[CRAC]

Hello,

I am modeling the supersonic flow around a bullet and I'm having trouble validating my domain size. I might be doing this wrong so please let me know...

To know if it converges, I am incrementally increasing my cylindrical domain size, (with a simple mesh for now) and checking for the total drag (Fx) on the bullet every time. My logic is that once the domain is big enough, the Fx should stop varying. I am now at a radius of 70" (versus the bullet radius of 0.2") and the drag force I obtain still fluctuates....not seeming to converge, which is why I feel I'm doing it wrong. I have allowed for 25" downstream from the bullet, which I feel should be enough...Any help? Is there a better way to check domain size convergence?

Thanks,

Chris

[ghorrocks]

You are on the right track, and it is good to hear you are checking these things.

The problem is that your results will only converge when everything in the simulation is close to the necessary values: Key things are mesh size, boundary proximity, convergence tolerance and time step size; but some simulations have other tunable parameters which need sensitivity checks.

So if the drag is fluctuating apparently at random (ie not converging) as you make the boundary proximity larger then it is likely something else is a long way from the necessary value. You say you are checking boundary proximity with a simple mesh - this is likely to be the problem. Your simple mesh is likely to be inadequate to capture the important physics, so nothing will converge until you generate a mesh good enough to be in the converging region of the simulation setup.