Air temperature fields inside refrigeration cabins: A comparison of results from CFD and ANN modelling

Abstract

In refrigerated spaces, the inside air is cooled by a heat sink operating either by forced or natural convection. The last situation is more frequently used in small apparatus, such as domestic household refrigerators. The inside air temperature is not usually monitored in these refrigerated spaces. Therefore, knowledge of the air temperature field inside these units is limited and large air temperature gradients often exist that can put the stored products at risk. This work studies temperatures in a commercial household refrigerator that were monitored with thermocouples located at several points. The measured temperatures were then compared with those obtained from two different simulation tools: the Fluent code and another method based on an Artificial Neural Network with supervised learning performed using a Genetic Algorithm. Results lead to the conclusion that, at least in this case, the second tool produced a lower absolute error (0.8 K) when compared with the first (1 K) and yielded modelled inside air temperature fields that are more consistent with reality.