Influence of Mesh Refinement on the Accuracy of Numerical Results for the Next-Generation High-Speed Train Aerodynamics

Authors:
  • IIIzuan Amin Ishak
    Izuan Amin Ishak
  • NONurnida Elmira Othman
    Nurnida Elmira Othman
  • SSSyabillah Sulaiman
    Syabillah Sulaiman

Abstract
A key concern in computational fluid dynamics (CFD) is the accuracy of numerical simulations, hinging not just on capturing accurate flow physics but also on grid generation. This research employs the Richardson extrapolation technique to examine how mesh refinement impacts the precision of simulation outcomes for aerodynamic loads and flow patterns around a next-generation high-speed train (NG-HST). The study utilizes the ANSYS software, using an SST-based delayed-detached eddy turbulence model for the numerical simulation. The DLR NG-HST train model is utilized for the simulation, involving three distinct mesh resolutions-coarse, medium and fine. Richardson extrapolation and the grid convergence index method are applied to assess the simulation results and these outcomes are subsequently compared with those from a prior study. According to the Richardson extrapolation results, the error for drag and moment coefficients is found to be less than 1.5%. However, for lift, the error was approximately 7% in comparison to experimental data. Additionally, the study revealed that employing a reduced number of mesh elements enhanced accuracy compared to the findings of the previous investigation, in which the GCI values of all parameters decreased with successive grid refinement. This could potentially address the issue of computational costs, such as high memory and computational time, in future studies.

This publication has 28 references indexed in Scilit: