[fluid23]

Occasionally, when running a case, I encounter an issue where the mass flow inlet I have defined (operating with a negative flow rate) will cause considerable instability in my model. Specifically, the density near the boundary condition will crash and approach zero. I have even heard of it going negative for someone else.

Has anyone ever experienced this and do you know what I can do to avoid it?

[ggulgulia]

negative flow rate implies ( assuming you are using CCM+ mass flow report) that the mass flow rate direction is opposite to the normal of the boundary, hence mass flow rate at inlet boundaries is usually negative.

In this case if the residuals are jumping too high, you should try checking the CFL number

[fluid23]

Negative mass flow rate was meant to imply a 'sucking' boundary condition. It is essentially serving the function of a compressor setting mass flow rate through an inlet design in a mixed internal/external aero analysis.

Its not an issue of residuals jumping too high. They do not behave well in general, but the cause is rooted in some kind of conflict between the boundary condition, turbulence model and prism layer mesh. I have been working with CD-Adapco support trying to resolve it, but so far we are mostly coming up empty.

The current consensus is that a better initilization may help stabilize the solution. I originally started with quiesent flow which is where the issue first appeared. I then tried grid sequencing (expert initilization) but that had the same issue, albeit delayed. The results of the grid sequencing were pretty much garbage though. If you have ever used it with any regularity, you can sympathize I am sure. Currently trying to initialize regions independently with somewhat appropriate velocity and so far so good.

If either the support people or myself come up with anything useful I will be sure to add it here.

[fluid23]

I should also mention that we tried CFL ramps with no luck as well.