This is the first couple of simulations I’ve ever run on Simscale and it seems that I have run into some stability issues, particularly with Uy and Uz.
After reading this post on how to check convergence, I realised I had instabilities.
I have checked my boundary and initial conditions, all of which seems okay to me. This leaves the mesh quality. The highest non-orthogonality I have is 83 degrees and there are only a handful of cells above a non-orthogonality of 70, most of which are not directly in the path of the flow, so I assumed these will not really affect convergence.
My question is, was I wrong to assume that these poor cells will not cause instability, or have I completely missed something in my simulation setup and is the instability due to something else?
I have provided some screenshots of regions of the geometry where the cells have a non-orthogonality above 70 and I have also shared my project link.
Project Link: SimScale Workbench
Maybe the poor cells in the pic above could be causing some issues?
Any help on how I can reach a stable simulation will be much appreciated!
I believe you are worried about the “wiggle” patterns in the residuals? Honestly, the patterns from your particular simulation are not bad. However, the parameters of interest, such as drag coefficients, are still shifting, so the simulation definitely would need more iterations to converge.
In any case, I like to run external aerodynamic cases with the k-omega SST model (instead of k-epsilon, which you currently have). Furthermore, under numerics, running with automatic relaxation is fine. For external aerodynamics, I also like to switch to manual relaxation, while maintaining the factors at 0.3 0.7 0.3 0.3 (for pressure, velocity, k, and omega, respectively).
Yes, however, I was more worried about the unstable convergence plots. I am getting quite a bit of oscillation, particularly with Uz and Uy. For example, Uz at the outlet seems to be diverging.
I will make the changes you’ve suggested, but before I go ahead and rerun the simulation, what do you make of the convergence plots? As in what could be causing the oscillations? Could it just be that more iterations are needed?
Since I haven’t got a reply to my question above I went ahead and implemented your suggestions then ran another simulation.
The settings you suggested seemed to have introduced significantly more instability.
Can someone please have a look at the project and suggest any improvements I can possibly make?
I would make changes and keep rerunning the simulation, however, I’m running low on the core hours I have available.