Development of heat transfer correlation for falling film absorber using artificial neural network model

Abstract

To build an eco-friendly and sustainable energy ecosystem, research pertaining to absorption systems that can utilize waste heat from various industries is garnering attention. Among the elements of the absorption system, the phenomenon of heat and mass transfer in the absorber is complicated, and the predictive performance of existing absorber correlations is insufficient for universal use. Artificial neural network models are being considered as alternatives to existing methods in various fields. However, there is still a need for conventional correlations that allow easy analysis of physical relationships. In this study, rather than developing a model to classify or predict with an artificial neural network, it is used to analyze the collected experimental data. This method is expected to contribute to the easier development of more accurate correlations. Using the learned artificial neural network model, off-trend data organization, key parameter determination, and trend investigation for each parameter are performed. Based on the results of this investigation, a universally usable correlation for predicting absorber heat transfer is developed. The correlation, which can integrate the experimental data of nine published papers, is reviewed; it is demonstrated that the investigation results are consistent with the experimental data, with an error of approximately 22%.