[caitao]

Hi，I came across a problem when I use the kEplison model to do the transient simulation of flow around a square cylinder.

A shock appears after several periods of the simulation, the shock vibrates every time-step, and the amplitude remains unchanged.

the attachments shows the phenomenon. point E is a monitoring point of the pressure in the wake area of the cylinder. the period is about 0.01s, and the phenomenon appears near the 27th periods.

1.png

2.png

[RodriguezFatz]

Hi,
We need more information.
1) What solver do you use?
2) What do you mean by "periods of the simulation"?
3) Is it possible that you just see a very poorly resolved vortex street?
4) some time steps at the end of your log file

[caitao]

Hi，RodriguezFatz
thank you very much.
the informations:
（1）the file of fvsolution
// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //
solvers
{
p
{
solver GAMG;
tolerance 1e-06;
relTol 0.1;
smoother GaussSeidel;
nPreSweeps 0;
nPostSweeps 2;
cacheAgglomeration on;
agglomerator faceAreaPair;
nCellsInCoarsestLevel 10;
mergeLevels 1;
}

pFinal
{
$p;
tolerance 1e-07;
relTol 0;
}

"(U|k|epsilon|R|nuTilda)"
{
solver smoothSolver;
smoother GaussSeidel;
tolerance 1e-07;
relTol 0;
}
}

PISO
{
nCorrectors 5;
nNonOrthogonalCorrectors 0;
pRefCell 0;
pRefValue 0;
}
and the file of fvschemes

// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //

ddtSchemes
{
default Euler;
}

gradSchemes
{
default Gauss linear;
}

divSchemes
{
default none;
div(phi,U) Gauss limitedLinearV 1;
div(phi,k) Gauss limitedLinear 1;
div(phi,epsilon) Gauss limitedLinear 1;
div(phi,R) Gauss limitedLinear 1;
div(R) Gauss linear;
div(phi,nuTilda) Gauss limitedLinear 1;
div((nuEff*dev(T(grad(U))))) Gauss linear;
}

laplacianSchemes
{
default Gauss linear corrected;
}

interpolationSchemes
{
default linear;
}

snGradSchemes
{
default corrected;
}

fluxRequired
{
default no;
p ;
}
(2)"periods of the simulation" means ：At first of the simulation, the simulation is correct（there isn't any shock）， while after a period of time，the shock begins to appear.
(3)Actually, the transient simulation is based on the result of steady simulation, and I use the software CFX simulated the case, the result is right.
(4)some information of the file "log":
Time = 0.04264

Courant Number mean: 0.122131 max: 1.2103
smoothSolver: Solving for Ux, Initial residual = 4.04218e-05, Final residual = 7.8952e-08, No Iterations 6
smoothSolver: Solving for Uy, Initial residual = 0.0019072, Final residual = 6.09567e-08, No Iterations 12
smoothSolver: Solving for Uz, Initial residual = 0.000173136, Final residual = 8.44166e-08, No Iterations 8
GAMG: Solving for p, Initial residual = 0.000218439, Final residual = 1.89013e-05, No Iterations 8
time step continuity errors : sum local = 4.78031e-10, global = 6.72354e-11, cumulative = -4.2685e-08
GAMG: Solving for p, Initial residual = 8.85436e-05, Final residual = 7.51789e-06, No Iterations 7
time step continuity errors : sum local = 1.90134e-10, global = -2.34216e-11, cumulative = -4.27084e-08
GAMG: Solving for p, Initial residual = 1.66438e-05, Final residual = 1.5961e-06, No Iterations 14
time step continuity errors : sum local = 4.0367e-11, global = -5.13811e-12, cumulative = -4.27135e-08
GAMG: Solving for p, Initial residual = 3.36047e-06, Final residual = 9.43844e-07, No Iterations 8
time step continuity errors : sum local = 2.38714e-11, global = -3.1315e-12, cumulative = -4.27167e-08
GAMG: Solving for p, Initial residual = 1.55151e-06, Final residual = 9.86122e-08, No Iterations 32
time step continuity errors : sum local = 2.49409e-12, global = -3.44097e-13, cumulative = -4.2717e-08
smoothSolver: Solving for epsilon, Initial residual = 1.33639e-05, Final residual = 3.68449e-08, No Iterations 5
smoothSolver: Solving for k, Initial residual = 4.62431e-05, Final residual = 9.74823e-08, No Iterations 6
ExecutionTime = 7549.61 s ClockTime = 8494 s

Time = 0.04268

Courant Number mean: 0.12213 max: 1.21029
smoothSolver: Solving for Ux, Initial residual = 4.03712e-05, Final residual = 7.88954e-08, No Iterations 6
smoothSolver: Solving for Uy, Initial residual = 0.001907, Final residual = 6.09236e-08, No Iterations 12
smoothSolver: Solving for Uz, Initial residual = 0.000173084, Final residual = 8.4416e-08, No Iterations 8
GAMG: Solving for p, Initial residual = 0.000218598, Final residual = 1.88857e-05, No Iterations 8
time step continuity errors : sum local = 4.77668e-10, global = 6.71387e-11, cumulative = -4.26499e-08
GAMG: Solving for p, Initial residual = 8.87166e-05, Final residual = 7.51532e-06, No Iterations 7
time step continuity errors : sum local = 1.9008e-10, global = -2.34088e-11, cumulative = -4.26733e-08
GAMG: Solving for p, Initial residual = 1.66435e-05, Final residual = 1.59465e-06, No Iterations 14
time step continuity errors : sum local = 4.03328e-11, global = -5.13204e-12, cumulative = -4.26784e-08
GAMG: Solving for p, Initial residual = 3.35917e-06, Final residual = 9.41966e-07, No Iterations 8
time step continuity errors : sum local = 2.38254e-11, global = -3.12867e-12, cumulative = -4.26815e-08

[RodriguezFatz]

Hi again,

1) I still don't get what you mean by "period". Do you see periodic vortices in the wake of the body and the period T = 0.01s?
2) In the log file you show information of time = 0.04264. In your upper picture the whole pressure-mess starts at time = 0.27s. Can you show the log output of a time step when - what you call a "shock" - is already there?
3) Your max. Co. > 1. Does this all happen if you use a smaller time step? Try to halve the time step!
4) What about changing to safer numeric schemes? Such as upwind and uncorrected laplacian?

[caitao]

Hi，RodriguezFatz
1) the period T = 0.01s is base on the st（Strouhal number）of Flow around a cylinder which value is 0.2. The diameter of my model is 0.2m, and the U∞=10m/s. So the f=0.01hz.
2)I show information of time = 0.04264 is another case, but it also have the same phenomenon at that time step.
The good news is that I change the mesh, the y+ is increased form 30 to 49. i didn't change any other sets, and the shock phenomenon disappeared. i think the Co may be the cause of the shock phenomenon.

[RodriguezFatz]

You say the log output is during what you call "shock". But the simulation is converged... this doesn't look like a CO-problem. For CO-problems the simulation diverges and explodes.

[caitao]

Thank you again.
I just change the mesh(do not change any other settings), and then the phenomenon disappears. I still look for the reason that lead to the phenomenon.