Numerical simulation investigation of heat exchangers for active chilled beams based on neural networks and a genetic algorithm

Abstract

The indoor thermal environment and air quality are critical components of urban living, making the energy efficiency and performance optimization of air conditioning and mechanical ventilation (ACMV) systems especially important. Active chilled beam systems, recognized for their energy-saving potential, have garnered significant attention. However, while existing investigations have focused primarily on design and control strategies, there has been a lack of in-depth exploration into the structural optimization of heat exchangers within active chilled beams. This investigation utilized computational fluid dynamics (CFD) simulations to examine the effects of fin spacing, tube spacing, and tube shapes on both pressure drop and heat transfer efficiency in heat exchangers. Subsequently, a further analysis was conducted to evaluate how these structural parameters impact the overall cooling capacity of chilled beams. By integrating neural networks and genetic algorithms, the investigation achieved a balance between pressure drop and heat transfer efficiency, resulting in optimal structural parameters to improve the cooling performance of active chilled beams. The results demonstrated that the cooling performance of the chilled beam system with the optimized heat exchanger was significantly improved, reaching a heat transfer rate per unit projected area of 4533.9 W/m², with a cooling performance enhancement of 30.6 %. Under temperature differentials between the heat exchanger and air ranging from 6 K to 22 K, the cooling capacity increased by 26.4–30.6 %.