Standing shock problem... I wrote a UDF to initialize a 1-D problem with the exact pre and post-shock states for a M=2 shock (used J. Anderson for T2/T1, u2/u1, p2/p1, etc.). First-order Roe's scheme should (if coded correctly) perfectly resolve the shock exactly with zero residual. However, Fluent does not and smears the shock over time. Why? There's no limiter. Does Fluent apply an entropy fix which might cause this?

1-D shock tube... a 1-D shock tube (Sod's problem with p41 = 10:1 and rho4:rho1 = 8:1) produces overshoots in the density for a FIRST order scheme... Again, why?! First order should be monotonic.

Thanks in advance

PS don't get me started on enability to modify Cp via UDFs... It would be so nice. errrrgh

I'm just curious, what kind of boundary conditions did you used to get a 1D solution from a 2D (or even 3D) one?

Thanks

The grid is 401x2 and I'm running Fluent in 2D mode. The bc's in the second direction are "symmetry" which should force the v component to remain zero. For Sod's problem the left and right boundaries are "walls".

Good catch on the stationary-shock problem... I think I left the l/r bc's at "wall" as a hold-over from the Sod case (I ran the stationary shock problem after seeing the overshoot issues with the shock tube). However, the bc's should be "pressure-far-field" and "pressure-outflow" for the stationary-shock problem. I'll re-run and see what happens...

Update... An exact Riemann problem was run using 3rd-order MUSCL & standard limiter (what is a "standard" limiter anyway? MinMod?)... Here's the 3rd-order (x,T) data at the discontinuity. Notice the large overshoots. This data should be monotonic. This issue propagates to Sod's problem and most all problems with shocks.

x T (K)

-0.0175 350.675

-0.0125 341.157

-0.0075 369.293 <----

-0.0025 210.045 <---- yikes!

--- shock is here ----

0.0025 591.436 <----

0.0075 587.304

0.0125 587.35

0.0175 587.353