[vikoel]

My best friends, please give me some advice. I am definitely in need of your help!

In a steady-state conjugate heat transfer simulation under the stationary grid (i.e., the water jacket), numerical fluctuations occur at the interface of the transition from coarse grid to fine grid in the inlet pipe, with refinement of scale 3 and layers 2 near the boundary wall, scalable wall function, and realizable turbulence model.*

I am curious about how to eliminate the spatial discretization error caused by this cell size transition, i.e., the small amount of non-orthogonality in the perfectly orthogonal grid. I have come up with three ways.

1) Refining the mesh size of the entire fluid region rather than using near-wall refinement, but this method is expensive in computation.
2) Continue increasing the number of near-wall refinement layers, like a buffer layer, to achieve a lower mesh growth ratio at the interface between the coarse grid and the fine grid.
3) Refining the inlet boundary of the pipe to more accurately capture the entrance effect.*


I very much await your response!!!

Vikoel

April 13, 2024

[drowinsk]

Hello, and thank you for your question. In addition to your 3 solutions provided, you may also consider these two approaches (1) using the inlaid-mesh option to create a boundary layer mesh for your region of interest. In this way, you can directly control the aspect ratio and growth ratio of cells near the wall; and (2) triggering adaptive mesh refinement (AMR) based on the quantity of interest in the near wall region. If you found that your solution was sensitive to the mesh resolution at adjacent jumps in grid level, you can decrease your SGS criterion used for refinement to increase the sensitivity of the AMR controls.