code_saturne User's Forum

Hello everybody.
I have been using Saturne for a couple of months now for transient simulations, and despite using Salome at first for postprocessing, quickly I went for Paraview and probabaly I'll start practicing with VisIt as it handles the same kind of files. However, I cannot make Paraview display all the timesteps of my calculations, no matter which settings I use, it will keep postprocessing only every 1 second ("real" time).
For example, let's asume that I need to make a transient simulation lasting 2seconds. I will perform those 2s in 200 steps, so each timestep will be of 0.01s. I then select in Saturne's GUI to postprocess at each time step (or every two, four, etc., I always get the same result). After the simulation was done I open the chr.case file in Paraview and instead of having 200 frames, I have only three, that is 0s, 1s and 2s.
Now if I make those 2s in 2000 steps I will need a timestep of 0.001s. Again, Paraview shows only 0s, 1s and 2s. Changing the animation options doesn't help at all. I usually make low precission simulations that are to be simulated along various seconds, so it hasn't been much of a problem, but probably soon I will need better time discretization.
Is there any way to make Paraview display all the postprocessed timesteps? Does VisIt allow to do so? Thank you in advance.
 
César

I experienced no troubles like your description, particularly with the last version 3.8.1 of Paraview. For example the computation posted here https://code-saturne.info/products/code-saturne/forums/flow-modeling-examples/868717903/884723847 works well. Could you confirm that you have the right number of time step record in the file CHR.case?

I just want to confirm the problems César is experiencing. A while back when I was experimenting with Code_S / SYRTHES, I had to revert back to Paraview 3.4.0 to view the individual time steps. Neither 3.6.2 nor 3.8.0 worked as expected. VisIT did, as far as I remember.
I've put my CFD experimentation completely on hold at the moment, so I can't even provide result files for debugging, but César, try downloading Paraview 3.4.0 - it should work. I have no files to test Paraview 3.8.1 with.
Regards,
Claus
 

Thank you both. It is indeed a Paraview 3.8 bug. I tried VisIt and all my timesteps are there. I suppose I'll use VisIt for animations and Paraview for computations (it feels easier to use).
Just for the record, I'm using an up to date Unity Linux 32bit (Mandriva based and similar to PCLinuxOS), and my Paraview was compiled from the sources provided with OpenFoam, but I got no errors during the process. The VisIt version I use is the 2.10 precompiled binary for 32bit systems, downloaded from their official site. I'll try then Paraview 3.4, but so far we can say this issue is solved :)

Hello,
Note that with ParaView 3.4, when opening an EnSight file created by Code_Saturne, you must set the open file dilalog option to "little endian" if the case was run on on Intel/AMD machines, unless you used Code_Saturne's postprocessing output options to force it to "big endian". By default, ParaView expects a "big endian" file.
ParaView versions 3.6 and above detect this automatically (as does EnSight), so I would recommend trying version 3.8.1.
Also, we have seen incorrect reading of EnSight files with polyhedra (after non-conforming mesh joining) in ParaView 3.8.0 (not retested with 3.8.1), while 3.6 and 3.9 (the current developpement version, future 3.10) work fine.
Best regards,
  Yvan