[thecommon_jack]

Hey all.

I'm lost, and since having started on a personal project to learn openFoam months ago, I've spent hundreds of hours reading, watching, and troubleshooting, only to feel closer, but somehow dumber. This isn't a thesis project, and I'm neither a student nor business.

But you didn't click on this to read a sob story, so what do I want?

I've attached a link to my case file without sHM run at the bottom, and I've only been able to get this to solve if I turn turbulence off in /constant/momentumTransport. I'd love to know the following:

1. Is my mesh too fine? I'm almost certain it is (nearly 11M cells), but I haven't been able to mitigate non-orthogonal and skew issues otherwise.

1a. If it is too fine, what the heck am I missing or doing wrong in snappyHexMeshDict to get non-ortho and skew issues? What would you change?

2. I'm trying to find how much lift the wing model generates at the given angle of attack (8 degrees). Is it reasonable to be using simpleFoam for this?

I'm only trying to use simpleFoam because that's what the motorbike tutorial uses, but I realize the solution may not be steady state if the wing is stalling (which it shouldn't be at this angle and velocity...).

2a. If it's not the right solver, which should I use, and which tutorial should I start with?

2b. Turing turbulence off is unrealistic, right? This is more of a sanity check.


3. My solution keeps diverging... I think. No matter what time step I use (that's deltaT, right?), be that 0.001 or 10, it gets to the 6th or 7th step, then goes from a few iterations to 1,000, hangs, then exits. What am I doing wrong?

3a. Does simpleFoam even use a time step?

Time = 25

smoothSolver:  Solving for Ux, Initial residual = 0.999993, Final residual = 0.0479919, No Iterations 8
smoothSolver:  Solving for Uy, Initial residual = 1, Final residual = 0.0415902, No Iterations 7
smoothSolver:  Solving for Uz, Initial residual = 0.999997, Final residual = 0.0418846, No Iterations 8
GAMG:  Solving for p, Initial residual = 0.068769, Final residual = 0.000509682, No Iterations 4
time step continuity errors : sum local = 8.00304e+06, global = 5607.69, cumulative = 5607.76
smoothSolver:  Solving for omega, Initial residual = 0.333333, Final residual = 3.79056e-11, No Iterations 1
bounding omega, min: -2.77508e+14 max: 1.91311e+24 average: 2.38377e+17
smoothSolver:  Solving for k, Initial residual = 0.234029, Final residual = 3.00826e-06, No Iterations 1
ExecutionTime = 48.92 s  ClockTime = 49 s

Time = 30

smoothSolver:  Solving for Ux, Initial residual = 0.770476, Final residual = 0.0348405, No Iterations 37
smoothSolver:  Solving for Uy, Initial residual = 0.685049, Final residual = 0.0282642, No Iterations 38
smoothSolver:  Solving for Uz, Initial residual = 0.811879, Final residual = 0.0372875, No Iterations 35
GAMG:  Solving for p, Initial residual = 1, Final residual = 118.374, No Iterations 1000
time step continuity errors : sum local = 2.50918e+15, global = 1.97202e+15, cumulative = 1.97202e+15
smoothSolver:  Solving for omega, Initial residual = 0.329909, Final residual = 6.71818e-08, No Iterations 1
bounding omega, min: -402621 max: 9.56555e+23 average: 2.33779e+17
smoothSolver:  Solving for k, Initial residual = 0.186, Final residual = 0.0032006, No Iterations 1
ExecutionTime = 279.37 s  ClockTime = 279 s

4. I've been running this set of commands. Am I missing one or more?

blockMesh
surfaceFeatures
decomposePar -copyZero
mpirun -oversubscribe -np 20 --use-hwthread-cpus snappyHexMesh –parallel -overwrite
mpirun -oversubscribe -np 20 --use-hwthread-cpus simpleFoam – parallel
reconstructParMesh –constant -noZero
paraFoam

I'm frustrated and frankly embarrassed to ask for help, but I'm losing hope. Let me know what you think and/or find about any or all of the above, and I'll do what I can to answer any questions. Thanks.

Files:

https://drive.google.com/drive/folde...a1?usp=sharing

- Jack

[arjun]

while I can't help you with openfoam because i am not an user, if you are interested in results only in painful manner then i could suggest you to use Wildkatze because it is designed for such cases (assuming the mesh etc isn't perfect and convergence is an issue).

It would be a days job with wildkatze ie set up and run. (even if your mesh is bad at certain places).