What’s the general step to validate grid convergence in CFD

Grid convergence validation in Computational Fluid Dynamic (CFD) is essentially about ensuring that the numerical solution of the simulation is independent of the grid size, and is thus sufficient in capturing the characteristics of the flow. Here are general steps for grid convergence validation:

  1. Pre-processing: Setup your problem, taking care to define boundary conditions, material properties and initial conditions.
  2. Coarse grid solution: Solve the problem using a coarse grid first. This will act as the starting point for comparison to assess whether the grid is sufficiently resolving the flow details or not.
  3. Intermediate grid solution: Create an intermediate grid that is finer than the coarse grid used above, and solve the problem again.
  4. Fine grid solution: Create the finest grid that is reasonable to use with regard to computational cost, usually making the grid twice as fine as the intermediate grid, and solve the problem once more.
  5. Results Comparison: Compare the results obtained from all the different grid sizes to check for the variations.
  6. Grid refinement: If the results exhibit significant variations, then refine the mesh in the region of disparities or in the entire domain.
  7. Convergence Study: Calculate error norms and the order of convergence. If the calculated and expected orders of convergence are commensurate, grid convergence is validated.
  8. Developing GCI (Grid Convergence Index): This is a factor which indicates the error in the results and it further confirms the grid convergence for your CFD model.

On observing a consistent result pattern in the calculations with the grid changes, it may be assumed that a grid independent solution has been derived.


作者: 常恭

略懂 OpenFOAM