Prediction of Optimum Parameters and Product Gas Composition in Gasification Using ANNs

Gasification is the cleanest and most efficient process of utilizing fossil fuels for energy production. Gasification efficiency and product gas quality depend mainly on Equivalence Ratio (ER), Steam-to-Fuel Ratio (SFR), and Gasification Temperature (GT). This study aims at developing an Artificial Neural Network (ANN) model for the prediction of optimum parameters and resulting syngas composition. Very few studies have been reported on ANN modeling with a limited scope of predicting the heating value and composition of syngas. However, none of the previous works are reported on the prediction of optimum operating conditions based on fuel properties so far. With the recent surge in the variety of feedstocks, it is increasingly difficult to determine optimum operating conditions and corresponding syngas composition in a conventional manner. In this scenario, ANNs would be handy in faster and accurate prediction of operating conditions. In the present work, a multi-layer feed forward back propagation network has been proposed for modeling gasification process. The first stage of study involves identification of suitable network architecture and its parameters such that the performance measures are achieved. In the second stage, the network is trained with experimental data available in the literature. The accuracy of the network depends on the amount of data. Since available experimental data is limited, training data is generated using Minitab software. The input parameters used in the model are feedstock composition from the ultimate analysis, and output parameters are operating conditions ER, SFR, GT, and syngas composition (amount of H-2, CO, CO2, and CH4). A 5-5-1 network is used for estimating optimum parameters and an 8-5-1 network is used for estimating syngas components. The predicted data is in good agreement with experimental results with a correlation coefficient of above 0.91 and mean square error less than 0.001.