Comparison of performance of three parallel-connected opposite flow ranque-Hilsch vortex tube systems using machine learning methods

In this research, a novel experimental setup was designed to investigate the thermal behavior of a cascade system composed of three counter-flow Ranque-Hilsch vortex tubes. The system operates without any moving mechanical components, apart from control valves used to regulate the volumetric air flow. Vortex tubes with varying nozzle configurations were employed to assess the influence of nozzle count on cooling performance. Theoretical analyses based on the first and second laws of thermodynamics were conducted to establish the relationship between inlet pressure and the resulting temperature separation.Furthermore, a comprehensive machine learning approach was implemented using Support Vector Machines (SVM), Regression Trees (RT), Elastic Net (EN), and Light Gradient Boosting Machine (LGBM) algorithms to model and predict the hot outlet temperature and the cold-hot temperature differential. The predictive capabilities of the models were comparatively evaluated, highlighting the most effective algorithm for estimating system performance under varying operational parameters.