The effect of mediator variable on the higher heating value estimation of low-rank lignite

OpenAlex İngilizce

There are many studies investigating the usage of lignite resources in Turkey by estimating the higher heating value (HHV) with ultimate analysis values of lignite for electricity production in thermal power plants. In this study, it was determined that the lack of investigation into causality while including ultimate analysis variables causes some statistical problems in the HHV estimation with multiple linear regression (MLR) models. In particular, the fact that the element H is not included in the MLR model and the coefficient of the element C contains a negative value creates a situation contrary to coal combustion chemistry. Therefore, an MLR model was developed by considering variables of ultimate analysis that were independent or mediator. As a result, fixed carbon content was found to be independent variable, while the hydrogen and nitrogen contents were found to be mediator variables, and the proposed HHVMMLR model was also proven to be statistically significant. This approach prevents the emergence of faulty regression coefficients and models that contradict the principles of coal combustion chemistry, and it minimizes the influence of elemental analysis data with large values on the multiple regression model.