[BigJoe]

I'm simulating M=0.9 compressible flow over a cone and cylinder with flow angles. The domain is a square with 100*100mm and cone diameter is 2.2mm, placed in the domain center. In order to have flow angle, I set two velocity inlets with x,y components at the bottom and the left, two pressure outlet at the top and the right. Y+ here is about 0.7. I'm monitoring the inlet Mach to see if it's 0.9

The result I got is very wierd. Does anybody know how to explain what happens near to the bottom? Also there is no wall in this case but why it seems that the flow will be choked?

Any suggestions on BC?

[LuckyTran]

The problem is you have unrealistic boundary conditions. You're imposing the velocity on the bottom and that results in funny business when the shockwave off your body reaches the boundary.

Try to find a better way of changing the direction of the flow. An example would be, keep the velocity inlet on the left, but use a periodic b.c.'s for the top and bottom walls. This might not be a good example because you'll get spatial aliasing if there are any shocks reaching the periodic b.c.'s

[BigJoe]

Thanks a lot bro.

I'm also using the same BC for subsonic case (M=0.3). Do you think I should also change BC for that case? Because the results from subsonic case are not this wierd..

Some people says I should enlarge my domain such as increasing the height of the square. But I guess I will have the same problem in transonic cases when imposing the same BC.

How to explain the flow will be choked in this case even without walls?

[BigJoe]

I use same BC for subsonic case and the result same Ok. But transonic case gave me wierd stuff above.

[BigJoe]

Besides BCs you mentioned. How about I keep velocity with components at the left but switching to pressure-far-field at the bottom and the top with M=0.9 and Ps=Ps(outlet)?

[LuckyTran]

Flow choking is just the flow becoming hyperbolic, which is what happens when your Mach number exceeds 1, which is what is happening. It's a consequence of your boundary conditions. The flow needs to accelerate around the body because of your constraints. Actually your velocity inlet is even worst than a wall. A wall only prevents the flow from freely expanding. A velocity inlet is even worse, as it constricts the flow even further, the flow will accelerate even faster around the body. I wouldn't call this flow choking (because of all the implications of internal flow and such). I would just call this high speed flow.

You will have similar issues even for low Mach number cases. The reason it's apparent in this case is because your inlet Mach number is ~0.9, it's so dangerously close to 1 that the flow is borderline elliptic-hyperbolic whereas a low Mach number case is mostly elliptic.

Increasing the size of your domain will certainly help. But even still, I would not use a velocity inlet on the bottom wall. You should look for robust ways of solving this problem, rather than a workaround.

[BigJoe]

Many thanks. I'm looking for a way to solve this without changing domain since I need to remesh(I'm using unstructured mesh) and the number of cells will increase. It for sure will take more time.

To find an appropriate BC is what I'm struggling for in these days. I also tried mass flow inlet but does not work.

[BigJoe]

Now I switch velocity inlet both at the left and bottom to pressure inlet with specifying direction(x,y). The Mach contour looks almost identical except for that wierd stuff near to the bottom.

I cannot figure out other BC which could work to have flow angles. Any idea guys?

[BigJoe]

Here is the picture

[BigJoe]

Any idea? I'm currently switching from velocity inlet to pressure inlet specifying normalized direction in order to have flow angles...