Evaluation of exergy destructions of different refrigerants in a vaccine cooling system with artificial intelligence

Abstract

Nowadays, low-temperature storage and distribution of many vaccines are as important as their production. In this study, the performance of a storage device operating in a vapour compression refrigeration cycle designed to provide low-temperature cooling between 201 K and 275 K using R134a, R1234yf, R502, and R717 fluids is evaluated by both thermodynamic and artificial neural network (ANN) methods. Levenberg-Marquardt, Bayesian regularisation, and scaled conjugate gradient algorithms are compared with thermodynamical calculations to predict the energy efficiency and exergy destruction of the cooling system. All the considered artificial intelligence algorithms are found to accurately predict the expected outputs with R2 values greater than 0.9.