Modelling and Optimization of a Pilot-Scale Entrained Flow gasifier using Artificial Neural Networks

Abstract

In this research, the construction and validation of both ANN and RNN models was presented to accurately and efficiently predict both steady state and dynamic performance of a pilot-scale gasifier unit. The corresponding ANN and RNN models’ performance were validated using data generated from a gasifier’s ROM. After validation of ANN and RNN models, optimization studies on the steady state and transient performance of the gasifier were performed under different scenarios. In the optimization studies at steady state, results show that increasing the peak temperature limitation of the gasifier can promote a high maximum carbon conversion. In the dynamic optimization studies, the results show that increasing the peak temperature limitation of the gasifier can lead to higher CO compositions at the outlet of the gasifier. These optimization studies further showcase the benefit of the ANN and RNN models, which were able to obtain relatively accurate predictions for the gasifier similar to the results generated by ROM at a much lower computational cost.