Application of Artificial Neural Networks and Adaptive Neuro-Fuzzy Inference Systems to Predict Activated Carbon Properties for Methane Storage

Abstract

BET surface area and micropore volume are important factors for improving methane storage in activated carbons (ACs). Specification and optimization of carbon structures are vastly examined by different researchers. However, because of complex relations between independent and dependent variables, the proposed statistical and mathematical models are not satisfactory. In this paper, the specifications of some ACs synthesized by chemical activation methods are predicted. The effects of parameters such as agent type, activation time, activation temperature, impregnation ratio and heating rate on the BET surface areas and micropore volumes of ACs are also analyzed. Two models of artificial neural networks and adaptive neuro-fuzzy interference systems are used. Later on, a number of data on other ACs reported by several researchers are used for the model validation. The obtained results from these two models are found to be satisfactory. The coefficients of determination for these models were 0.982 and 0.984, respectively. Through this modelling of AC production process, which was the main purpose of this study, the specifications of ACs may be obtained without spending extra time and expenses.