Artificial neural network models for heat transfer in the freeboard of a bubbling fluidised bed combustion system

Abstract

The heat transfer in the combustion chamber (freeboard) of a fluidised bed combustor is of critical importance, due to the need for thermal efficiency, the possible environmental impacts, and the increasing costs of fuel and electricity. In this study, an artificial neural network model was used to analyse the heat transfer in a bubbling fluidised bed combustion system. The heat transfer of the system was investigated using temperature measurements of the combustion chamber and side wall of a 0.3 MWe fluidised bed combustor, the fly ash contents (%) of 40 types of coal, the gas absorption coefficients (CO2 and H2O vapour), and the enthalpy of the flue gas. The whole system was analysed using a backpropagation algorithm (10 hidden layers with 10 neurons). It was found that the most important factor when calculating the heat transfer in the combustion chamber was the height of the chamber, and the influence of the flue gas enthalpy and the ash particle content on the heat transfer accounted for 4% and 8.6%, respectively. The accuracy of the results from the proposed multilayer perceptron algorithm is high, and analyses could be performed for different values of the input data.