Oscillating Convergence Plots

Hey guys,

So I am working on a incompressible CFD Simulation of a building complex and still have a convergence issue since in every plot the values seem to be oscillating. It seems to be stable around a median so my question is do you have any ideas how to smooth the fluctuation out and also the general question if it’s “safe” to work with the median if nothing else works?

I am talking about Run 12, Mesh 11 inside the following project:

I already tried a lot of things like increasing the simulation area, adding region refinements, changing pressure outlet between mean and fixed value, playing with absolute tolerance but nothing seems to work the way I want so I would very much appreciate any help :slight_smile:

here are pictures of the convergence plots:



Hi there,
Thank you for posting on the forum!
We will take a look and get back to you :slight_smile:

1 Like

Hi there,
So I have been going through the project. I was able to remove the fluctuations in the residual plots by removing the boundary condition for no-slip. By default, any walls that do not have any specific boundary conditions, are assumed to be no-slip. So removing this boundary condition from the project helped in stabilising the residuals.

You can try different settings with the relaxation factors as well

3 Likes

Hi @nludwig:

Your last run (12) in Knipex Flow /…/ is based on that TET mesh:

I wonder on which premises you conclude, that you did everything but it does not work.

Please read that thread about ‘Urban Wind’ posted in 2022: https://www.simscale.com/forum/t/simulation-failed-urban-wind/95267/2

Mind as well, that ‘sky’ over your boxes is not the same (wall) as the ground.

Take care,

Retsam

Thanks so much for having a look! It makes so much sense. It solved a huge false narrative I had about the walls. Have a great day!

Well I did everything I personally could find in forums and threads but as this is my very first simulation I would never assume I did everything. That is the very reason I ask here in the forum for help. So thank you for your suggestion and the thread. We are never done learning :slight_smile:

:clap:

In my first CFD simulations I found ‘artistic’ inspiration when visualizing the colorful results of my failed attempts. Well, it was unintended by SimScale tool, however I understood that I need to overcome my ‘cognitive bias’ about ‘CFD is for everyone’. So called “automatic mesh” is useful, but it is perhaps 10% of simulation success.

Take care,

Retsam

1 Like