In a previous work (Quintero et al., 2007), a simulation of the ethanol dehydration process in addition to an evaluation of energy and capital costs was performed. For this, several representative configurations were compared under the same input conditions as the case of a fermented mash with a given ethanol concentration (11.4%) and considering the physical-chemical properties of the components of the culture broth. To this end, commercial process simulators were utilized for solving mass and energy balances and performing the economic evaluation.

The simulation of mass and energy balances was accomplished by using the commercial package Aspen Plus v11.1 (Aspen Technology, Inc., Burlington, MA). The composition of the feed stream to the separation and dehydration steps cor­responded to the culture broth resulting from the fermentation step after the cell biomass removal by centrifugation. The composition of this stream for the particu­lar case of fermentation of starchy materials is as follows (in % by weight): water 63.70, ethanol 11.40, CO2 10.80, protein 7.44, linoleic acid 1.71, hemicellulose 1.50, cellulose 1.12, oleic acid 1.40, ash 0.38, glucose 0.22, dextrins 0.20, lignin 0.05, starch 0.01, and others. The flowrate of this feed stream was set to 152,153 kg/h for all the separation and dehydration schemes analyzed. Part of the data for simulation of physical properties was obtained from Wooley and Putsche (1996) who compiled information from the literature, estimated the properties when necessary, and deter­mined a consistent set of physical properties for some key components of fuel etha­nol production process. The remaining data were obtained from secondary sources of information (handbooks, monographs, papers, presentations, etc.). During the simulation, the nonrandom two-liquid (NRTL) thermodynamic model was used to calculate the activity coefficients in the liquid phase and the Hayden-O’Conell equation of state was used to model the vapor phase. The biological transformations were simulated based on a stoichiometric approach as shown in the Chapter 7.

Before the definitive simulation of the schemes, a preliminary simulation of the processes involving distillation columns was performed. The preliminary simu­lation of such columns was done by using the DSTWU module of Aspen Plus, which employs a short-cut method based on the equations and correlations of Win-Underwood-Gilliland. This module was chosen taking into account the pres­ence of the binary ethanol-water azeotrope. This module also provides an initial

estimation of the minimum amount of theoretical stages, minimum reflux ratio, location of the feed stage, and component distribution.

Once performed, the preliminary simulation distillation processes were analyzed using rigorous thermodynamic models. Thus, the rigorous calculation of operating conditions in distillation columns was carried out using the module RadFrac of Aspen Plus based on the method inside-out that employs the MESH equations. This method implies the simultaneous solving of mass balance equations (M), phase equilibrium equations (E), expressions for summation of the compositions (S), and heat balance equations (H) for all the components in all the stages of the given dis­tillation column. Other than the results obtained by the DSTWU module, the infor­mation required by the simulator for the specification of input data of distillation columns was obtained with the help of short-cut methods based on the principles of the topological thermodynamics, especially the analysis of the statics (Pisarenko et al., 2001). Sensitivity analyses were performed to study the effect of main operating variables (reflux ratio, temperature of feed streams, ratio between the solvent and feed, etc.) on the purity of ethanol obtained and the energy consumption. The final result was the establishment of the operating conditions that allow accomplishing more efficient ethanol dehydration processes. The estimation of the energy con­sumption was carried out based on the results of the simulation, taking into account the thermal energy required by the heat exchangers and reboilers. Capital costs, overall operating costs, and other related expenditures were calculated by using Aspen Icarus Process Evaluator™ v11.1 (Aspen Technology, Inc.).

Pressure-swing adsorption was the studied technology for the analysis of etha­nol dehydration by adsorption with molecular sieves. The simulation of this process considered that the adsorption was carried out in vapor phase so the distillate from the rectification column was not condensed and its temperature was raised to 116°C before sending it to the adsorption unit. The technology simulated corresponded to the PSA. The desorption cycle considered vacuum conditions at 0.14 atm. The vapors from desorption were recirculated to the rectification column where the used ethanol was recovered (see Figure 8.7). Vacuum distillation, azeotropic distillation using benzene as the entrainer and extractive distillation using ethylene glycol as the solvent were also analyzed as ethanol dehydration configurations. Each one of the schemes included a scrubber for recovery of 98% of ethanol volatilized and a preheater that brought the stream to be fed to the first distillation column until its saturation point. To concentrate the wine, two distillation columns were consid­ered. In the first (concentration) column, ethanol content was raised up to 50%, while in the second one (rectification column), the ethanol concentration reached 91%. Both columns, the scrubber, and the pre-heater were included in all the dehy­dration schemes simulated (see Figure 8.1).

Obtained results showed the inconvenience of using pressure-swing (vacuum) distillation for dehydrating ethanol. The simulation indicated that large distillation columns with many stages (above 40) are required to obtain a high purity product. In addition, high reflux ratios are needed. These conditions imply high energy costs due to the high heat duty of the column reboilers and to the maintenance of vacuum conditions in the second column having a large amount of trays. Thus, the energy consumption of this dehydration scheme reaches 12.17 MJ/L of ethanol. According to the performed calculations, the distillate of the vacuum column has an ethanol content of 99.3%. The capital costs of this scheme along with the corresponding costs of the other dehydration schemes are presented in Table 8.1.

TABLE 8.1 Capital and Operating Costs (in US$) for Different Separation and Dehydration technologies used for Fuel ethanol Production Vacuum azeotropic extractive Molecular Item units Distillation Distillation Distillation sieves References Ethanol produced kg/yr 141,560,084 142,609,349 141,897,940 142,726,998 Montoya et al. (2005); Total capital costs US$ 14,156,063 9,547,963 9,525,920 12,809,706 Quintero et al. (2007); Total operation costs US$/yr 11,539,808 8,943,642 8,023,714 7,730,563 Sanchez(2008) Utilities US$/yr 9,063,508 7,113,850 6,266,715 53,821,429 Labor US$/yr 600,000 600,000 600,000 600,000 Maintenance costs US$/yr 381,000 78,000 75,100 191,000 Other US$/yr 1,495,300 1,151,592 1,081,899 1,118,134 Unit capital costs US$/kg 0.1000 0.0670 0.0671 0.0897 Unit operation costs US$/kg 0.0815 0.0627 0.0565 0.0542

In contrast, adsorption with molecular sieves showed the best results regard­ing operation costs, i. e., this technology presents lower energy costs (7.68 MJ/L). Elevated capital costs for the configuration involving adsorption are related to the complexity of the automation and control system inherent to the pressure swing adsorption technology. The higher energy consumption for azeotropic (9.77 MJ/L) and extractive (8.44 MJ/L) distillation is explained by the presence of two addi­tional distillation towers that increase the energy costs. The product for these lat­ter schemes contains traces of the entrainer or the solvent unlike the dehydration by adsorption where these third components are not utilized. Simulation results indicate that the extractive distillation can be competitive compared to azeotropic distillation from an energy point of view, as pointed out by Meirelles et al. (1992). According to the simulations performed, the amount of ethylene glycol required to attain the desired dehydration of ethanol was 17,900 kg/h, but it is necessary to cre­ate only 50 kg/h of this solvent thanks to the recirculation stream from the recovery column. The amount of benzene required for azeotropic distillation was 19,980 kg/h, but the amount of fresh benzene in the make-up stream was only 17 kg/h. It is worth noting that the convergence of the simulation of this dehydration scheme was a difficult task that required numerous successive simulation runs. This behavior can be explained by the appearance of multiple steady states and the presence of a parametric sensitivity with small changes in the column pressure when benzene is the entrainer used, as indicated by Wolf and Brito (1995). The obtained results represented a suitable approximation of the results published in different sources (Chianese and Zinnamosca, 1990; Luyben, 2006) as well as the predictions of the thermodynamic-topological analysis. It should be emphasized that the azeotropic distillation using benzene is not an environmentally friendly process and the opera­tion of such dehydration schemes imply the utilization of a carcinogenic substance that can involve potential risks for the operating staff.

From outcomes achieved, it is evident that process simulation represents a pow­erful tool to design the downstream processes of such biotechnological processes as fuel ethanol production. Similarly, the importance of applying a suitable ther­modynamic approach to study the separation operations is another crucial factor influencing the success of the simulation procedures.