Artificial neural network-based modeling for the prediction of heat and mass transfer coefficient of the adiabatic liquid desiccant system

Abstract

This work, based on the data obtained from the literature reported by (Varela et al., 2018), aims to use the artificial neural network approach to predict the heat and mass transfer in a dehumidifier system, using lithium chloride as a liquid desiccant. A neural network model was developed in MATLAB environment based on multilayer perceptron that included an input, hidden, and output layer. The network input parameters are air velocity, air temperature, air humidity ratio, liquid desiccant temperature, liquid flow rate, and liquid desiccant concentration. The network output includes two variables which are the heat transfer coefficient (Kh) and mass transfer coefficient (Km). The performance of the ANN model was evaluated using the statistical parameters between the prediction results and experimental values. The performance regression yields R2 and MSE values of 0.9344 and 9.0032, respectively, for the test data set of heat transfer coefficient (Kh). Moreover, for the mass transfer coefficient (Km), the regression parameter R2 and MSE values for the ANN tests were found to be 0.9657 and 2.0414, respectively. In addition, air velocity, air temperature, solution mass flow rate, and solution concentration are the most influential parameters on the heat and mass transfer between the air and liquid desiccant.