Performance prediction for a parallel flow condenser based on artificial neural network

This paper reports the use of artificial neural network (ANN) to predict the thermal performance of a parallel flow (PF) condenser with R134a as working fluid. The ANN predictions were compared with the distributed parameter model (DPM) calculation data, which had been validated by experiments at steady-state conditions while varying the air inlet temperature and velocity, refrigerant inlet temperature, pressure and mass flow rate. Based on the data deduced from DPM, ANN was built to predict heat exchange capacity, outlet refrigerant temperature and pressure drop for both air side and refrigerant side. The ANN was optimized for 6-9-5 configuration with Lavenberg–Marquardt (L–M) algorithm, which showed good performance with the root mean square error (RMSE) in the range of 0.00149–0.00605, correlation coefficient (R2) in the range of 0.99989–0.99999 and mean relative error (MRE) in the range of 0.24143–1.31947%.