We want to simulate the supersonic flow field over 3d blunt body using the shock-fitting technique and a finite-difference scheme. My question is: How can the grid singularity in front of the body nose be removed. Any ideas?

(1). What do you mean by "grid singularity" in front of the body nose ? (2). Why not take a look at the Joe Thompson's book on "numerical grid generation".

Many people have done shock fitting in 2D. Most of what I've seen in the last 5-6 years has been releated to shockwave induced detonations around various blunt objects. (See experiments by Lehr et al. in the early 70s, you get very interesting combustion instabilities depending on the size and mach number of the projectile).

Is there some fundamental difference in 3D that creates a grid singularity? I don't see it. Maybe you just need a different grid?

Dan.

Hello, if you generate a grid for a 3d axisymmetric body, i.e, the (eta) body-normal grid lines fall into each other at the nose. This is a GRID SINGULARITY. Therefore you cannot obtain the metric coefficients in a common way. Maybe you can to some interpolations around this line. I only wanted to hear other ideas.

Hi, Grid singularity can me eliminated by considering grid method for Doubly connected regions mentioned in the Thompsons book. I hope this info is usefull for u. by A.Rajani kumar

Use a grid with multiblock topologie.

Try an "O-GRID" topology. Make the interface in the wake. O-Grids are normally suited for subsonic simulations but you can try it in supersonic flow as well. Another option is to assume a small cyliderical spike-similar to the PITOT in front of the nose. make 3D averaging on the small cylinder which runs from the nose to the FAR FIELD. If you have some validation to make, we can collaborate.

Best of luck

"Use the O grid Luke!" - Obi Wan

If you look at the problem in a finite volume context, there is no singularity, because the flux can only travel AROUND that line. Thus for a finite volume Euler code, there is no problem, and shock capturing works just as well. It get's a little more tricky when you try to define the stress tensor or heat flux vector in the vicinity of that line.

I addressed this singularity problem in great detail --from a finite volume and finite difference point of view -- in my dissertation (viscous compressible flow over rotating disc), which can be downloaded from my website,

www.cfd4pc.com/papers.htm

Viel Spass!

Axel