I’m using CodeBlocks with mingw gcc 4.7.1.
I’ve built tutorial cavity2d successfully and it worked completely fine!
Then, I’ve build successfully the tutorial “externalFlowAroundObstacle”,
but at this time, calculation seems to be diverged.
The standard output of Palabos.exe is as follows;

Reading STL data for the obstacle geometry.
tau = 0.552667
dx = 0.0113924
dt = 0.00227848
Number of iterations in an integral time scale: 438

Voxelizing the domain.
Size of the multi-block: 260-by-79-by-79
Number of atomic-blocks: 1
Smallest atomic-block: 260-by-79-by-79
Largest atomic-block: 260-by-79-by-79
Number of allocated cells: 1.62266 million
Fraction of allocated cells: 100 percent

Generating the lattice, the rhoBar and j fields.
Using BGK dynamics.
Generating boundary conditions.
Periodic lateral boundaries.
Generating an outlet viscosity sponge zone.

Starting simulation.
At iteration 0, t = 0
Force on object over fluid density: F[x] = 3.96586e-314, F[y] = 3.97275e-314, F[
z] = 3.97963e-314
Average kinetic energy over fluid density: E = 0

Writing VTK at time t = 0
Writing PPM image at time t = 0
At iteration 9, t = 0.0205063
Force on object over fluid density: F[x] = nan, F[y] = nan, F[z] = nan
Average kinetic energy over fluid density: E = nan

At iteration 18, t = 0.0410127
Force on object over fluid density: F[x] = nan, F[y] = nan, F[z] = nan
Average kinetic energy over fluid density: E = nan

Hi, DavidA
Thank you for your reply. Please tell me how to change tau value?
I think to change tau value, I must change omega, then invCs2 or nuLB…, it’s little bit confusing for me.

I’ve just execute original settings of the externalFlowAroundObstacle.
And in my Linux system, same condition worked perfectly.

I’ve increased resolution from 80 to 100, them tau was increased from0.55 to 0.56.
But the results was not so changed.

On my Linux system, fine sphere shape was observed in image in tmp directory,
but on Windows system, very distorted shape of obstacle was observed even in
increased resolution case.

I’d like to show these images, but I can’t attach images in this forum.

0.56 is still very far try to get it close to 1 but do not increase only the resolution, lower also the time-step. I have no idea about the OS-dependency, you could try to upload photos on a external website and provide a link and someone may help you with that.

I’m sorry. I can’t agree with you.
Following message is output on my Linux system.

Reading STL data for the obstacle geometry.
tau = 0.552667
dx = 0.0113924
dt = 0.00227848
Number of iterations in an integral time scale: 438

Voxelizing the domain.
Size of the multi-block: 260-by-79-by-79
Number of atomic-blocks: 1
Smallest atomic-block: 260-by-79-by-79
Largest atomic-block: 260-by-79-by-79
Number of allocated cells: 1.62266 million
Fraction of allocated cells: 100 percent

Generating the lattice, the rhoBar and j fields.
Using BGK dynamics.
Generating boundary conditions.
Periodic lateral boundaries.
Generating an outlet viscosity sponge zone.

Starting simulation.
At iteration 0, t = 0
Force on object over fluid density: F[x] = 0, F[y] = 0, F[z] = 0
Average kinetic energy over fluid density: E = 0

Writing VTK at time t = 0
Writing PPM image at time t = 0
At iteration 9, t = 0.0205063
Force on object over fluid density: F[x] = 0, F[y] = 0, F[z] = 0
Average kinetic energy over fluid density: E = 1.77546e-11

In this way, tau=0.55 is still valid on Linux system.
I think there is some bug in handling STL on Windows.

I see your point now, I didn’t know that you have tried the same case under Linux with different result. Anyway that value of tau can give trouble later on in the simulation process. I have succesfully run the externalFlowAroundObstcacle with different STL files under windows so I do not think is a bug, maybe it is related with your stl file. If you upload it I can give it a try but if you have the chance I would run it in linux.