LBM turbulence

[H.Yu]

Hello
I am trying to calculate aerodynamic performance using OpenLB.
I studied some examples. In the 3D cylindrical laminar flow example, I am trying to proceed with calculation by defining it as turbulent flow, but some problems occur.

In the example of a 3D laminar flow cylinder, the result values ​​such as pressure 1, pressure 2, pressure drop, drag force, and lift are not calculated as -nan.

I try to calculate by adjusting the physDeltaX and time step, but I can’t overcome -nan.

What methods can I try?

The code of the part defining the turbulence is also attached.

CODE : bulk dynamics = SmagorinskyBGKdynamics
// Material=1 –>bulk dynamics
sLattice.defineDynamics<BulkDynamics>(superGeometry, 1);

// Material=2 –>bounce back
sLattice.defineDynamics<BounceBack>(superGeometry, 2);

// Setting of the boundary conditions

//setLocalVelocityBoundary(sLattice, omega, superGeometry, 3);
//setLocalPressureBoundary(sLattice, omega, superGeometry, 4);

//inlet
sLattice.defineDynamics<BulkDynamics>(superGeometry, 3);
setInterpolatedVelocityBoundary<T,DESCRIPTOR>(sLattice, omega, superGeometry, 3);
//outlet
sLattice.defineDynamics<BulkDynamics>(superGeometry.getMaterialIndicator({4}));
setInterpolatedPressureBoundary<T,DESCRIPTOR>(sLattice, omega, superGeometry.getMaterialIndicator({4}));
//setInterpolatedPressureBoundary(sLattice, omega, superGeometry, 4);

// Material=5 –>bouzidi / bounce back
//airfoil
#ifdef BOUZIDI
sLattice.defineDynamics<NoDynamics>(superGeometry, 5);
setBouzidiZeroVelocityBoundary<T,DESCRIPTOR>(sLattice, superGeometry, 5, stlReader);
#else
sLattice.defineDynamics<BounceBack>(superGeometry, 5);
#endif

help!

[FBukreev]

Hello,

I think you have not defined the Smagorinsky Constant there
sLattice.setParameter<collision::LES::Smagorinsky>(T(0.15));

Greetings
Fedor

[H.Yu]

None of the above, but I already defined it as the Smagorinsky constant 0.1

[H.Yu]

Is the meaning of ‘-nan’ uncalculated or diverged?

[H.Yu]

What is the relationship between voxel size and stl size if I use stlreader? voxel size =<stl size ??

[H.Yu]

I attach a photo.
https://ifh.cc/g/n5rT5x.jpg

This picture is a speed contour. What are the reasons why it looks like that picture?

I am testing various things such as voxel size, stl size, maxVel, resolution, and time discretization refinement, but it is still calculated at first, and -nan occurs as time passes. There is also the problem that yPlusMax keeps increasing.

Anything else to consider?

[FBukreev]

Hello,

you can look how the stl part is implemented in particles/bifurcation/eulerEuler example with IndicatorLayer function. It must ease the stl inclusion for you.
You can use the UnitConverterFromResolutionAndLatticeVelocity and set the lattice velocity to 0,1 there or with some smaller value. It can make the simulation more stable.

Greetings
Fedor

[H.Yu]

thank you!
I have one more question. Are there any usage examples to learn how to use IndicatorLayer?

[FBukreev]

You can also look into examples/laminar/cylinder3d.
This indicator just enlarges the stl geometry to lattice points depending on cell size.

[H.Yu]

thank you.
I set it up like the example above. Also, when I set ‘UnitConverterFromResolutionAndLatticeVelocity’, ‘-nan’ does not occur anymore.

[H.Yu]

I have one more question.
I want to plot various things using paraview for post-processing, but even if I load the calculated ‘.vtm’ file into paraview, only cuboid is visible. What can I think about to solve this problem?

• This reply was modified 1 year, 6 months ago by H.Yu.

[FBukreev]

You need to use the pvd file to see the results.

[Author]

Posts