respectively. The equilibrium system was immersed in an air-bath controlled to a ≤ 0.1°C using a PID. The temperature inside the equilibrium cell was measured using a platinum resistance probe (ChinoPt100 Class A, accuracy: 0.1 K). The pressure was measured using a digital gauge having 0-3 MPa range (PAA-33XEi, Keller AG, accuracy: 0.1% of the full scale).

The given amount of ethanol was provided for the equilibrium cell using a liquid pump and the liquid phase of mixed butane (Siphonic cylinder) was pressurized to the cell using a diaphragm pump. When equilibrium was achieved, the vapor and liquid phases were analyzed using the process gas chromatograph (Shimadzu GC-8A, TCD, Column Temp. 170°C, Porapak T). Each concentration data set was measured at least 3 times and the reproducibility was accurate to less than 0.001 mole fraction.

Modeling

The modified RK EOS proposed by Twu et al. [17,18] was selected for modeling the experimental pxy data, which can represent polar/ non-polar systems. Tables 2 and 3 show the critical properties of each component, and the binary parameters for the modified RK EOS, which were determined in previous reports [15] respectively. The butane+isobutane parameters were set to zero. Connolly et al. reported the ideality of the butane+isobutane system by measuring the phase boundary pressure; the results showed that the deviation between the ideal and measured pressures is 0.2% [19]. The PSRK EOS was also used for comparing with the experimental data, which is a group contribution EOS based on the Soave-Redlich-Kwong EOS with the UNIFAC method [16] for mixing rule (Tables 2 and 3).

Table 2: Pure component parameters of the modified RK EOS.

Table 3: Parameters of the modified RK EOS for the butane+ethanol and isobutane+ethanol systems.

The modified RK EOS is described by the following equations.

(1)

a(T)=a(T)a(T_c) (2)

(3)

b=0.086640349965RT_c/P_c (4)

α=α⁽⁰⁾+ω(α⁽¹⁾-α⁽⁰⁾) (5)

(6)

(7)

The mixing rule is described by the following equations:

a_m=Ʃ_iƩ_j x_ix_ja_ij (8)

b_m=Ʃ_i x_ib_i (9)

(10)

H_ij=k_ij-k_ij (11)

G_ij=exp(-β_ijH_ij) (12)

K_ij=ka_ij+kb_ij/T (13)

Pseudo-binary phase equilibrium (mixed butane+ethanol) calculations were carried out using the T-x flush calculation method.

Given figure compares the experimental data for the isobutane+butane binary system obtained by Connolly [19] with those calculated using the modified RK EOS. The calculation results showed good agreement with the experimental data; thus, the interaction parameters for isobutane+butane can be considered zero (Table 4; Figures 2 and 3).

Figure 2: Phase equilibrium data of the isobutene/butane system with the modified RK EOS at 344.26 K.

Figure 3: Phase equilibrium data of the mixed butanol/ethanol pseudo-binary system with the modified RK EOS and PSRK EOS, Mixed butane: isobutene/butane=30/70.

Table 4: Vapor–liquid equilibrium data for the isobutane (1)+butane (2)+ethanol (3) systema.

The pxy data for the isobutane+butane+ethanol system are shown in Table 4 and plotted in Figure 3 with the calculation results of the modified RK EOS and PSRK EOS. The data in Figure 3 correspond to the mixed butane+ethanol pseudo-binary system. Thus, the modified RK EOS can represent the mixed butane+ethanol system. For the PSRK model, the composition of the liquid phase is slightly different, and these results are similar to those obtained in our previous reports for the butane+ethanol or isobutane+ethanol binary systems [15] (Table 5).

Table 5: Absolute average deviation (AAD) between calculation and experimental data.

Table 5 shows the average deviation of the calculated and experimental data. The modified RK EOS shows better agreement with the experimental data in terms of pressure and ethanol vapor phase concentration, although agreement in the vapor phase concentrations of butane and isobutane is worse than that of PSRK. Gardeler et al. [20] reported the phase equilibrium data of alkane+alcohol systems and experimental data were compared with PSRK EOS. They concluded the reliability was not as good in the near critical point. In our experimental condition, the critical temperature of isobutane (408 K) and butane (425 K) are close to experimental condition (403 K), so the PSRK model may show systematic differences (Figure 4).

Figure 4: Comparison of phase euilibria data for butane, isobutene, and mixed butane (isobutene/butane=30/70) at 313 K with the modified RK EOS.

Figure 4 compares the phase equilibria data of the butane+ethanol, isobutane+ethanol, and mixed butane+ethanol systems. The data for the mixed butane+ethanol system were between those of the isobutane+ethanol and butane+ethanol systems. The butane+ethanol system has an azeotropic point, but the isobutane+ethanol system does not. For the mixed butane+ethanol system, the appearance of an azeotropic point depends on its ethanol free butane concentration. From the viewpoint of the bioethanol dehydration system, mixed butane has a lower azeotropic region than the pure butane solvent, so it leads to lower ethanol loss from the process. Thus, since mixed butane is also less expensive than butane or isobutene, it may be a suitable solvent for ethanol dehydration.

Vapor-liquid equilibria for the ternary system butane+isobutene+ethanol were obtained in the 313-403 K range. The modified RK EOS with our previously determined parameters corresponded well with experimental one, though in the PSRK EOS, liquid phase concentration showed a systematic deviation. These results may depend on the trend of PSRK model whose calculation results show the deviation in the near critical point.

This work was performed under “Development of Alcohol Dehydration Technology,” funded by the JX Engineering Corporation.

ka, kb parameters of the modified RK EOS

K equilibrium ratio (=y/x)

P total pressure (MPa)

R gas constant (m²kg/s²∙Kmol)

T temperature (K)

V molar volume (m³/mol)

X liquid-phase mole fraction

Y vapor-phase mole fraction

α alpha function defined in eqn. [2]

β_ij binary interaction parameter in eqn. [12]

ω acentric factor

C critical property

i, j property of component i, j

ij property of the i-j interaction

m mixture property

r reduced property