Learning Forum

[SV]

Dear Sir/Madam

I am trying to predict the location of shock wave inside a converging- diverging nozzle. My program can see the supersonic flow at the outlet. I expect to see the variation of location of shock wave as I increase the outlet pressure, but as I increase the outlet pressure the program start diverging.

I would like to see the variation of shock wave through nozzle. Please let me know if you can provide help.

Thank you

Saeid Vafaei 

[Rob]

Please post some images so we can see what's going on.

[SV]

Would you like to send you the files that you can run the code and see the issue

[SV]

please look at the attached files to see residuals when outlet pressure is 3000 pa and 50000 pa. please let me know how i can send you the code that you can run by yourself. Thank you

[Rob]

/forum/discussion/23093/why-ansys-employees-dont-download-attachments

[SV]

Please let me know what do you want me to do and how you want to help

[SV]

[SV]

please look at the pictures to see residuals when outlet pressure is 3000 pa and 50000 pa. you can see the code converge when the outlet pressure is 3000 pa and diverge when the outlet pressure is 50000pa

[Rob]

What's the inlet boundary condition?

[SV]

Please look at the following images to see inlet and outlet boundary conditions

[Rob]

What is the flow field doing just before the model fails?

[SV]

Hello Rob
I do not know what you mean? this model can predict flow if flow is supersonic at outlet. When outlet pressure is 3000 pa, we have supersonic flow at outlet and model works very well. As you increase the outlet pressure the shock wave starts moving toward throat. This model can see the shock wave close to outlet as well but it cannot see further back toward throat.
Thank you

[Rob]

If you plot & share what's going on in the model we may be able to help. The boundary conditions don't show anything other than it's likely the flow is going in the right direction, so more info needed.

[SV]

Please look at following figures for outlet pressure =39000 pa.

[Rob]

And the flow field? What's the Reynolds Number?

[SV]

What you mean by flow field? Re changes because density changes, we are in turbulent flow . Can you please let me know how to plot Re as a function of x

[Rob]

We don't plot Re, you need to calculate it before you start. Which turbulence model are you using?

[SV]

[SV]

i use inviscid model

[SV]

can Ansys predict the location of shock wave at all?

[Rob]

Yes it can. Why inviscid if you're modelling shocks in a confined system?

[SV]

i tried k-epsilon method and we have same issue. iviscid is slightly better, because we do not care what's happening next to the wall. let me know if you want me to send you results for k-epsilon method.

[SV]

please look at following figures to see results for k-epsilon method

[Rob]

Please can you post some contour & vector plots? Ie enough information to actually figure out what's going on? An image of the mesh will help too.

[SV]

Please look at following figures

[SV]

[Rob]

Extend the domain by at least 5-6 diameters in each direction. Your shocks are far too close to the boundary and using "no reverse" flow is causing you problems. Post process in Fluent too, you may need some of the data that's not moved to CFD Post by default.
If you look under courses at the top of the page you'll see some that are very relevant to this problem.

[SV]

Hello
I removed the "no reverse" flow. Still I have problem and data doesn't move to to CFD Post by default. Can you please send me the link you mentioned. I would rather to keep the geometry as it is.

[Rob]

Scroll up, just to the right of "Ansys Learning Forum" look at Courses.
Depending on where the shock is, you're forcing the solution with the boundaries being too close: it's not a preference to retain the current domain but a need to extend it. Or revise the boundary condition to be exactly what the solver would calculate at that location.

[SV]

Please let me know if you have suggestions for boundary conditions. I am keep changing the boundary conditions to see if i can fix the issue. Changing boundary conditions is a better idea

[SV]

Hello
I kept the ratio almost same and reduced the inlet and outlet pressures. Still I have same issues.

[aitor.amatriain]

As suggests, you should increase the size of the domain. It is important to include the whole nozzle, plus a zone such that the pressure at the outlet is of the order of ambient pressure. It is also a good practice to include a boundary layer on the walls of the nozzle, in order to predict accurately the separation point that leads to the generation of shock waves in the diverging part of the nozzle (as you may know, this happens in some interval of the ratio between stagnation pressure and ambient pressure).

[SV]

Hello
i increased the domain five times as it was suggested. I cannot see the shock wave correctly now. I will send you some figures, let's start to check step by step and i will follow your suggestions

[SV]

[SV]

please look at variation of mach number carefully.

[aitor.amatriain]

You have modified the geometry of the nozzle, and that is supposed to be fixed. My suggestion was to extend the geometry outside nozzle outlet.
Your problem will not be in hypersonic regime, so I would choose the pressure-based solver with the default settings (k-omega SST, second order discretization)

[SV]

Please look at data for k-omega

[SV]

Please look at data for k-Ep

[SV]

Hello
Please look at data for pressured based, k-omega
Can you please explain when we can use density based and when pressure based

[aitor.amatriain]

You have done practically the same thing, just increase the length of the nozzle. What I suggest is the following:
You should always extend the domain until the solution does not change with the size of the domain. The solvers do not like high gradients at the outlets.

[Rob]

Thanks

[aitor.amatriain]

You are welcome!
EDIT: I did not see the question "Can you please explain when we can use density based and when pressure based".
Fluent's User Guide give some information about this topic:
32.1. Overview of Using the Solver (ansys.com)

[SV]

Thank you. Seems to it's working now, if i use pressure based.
Please look at following figure to see the wall and outlet

[aitor.amatriain]

Good!

[SV]

Hello
I am looking for couple of clips to explain 3D, single flow (like t-fitting, elbow, etc) simple geometries for educational purposes. i would rather to have clips to start explaining from making geometry to meshing, solving and post processing. Please let me know if you know if you know where i can find.

Thank you

[Rob]

Other than the AIC and YouTube content plus Help there isn't anything as we mentioned here /forum/discussion/30621/effective-video-clips-to-learn-about-new-spaceclaim-geometry#latest For simple examples you may be better with the Cornell materials?

[SV]

Hello
I have checked with Cornell University, They are 2D. Thank you

[SV]

I have a T-pipe with two inlets and one outlet. It diverges. Would you like to help with that?
Thank you

[aitor.amatriain]

Please create a new post and give details of the simulation (Models, solver settings, screenshots of the mesh and contours...)

[SV]

Hello
Thank you for helping on converging-diverging nozzle. Please look at figure to see my solution methods for converging-diverging nozzle. May i ask you where i can find information about Scheme (coupled), Gradient (least squares cell based), Pressure (second order) Density (Second order Upwind) etc.

[aitor.amatriain]

Practical aspects are covered in the User's Guide Chapter 32: Using the Solver (ansys.com)
while additional aspects related to the theory are given by the Theory Guide:
Chapter 24: Solver Theory (ansys.com)