Artificial Neural Networks Model for Predicting Ultimate Analysis using Proximate Analysis of Coal

In the fossil fuel (coal) based power plants, for estimating the combustion air requirement and for ensuring effective combustion of coal, it is very essential to know the elemental composition of the coal that is fired. Ultimate analysis is the process to be performed to know elemental composition of the coal collected. The ultimate analysis is costly, time-taking and also cumbersome in nature and therefore at the power-plants only gross-level coal compositions are estimated which is called proximate analysis. Based on the gross-level compositions of the coal, the elemental compositions are estimated using standard empirical formulae. The relationship between the gross level composition (i. e. proximate analysis) and the elemental level composition (i. e. ultimate analysis) is nonlinear, whereas the empirical formulae are linear assumptions which may lead to erroneous estimations. The empirical formulae based erroneous estimations lead to variation in the combustion behavior and thereby leading to suboptimal performance of the boilers. To achieve better control on the boilers and thereby to achieve better performance, accurate computation of elemental composition is required. In this article, we suggest a method to compute ultimate analysis based on the proximate analysis information using Artificial Neural Network model (ANN). The predictions of ANN and empirical models have been compared. It is found that the ANN prediction is in very good agreement with lab data than the predictions of empirical model.