Learning Forum

[jingyuliu]

Hi,

 

I am running the CFD in ansys fluent. The model geometry is as follow.

 

I'm using Fluent Meshing to generate the mesh. During the meshing process, I need to set the minimum and maximum mesh sizes, along with defining the boundary layers. I want to test the effect of the maximum mesh size with 13 boundary layers and a minimum mesh size of 0.01 inches. The maximum mesh sizes I plan to test include 0.9, 0.5, 0.3, 0.2, and 0.1 inches.

Initially, I'm running a k-epsilon simulation for stability. However, for these models, it's challenging to achieve the residual criteria: continuity at 1e-5 and x, y, z velocities at 1e−6. The continuity residual often stalls at 1e-3 and won't decrease further.

When this issue occurs, I regenerate the mesh. Occasionally, a well-meshed model allows the continuity residual to reach 1e-5, but in most cases, it remains difficult to achieve this criterion. 
I am using the polytrada mesh. My mesh is as follows:

[jingyuliu]

Thanks in advance.

best regards,

Jingyu Liu

[SRP]

Hi,

When increasing the number of processors, parallelization effects can sometimes lead to inconsistencies. Increasing the number of cores improves computation speed, but it can also cause more numerical instability due to partitioning effects, especially in complex geometries or with polyhedral meshes.I suggest you to try reducing the under-relaxation factors, particularly for pressure and momentum. This can slow the progress of each iteration slightly but can help with stability. Ensure that the boundary layer mesh near walls provides the necessary y+ values suitable for the k-ε model

[jingyuliu]

Thanks for your response. 

I think the mesh generation cause the problem. COuld you give me more clues about the mesh generation specially for my geometry (specially for the sharp connection between two cylinder). I am planning to generate the boundary in 16 layers.