Suspicious results
3 Attachment(s)
why this suspicious result in p field has obtained?
what may be the cause?what experiences are there? |
Uhm... I already answered to you here: http://www.cfd-online.com/Forums/ope...tml#post449229 post #28.
|
1 Attachment(s)
Hi Bruno
above images was for kOmegaSST case,but for the case with realizableKE the results are smooth as attached at the same snapshot time. maybe this model has damped the variations more strongly? this site should be renamed to:www.Bruno-online.com!:D ;) |
Hi Eshan,
I think I had some problem similar to this in rhoCentralFoam running on a supersonic channel flow: http://www.cfd-online.com/Forums/ope...nnel-flow.html By applying near wall refinement I was able to reduce those strange waves. However, I was never able to get a nice turbulent flow... |
Hi Felipe
thanks for consideration, then you used LES and still this problem existed? in the other run I did with realizableKE as you observe above,its smoother in high pressure parts of the tube. so what may be the cause in your opinion? |
Honestly I do not know. I might have done some mistake setting up the case, but I highly doubt it since I was able to get reasonable results with rhoPimpleFoam.
My whole point of using rhoCentralFoam was to try and get the implicit dissipation to work as my SGS model, in the end, as I mentioned, the flow was rather laminar so adding an explicit model did not change anything. My case was supposed to have almost uniform pressure due to the presence of the forcing term, but still I got the wavy pattern as you see in my other post. What exactly is your case setup? Maybe we are computing completely different things, rendering this discussion pointless :P Quote:
|
Hi
my case is a tube like the tutorial of shockTube,but with unsteady boundaries in both sides,a shock originate like shock tube from one side and moves to the other(front) side. but in times has shown in images above,shouldn't remain strong shocks. I only changed the turbulent model,do you think its because realizableKE always use wall functions while in kOmegaSST case may it use values on the wall directly sometimes during the run? |
Hi Eshan,
The shockTube tutorial is 1D. The plots in the first post do not seem to represent a 1D case. Quote:
|
Hi Felipe
no I said its general shape,I converted it to my problem needs,then my case is 2D as you told.has it any effect if I use cellMDLimited in grad schemes? |
what boundary conditions do you use for the top and bottom boundaries?
have you tried an clean case (without any limiters and whatsoever)? this should shed some light on what is going on. Quote:
|
upper and lower are walls but right and left are patches,yes I used schemes without any limiters till now,but maybe a limiter can result a smoother solution I think,whats your opinion?
|
Hi Eshan,
By walls you mean you impose no-slip conditions right? Have you tried refining the mesh (still without limiters) to see what happens? In my problem with rhoCentralFoam (which I posted above), I was able to remove the waves by applying near wall refinement and using a rather fine mesh. I would see to try refining and identify if the pattern you see is related to the mesh size, or if it stays constant with refinement. In case of the former, then the issue is related with your numerical model (in which case limiters might help), in case of the later, the issue should be related to you physical model. Keep me updated. Cheers, Felipe Quote:
|
but if I make mesh finer near wall its y+ will be too low,I want it be more than 30 till I have a high Re mesh.
edit: yplus turns below 30 in many snapshot times like the attached one,so I use mutUSpaldingWallFunction that someone said to me that its fine for these situations,is it OK in your opinion? |
My point in the previous post is that if you refine (to near DNS resolution) then you shouldn't need limiter or explicit models (someone correct me if I'm wrong). This is a nice way to isolate the problem, I would say.
Quote:
|
All times are GMT -4. The time now is 19:20. |