[Dexonin]

Hello Guys,
I want to simulate air running trough a pipe with a cylinder inside working as a baffle. I want to determine the pressure drop between right before and after the cylinder. The pipe has a inner diameter of 254,15mm. The narrow pass where the air deviates is 1,5mm big. I created my mesh using the models Prism Layer Mesher, Surface Remesher und Thin Mesher. With volumetric controles I was able make the mesh right before and after the cylinder finer. I have 727033 cells in total.
For my physic model I use Coupled Flow, Gas, Ideal Gas, Turbulent, k-omega (SST) turbulence and Steady.
I want to have a set inlet pressure of 1bar so I use a stagnation BC as an inlet and set the pressure from it to 1bar. I also set the initial velocity with my physic model to 0.241 m/s. I am also using a pressure outlet for the outlet.

Now when i simulate the problem the residuals go up as soon as I get a reversed flow.
Do you guys know if the reversed flow is the problem? And if so how can i fix it?


https://imgur.com/a/Fph6WHz

Thanks in advance

[me3840]

You need to re-think your boundary conditions, mesh, and initial conditions.

What is the expected velocity upstream, in the narrow gap, and downstream? The expected pressure distribution? How far is your initial condition from that distribution? Thinking about this, and combine the fact that you're running this steady, what do you expect the solver will be doing under these circumstances?

Consider again the distributions of pressure and velocity. Where will the gradients be steepest? Does your mesh reflect that?

[Dexonin]

Thanks for the quick answer.
To the first question:
I dont know what the exoected velocity will be. The only value I got is the initial velocity of 0,2417818066 m/s. I would assume that the velocity stays at this value until reaching the narrow gap. In the gap the velocity will increase and eventually fall down in the downstream.
To the second question:
I was given a pressure drop for this velocity inlet. It´s 659.8301454082 Pa with the whole upstream being at 1bar and dropping in numbers after the narrow gap.
To your third sentence:
Looking at the STAR CCM+ Tutorial once again and reading a line like: "One example is the solution of flow past a cylinder. At certain Reynolds numbers, the flow is unsteady because of vortex shedding, and it is incorrect to use a steady solution approach." I may switch to unsteady. Is that change correct?
To your third question:
I think the gradients will be changing the mos tin value in the narrow gap. Meaning that I should make my mesh finer in this area?

And once again, thank you for helping me

[me3840]

Originally Posted by Dexonin

Thanks for the quick answer.
I dont know what the exoected velocity will be.

Sure, but can't you make some good guesses with hand calculations?


Now that you know what the pressure drop is, can you think about your initial conditions? How far are they from what the final solution should look like? Consider the initial conditions you're giving the computer. What would happen if you ran an experiment that way?

That sentence you saw about the flow around a cylinder is in the right direction. Do you think a steady solution exists for this flow? If so, considering what you learned from the thoughts on an experiment above, is there anything you can do to make things easier on the solver? There's more than one answer to this.

I don't know if this is considered a capillary or microchannel flow, those are out of my area of expertise, but initially I'm not convinced the steepest gradients are in the gap. Sure, the flow changes a lot there, but it's over quite a distance. To me it seems there's a more obvious candidate. But you're on the right track in thinking about the mesh refinement.