Predicting kinetic parameters for coal devolatilization by means of Artificial Neural Networks

Jiangkuan Xing,Kun Luo,Pitsch Heinz,Haiou Wang,Yun Bai,Chunguang Zhao,Jianren Fan
DOI: https://doi.org/10.1016/j.proci.2018.05.148
IF: 6.535
2019-01-01
Proceedings of the Combustion Institute
Abstract:The chemical percolation devolatilization (CPD) model has been shown to represent the devolatilization process of different coals and heating conditions with good accuracy. However, its use in computational fluid dynamics is limited because of its relatively high computational cost. Here, an Artificial Neural Network (ANN) based model for predicting coal devolatilization kinetics is developed based on a database constructed with the CPD model for a wide range of coals and heating rates. The heating rates and the information of ultimate and proximate analysis are chosen as inputs of the ANN model to consider the effects of coal types and heating conditions on coal devolatilization; the outputs are the kinetic parameters for the two-step kinetic model. The learning, validation, and application results show that the proposed ANN model has a competitive prediction capability on both the total volatile release and release rates when compared with the CPD model, but has obvious computational efficiency advantages. Furthermore, the relative impact of the coal type and heating rate on each kinetic parameter for coal devolatilization is quantitatively evaluated through the Garson equation. It is found that the heating rate has the strongest effect on the pre-exponential factor, while the coal types show significant influence on the activation energy and final yield of the two reactions in the two-step model.
What problem does this paper attempt to address?