hi,

i need to set boundary condtions for inlet and oulet in the case of compressible computations: for instance for a transonic nozzle....

noramlly in this kind of problem stagnation conditions should be fixed up at inlet : total pressure and temperature : how to do so in phoenics, i was disapointed that such classical conditions are not built in ... for my case i solve static form of enthalpy (that is adviced in the documentation): from this stagnation condition , the mass flow rate should be computed in fly along iterations (and should become shocked in case of transonic condition in the nozzle)

for outlet condition, i dont want to set a constant pressure because if the flow is supersonic, pressure should be extrapolated. but if the flow is subsonic pressure should be actually imposed... how to deal with that: deos it require a great amount of coding... why not built in??

the different libray cases on this subject are not realistic because all is imposed (pressure, velocity, and so mass flow ) at the inlet and a compatible pressure is set at outlet to be compatible with the geometry and flow conditions...perfect design conditions...

if cham ssupport or other can help , thanks because these are basic problems in compressible CFD

thanks

If you are a maintained user it is best to send your q1 and ground files to your local support centre for investigation together with a precise statement of the problem you are trying to simulate together with required the boundary conditions. See also our response to your earlier posting re: NAMGRD=CONV.

yann,

I had struggled a lot with a problem of a highly underexpanded nozzle flow. The solution to this problem includes a complex shock structure, including a Mach disk. In my inviscid solution I had all kinds of pecularities, including back flow after the Mach disk. I tried every trick possible, including those suggested by CHAM, but was never satisfied. Only by adding substantial viscosity using k-e turbulence model (as suggested by Dr. Malin) I could get some sort of agreement (but very smeared) to the experimental results from the literature.

By that time (1991 or so) I was promised by CHAM that they will consider revision of the treatment of compressible flow, including a better solution to normal shocks. As far as I can judge from your postings, the same problems are still there. Hopefully I am wrong. If not - maybe it would be revised now...

Rami

Rami,

The case you are describing refers to the external jet flow downstream of the nozzle discharge plane, whereas I believe Yann is referring to the internal flow inside a convergent-divergent nozzle. The default treatment of the algebraic convection fluxes in the momentum equations does not predict the correct location of a normal shock in such a nozzle, but this can be remedied by using the alternative treatment coded in the file gxconvec.for.

Our experience was that both treatments produced fairly similar results for the Mach disc problem, which I think supports your statement. Fairly good agreement was obtained for both underexpanded and overexpanded jets, although significant smearing of the second Mach disc was noted for an underexpanded case. You are correct about very low velocities or even flow reversal behind the disc for inviscid calculations. It was felt that the flow is unlikely to be inviscid in this region because of the turbulent transport of higher momentum fluid into the region behind the disc.

Dr. Malin (I presume),

> The case you are describing refers to the external jet flow downstream of the nozzle discharge plane, ...

You are right (I am amazed how you remembered it after so many years!).

Regarding the second paragraph of your posting. It is disappointing to find out that this defficiency, known for such a long time, is still not rectified in a general purpose CFD package. I tried at that time to compare my results to another numerical solution of this problem, which solved the inviscid flow (Euler equations) successfully with very sharp shocks captured. Therefore I wished to do the same with PHOENICS, and avoid turbulence model issues and viscous effects altogether. I further assume it is desirable in a general purpose package to be able to solve inviscid flows with strong normal shocks (e.g., blunt body bow shocks, etc.).

Rami

the inlet and outlet boundary conditions for unsteady, compressible flow are causing problems for me. i am using Simple algorithm. at the inlet the total pressure and temperature are known, while at the outlet the static pressure is known. please help me to solve the problem. ajmal.