MODELING A BATCH DISTILLATION COLUMN BUTANOL/ETHANOL/WATER SEPARATION
FOR
C. L. da SILVEIRA1, N. P. G. SALAU1 1
Universidade Federal de Santa Maria, Departamento de Engenharia Química E-mail para contato:
[email protected]
ABSTRACT – The main goal of this work is to reproduce the behavior of a ternary mixture of ethanol/1-butanol/water in a batch distilation process. The modeling is based on a simple approach of material balances and the vapor-liquid equilibrium thermodynamic model used was the NRTL. The distillation column model, together with the NRTL, has presented results with good agreement to what was expected as a behavior of the components along the distillation process. Although with some inherent limitations in predictions for ternary mixtures, the use of the NRTL model seems to be reliable for the studied system.
1. INTRODUCTION The modeling in phase equilibrium process has been used as a manner to obtain the optimal design and control, and best operation conditions, reducing energetic costs of process operation and equipment maintenance. Even though there is wide data and many works available for binary mixtures predictions of vapor-liquid equilibrium (VLE) with well-established models - such as NRTL (Non-Random Two-Liquids), UNIFAC (UNIQUAC Functional-group Activity Coefficients) and UNIQUAC (Universal Quasi Chemical) – the predictions in ternary mixtures are more restrict and, frequently, less accurate using the referred methods (Marcilla et al. 2015, Marcilla et al. 2016). Luyben (2015) has performed simulations of a ternary mixture composed of benzene, toluene, and o-xylene with the software Aspen ® using the NRTL model. Also, Luyben has tested different control structures available in the software library to evaluate their efficiency. Faúndez et al. (2006) compared the accuracy of four different models – PSRK, UNIFAC, UNIQUAC, and NRTL) in describing the VLE of a mixture composed of water/ethanol/congener for wine distillation. The authors concluded that the non-empirical models (PSRK and UNIFAC) are not reliable for the vapor phase concentrations prediction, and that the semi-empirical models (NRTL and UNIQUAC) predicted more accurately the temperature behavior. Marcilla et al. (2015, 2016) report, however, that the NRTL model has some inherent
limitations in predictions for ternary mixtures, mainly for azeotropes. Kosuge and Iwakabe (2005) presented a parameter estimation for VLLE of ethanol/water/(1 and 2)-butanol, which is a partially miscible mixture, using NRTL and UNIQUAC, claiming that both models can be used for this mixture. Li et al. (2016) also reported that both NRTL and UNIQUAC models were sucessfully used to predict a LLE for the ternary system isopropyl acetate/2-propanol/glycerol. The main goal of this work is to present a ternary (ethanol/1-butanol/water) distillation column simulation based on the NRTL model. The parameters used for the model were the ones presented on Kosuge and Iwakabe (2005) work.
2. MODELING AND SIMULATION In this work a phenomenological model is used, it comes from the mass balance in a batch distillation column where all the components are fed to the bottom of the column before the process starts. In this manner, the mass balances for each component results in Equation 1 for the condenser, Equation 2 for the stages, Equation 3 for the boiler, and Equation 4 for the global material balance.
(1)
(2)
(3)
(4)
Where x i and y i are the molar fraction of the component in the theoretical stage i in the liquid and vapor phase, respectively; M D , M P , and M B are the molar hold-up in the condenser, in the stages, and in the boiler, respectively; L , V , and D are the liquid, vapor and distillate flow-rates; N T is the total number of mols fed to the column; and NS is the total
number of stages of the column. The parameters used in this work are listed on Table 1. The vapor flow-rate is calculated using the boiler potency and an experimental value of mols vaporized in the boiler per second. The initial conditions are also provided by a realized experiment.
Table 1 – Parameters and initial conditions used for simulation Parameter
Value
Units
Description
x ethanol,NS ( t= 0 )
0.3773
Dimensionless
Initial ethanol molar fraction
x butanol,NS ( t= 0 )
0.1006
Dimensionless
Initial butanol molar fraction
x water,NS ( t= 0 )
0.5220
Dimensionless
Initial water molar fraction
NT ( t= 0 )
239.84
Mols
Initial number of mols
NS
11
Dimensionless
Number of stages
R
6
Dimensionless
Reflux ratio
q
488.6
J/s
Heat
c
4 .1772×10−6
Mol/J
Mols vaporized in boiler
0.3
Dimensionless
Condenser molar hold-up
0.1
Dimensionless
Stages molar hold-up
2.0
Dimensionless
Boiler molar hold-up
P
760
mmHg
Pressure
R
8.314472
J/mol.K
Universal gas constant
Also, for the thermodynamics calculations, it was used the Antoine coefficients and the NRTL parameters, as can be seen in Table 2.
Table 2 – Antoine coefficients and NRTL parameters A ethanol
7.6811
α 1,1
0
b1,1
0
B ethanol
1332.0400
α 1,2
0.3038
b1,2
38 .0723×R
C ethanol
199.2000
α 1,3
0.2
b1,3
−223 . 2760×R
A butanol
7.9306
α 2,1
0.3038
b2,1
−32 . 9414×R
B butanol
1738.4000
α 2,2
0
b2,2
0
Cbutanol
226.6060
α 2,3
0.2
b2,3
−211 . 9310×R
A water
8.1402
α 3,1
0.2
b3,1
888 .3400×R
B water
1810.9400
α 3,2
0.2
b3,2
1699. 5870×R
C water
244.4850
α 3,3
0
b3,3
0
The NRTL model equations can be seen in Kosuge and Iwakabe (2005) work. The model was programmed in Matlab and all the simulations were performed in a Intel® Core™ i7-2670QM with 2.20 GHz processor with 8 Gb of RAM memory, running in a Ubuntu 14.04 LTS 64 bits Operating System.
3. RESULTS AND DISCUSSION The pure boiling point of the components, at normal pressure and temperature conditions, is 78.03, 117.34, and 100.9 for ethanol, 1-butanol, and water, respectively. The ternary mixture boiling point, at its initial condition, calculated using a COSMO-SAC model (Lin and Sandler, 2002), is 104.70 ºC. In this manner, it is expected that ethanol and a little fraction of water comes
out in the condenser as distillate, enriching the mixture in 1-butanol, which is supposed to leave the column later. Figures 1 and 2 depicts this behavior inside the distillation column.
Figure 1 – Ethanol profile in the distillation column through time.
Figure 2 – Butanol profile in the distillation column through time. As it can be seen in Figures 1 and 2, firstly the ethanol rises to the condenser, leaving the column and diminishing its content inside of it. As the ethanol leaves, the 1-butanol relative
content rises, and the equilibrium changes, allowing the component to leave the column after the ethanol.
4. CONCLUSIONS Modeling and simulation is a well-known procedure to obtain previous results to visualize the process behavior. Also, it could be used as a manner to optimize design and operation conditions in processes, and develop control strategies, leading to several advantages such as energy economy or more pure products. Although the NRTL inherent limitations in predictions for ternary mixtures, as referred in literature, the model could depict the expected behavior of the distillation process of the ternary mixture of ethanol/1-butanol/water. Better results could be obtained by using UNIFAC, UNIQUAC, or even COSMO-SAC methods. All of these thermodynamic models will be compared for the studied system in a further work.
6. REFERENCES FAÚNDEZ, A. C., ALVAREZ, V. H., VALDERRAMA, J. O. Predictive models to describe VLE in ternary mixtures water + ethanol + congener for wine distillation. Thermochimica Acta, v. 450, p. 110-117, 2006. KOSUGE, H., IWAKABE, K. Estimation of isobaric vapor-liquid-liquid equilibria for partially miscible mixture of ternary system. Fluid Phase Equilibria, v. 233, p. 47-55, 2005. LI, Y., XU, Q., LIU, S., LI, H., ZHANG, F., ZHANG, G., XIA, Q. Liquid-liquid equilibrium for the ternary system of isopropyl acetate + 2-propanol + glycerol at different temperatures under atmospheric pressure. Fluid Phase Equilibria, v. 412, p. 199-204, 2016. LIN, S., SANDLER, S. I. A priori phase equilibrium pediction from a segment contribution solvation model. Industrial & Engineering Chemistry Research, v. 41, p. 899-913, 2002. LUYBEN, W. L., Aspen Dynamics simulation of a middle-vessel batch distillation process. Journal of Process Control, v. 33, p. 49-59, 2015. MARCILLA, A., OLAYA, M. M., REYES-LABARTA, J. A. Simultaneous VLLE data correlation for ternary systems: Modification of the NRTL equation for improved calculations. Fluid Phase Equilibria, p. 1-9, 2015. MARCILLA, A., OLAYA, M. M., REYES-LABARTA, J. A. Comments on the correlation of vaporliquid equilibrium (VLE) data in azeotropic ternary sytems. Fluid Phase Equilibria, p. 1-9, 2016.
