Application of artificial neural network for kinetic parameters prediction of biomass oxidation from biomass properties. Issue 1 (February 2017)

• Record Type: Journal Article
• Authors: Sunphorka, Sasithorn; Chalermsinsuwan, Benjapon; Piumsomboon, Pornpote
• Is Part Of: Journal of the Energy Institute. Volume 90:Issue 1(2017:Jan.)
• Page Start: 51
• Page End: 61
• Publication Date: 2017-02
• Subjects: Artificial neural network -- Biomass -- Kinetics -- Modeling -- Oxidation; Power (Mechanics) -- Periodicals; Power resources -- Periodicals; Fuel -- Periodicals; 621.04205
• DOI: 10.1016/j.joei.2015.10.007
• Languages: English
• ISSNs: 1743-9671

Abstract

Abstract: The development of mathematical models for biomass oxidation is required to understand its decomposition behavior. This work is attempted to develop models for proper process design and monitoring. An artificial neural network technique was applied since it is widely used for modeling a complex non-linear system. A set of data containing one hundred points was selected, and the kinetic parameters of each were determined. ANN models were presented to predict kinetic parameters from biomass compositions. The proposed models could quickly predict the kinetic values and provided the comparable oxidative decomposition trends to experimental data ( R 2 > 0.9). In addition, the relative importance of input parameters on the predicted output was investigated. Ash content had the most effect on frequency factor and activation energy. Fixed carbon content also influenced the frequency factor, while the oxygen concentration had the most effect on reaction order. Graphical abstract: Highlights: The correlation between biomass constituents and oxidation kinetics were proposed. Artificial neural network models were generated to predict the kinetics. 100 data of different sources were essentially selected. The relative importance of biomass constituents was evaluated.