Modeling and Optimization of Anode- Supported Solid Oxide Fuel Cells on Cell Parameters via Artificial Neural Network and Genetic Algorithm S. Bozorgmehri 1,2 , and M. Hamedi 1 * 1 School of Mechanical Engineering, College of Engineering, University of Tehran, North Kargar at Jalal-Exp Way, Tehran 1439957131, Iran 2 Renewable Energy Department, Niroo Research Institute (NRI), End of Dadman Blvd., Shahrak Ghodes, Tehran 14665517, Iran Received August 21, 2011; accepted December 21, 2011 1 Introduction Solid oxide fuel cells (SOFCs) are recognized as one of the most promising clean power technologies. They directly con- vert the chemical energy of fuels into electricity and thermal energy and are known for their high efficiency, low pollution emissions, modularity, and fuel flexibility [1]. SOFC technol- ogy has attracted a considerable amount of research activities, in order to overcome challenges in SOFC development. The main challenge is commercializing SOFCs on account of their high cost; and their technical problems such as thermomecha- nical issues, sealing requirements under high temperature operating conditions, rapid degradation of elements causing decrease in performance, the requirement of a large amount of extra equipment, and a long start-up time [1, 2]. Modeling and optimization methods can assist in making SOFC tech- nology cost-effective. Mathematical models of SOFCs, developed to employ a system of multi-physical, chemical, and electrochemical equa- tions, have been extensively applied to improve SOFC perfor- mance and design [3–7]. The mathematical models are gener- ally derived from thermofluid and thermomechanical relationships, the Nernst equation, and equations governing activation, Ohmic, and concentration polarizations. There- fore, these methods require knowledge about many parame- ters of the microstructure and electrochemical properties of the component materials, the operating conditions, the exact multi-physicochemical processes and the numerical solu- tions, making for a complicated problem. Advanced approaches such as artificial neural networks (ANNs) [8–18] and genetic algorithms (GAs) [18, 19] can be applied to modeling SOFC systems and used to improve their performance. An ANN can be used as a black-box tool to simulate systems without solving the physical equations. – [ * ] Corresponding author, mhamedi@ut.ac.ir Abstract An artificial neural network (ANN) and a genetic algorithm (GA) are employed to model and optimize cell parameters to improve the performance of singular, intermediate-tem- perature, solid oxide fuel cells (IT-SOFCs). The ANN model uses a feed-forward neural network with an error back-pro- pagation algorithm. The ANN is trained using experimental data as a black-box without using physical models. The developed model is able to predict the performance of the SOFC. An optimization algorithm is utilized to select the optimal SOFC parameters. The optimal values of four cell parameters (anode support thickness, anode support poros- ity, electrolyte thickness, and functional layer cathode thick- ness) are determined by using the GA under different condi- tions. The results show that these optimum cell parameters deliver the highest maximum power density under different constraints on the anode support thickness, porosity, and electrolyte thickness. Keywords: Artificial Neural Network, Genetic Algorithm, Modeling, Optimization, SOFC FUEL CELLS 12, 2012, No. 1, 11–23 © 2012WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 11 ORIGINAL RESEARCH PAPER DOI: 10.1002/fuce.201100140