Integrated Experimental and Hybrid CFD–ANN Analysis of Transient Cooling Behaviour in Jet Mixing Systems for Reactor Thermal Stability

Efficient heat removal is essential for maintaining thermal stability and operational safety in chemical reactorsand mixing vessels. This study investigates the transient cooling performance of a jet mixing reactor system using an integrated experimental, Computational Fluid Dynamics (CFD), and Artificial Neural Network (ANN) modelling approach. Experiments were conducted in a cylindrical stainless-steel reactor using water as the working fluid at a Reynolds number of Re = 4300 for two jet configurations (45° side jet and a 90° axial bottom jet). The validated CFD model was then employed to explore six jet configurations, including side jets at 5°, 25°, 45°, and 65°, as well as axial jets at 90° injected from the bottom and top of the reactor. Cooling performance was quantified using normalised temperature decay and two characteristic cooling indicators, t50 (time at 50% cooling) and t90 (time at 90% cooling). The results show that jet orientation significantly affects cooling efficiency. The 90° axial jet top side achieved the best performance with t50 = 325 s and t90 = 1051 s, improving cooling time by approximately 14–16% compared with the worst configuration (5° side jet). An ANN surrogate model trained using CFD-generated data achieved mean square error (MSE) on the order of 10-8, demonstrating high predictive accuracy within the parameter ranges. The integrated experimental and hybrid CFD–ANN framework provides an efficient tool for analysing and optimising jet-induced cooling systems in reactor applications for thermal management strategies.