Construction of a fast 2D simulation model from 3D for subway tunnels with mass flow conservation under information entropy guidance

Rapidly simulating aerodynamic phenomena such as piston wind in subway tunnels is of significant importance for passenger comfort, air quality, and system operational efficiency. While three-dimensional (3D) Computational Fluid Dynamics (CFD) simulations can accurately capture flow field characteristics, their high computational cost severely limits engineering applications, especially in scenarios requiring rapid pollutant dispersion analysis and real-time emergency response. To overcome this bottleneck, this study introduces information entropy theory to analyze the information distribution patterns of 3D subway tunnel flow fields. By combining this with mass conservation and blockage ratio equivalence principles, a two-dimensional (2D) subway tunnel model was constructed. The results show that the flow field entropy is highest in the longitudinal (x-direction), followed by the transverse (y-direction), and lowest in the vertical (z-direction), providing a scientific theoretical basis for dimensionality reduction to the x-y plane. The constructed 2D x-y model precisely retains the mass flow rate information of the 3D flow field (relative error ≤ 5 %). This study then proceeded to replicate the velocity and pressure trends, finding that its accuracy in reconstructing velocity is superior to that of the 2D model based on hydraulic diameter. This method reduces computation time by 99 %, from 17 h to 0.16 h, offering a novel and efficient computational method for subway tunnel aerodynamics research and offers theoretical support for optimizing subway system performance.