II'm doing a transient flow calculation around a bridge and I need to get the vortex street and calculate the Strouhal number. To do this, I made a structured mesh. At the beginning of the calculation, a flow was established and fluctuations in the lift force were visible, but then, after about 30 seconds, the vortices disappear and a non-separated/non-vortex flow is obtained.
Could it be related to grid resolution? When I make an unstructured mesh, the vortices are more stable over time.
I am not familiar with the specific vortex you mentioned, but in general terms, a structured mesh refers to an organized and less complex mesh structure, while an unstructured mesh is characterized by random and complex shapes. If any other researchers have different insights, please feel free to correct me.
That seems due to a large numerical diffusion.
what about the discretization?
are you using some model for turbulence?
K-w SST model, for time discretization I used time step size 0.005, same for both mesh. The mesh discretization near structure (about 2-3 character size) same. The different is size behind structure.
Well, first you have to consider that RANS provides a smooth solution by definition. You cannot resolve the details of vortical structures.
In statistical sense, it is correct your solution become steady. An excess of diffusion will smooth again the solution.
How about the grid resolution?
could you show the results?
Yes, of course, I know that RAND solver smooth the solution, it's ok.
Yes, after some time, solution have a steady time dependence of Cy and etc. And , problem, that steady solution different for struct and nonstr mesh.
I added picture with mesh resolution. If you not understand, ask me, please)
And for result, I added graphic with frequency of vortex. It is different for two mesh. Mean value the same for both mesh, amplitude is different.
Structured and unstructured grids have differences not only for the size and shape of the mesh but also on the resulting numerical discretization.
The comparison between two different grids requires to be sure that also the schemes can be comparable.
Furthermore, you have to check the comparison by means of a grid refinement study for both.
You have also to compare some other variable, not only the characteristic frequency.
- Badges
- Science topic
- Similar topics
- Mathematics