[zacko]

Hello everyone,

I am simulating a centrifugal compressor that I designed for a pipeline with methane.
After meshing it with TurboGrid and setting up the Boundary Conditions and solver options in CFX I saw that the solver is placing walls because of backflow. But unusual here is, that it is not at the OUTLET but at the INLET!
And the percentage is around 40 % which is not neglect able I would say.
The efficiency values did converge as well as one additional monitor point.
In one image the streamlines show such backflow near the inlet.

My assumptions are that I either messed up the BC or the INLET is really to "short" and that I need to extend it much further. But extending a bit further only decreases the Wall % a bit (as a tried). So I ask myself whether this is correct to extend so long until no wall (or little %) will be placed.
Also considering changing my INLET to an Opening is a question i ask myself whether this is the right way.

(Sadly I can't fully increase mesh density because I only have a student version which limits mit to 512K nodes. But TG didn't show an mesh errors. Y+ might be a problem here.)

Fluid:
CH4 Ideal Gas

BC INLET:
Total Pressure - 20 [bar]
Total Temperature - 293.15 [K]

BC OUTLET:
Massflow (Total to all sectors) - 200 [kg/s]

If I should provide further information, let me know

I hope someone has a hint for me

Thank you very much!

Daniel

[ghorrocks]

the INLET is really to "short" and that I need to extend it much further.

Yes, that is right. Extend it further upstream. It is best to use the actual upstream profile rather than just simply extrude it.

[zacko]

Then I will try that again and run the simulation with much longer inlet.

Thank you for the hint!

[jmenendez]

I would also try to modify the inlet to stabilize the flow

[Opaque]

Originally Posted by zacko

Hello everyone,

After meshing it with TurboGrid and setting up the Boundary Conditions and solver options in CFX I saw that the solver is placing walls because of backflow. But unusual here is, that it is not at the OUTLET but at the INLET!
And the percentage is around 40 % which is not neglectable

Let us try an analogy to understand the backflow at the inlet. Imagine yourself with a fixed amount of available energy (represented as Total Pressure/Total Temperature), and you have to challenges: climb up the "pressure hill" (compressor), or going down the "pressure hill" (turbine). If you do not have enough available energy across the full inlet there is a point where the amount of mass that cannot be pushed "upwards" has to leave the domain because the static pressure downstream has increased above the available static pressure at the inlet; therefore, backflow.

Also, the % reported must be understood in the local context. Is it 40% of the mesh faces, or 40% inlet area? The latter is not tolerable; however, the former must be visualized in the post-processor to see what it means. Definitely, no backflow is ideal, but compromises are needed at some point.