[Vishnu Sankar]

Hi friends,

I am trying to do a simulation of a perfectly expanded nozzle (Euler flow) with only the divergent section and I have extended the mesh to capture the flow after it leaves the nozzle exit.

Inlet conditions
-> Mach number = 1 and inlet pressure and temperature are mentioned in the .cfg

Outlet is far from the nozzle exit and has an outlet pressure which is the same as the nozzle exit pressure for ideal optimum expansion.

-> Numerical scheme = JST(I tried Lax-Friedrich, ROE and HLLC schemes but the residual didn't converge even after using the highest MG level).

-> Structured mesh with good enough mesh metrics

-> linear solver is used along with the dual time stepping 2nd order for unsteady simulation(steady simulation didn't work out too).
-> Time step = 5e-3
-> A total of 200 iterations were performed with each iteration consisting of 50 inner loops.
But the residual oscillates and doesn't converge. The simulation results were bizarre and don't make any sense. I have attached the .cfg file, mesh file(meshed using gmsh), the resultant mesh contour, and the boundary conditions. Can someone please explain what I am missing in this simulation?

Note - After reading some posts in this forum, I have turned off the CFL adaption and manually tweaked the CFL value to 0.1, 1, and 10. I also tried with the highest MG level. But the results were still bizarre . I tried the simulation without extending the mesh beyond the nozzle exit and this produced accurate results with converged residual.

[pcg]

Hi,
MG either helps or it really doesn't, i.e. more levels rarely improve a diverging solver (in SU2).
I never did that type of simulation but I would use an inlet boundary condition for the nozzle and the farfield for the outlet, since the outlet BC does not have any mitigation for recirculating flows.
I would change this: INIT_OPTION= REYNOLDS to TD_CONDITIONS.
Regarding the choice of convective scheme options and linear solvers I recommend you take a look at these pages:
https://su2code.github.io/docs_v7/Convective-Schemes/
https://su2code.github.io/docs_v7/Li...econditioners/
With JST you should be able to use much higher CFL but you will need to use the ILU preconditioner.
None of the MUSCL-type options matter for JST, and when using an upwind scheme (like Roe) the Venkatakrishnan limiter should only be used for non-dimensional simulations.

[Vishnu Sankar]

Hi Pedro,
Thank you for your prompt reply. I changed the INIT_OPTION from REYNOLDS to TD_CONDITIONS. After reading the information from the URLs that you had sent, I tried various methods and one of them is working, the good thing is that the residuals are converging but very slowly. Though the results are still unsatisfactory, it has improved a lot. These are the changes I have made in the code:
->Convective scheme - ROE
->MG level - 3
->No Venakatakrishnan limiter (as it is a dimensional simulation)
->MUSCL - ON
->preconditioner - LU_SGS
->CFL - 10
->CFL adaptation is turned OFF
I have attached the link to the image of Mesh contour. The Mach number inside the nozzle matches pretty well with the theoretical values. What I don't understand is that the Mach number near the top and left outlets have the same values as that of the nozzle exit Mach number. But the expected mach numbers at the top and left outlets are close to zero.
[IMG]
https://drive.google.com/file/d/1lfs3ztCqn-PvkoE-0g0N2OJ7lvV33CHs/view?usp=sharing[/IMG]
Many thanks,
Vishnu

[atelcikti1]

Outlet with 3 atm does not seem logical, try to open 1 atm or known realistic altitude.