A Study on the Development of a Heat Transfer Coefficient Correlation for R1336mzz(Z) in the Supercritical Region Using Deep Learning

This study aimed to develop a heat transfer coefficient correlation for R1336mzz(Z) in the supercritical region using deep learning. The unique thermal behavior of supercritical fluids, especially around the pseudo-critical point where property variations are significant, poses challenges for traditional heat transfer coefficient correlations. To address this, an intermediate model was first generated based on experimental data to reduce overfitting risks, followed by deep learning model training. The deep learning approach improved the prediction accuracy for heat transfer coefficients over conventional correlations, effectively capturing rapid changes near the pseudo-critical temperature. The resulting model had a good agreement with experimental data across various conditions, showing a higher coefficient of determination than existing methods, indicating its potential for broader applications in supercritical fluid heat transfer prediction.