[heksel8i]

Hey!

I am learning the characteristics of supersonic flows in de Laval nozzles and here I have a tricky question:

I have a strongly underexpanded flow in a relatively long nozzle. My simulations are done by Fluent implicit density based solver with 2nd order upwind scheme for all variables and with AUSM scheme. For laminar and turbulent case I get otherwise really nice looking flow field but there are two oblique shock waves starting near the nozzle throat. I have references about similar cases and there only one shock wave has been detected (similar shape as I have). I have been trying to get rid of the second shock by refining the mesh, changing Courant number and discretization scheme (MUSCL) but without success.

Temperature gradient magnitude plot :

Mach contours:


Mesh quality should be at least adequate:
-200000 cells inside the nozzle, 78000 in a farfield
-max equiangle skewness 0.1 (250k cells below 0.05 and 100k cells below 0.02)
-max aspect ratio 200
-max length ratio 1.12
-y+<1

Thougths:

i) Implicit solver fails to catch a single shock wave? Has worked in some other references generally for shocks though.
ii) second shock wave indeed would exist in reality. After the first shock wave the flow feels again a Prandtl - Meyer expansion due to the geometry of a nozzle creating a second shock wave. No references about this though...
iii) Mesh / nozzle wall geometry problem.

Ideas? Questions? Free chat...

[heksel8i]

To answer for myself:

The reason was a distorted geometry of the nozzle wall. Less than a millimeter "bump" created a local expansion-compression effect which propagated as a weak shock wave through the whole nozzle. I was stunned how sensitive the flow is for nozzle geometry.

[newid]

Great job! Heksel8i
I have stuck with a question which about atomizer with gas-liquid two phases flow ,
Which model in FLUENT can simulate the spray angle?even the hollow-cone created by pressure-swirl atomizer?
Please help me,Thanks!