Kinetic, thermodynamic and artificial neural network prediction studies on co-pyrolysis of the agricultural waste and algae

The thermal decomposition behaviors of Maize straw (MS), algae (AL), and their blends were studied in a thermogravimetric analyzer to evaluate the bio-energy potential in this research. The blends neutralized violent reactions of each other at the second pyrolysis stage, and S and CPI respectively achieved the lowest points at 40%MS-60%AL and 60%MS-40%AL. The average E-alpha calculated by OFW, KAS, Starink and Friedman methods changed from 6.94 to 68.03 kJ/mol, 14.61-85.76 kJ/mol, 14.31-85.12 kJ/mol, respectively. When alpha was in range of 0.2-0.6, the Coats-Redfern method was calculated within 19.39 kJ/mol. KAS and Starink methods seemed more reliable. Reaction mechanism analyzed by Criado method show that AL was significantly different from that of MS. In addition, thermal degradation input by biomass types, heating rate and temperature were well predicted by artificial neural network (ANN). ANN 19 modal (3-5-15-1) was obtained the best (MSE = 0.0978 and R-2 = 0.9999). The modal was utilized to predict the thermogravimetric curve without experiment, DTG(max) of 70%MS-30%AL was the highest. While overfitting occurred in the high-temperature ranges (>700 degrees C). This study proves the potential of ANN modeling for producing derived energy from co-pyrolysis of biomass and other residues.