How many probes can I have in a simulation ?
Good afternoon everyone,
I run a flat plate simulation, RANS with rhoCentralFoam, with a 600 probes to compute some boundary layer parameters. It was done in parallel processing, and when I reconstructed the fields...there was no fields. And the probes working were not all, only the first 334 probes defined in the controlDict, that is the folder exists but no sub folders with the starting time and fields. Summarizing, not all the probes worked (and no warnings was given) and I can not view the fields in paraview. And the program runs without any other problem. Thanks in advance. |
Similar problem
I run other simulation with fewer probes, they did not work at all. Some of them do not have all the time steps.
Does anyone know what is the problem ? I define the probes in the controlDict in the following way (at the bottom of the file): Code:
functions |
Hi guilha,
I got your private message. I suspect that the problem might be simpler than you expect: you might be running out of disk space. To check this, run: Code:
df -h On the other hand, if disk space isn't the problem, then I need some more information:
Best regards, Bruno |
Thank you Bruno,
If the problem is the lack of memory, should not it give this error http://www.cfd-online.com/Forums/ope...-2-errors.html - my OpenFOAM version is 2.0.1; - debian - I do not know, it might be impossible to know (I am on the university computers) - 3.1 million cells - 900 gb but only 100 gb available - I used 32 nodes and 64 too - I do not know how to use paraView to view the probes, I found a filter with the option of Probe Location but I could not find out anything, but I suppose the problem is not the location - the checkMesh (with the two options) said the Mesh is OK but it gave an error Code:
--> FOAM Serious Error : - my controDict file is (I omitted the probes because I have already put an example) Code:
/*--------------------------------*- C++ -*----------------------------------*\ Code:
Mean and max Courant Numbers = 0.04979834 0.1448724 When I run simulations I can only run 24 hours, then the computer kills the run. When I rerun, from the previous saved time, a new folder is built in the probe's folder. It was on one of those folders that it did not wrote on every probes. For instance if my signal is from 0-0.01 from 0.008 to 0.009, in some probes there is very few results. Edit: I can view the results in paraview. |
4 Attachment(s)
Hi Guilha,
Quote:
Quote:
Quote:
Quote:
Quote:
Quote:
But have a look at the contents of the file "0.0144/uniform/time", to check if it's empty or not. Quote:
Quote:
Nonetheless, everything seems fine with the settings in "controlDict". By the way, I strongly suggest that you use the following settings, if you're trying to save disk space: Code:
writeFormat binary; Quote:
To check the OpenFOAM version, check the beginning of the output. Actually, we already have gotten the confirmation of the version number, from the error message that checkMesh gave you. Quote:
As for 24h, there are some job schedulers in clusters that allow requesting for more runtime than the default. Best regards, Bruno |
Thank you Bruno for your reply,
When I said that the available memory is 900 gb, it is 900 gb to storage data, the RAM is 32 gb. And the memory RAM that this case needs is about 8 gb. When I used the decomposePar, I used the preservePatch however it gave me random errors again on the first iterations the ones from the printstack thread. The checkMesh output is: Code:
/*---------------------------------------------------------------------------*\ The probes, I could not do csv file because there is too many probes but they are definitely in the domain. And I say it again that it only did not save some iterations from one time file. Off-topic 1 I had this output in the checkMesh (other case, RANS k-omega) Code:
/*---------------------------------------------------------------------------*\ Off-topic 2 when I use paraview to view the colour plots I have always the average on the cells, I do not have the option to select the color by interpolation. Is this why I can not plot the streamlines nor isosurface and contour plots ? |
Hi guilha,
Quote:
Quote:
I'm starting to feel that you should try and upgrade to the latest OpenFOAM version, because I think this is a bug that has been fixed a long time ago. Quote:
What does the file "0.0144/uniform/time" contain? Again I suspect this is another bug that has already been fixed. Quote:
The problem is only when you continued the simulation? Then I suspect that it's a bug that has already been fixed in the latest versions. Keep in mind that you are using 2.0.1 and the latest is 2.2.2. Quote:
Quote:
And which file extension? ".OpenFOAM", ".foam" or ".VTK"? Best regards, Bruno |
Hello Bruno, thanks for your replies.
I think the same about the OpenFOAM versions, but that does not depend on me. Does newer versions accept the same files ? About the off-topics, the 1st it is the same mesh (blocks and boundaries) of another case I ran (RANS k-epsilon, but I put the first cell farther the k-omega) so it seems strange to me that it is badly built. The second, I execute the command foamToEnsight then I simply open ParaView with ./paraview, the file extension is .case |
Hi guilha,
Quote:
You can at least try building/using OpenFOAM 2.0.x - http://www.openfoam.org/download/git.php but replace 2.2.x for 2.0.x - which has several bug fixes in the 2.0 series and should be directly compatible with 2.0.1. The other possibility is to talk to your supervisor and/or cluster administrator and up the time constraint for your runs to 48h or 72h. It's LES, therefore 24h is waaay too little time to solve anything. ;) Quote:
In other words, what was the change you made in "blockMeshDict"? Quote:
If you are using 3.10.1 or 3.12.0, run: Code:
touch case.foam Best regards, Bruno |
The meshes only have one difference that is on the walls.
My first point of the block which contains the wall in one case is closer to the wall than the other. Something like this: ....................................... ______________________ \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\ ....................................... ______________________ \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\ This is to respect the y+. The paraview version I have is the 3.14.1 64 bit. |
Quote:
Quote:
|
Sorry, of course it was not only one point, what I meant was that I only changed one thing which was the position of the points just next to the wall. I did not know I had my mesh with a bad topology, I only realised it now :(
However, I found out that increasing the distance it does not give any problem in the determinants. |
hello ,
i am facing similar problem i have check the storage space. its fine the problem i dectected is 0 directory missing in processor folder. and while decomposing . i have seen a message stating First token could not be read or is not the keyword 'FoamFile' thanks in advance manikanth |
Quote:
|
All times are GMT -4. The time now is 15:14. |