Machine learning approach for optimization and performance prediction of triangular duct solar air heater: A comprehensive review

This paper presents a comprehensive review of various kinds of distinct artificial roughness employed in rectangular and triangular duct solar air heaters to aid prospective researchers in finding a critical gap in the domain of solar air heaters. A Machine Learning (ML) model is developed using 72 distinct rib combinations compiled to 454 datasets and trained using an Artificial Neural Network (ANN) to predict the performance of ribbed triangular duct Solar Air Heater (SAH). The developed ML model predicts the data with an average deviation of <3%. Owing to reasonably accurate predictions, the same could be increased when more cases (geometric or operating parameters) are added to the databases by retraining the ANN. Further, a second law analysis of the rib configurations features collector efficiency and entropy generation variation with Re for various rib parameters. For the Re range of 4000 to 18000, optimum parameters such as rib height, pitch, chamfer angle, and inclinations are obtained for triangular duct SAH. This could help design engineers obtain the performance parameters of ribbed triangular duct SAH with other artificial roughness designs, possibly with a combination of different geometrical and operating parameters, without having to perform tests.