[kkpal]

dear all
I've come across a strange phenomenon concerning the drag and lift coefficients of a cylinder in uniform flow.
The trend of the two lines seems ok, but it seems that there is big jump within every single timestep, which is shown in the image below.
Why is this happening? is it related to the mesh or algorithm that I use?

[wyldckat]

Greetings Kai,

Please provide more information about the case set-up. Because from what I can see, "deltaT" doesn't have a good enough value and you're only seeing part of the effects of small vortices.

Best regards,
Bruno

[kkpal]

Dear Bruno

this is a LES of flow around cylinder at Re=3900, and I'm using the settings exactly the same with the benchmark case provided here

http://openfoamwiki.net/index.php/Benchmark_ercoftac_ufr2-02

the difference is that mine is a circular cylinder made by GMSH.

I am using pimpleFoam as the solver and I set the maxCo to 0.5 and let the program decide the timestep automatically.

At first I used very coarse mesh, the result from which was mapfielded into better meshed one.
the wiggle was found in the coarse one and I hoped that it would disappear in the better mesh one only it didn't

here are some related pictures
the first one is the geometry of the case, diameter of the cylinder is 1m and the diameter of outside boundary is 100m;
the second one is a closer look of the mesh in the vicinity of the cylinder;
the third one is the Cd and Cl plot, as can be seen the trend seem ok;
the last one is a zoom in for the Cl, which looks ugly

[wyldckat]

Hi Kai,

Mmm... well, the case in the wiki also has a lot of floating around in the plots, since it has a coarse grid...

Have you looked at a cross-section of the flow, to see how the vortices are developing around the cylinder?
I think it's probable that you don't have enough resolution near the cylinder wall, which leads to giant blobs of kinetic energy to flow into the vortices, leading them to float around like a hammer is hitting the fluid from both sides of the cylinder!

Best regards,
Bruno

[kkpal]

thank bruno!

before this I had already done some similar cases which are densely meshed at the first place. In those cases the Cd values are not very close to what I had expected them to be, but at least there is no such wiggles happening in this post.

Just now I modified the blockMeshDict in the wiki case to be a circular cylinder one and from the first few steps there is no wiggle phenomenon this time.
the mesh around the cylinder is still coarse, but I suppose the wiggle phenomenon dose have something to do with the mesh.

I will let you know if I find more.

one more thing occurred to me. Weeks ago I posted a thread asking a similar question

PHP Code:

http://www.cfd-online.com/Forums/openfoam-solving/116839-wiggling-cd-value.html 


in that post the Cd plot also had wiggles, but that wiggle seems more organized than the current one

CFD is so difficult!!

[wyldckat]

Yes, the mesh plays a very important role in the results! It's where the simulation is being conducted

Originally Posted by kkpal

CFD is so difficult!!

It's not difficult! But it's very, Very challenging

I vaguely remember reading the other thread you mentioned... I'll answer it there.