SimScale CAE Forum

Overcome divergence in the simulation - Passive Scalar


#1

Hello Folks,

I am working on the below project. the link can be found here.

https://www.simscale.com/workbench/?pid=5966820855843962607&rru=fd8cac2e-ee7a-45c5-808b-2a6d50e4b8d1&ci=b8eb06df-8132-4612-9094-0a5bfa584c30&ct=CONVERGENCE_PLOT&mt=SIMULATION_RESULT

i believe my mesh seems to be fine, will anyone please suggest modifications to the existing simulation in terms of mesh modification or passive scalar definition or suggestions in numeric section. i request to please let me know what is the reason for divergence of the solution.

thanks for your time and consideration,

thankyou,
Sam.


#2

Hi Sam,

The additional velocity inlet you defined might cause this issue, will try another setup later on. @CFD-SQUAD, feel free to jump in here in case you have an immediate solution.

Jousef


#3

Hello again, Sam,

IMO the mesh has some room for improvement (well, it always does :grin:). You have a bunch of cells with high non orthogonality, as per the meshing log:

image

And another potentially worse issue is not having enough refinement in the pollutant inlet patches. If you pay close attention, the faces are not captured smoothly, which could cause some convergence issues/wrong results:

Despite these remarks, I tried to run some simulations with tweaked numerics (some variations of limited schemes, different URF) but all of them were giving unphysical results.

I got a simulation running right now using limited schemes for gradient/laplacian terms and also non orthogonality correctors. I’ll let you know tomorrow if this was enough to prevent the simulation from diverging. It’s looking good so far :stuck_out_tongue_winking_eye:

Edit: this was the run https://www.simscale.com/workbench/?pid=8374857046360727853&mi=run%3A10%2Csimulation%3A1&mt=SIMULATION_RUN. It ran for 1k iterations, which wasn’t enough to reach convergence just yet (see the area average result control), but it’s safe to say this numerics set works nicely for this specific mesh. Feel free to copy the numerics and try it in your project.
PS: Following Jousef’s hint, I also changed the pollutant source approach from velocity inlet to volumetric sources. Today, when I was inspecting the results, I noticed that I skipped one of the inlets by mistake :man_facepalming: If you decide to use volumetric sources aswell, doublecheck that you have one geometry primitive for each of the inlets :no_mouth:
image


#4

Dear Ricardopg and Jousefm,

i would like to thankyou for all of the above efforts and support. i will try to refine the mesh near the velocity opening vicinity area and rerun again, also i will try to solve the non orthogonality issue right now. However if the velocity approach is not going to help me, i will try to follow the Ricardopg approach. for now i need to work on these things, thanks for the support folks. i will update if i get proper results. thanks for your time and consideration.

Sam.


#5

All props to Ricardo!! :sunglasses: Keep us updated!

Jousef