[Bob Tipton]

Hello all,

New to using OpenFOAM, but experienced with FEA,CFD and CAE.

I'm designing a passive device which uses an ejector slot to energize its trailing ring vortex. The normal ring vortex is stable and stays attached but the flow isn't well attached to the back of the device.


The goal is to maximize the induced drag of the device and its Cd.


The theory seems to be working. I've gotten some very large Cds (values I would have thought were not possible) so they're probably wrong.

I'm using simpleFoam, I suspect I should be using a method with adaptive meshing so the mesh is improved near the interesting parts.

The models are very difficult to mesh because I need a very large (3 x model dimensions) volume, but the model has a very narrow slot ( w < .05 x model dimensions)

The models require tuning in order to get the boundary layer to attach well in the slot and to get as much flow as possible into the slot. These are contradictory requirements.

There are 3 or 4 critical design parameters which need to be optimized as a set.

I have no success modeling the problem axisymmetrically. That approach produces an isolated vortex core which shows no fluid entering or leaving the vortex. I've had similar issues with a planar analysis. It appears that the 3D effects _may_ be required to make the device work.

Streamlines from one of the better cases.


My problems are this:

It appears that when the device is working well, the vortex is large relative to the device. That requires a larger analysis volume. Yes/no?

The convergence of each run is similar. It takes about 50-200 steps to arrive at a false convergence. The forces aren't changing but the residuals aren't converged. Around step 250 to 1000, the residuals get noisy and go up. At this time the force increases rapidly. Sometimes there is a convergence, but the residuals are 'noisy'. In these cases I get forces on the device as high as 1.6 x the non-energized version. In most cases the vortex grows like a Kansas tornado and engulfs the entire device.

Looking at the images the instability begins at the downstream boundary of the volume - the outlet. I thought I had saved an image of the case, but I'm not able to locate it now. I'll try to reproduce it. What it shows is the calculations exploding near the point where the center of the trailing vortex meets the free stream boundary. Exploding means that the first the pressures (as shown by their false colors) get very large and within a few steps the p values go over 1e20.

I'm working on reproducing the cases.

[Bob Tipton]

Image which would not upload as imbeded image.