Learning Forum

[wheeler.ky]

Hi all,

I am simulating channel flow with a textured wall using periodic boundary conditions and am running into an issue with post-processing my results. See the photo below for reference. If I run a force report, Fluent calculates that the total pressure and viscous forces on the wall are 6.6836523e-5 and 4.1757701e-5 respectively, as shown below (the wall is broken up into three sections but I'm looking at the total). When I export the solution data, I export the static pressure, periodic static pressure, and x-wall shear stress, x-face area, and y-face area. Since I am looking for the forces in the x-direction (along the channel), I compute the pressure forces by summing the static pressure*x-face area for each cell and compute the viscous forces by summing the x-wall shear stress*y-face area for each cell. 

The issue is that my calculated pressure forces match the results from Fluent (I get 6.6837e-5), but my viscous forces do not (I get 3.9639e-5 vs. Fluents 4.17577e-5), which is about a 5% error. Any ideas on why there would be a discreptancy for the viscous forces but not pressure? I'm exporting the data at the cell center, could it be something with extrapolation since the cell center isn't directly on the wall?

Finally, I am also not quite sure why Fluent calculates the pressure forces based on the static pressure and not the periodic static pressure. The static pressure is the periodic static pressure without the background pressure gradient, but to me it seems like a more realistic result would be to calculate the pressure forces with the periodic static pressure since the flow in reality would have that pressure gradient driving the flow.

Any advice or help would be greatly appreciated!

Thanks,

Kyle

 

[Rob]

It could be a facet-cell issue. If you adapt (refine) the near wall cell what happens to the numbers?

[wheeler.ky]

Hi Rob, thanks for the reply!

I performed some mesh refinement and ended up decreasing cell size on the walls by a factor of ten. Here's a picture comparing the old mesh on the wall (on the right), and the mesh after decreasing cell size 5x, so the final mesh I tested has a wall cell size that is half of the left side of the picture. And all walls were refined, the image is just for a comparison.

After running at the 10x refinement, Fluent gave me a net pressure force of 6.839022e-5 N and net viscous force of 4.0280261e-5 N. Exporting the solution data and performing the calculation described above gives me a net pressure force of 6.83902e-5 N and net viscous force of 3.95294e-5 N. This is about a 2% difference between the value I calculed and the value from Fluent, which is an improvement from the 5% from above. However, I still find it odd that I can only calculate the pressure forces (nearly) exactly and not the viscous forces. I also find it odd that the value seems to be converging towards the value I get when I export and calculate viscous forces, ie. the value from Fluent changed much more than the value I calculated when performing the mesh refinement. This seems to imply that exporting and calculating the forces is a better method since it might not require as fine of a mesh and it's starting to take a while to create the mesh with the refinements.

Again, thank you for the help and let me know if you have any more ideas about why I can't get the values to match closer.

Best,

Kyle

[Rob]

Viscous will be a drag effect, so if you're not resolving the flow boundary layer you'll not get the right answer. Mesh dependency studies are there for a reason, but also aren't just for refining the whole domain. Run another level of refinement and see how that ties up. You may need a couple more cycles until the value stops changing.