Artificial Neural Network Modelling of a Mobile Air Conditioning System Using Refrigerant R1234yf

This study investigates modelling of various performance parameters of a Mobile Air Conditioning (MAC) system using artificial neural networks (ANNs). In order to have data for the proposed model, a laboratory MAC system made up from the original components of the air conditioning system of a compact size automobile was set up. Although the MAC system was originally designed for the refrigerant R134a, it was charged with alternative refrigerant R1234yf with considerably less global warming potential. Then, the experimental MAC system was tested in a wide range of operating conditions, i.e. inlet parameters, namely the compressor speed, air stream temperatures entering the evaporator and condenser as well as the relative humidity of the air stream at the evaporator inlet. Using experimental results and the first law of thermodynamics, various performance parameters, i.e. output parameters of the system, were evaluated. The considered output parameters were the cooling capacity, power absorbed by the refrigerant in the compressor, condenser heat rejection rate, coefficient of performance, conditioned air temperature, compressor discharge temperature, refrigerant mass flow rate and pressure ratio across the compressor. Some of the input-output data pairs were used for training the proposed ANN model, while the remaining pairs were employed for testing the prediction performance of the developed model. Yielding correlation coefficients in the range of 0.9159–0.9962 and mean relative errors in the range of 2.24–7.46%, it is revealed that the ANN model performed quite accurate predictions. Because the results of the ANN model are in good agreement with the experimental data, instead of dealing with complex mathematical modelling processes, ANN models can be used for predicting the performance of MAC systems with new environmentally friendly refrigerant R1234yf.