Parametric analysis of a three-stage cascade refrigeration system for ultra-low temperature applications and performance prediction using multilayer perceptron algorithm

Abstract

In this study, the thermal design of a three-stage cascade refrigeration system capable of operating at ultra-low temperatures wasconducted. In the first part of the study, a parametric performance analysis was performed using six refrigerant groups: R1150/R170/R717, R1150/R170/R1270, R1150/R170/R1234yf, R1150/R744/R717, R1150/R744/R1270, and R1150/R744/R1234yf. The resultsshowed that the most efficient refrigerant group was R1150/R170/R717, and with this group, a COP of 0.65 and an exergy efficiencyof 0.35 were achieved at an evaporator temperature of −85 C. In the second part of the study, a dataset was created using the mostefficient operating parameters identified from the parametric analysis, and a prediction study was conducted using the MLP algorithm.It is important to note that, for the first time in the literature, machine learning was applied to a three-stage cascade refrigerationsystem. The findings revealed MAE values of 0.0006, 0.0193, 0.0003, and 0.0194 for COP, total compressor power consumption,exergy efficiency, and total exergy destruction predictions, respectively, in the test set. It is important to highlight that thethermodynamic and artificial intelligence analyses performed for the three-stage cascade refrigeration system in this study will serveas a model for ultra-low temperature refrigeration system designs.