[biscarri]

hi community,

I'm running my first models using OpenFoam, the problem is a confined water flow, initial velocity is 0.1m/s, the solver is simpleFoam stationary

when postprocessing and animating the results through time steps I observe:
- first 6 to 10 time steps show big oscillations in the velocity field, with changes in velocity even reversing directions
- then these oscillations reduce and the flow field evolves to an each time more stable solution and reaches a visually quasi-converged solution
- after that, additional time steps produce an instabilization that yields in and error stop after 4 - 5 time steps
- cumulative time step continuity errors dramaticaly rise to e+60 order values

I attach the solver log file for the last iterations

thank's in advance for your comments

[Aurelien Thinat]

You have an issue with your turbulence model. Look at epsilon. You are already diverging when you are considering the "solution quasi converged...".

[biscarri]

thank you very much, Aurelien, for your help

can you give me any orientation about what I'm doing wrong?

a) I prepare the case importing the mesh from a UNV file into Discretizer GUI
b) in Discretizer I specify boundary conditions
c) and set several parameters:
- viscosity NU
- RNGkEpsilon model
- turbulence ON
- WALLKAPPA 0.4187 (default value)
- WALLE 9 (default value)

Discretizer offers further wall models: kEpsilon, realizableKE, LaunderSharmaKE, LamBremhorst; for all of them default values are the same as above

[Aurelien Thinat]

I didn't know Discretizer.

But what you can do :
- Launch a checkMesh : The command is "checkMesh > log.checkMesh" (or whatever you want for the output log). It will check the quality of your mesh (upload the log file if you don't understand the output) : skewness, cell's aspect ratio, nonorthogoality...

If the quality of your mesh is low, you'll have to change your fvScheme file (in the system sub-directory).

[biscarri]

I use salome for geometry construction & meshing, then I export to UNV file that I import into Discretizer...
I try to use as much as possible open-source tools with Graphic User Interface in order to work in a more friendly environment
I'm familiar with mesh distortion parameters, but don't know where to launch the command checkMesh, I've tried it in a linux shell, at the open-foam case directory, but command is not found
How do you set your openfoam cases up, what tools do you use?
thank's

[Aurelien Thinat]

1/ You have to go in a linux shell to your case directory.
2/ Then you have to source OpenFoam :
"source ~/OpenFOAM/OpenFOAM-2.0.0/etc/bashrc (adapt the command)
3/ And finally type : "checkMesh > log.checkMesh"

I'm doing all the stuff within a simple linux command shell. It's a bit difficult at first, but I think you learn faster how to set up cases properly by this way than with a GUI.

[biscarri]

OK, it worked, thank you
I attach the log.checkmesh of my model
it seems everything is correct
what do you think?

[Aurelien Thinat]

Yes, everything is good with your mesh. Upload your boundary condition files (from the folder 0/, you should have something like U, p nut, k and epsilon).

[wouter]

hello,

Is your domain 250 m wide, 200 m high and 200 m deep? Are you using the right units?
I needed to change in constant/polyMesh/boundary for the wall the patch type from patch to Wall. I do not know discretizer does this.

hope this helps

Wouter

[biscarri]

thanks again, Aurelian
here I upload a the openfoam case files, I've omitted those big files listing points, faces,...

thanks as well, Wouter
model units are mm-g-s
I use consistent units for NU=MU/RHO, being for water at 20ºC:
MU = 1e-03 g/(mm·s)
RHO = 1e-03 g/mm3
then NU = 1 mm2/s = 1e-06 m2/s (SI)

all the case files uploaded are generated by Discretizer, where you enter the values through its GUI

hope this may help

Lluís