Modeling the kinetics of pyrolysis of date seeds using artificial neural networks

Abstract

Ground date seeds were subjected to thermal analysis in a stream of Nitrogen at four different heating rates (5, 10, 15 and 20oC.min-1) and their TG -DTG patterns were obtained. Two peaks showed up for the degradation of lignocellulosic components. Three iso-conversional methods were used to obtain the activation energy of these steps: the Flynn-Wall-Ozawa (FWO), the Kissinger-Asahira-Sunoze (KAS) and the Friedmann methods. The results show that the values of activation energy for the first step of degradation varied from 113.76 to 117.80 kJ.mol-1, depending on the calculation method. For the second step, the corresponding values were 130.99, 123.07 and 127.52 kJ.mol-1. At the end of the second peak, biochar was formed that went on cracking off its more volatile constituents at higher temperatures. An artificial Neural Network simulation was carried out for the first degradation step. The values obtained from that simulation for conversion -temperature curves and for biochar content were in excellent agreement with the corresponding experimental figures. However, the simulated values obtained for activation energy at different conversion levels were higher.