ANN for Solar-Assisted Adsorption Cooling Device Performance Prediction with PCM as Thermal Energy Storage

Abstract

The paper deals with use of an artificial neural network for predicting the performance or efficiency of a single bed solar-assisted adsorption cooling (SAC) device with activated carbon-ethanol as adsorbent-adsorbate pair. A single bed harmonize with PCM as an energy storage system has been developed and tested according to the weather conditions (Latitude-8.890 N, Longitude-76.610 E) of Kollam, Kerala. The cooling system features an adsorbent bed, condenser, evaporator, expansion valve and two storage tanks. In this system, the second storage tank has been used for the storing phase change material (PCM) which acts as the latent heat storage device so that the system can be used during night time also. The usage of PCM for thermal energy storage will help to boost the adsorption refrigeration system performance. The 35kg PCM mass has been used for storing solar energy 6.5MJ during the day time for four hours and is used to improve the adsorption cooling system working time by two hours during the night time. The system has been designed for cooling of six litres of water from 30oC to 5oC within one hour. The cooling system produces a refrigeration effect of 4500kJ in each operating cycle with a coefficient of performance (COP) 0.413 and specific cooling power of 26W/s-kg. It is suggested that the Artificial Neural Network be used to predict or forecast the system’s COP and SCP. The Levenberg-Marquardt (LM), Scaled Conjugate Gradient (SCG) and Pola-Ribiere Conjugate Gradient (CG) backpropagation algorithm has been used to evaluate and determine the best approach for predicting the efficiency of the adsorption cooling system. The analysis shows that the findings obtained from ANN are well in line with the result of experiments.