Clarification of the flow field

[elinlarsson744-ux]

Hi, i am running a simulation of flow over two cubes, at moderate Reynolds number around 5000. I have run the code in DNS mode with polynomial order 3, and explicit filtering with these setting "type": "explicit",
"order": 4,
"cutoff_ratio": 0.01
When I look at the flow field I see some fringe like patterns, which I am attaching below for your review. I am sure this effect is a not physical, so I am asking is it some kind of numerical artifacts as I don't have fine enough mesh near the walls or my polynomial order is low. The residuals converge quite okay though. Have someone experienced this and can help me with how to make correct this.

Thanks,

[Shiyu-Sandy-Du]

It will be more helpful if you could also share your case file :-) In Neko, the explicit filtering stabilisation in Nek5000 is not implemented so I am quite confused of which filtering your are actually using. (And FYI, for simulations for incompressible flows, we have implemented gradient jump penalty as a stabilisation alternative.)

[elinlarsson744-ux]

I read the documentation again and now I have defined the stabilization as the source terms and now my case file looks like this :

{
"version": 1.0,
"case": {
"mesh_file": "test_mesh.nmsh",
"restart_file": "fluid00001.chkp",
"output_at_end": true,
"output_boundary": true,
"output_checkpoints": true,
"checkpoint_control": "simulationtime",
"checkpoint_value": 1.0,
"numerics": {
"polynomial_order": 5,
"time_order": 3,
"dealias": true
},
"fluid": {
"scheme": "pnpn",
"rho": 1.0,
"mu": 3.125e-4,
"source_terms": [
{
"type": "gradient_jump_penalty",
"enable": true,
"scaling_factor": 0.8,
"scaling_exponent": 4.0
}

   Is it okay now ?

[elinlarsson744-ux]

@timofeymukha, true my this mesh is really coarse to be honest, but I am running for testing purposes only.

[timofeymukha]

Shiyu might have a comment on the parameter values, but that is a valid attempt.

[Shiyu-Sandy-Du]

it is a good parameter to start first

[elinlarsson744-ux]

Allright, thanks very much for the help.