CFD colleagues,

Recently I used the Harten-Yee 2nd order upwind scheme to study inviscid transient flow development in blow-down wind tunnel geometry (C-D nozzle, test section, C-D diffuser) for M=2 and M=4. While known flow features do appear in the simulation, there are other features such as shock oscillation across diffuser throat etc. which appear suspicious. Can anyone suggest experimental / computational literature on the subject

KSR

Hi Ravi,

I discussed this problem with Sanjay Mittal of IITK. He wondered how you could get oscillations predicted just by the Euler code. It is generally held that driving mechanism for shock wave oscillations is the upstream communication of pressures through the viscous sublayer. Please comment.

Selvarajan

it may be that his grid is too coarse? that all I can think of unless he made an error in the code

Selva!

Thanx for your comment and Mittal's observation. It may be true that upstream communication through sublayer is one of the causes of shock oscillations. I must however tell you thet the supersonic flow upstream of the shock is steady to specified residual levels (1e-5) despite the shock oscillation. To my mind, a steady shock will be admitted by a decelerating supersonic flow at any station where locally the R-H conditions can be satisfied with shock speed=0. If this is not possible either because of the flow conditions or because of the numerical scheme, the shock will oscillate because the downstream BC demands a subsonic flow. Problem can arise with numerical scheme if it does not admit a stationery shock as a steady solution. Among Godunov type schemes, the Roe solver with Roe averaging is known to admit steady shocks. Though we have used the Harten-Yee scheme, we have employed arithmetic averaging. I am not however sure that this was at the source of the large scale oscillation observed in the calculations. We may of course employ Roe-averaging and repeat the calculations to clear this lingering doubt.

Thanx again

Ravichandran

Clifford!

Thanx for your comment. The calculations were done with 100, 200, 400 and 800 grid points in axial direction and for second order MUSCL as well as first order Roe scheme with aritmetic averaging. Shock oscillations were present in all cases. Secondly shock oscillations are known to occur in blow-down wind tunnels. The question here is (i) can it occur in an inviscid calculation? (ii) can it oscillate across the second throat or should it be confined to downstream of the throat?

Ravichandran