Diffuser reversed flow
 

Hi,

I know the subject of ‘reversed flow’ has been brought up several times, and I have diligently searched previous threads, however I cannot find a solution to my issue.

I am trying to simulate an eductor that uses gas as the motive inlet, and a liquid phase as the suction inlet, both of which are pressure inlet BCs. Due to the study parameters, I have supersonic flow just following the nozzle, but I am continuing to get reversed flow at the end of the diffuser which has a pressure outlet BC.

“reversed flow in XXX faces on pressure-outlet YY”

So far I have tried the following solutions:

• Extending the outlet further away (6-7x the diameter of end diffuser)
• I cannot use ‘outlet’ as a BC because I have compressible, two-phase flow
• Closer matching of each phases’ temperature specification (which you need to specify when using pressure outlet) to the temperatures earlier in the diffuser
• Lengthening of the elements near the exit
• Running the diffuser with a less structured mesh
• Lowering the angle of the diffuser (from 7deg, to 5deg, to 3deg)

Model information:

• Standard K-epsilon, per phase
• Ideal gas compressibility
• 2-phase eulerian mixture model, with implicit body forces
• Pressure-velocity coupled solution
• First order upwind for density, momentum, volume fraction, turbulent kinetic energy, dissipation rate, energy
• Low courant number (=20), and have halved the URF for all other factors

Any help/advice to solve the reversed flow issue would be amazing!

Cheers,
Macdhui

There are two possible causes:

1. Fluent gives the warning without reason (mostly for external flows), where velocity is almost zero.

2. For internal flow, best choice is to place the outer boundary faraway from the main flow.

Thanks Sijal.

I have placed the outlet boundary 10x the internal diameter away from the exit of the diffuser, but I still get this issue. Should I extend it even further or do I need to investigate other sources of the reverse flow?

Macdhui

can you show some pics at the outlet domain (vector plots )

Sijal - I have PM'd you.

Many thanks