[sangrampp]

Hi there,
thought that my problem was pretty straight forward so may find many posts related to it but after much searching i am posting it here for assistance.
I have a CD nozzle (De-Laval) exhausting to open atmosphere - one similar to NASA rocket testing. So i set pressure inlet as boundary condition to inlet of nozzle and modeled a large rectangular space at exit to the nozzle for atmosphere (it is fairly large) and i setup pressure far-field on the rectangle sides with Mach = 0 since the atmosphere will be standstill. I used k-omega turbulence model and set inlet and far-field at 1% turbulent viscosity with respective hydraulic diameter.
When i run the simulation, the results are spurious with no flow or erratic flow.sometimes i even get un-handled exception in fluent which i reckoned would be due to mesh so i refined it and ran the simulation again. but to no avail. am i doing something wrong here. can anyone please guide me with the setup for getting accurate results.

[sangrampp]

guys i really need some help...i have been banging my head for 4 weeks to no avail until i decided to post here.
still searching for a life saver... any input would be highly appreciated. please give me some leads.

[jonny_b]

sangram,

I too am having a huge issue with getting FLUENT to work well for CD nozzle exhausting to ambient. I have been seeing the same results you have with spurious contours and floating point exceptions and have yet to figure out the problem.

I think my biggest concern is that I cannot create the mesh that I want using their unstructured meshing inside of FLUENT.

Did you happen to resolve this issue yet? I will let you know if I find some resolution on my end.

[sangrampp]

Originally Posted by jonny_b

sangram,

i too am having a huge issue with getting fluent to work well for cd nozzle exhausting to ambient. I have been seeing the same results you have with spurious contours and floating point exceptions and have yet to figure out the problem.

I think my biggest concern is that i cannot create the mesh that i want using their unstructured meshing inside of fluent.

Did you happen to resolve this issue yet? I will let you know if i find some resolution on my end.

dear johny,
i have been able to resolve the problem. Seems that the problem was two-fold.
First is ambiguous problem statement - you might want to use mass-flow inlet to specify the pressure inlet - for this use the standard compressible flow equation to calculate the choked flow mass flow rate for given total pressure.
Second when you are having a nozzle discharging to atmosphere, more than probable that the flow will be under or overexpanded since the outlet pressure of 1 atmosphere is hard to be maintained by the nozzle since you may have a typical pressure inlet to the nozzle. So there will be a shock somewhere around the flow exit (along the divergent part of the nozzle) so you should use an extremely fine mesh around this region to capture the shock.
When you take this care, you will get accurate solution.
Hope this helps.
If you have any other query pm me.

[jonny_b]

Thanks for the tips sangram,

I figured I need better resolution. It's crazy b/c I come from the structured world using a NASA CFD code. Create meshes with the resolution I currently have usually works fine in that code, but when I apply the same methodologies to the unstructured case it's not as robust.

Also, what meshing tool did you use? I am using the Workbench mesher in ANSYS 13 and I currently to create a quad dominant mesh. Im findting it's quite tough to get it to behave like I want.

[sangrampp]

Dear jonny
i used icem cfd. I have tried ansys 13 mesher with cutcell meshing but more than half the time the process aborts and says that cutcell meshing failed.
I suggest u try various combinations of minimum and maximum size this did the trick for me.
Also you must use program controlled inflation so that the solution is more accurate.

[jonny_b]

Another somewhat related question. In terms of a best practice when creating geometry models of CD nozzles, do you find that it's best to keep the entire geometry as one body or do you separate your domain into multipart bodies?

[sangrampp]

The answer to that question is - somewhat depends on the situation. I enlist them here:
1. If you are modelling a large space to exhaust the cd nozzle into you will need the mesh to be very fine near the throat and divergent portion but relatively large in the ambient space. So from that perspective, to be able to manage the mesh easily, you may want to divide the domain up in parts - depending upon your mesh quality.
If you are using a multi phase (or species) model for your your domain, you may want to divide the domain based on the phase present (or the species) present so that you may want to form a transition zone for mixing or other phenomenon between these zones.
So my advice is if you find it better to divide the domain into zones you may do so.
Either way it wont matter to the final solution accuracy. I have no inputs regarding the issue of computation speed wrt the division of zones.

[jonny_b]

Thanks for your feedback Sangram, you have been more helpful than ANSYS's technical support. These guys are really frustrating me b/c I cannot get the answers I need and they take a long time to get back to me.

[sangrampp]

Dear jonny,
you are very welcome. I have come to believe that people are always there to help, all one needs to do is just ask.
In future if you have any query just pm me i will be more than glad to help u - if i can that is.

[jonny_b]

Thank you very much and if I learn any tips along the way I'll be sure to pass them along to you.