[thorbo]

Hello,
I have a problem with my residuals not converging. For my Master Thesis I need to simulate the Flow of a transformer housing. Im doing a steady simulation using realizable k-epsilon turbulence model with capable wall functions. As you can see, after some time the residuals, especially the continuity residual are not converging anymore. I also put some monitor points in the domain. The velocitys at the monitor points are oscillating around a value.
It seems like the problem is transient and not steady, am i right? I already tried refining the mesh, but no difference.
Does somebody know, what might be wrong with my simulation or do i have to do a transient simulation? The steady simulation already took 20 hours so a transient simulation would take a lot more time I guess.
I´m quite new to cfd so please dont hate if it is a stupid question.

[Shamoon Jamshed]

Originally Posted by thorbo

Hello,
I have a problem with my residuals not converging. For my Master Thesis I need to simulate the Flow of a transformer housing. Im doing a steady simulation using realizable k-epsilon turbulence model with capable wall functions. As you can see, after some time the residuals, especially the continuity residual are not converging anymore. I also put some monitor points in the domain. The velocitys at the monitor points are oscillating around a value.
It seems like the problem is transient and not steady, am i right? I already tried refining the mesh, but no difference.
Does somebody know, what might be wrong with my simulation or do i have to do a transient simulation? The steady simulation already took 20 hours so a transient simulation would take a lot more time I guess.
I´m quite new to cfd so please dont hate if it is a stupid question.

Results you mentioned never say that they are not converged. I do not know what made you think about this? your resdiuals just indicate that they are not going further down. There is mesh limiation that is not enabling the solution to converge more. Therefore, it is recommended to monitor other points for convergence that you already placed. If they follow a constant behaviour that they are observing (as indicated by your pics), then they are converged. So no need to see residuals further. Rest, if the phenomenon is itself time dependent then transient simulation is obvious. o/w no need.