SRFSimpleFoam vs SimpleFoam (MRF)
 

Hello Foamers,

i have a couple of questions about the rotating reference frame solvers in OpenFoam.

i'm simulating a blade in half a cylinder, now the blade is static (represented by a force),

and the entire domain has to move.

i've been reading the threads and OpenFoamWiki,

so from my understanding it would be the same if i use SRFSimpleFoam or if i use SimpleFoam.

so my questions are:

In SRFSimpleFoam i can define the axis of rotation but i'm unable to define the origin of my rotation, so is the origin going to be (0,0,0) all the time? or if there a way to set it?

when i see SimpleFoam with the MRF option (MRFSimpleFoam), i can define the axis and the origin.

would it be the equivalent of SRF if i define the entire domain as the cellzone that rotates?

also when i use simpleFoam (domain with no obstacles in constant rotation), my inlet bc is fixedvalue (5,0,0), and my omega is 700rpm, but when i see my boundary in parafoam after a couple of iterations it becomes (0, something, something).
i understand if it adds the rotational forces but why is setting my axial inflow to 0 when it's suppose to be fixed to 5?

and in the internal field my axial velocity tends to 0 and gets negative values even.


if someone could throw some light into this i would really appreciate it

thanks

Oscar

Hi Oscar,

If you put all the cells in the MRFzones, theoretically SRF and MRF will give you the same result.

I did a very simple test before, that is using simpleFoam, SRFSimpleFoam and MRFSimpleFoam to run the same case. That means three different solvers solving the same problem, though the setup for each solver is different. Unfortunately, MRFSimpleFoam cannot give me the correct result. The results from simpleFoam and SRFimpleFoam are the same, and agree with my analytical solution. But the result from MRFSimpleFoam does not. And right now I still don't know the reason why the results are different. If you have any news, please update me.

Thanks,
Pengchuan

Hi Pengchuan,

I found a solution to my problem,

first i discarded SRFSimpleFoam, the solution is too unstable at least for my case.

i used SimpleFoam with the MRF option.

now as i said before if you set all the cells in the rotationg frame it will crush your boundary conditions, i think it's because MRF is made for more than one frame, so if you put only one frame these kind of problems arise.

the solution for me was to leave at least a one cell distance between the rotating frame and the BCs, this way your BCs are respected and the overall solution is not affected (it behaves as SRF).

hope it helps,

Oscar

Hi Oscar,

Thank you for your reply. Then your results are exactly the same? I mean the results with MRF and without MRF. For me, the results are different, and without MRF is better than the result with MRF.

Thanks,
Pengchuan

Hi,

i think they are suppose to be different :)

in simpleFoam without MRF you only have the inertial frame, and in simpleFoam with MRF you have the inertial frame plus the rotating frame in other words you add the centrifugal and coriolisis forces.

the only case in which both will have the same results is if you have a domain with no obstacle cause then the forces won't have any effect over the flow.

as soon as you add an obtacle let say a force, the solutions will be different.

you should read this article in openFoamWiki, here they explain how the MRF works.

http://openfoamwiki.net/index.php/Se...RF_development

hope it clears out some doubts,

good luck

Oscar

Hi Oscar,

Thank you for your reply. Yes, you are right, they use different method to solve the equation. However, though they use different methods, they are actually solve the same PDE, since they are simulating the same problem. No matter what method you use, the results should be the same. But right now, I got different results from them, that's what makes me confused.

Thanks,
Pengchuan

Hi Oscar,

in your first Message you wrote:

Did you find an answer to the question about setting the origin in SRFSimpleFOam?

Cheers, Dennis

I am also getting different results in SRF and MRF.

I am simulating a 3-D stirred tank, in which the propeller is on the bottom. In the SRF case, the liquid is stirred along the whole tank, whereas in the MRF case, the liquid is not getting stirred not too far away from the propeller. I have tried many ways to solve the MRF, messing up with different solution algorithms, interpolation schemes, and mesh grades, but the result is always very similar. On the other hand, in the SRF case, I did a test where I deleted the rotor, but it resulted in the whole tank equally stirring (besides the top and the bottom, of course). I guess, the rotor is not essential for the rotating the system in SRF, because the source terms are added everywhere. Interestingly, I did a test in MRF, where I elongated the rotating cellZone up to the top of the tank, but the result was the same as in all other MRF tests, instead of being similar to the SRF case.

I am not too sure that SRF cannot be used in steering tanks, where the rotor is not as long as the tank itself. It seems to me that it cannot. I can't be sure, because I don't have access neither to a CFD expert nor to experimental data to validate my case.

Could anyone clarify that?

Attached is the SRF case without propeller. Boundary conditions are, side and bottom walls U (0 0 0), and top wall slip.

Hi Rudolf,

I am currently working on SRFSimpleFoam, The case you have mentioned is right. cozz we have only one cellzone in this case and that whole set of 3D meshes keep rotating. The blade or propeller are kept in the way of the rotating fluid to simulate the results. In realtime, the blade rotates and the fluid is stationary, for simulation purpose, we keep the blade stable and rotate the fluid for calculation of effects.

testing simpleFoam using MRF and leaving one cell outside to define the stationary zone, gives me non-reliable values. The y+ increases up to 5000!