'Test004' simulation project by andrew_masters


#1

I created a new simulation project called 'Test004':

Mo003


More of my public projects can be found here.


#2

@andrew_masters,

I just had a quick look at your simulation setup - looks good, that will work. However I’d suggest that you first create finer mesh to run the simulation. Actually the most parts of the geometry seem to be reasonable resolved, however the thin connections across the two bigger parts of the flow domain are not resolved well, see screenshot below:

Some of the connections are completely missing, some of them are resolved only with a few cells across the flow cross section, which will lead to artificial blockage. The mesh clip shows it very good:

You can resolve that quite simple by increasing the fineness of the mesh generated. You currently have it on “Coarse” - I guess moderate or fine will resolve all these issues.

Hope I could help,

David


#3

@andrew_masters - new mesh looks better, even though those small channel could use maybe even a couple of more cells across the diameter to capture the high velocities + gradients there. Anyways, I now had a more detailed look at your sim setup. What appeared to me:

  • Is there a specific reason why you’re running the sim laminar? I didn’t do the math now completely regarding the Reynolds number but I guess with 1m/s inlet speed with those small pipes in the middle, I’d say we’re in the turbulent

  • The current mesh has 2 million cells and you’re using 4 cores, which means each core has 500k cells to cope with. My general rule of thumb is always that up to 100k cells per core, you’ll see a significant speed up, so I’d recommend going with a bigger machine (I mean that’s what cloud power is for :smile: )

  • For steady-state problems, I’d recommend using the “initialize with potential flow” option which will give you a faster turn-around time as well

I copied the project, tweaked a bit the setup and ran an analysis myself, check it out here: https://www.simscale.com/projects/dheiny/test004_1

Pressure loss across the device:



Velocity field:



Is that what you would expect? Deriving the mass flow per channel would from this also be quite straightforward.

What’s the application about?

Best,

David