[FBukreev]

Hello,

you can look into the cylinder example how the pressure difference is calculated locally. You can use this command each time step and save it to GNU plot. The GNU plt writer is also applied in that example.

[FBukreev]

Hello,

we are actually working on mesh refinement and wall functions. If you want to simulate a moving body, I can recommend you to test the settlingCube3d example in examples/particles/ where our homogenized LBM ansatz is used. It is the most efficient way to simulate the moving an changing geometries, because they are treated as porosity fields. We are also always ready for a new cooperative projects if you want to develop such a case with HLBM, mesh refinement and wall function.

[FBukreev]

Dear Torsten,

your modification shows now the 3 components of the WSS in the global axes directions.

If you need somewhere the normal or distance to the STL surface, than you can use the surfaceNormal() and distance() function of the indicator, which ist STL as well. You can find these functions in the developer guide and in the setBouzidi file.

Best wishes,
Fedor

[FBukreev]

Hello,

I mean

ParticleManager particleManager(
particleSystem, superGeometry, sLattice, converter, accExt,{false, true});

you can add this {false,true} vector if you want periodicity in the y-direction.

Best regars
Fedor

[FBukreev]

Dear Fuxin,

sorry, for more help from our side you need to come to the spring school.

Best regards
Fedor

[FBukreev]

Dear Fuxin,

for me it seems ok.
If you want more help from us, we need a common project or come to the spring school.

Best regards
Fedor

[FBukreev]

Dear Fuxin,

yes, you can simulate your case using only velocity validation.
The indicator is the wall where you want to calculate y+. It would be the cuboid in your case. IndicatorCuboid3D than.
For U+ you need to write a new lattice functor starting with the existing y+ functor.
If you need mor help for your project you can come to our spring school in Heidelberg.

Best regards
Fedor

[FBukreev]

Dear Fuxin,

I have not investigated this topic about pressure detailled myself, so you nead to look for the reference values in some literature.

Best regards
Fedor

[FBukreev]

Dear Fuxin,

yes, I had these values in converter. I would use the first setup because it is more stable.
The negative values in the turbulent vortexes are ok. You need to perform a convergency study to be sure that the pressure values are correct.

Best regards
Fedor

[FBukreev]

Dear Fuxin,

The 0 material number has no influence on the simulation, so you don’t need to check it.

I have now set the Resolution of 50 and the charPhysU I set in the converter instead of 1 m/s. In this case the simulation remains stable. But if you want to make a real LES I would calculate a finer resolution, check the Kolmogorov scale by the current Reynolds number and make the resolution 3-10 * Kolmogorov.

Best regards
Fedor

[FBukreev]

Dear Fuxin,

I tested the channel3d case with your ReTau, U and Nu with the following wall model parameters:
const int rhoMethod = 2; // method for density reconstruction
// 0: Zou-He
// 1: extrapolation
// 2: constant
const int fneqMethod = 3; // method for fneq reconstruction
// 0: regularized NEBB (Latt)
// 1: extrapolation NEQ (Guo Zhaoli)
// 2: regularized second order finite Differnce
// 3: equilibrium scheme
const int wallProfile = 1; // wallfunction profile
// 0: Musker profile
// 1: power law profile

In olb 1.5 and 1.6 I have seen the same results. The simulation diverges after some time, but if you choose a fine enough resolution it will converge I think.

Best regards
Fedor

[FBukreev]

Dear Fuxin,

I tested the channel3d case with your ReTau, U and Nu with the following wall model parameters:
const int rhoMethod = 2; // method for density reconstruction
// 0: Zou-He
// 1: extrapolation
// 2: constant
const int fneqMethod = 3; // method for fneq reconstruction
// 0: regularized NEBB (Latt)
// 1: extrapolation NEQ (Guo Zhaoli)
// 2: regularized second order finite Differnce
// 3: equilibrium scheme
const int wallProfile = 1; // wallfunction profile
// 0: Musker profile
// 1: power law profile

In olb 1.5 and 1.6 I have seen the same results. The simulation diverges after some time, but if you choose a fine enough resolution it will converge I think.

Best regards
Fedor

[FBukreev]

Hello,

the application of the force takes place in the dynamics, in that case SmagorinskyForcedBGKdynamics. You can find these definitions in src/dynamics/smagorinskyBGKdynamics.h
The force is then taked from the force fiels and applied in the collision.

[FBukreev]

Hello Mike,

the heat flux in SI units is kg/s³. Your idea is to multiply the heat flux with the cell face area?

Regards
Fedor

[FBukreev]

Sorry, we can not give such detailed support outside our spring school framework or a common project.

[Author]

Posts