Investigation of chemical looping pyrolysis characteristics of biogas residue through experiments, kinetic modeling and machine learning

Chemical looping pyrolysis (CLPy) is an innovative thermochemical conversion technique for transforming solid waste into valuable resources. In this paper, thermogravimetry (TG) experiments were conducted on blends of biogas residue (BR) and Fe2NiO4 oxygen carriers (OCs) at ratios (B:O) of 1:0, 0.7:0.3 and 0.5:0.5, under an N2 atmosphere at heating rates of 10, 15, 20, and 25 °C/min. Beside, a neural network model was employed to predict the mass loss curves under various conditions. Experimental results revealed that the pyrolysis process of BR during CLPy occurs in three main stages, with the most prominent peak in the DTG curve emerging in the second stage. Higher heating rates resulted in delayed pyrolysis reactions, and ultimately increasing the BR mass after pyrolysis. The BR and OCs blends exhibited a suppresses effect on the GLPy process, resulting in a decrease in the comprehensive pyrolysis index from 1.52 × 10−4 to 3.04 × 10−5, and a decrease in the maximum DTG from 5.41 %/min to 2.90 %/min as the B:O ratio increase from 1:0 to 0.5:0.5. The average activation energy calculated by FWO method and KAS method is 161.65 kJ/mol and 176.93 kJ/mol, respectively. In particular, the optimized artificial neural network (ANN) model, with 10 hidden layer nodes, a learning rate of 0.01 and minimum error of training target of 1.0 × 10−5, achieves highest R2 of 0.9997 in cross-validation. This model demonstrated superior performance in predicting TG data. These findings provide essential technical support and a scientific foundation for the industrial application of BR energy and the optimization of the CLPy process.