Problem running perturbUCyl
The question might seem trivial to some, but I am only a beginner with OpenFoam, just 2 weeks into it.
I am trying to simulate pipe flow using LES. I generated the mesh using ICEM CFD and then imported it as Fluent Mesh (.msh file) and then converted it using fluentMeshtoFoam. I then try to generate turbulence quickly, and hence I try using PerturbCylinder. When I try to run PerturbUcyl on the terminal, the following error message is generated. Any idea, as to how it can be fixed. I am modifying the ChannelFoam inbuilt solver to set up my code. Quote:
|
PeturbU
Hi sen,
I am also a newie in OpenFOAM, and I am also simulating the channel flow in OF using LES. I am also trying to use this utility PeturbU, to generate turbulence in the initial field, but did not get a correct version of it, that can run on my machine. Can you tell me the link? Quote:
|
Quote:
http://http://www.cfd-online.com/Forums/openfoam-solving/58905-les-turbulent-pipe-flow.html It seems ok to me. Though I have solved the problem that I had stated previously, which was caused due to wrong initialisations in the 0 directory, I am still trying to find my way. As it stands now, I have extremely high Courant numbers. The time step that I deduced from the experimental data is 2d-5. Does someone have a suggestion as to how one might go about it, or maybe, what one might try? |
Courant number
Hi,
Thanks for the reply. Regarding the Courant no, did you fix the maxCo in controlDict? It happened in my case too, when I ran the calculation, but then I had fixed it, and it was stable. Regards, dhruv. Quote:
|
PerturbU
Hi,
Can you tell me how did you compile it. I am using wmake perturbU, and it gives in the following error. Quote:
Dhruv Quote:
|
Quote:
What you can or rather should do is write 'wmake' only on the terminal. Make sure you are in the 'perturbU' folder, while doing that. It should work, I guess. After that, come out of the perturbU folder and write 'perturbU' on the terminal. The turbulence will be generated I guess. |
I found the source of my problem, somewhat though. The modified 'U' field, after I run perturbUCyl is quite unreasonable. Before running perturbUCyl, my internal field is defined uniformly as (1,0,0). After running perturbUCyl, the non uniform internal field varies somewhere in the range of (10e6 - 10e7, 0, 0). Apart from the unusually high Ux, the 0 field for Uz is not expected. Uy is not altered, that is fine, but why Uz?
Anyone with any suggestions? |
PerturbU Compilation error
Hi sen.1986
Thanks for the input. i tried to compile it. However it again throws up this error. Quote:
Regards, Dhruv. Quote:
|
Quote:
|
Ok, so I found this. The perturbUCyl is functional when I keep the inlet values at zeroGradient. But when I try to put them as cyclic (which is what I need), the following error message is flashed.
Quote:
|
perturbU compilation
Hi
I got the correct answer from this link. Now it works. http://www.cfd-online.com/Forums/ope...m-2-0-0-a.html Quote:
|
Quote:
|
Hi sen,
I was a little late in posting, but the above post is related to the compilation and the problem I was facing with physicalConstant. The link given in my post addresses that. Secondly, I am using cyclic condition for my inlet and outlet. I ran perturbU but did not encounter these errors. However, I did have error like this, when I tried to run my case having cyclic patches in parallel. The only thing I can recommend you is to try to use createPatch to generate your cyclic patches. I hope it helps. Quote:
|
Hello Dhruv,
I got rid of the previous problem. I redid my meshing and modeling, this time using blockmeshdict. The reported problems are mainly caused due to orientation problem of faces. Anyway, my problem with perturbU is far from being over. The 'U' field that perturbU generates, is very unreasonable. The Ux are in the range of ~4000, Uy is 0 (understandably so), Uz is also 0 (Why?). Do you have any idea, what might be causing this? Also, it would be helpful if you could attach three files, your 'transportproperties', 'U' after you run perturbU and the script, perturbU.c. Sorry for sounding too demanding. |
PerturbU
5 Attachment(s)
Hi Sen,
Sorry for the late reply. Can you tell me the following: 1. What is the velocity that you impose at the beginning of the channel? 2. How did you calculate the expected U_tau and Re_tau? I am also having the problems with perturbU, that it does not generate the fields in my whole geometry. I am pretty sure, that I am not wrong in giving the dimensions for the channel case, that I have. I am attaching the pictures of the fields getting generated. As you can see, Uz is being generated only in a part of the channel, and the whole thing. My transport properties are Code:
/*--------------------------------*- C++ -*----------------------------------*\ Code:
/*---------------------------------------------------------------------------*\ Quote:
|
Quote:
I am replying here on a post of already more than five years ago but I hope that you can help me out. I have a similar problem with using PerturbU to generate a U initial field for a cylinder with cyclic boundary conditions. The values for Ux (streamwise direction) are very large ~11000 and Uy and Uz are zero. I tried different Retau but the results are the same. I was wondering if you have found a way to generate a U initial field, either with PerturbU or with another tool? Something that works for pipe flow with periodic conditions. Thank you very much for any suggestions! |
I was searching the forums for something different entirely, but came across this post. In case someone else is searching and looking for an answer to the above problem, I believe it is because the Retau and diameter values are hard-coded in perturbCylinder. You have to edit the perturbU.C file and either change the const scalar Retau and d values to what is correct for your case, or edit the code to read in these values from somewhere (such as the transportProperties file, like the code does for Ubar).
|
I have seen that perturbCylinder only works for flow in the x direction, is it rght? And how can I calculate Ra_tau rightly?
|
All times are GMT -4. The time now is 09:20. |