(647c) Process Synthesis of Biodiesel Production Plant Using Artificial Neural Networks As the Surrogate Models

Biofuel from biomass is very appealing due to its abundance and renewability to replace the usage of the depleting fossil-based fuels. Among many types of biofuels, biodiesel is probably one of the most attractive one because it can be used directly with the current traditional petro diesel engine, either as a substitute or as a blending aggregate [1-3]. Biodiesel is commonly obtained through transesterification of bio oil, either in a form of triglycerides or hydrolyzed fatty acid. There are several alternatives that can be considered for biodiesel production: transesterification with homogeneous catalysts [2, 4] or solid base catalysts [5-7], or by using supercritical methanol without the presence of catalyst [8]. In this work, superstructure optimization is performed to synthesize the optimum biodiesel production plant, i.e., the one that gives the minimum net present sink. Given the high computational resource required to solve the resulting Mixed Integer Nonlinear Programming (MINLP) formulation, using surrogate models is suggested [9, 10]. Among many kinds of surrogate models, Artificial Neural Network (ANN) has been shown to be a powerful tool both for complexity reduction and accurate mapping-function modeling [11, 12]. In this presentation, the feasibility of synthesizing the optimum biodiesel production plant using ANN as the surrogate models for the required unit operations is investigated. The optimum solution of using the alkali-catalyzed reactor with the total cost of USD 41 million was obtained in about 5 CPU seconds. When the solution was modeled in a process simulator, the resulting total cost of the simulation only differs about 1% attesting to the accuracy of the ANNs predictions. However, this comes with a cost of having to provide a large amount of training data for the ANN.

Keywords: Biodiesel Production, Artificial Neural Network, Optimization, Process Synthesis, Surrogate Models

References:

[1]          D. Tapasvi, et al., "Process model for biodiesel production from various feedstocks," Transactions of the American Society of Agricultural Engineers, vol. 48, pp. 2215-2221, 2005.

[2]          Y. Zhang, et al., "Biodiesel production from waste cooking oil: 1. Process design and technological assessment," Bioresource Technology, vol. 89, pp. 1-16, 2003.

[3]          J. M. Lujan, et al., "Comparative analysis of a DI diesel engine fuelled with biodiesel blends during the European MVEG-A cycle: Preliminary study (I)," Biomass and Bioenergy, vol. 33, pp. 941-947, 2009.

[4]          L. F. Bautista, et al., "Optimisation of FAME production from waste cooking oil for biodiesel use," Biomass and Bioenergy, vol. 33, pp. 862-872, 2009.

[5]          S. Yan, et al., "Oil transesterification over calcium oxides modified with lanthanum," Applied Catalysis A: General, vol. 360, pp. 163-170, 2009.

[6]          P. D. Patil and S. Deng, "Transesterification of Camelina Sativa Oil Using Heterogeneous Metal Oxide Catalysts," Energy & Fuels, vol. 23, pp. 4619-4624, 2009.

[7]          D. E. Lopez, et al., "Esterification and transesterification using modified-zirconia catalysts," Applied Catalysis A: General, vol. 339, pp. 76-83, 2008.

[8]          K. T. Tan, et al., "Production of FAME by palm oil transesterification via supercritical methanol technology," Biomass and Bioenergy, vol. 33, pp. 1096-1099, 2009.

[9]          C. A. Henao and C. T. Maravelias, "Surrogate-Based Process Synthesis," in Computer Aided Chemical Engineering. vol. Volume 28, S. Pierucci and G. B. Ferraris, Eds., ed: Elsevier, 2010, pp. 1129-1134.

[10]        N. V. Queipo, et al., "Surrogate-based analysis and optimization," Progress in Aerospace Sciences, vol. 41, pp. 1-28, 2005.

[11]        H. Seung-Soo and G. S. May, "Using neural network process models to perform PECVD silicon dioxide recipe synthesis via genetic algorithms," Semiconductor Manufacturing, IEEE Transactions on, vol. 10, pp. 279-287, 1997.

[12]        I. Sabuncuoglu and S. Touhami, "Simulation metamodelling with neural networks: an experimental investigation," International Journal of Production Research, vol. 40, pp. 2483-2505, 2002.