Investigation of design parameters of a domestic refrigerator by artificial neural networks and numerical simulations

Abstract

This study presents an application of artificial neural networks (ANNs) to predict the design parameter's values of the static type domestic refrigerator. The interior air volume of refrigerator was modeled using computational fluid dynamics and heat transfer (CFDHT) method and analyses were made. The numerical results were validated by comparing with the experimental results and then inner design parameters were determined. Data sets for training and testing ANN model were acquired by numerical results. The ANN was used for predicting design parameters' values, namely the gap between evaporator surface and glass shelf, evaporator height and surface temperature. ANN predictions demonstrate us a good statistical performance with the average correlation coefficients of 1.00453 and maximum relative error of 2.32%. It is suggested that ANNs model is a successful method for the designers and engineers to obtain preliminary assessment quickly for design parameter modifications of the static type domestic refrigerators.