Predictive model for assessing and optimizing solar still performance using artificial neural network under hyper arid environment

Abstract

A mathematical model to forecast the solar still performance under hyper arid conditions was developed using artificial neural network technique. The developed model expressed by different forms, water productivity (MD), operational recovery ratio (ORR) and thermal efficiency (ηth) requires ten input parameters. The input parameters included Julian day, ambient air temperature, relative humidity, wind speed, solar radiation, ultra violet index, temperature of the feed and brine water, and total dissolved solids of feed and brine water. The developed ANN model was trained, tested and validated based on measured data. The results showed that the coefficient of determination ranged from 0.991 to 0.99 and 0.94 to 0.98 for MD, ORR and ηth during training and testing process, respectively. The average values of root mean-square error for all water were 0.04L/m2/h, 2.60% and 3.41% for MD, ORR and ηth respectively. Findings revealed that the model was effective and accurate in predicting solar still performance with insignificant errors.