CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM Bugs (http://www.cfd-online.com/Forums/openfoam-bugs/)
-   -   Lagrangian and Continous field donbt match after formToVTK (http://www.cfd-online.com/Forums/openfoam-bugs/62431-lagrangian-continous-field-donbt-match-after-formtovtk.html)

sebastian_vogl August 28, 2008 10:35

Hallo everyone, description
 
Hallo everyone,

description of problem:
I simulated the evaporation of single droplets of C7H16 in air. I used the dieselFoam (OpenFoam version 1.4.1 dev) solver. The geometry slightly differs from the aachenbomb test case (see attached files). The input files were taken from this test case however changed to make single droplets.
I chose a simulation time step of 1e-6s and wrote every time step in the processor folders. I decomposed the case in 4 parts and reckoned it on 4 CPU's. As I didn't want to reconstruct and transform to VTK every single time step, as it would have taken too much time and memory on my disk and isn't necessary for me, I decided to write only every 10th time step into the processor folders, reconstruct and transform them to VTK. When I watched the reconstructed time steps in paraview (version 3.3) I found out that the time step of the Lagrangian field (i.e position, temperature, diameter etc. of the droplets) is at least one reconstructed time step before the surrounding air (i.e temperature of the air, amount of evaporated C7H16 in the air, etc.). That means when I reconstruct every 1e-5s, the lagrangian field is 1e-5s before the continuous phase and when I reconstruct every 1e-4s, the difference between the two phase is 1e-4s. This problem doesn't appear, or at least I can't see a difference in the development of both phases, when I reconstruct and transform into VTK every reckoned time step.
Then I decided to write all time steps in the processor foldes, but only reconstructed and transformed to VTK those of interest, however the result is the same. Then I wrote all time steps (1e-6s) and reconstructed them all but only transformed to VTK every 10th time step. Nothing improved which led me to the conclusion that it doesn't have anything to do with the reconstructing process (as I asumed) but with the foamToVTK process. My assumtion was confirmed when I reckoned the case serial (on one CPU) with a reduced simulation time step on 1e-4s whereas I wrote into the processor files, reconstructed and transformed to VTK only every 50th simulatied time step. The situation didn't change.

For me it seems that there's is something wrong with the foamToVTK module.
I would be glad to hear your opinion about that.
Thank you very much in advance for your time!

Sincerely,
Sebastian Vogl


Solver: dieselFoam
version: OpenFoam 1.4.1 dev; paraview 3.
platform: openSuSe 10.3


picture with the described proble:

http://www.cfd-online.com/OpenFOAM_D...s/126/8906.jpg


I attached some files of my case:

0-files:

http://www.cfd-online.com/OpenFOAM_D...hment_icon.gif T

http://www.cfd-online.com/OpenFOAM_D...hment_icon.gif U

http://www.cfd-online.com/OpenFOAM_D...hment_icon.gif p


constant files:

http://www.cfd-online.com/OpenFOAM_D...hment_icon.gif environmentalProperties

http://www.cfd-online.com/OpenFOAM_D...hment_icon.gif injectorProperties

http://www.cfd-online.com/OpenFOAM_D...hment_icon.gif sprayProperties

http://www.cfd-online.com/OpenFOAM_D...hment_icon.gif blockMeshDict

system file:

http://www.cfd-online.com/OpenFOAM_D...hment_icon.gif controlDict


All times are GMT -4. The time now is 19:05.