Application of Artificial Neural Network (ANN) Model for Performance Prediction of a Desiccant-coated Heat Exchanger via COP and Vapor Removal Capacity

Abstract

Purpose: This study has been carried out to assess the performance of a desiccant-coated heat exchanger (DCHX) by the application of an artificial neural network (ANN) model, where its performance was evaluated via water vapor removal capacity and coefficient of performance (COP). Method: The DCHX was prepared by coating the surface of finned tubes using adsorbent powder. Each tube had the dimensions of 200 mm x 150 mm x 22 mm and 0.1 mm thick with a spacing of 1.5 mm. Four tubes pass through a fin, where a tube of 9.5 mm in diameter is used. As for the input data of the ANN, different conditions (parameters) were used for air and water streams. Especially, two different regeneration temperatures (50oC, 80oC) were tested to explore its effect on the development of ANN model. The ANN model was trained by employing 162 data samples from the previous experimental study. To study feed forward and backward propagation, MATLAB code was extensively used as appropriate. For the training of the ANN model, three-fourths of the experimental data was used and the remaining was used for its test and validation. Result: The results show the maximum difference of 0.05 for COP and 0.01 for water vapor removal rate between the ANN model and experimental data. Also, the difference in the regeneration temperature has little effect in affecting the development of the ANN model. This indicates the possible development of a universal ANN model applicable to different operating conditions. The present analysis could be further extended to explore the performance of the DCHX in the context of the 2nd law of thermodynamics.