Fast prediction model of three-dimensional temperature field of commercial complex for entrance-atrium temperature regulation

Abstract

Based on the status quo of poor winter thermal comfort of commercial complexes, we developed a three-dimensional temperature prediction model for the typical space of commercial buildings, the “Entrance-Atrium”. A database of 4500 samples was built by using computational fluid dynamics to simulate different spatial deformations and wind environments based on typical models. The simulation method was verified by on-site measurement. Two indicators, Cr (Compliance rate of temperature) and Tr (Temperature resistance index), were defined to evaluate the thermal environment performance of the spatial form. And S (spatial performance score) was used for the comparison of architectural design schemes. In order to achieve rapid temperature prediction, the simulation database was reconstructed and expanded by artificial neural network algorithm, and then the three-dimensional temperature prediction model was constructed. The prediction model may be applied for thermal environment prediction and architectural design optimization for key spaces. Meanwhile, we have expanded its application in air conditioning supply volume adjustment and verified its reliability through practical cases. This study may provide a reliable basis for architects and engineers to make critical decisions, helping optimize the thermal environment through a combination of active and passive methods in the entire process from building design to operation, and improve the utilization efficiency of building operation energy consumption.