K and Epsilon Do Not Change
I've been working on a 2d case and I feel making good progress improving my mesh, but I've run into a fairly substantial issue.
I'm using a realizable k-epsilon turbulence model and inputting experimental data as a boundary condition on one side of my 2d channel.
The solution converges, but in the solution k and epsilon both are their proper values on the BC inlet previously described and drop immediately to whatever I defined as the internal field in the /0 directory and remain that way.
I have a case I've been developing in 3d and it (though not yet converged sufficiently) it looks to have the same issue.
Does anyone have any experience with an issue like this?
Some more information on the case:
I'm using timeVaryingMappedFixedValue as the inlet boundary condition for the channel. I have experimental data for the velocity at the inlet and use empirical relationships to find k and epsilon at these points.
Here's a schematic diagram of my case and boundary conditions (2D):
| | <--C
A: Inlet Boundary Condition
B: End of simulation wall
C: Outlet Boundary Condition
D: Test Chamber Wall
p: fixedValue (0)
U: fixedValue (0)
Additionally, I checked and the solution converges more quickly using upwind instead of upwind linear, but the problem persists.
Thank you for any advice/tips on things to try,
Which solver are you using?
Could you post the log-output of a few time steps?
I'm using simpleFoam.
I still have solution instability, but I realized the reason for the zeros for k and epsilon was because of an error calculating them at the inlet.
I'm using timevaryingmappedfixedvalues to map experimental values to the inlet of my flow. I realized the problem is that I'm using empirical correlations to find k and epsilon at the inlet (I only have hotwire data). What's happening is that they're not exactly correct and so the solution is unstable in that region.
Do you have any experience with this boundary condition? Is there a way to "smooth" the boundary condition and make the solution more stable? I hope to expand this simulation to three dimensions, so a more stable solution would be incredibly helpful.
|All times are GMT -4. The time now is 14:58.|