Modeling of Experimentally verified Effect of Basic Input Parameters on Fuel Consumption of oil fired furnace using Machine Learning

Abstract

The authors have attempted to create a model, based on actual experiments conducted on an oil fired Rotary Tilting Furnace of 200 kg melting capacity installed in a foundry unit of Agra. Multiple regression machine learning has been considered as a suitable and novel tool for Modeling of basic input process parameters of rotary tilting furnace. The basic process parameters in a rotary furnace considered are RPM (rotational speed per minute), Time of melting of one charge of 200 kgs (minute) ,melting rate of furnace (kg/hr) and fuel consumed for melting of one charge need to be controlled during whole process. In this paper the relation between input parameters such as rotational speed of the furnace, time, and melting rate have been attempted to be established with output parameter fuel consumption. This model of multiple regression machine learning may found its practical applicability in foundry industry to predict the fuel consumption of furnace before putting it in actual operation and accordingly the input parameters can be controlled for desired optimal fuel consumption. The methodology consists of experimental investigations, modeling of process parameters using machine learning, followed by result analysis. The fossil fuels may not last forever and till no other alternate fuels are developed for melting the foundry industry need to optimize it. The optimized results obtained by this model and computational technique are in line with the results of actual experiments.