[alainislas]

Hello SU2 community,

I am modeling a rocket (fuselage only) at supersonic flow (Ma = 2.089) conditions.

- Models:

Turbulence: kw-SST
Viscosity: SUTHERLAND
Conductivity: CONSTANT_PRANDTL
EOS : IDEAL_GAS

- Fluid properties:

Ref. Viscosity = 1.716e-05 N.s/m^2
Sutherland Temp. = 273.15 K
Sutherland Const = 110.4 K

Prandtl (Lam.) = 0.72
Prandtl (Turb.) = 0.9

Gas Constant = 287.058 N.m/kg.K
Spec. Heat Ratio = 1.4

Initial and free-stream conditions:


Static Pressure = 84257.3 Pa
Density = 1.04824 kg/m^3
Temperature = 280.011 K
Total Energy = 453854 m^2/s^2
Velocity-X = 700.769 m/s
Velocity-Y = 0 m/s
Velocity-Z = 0 m/s
Velocity Magnitude = 700.769 m/s

Viscosity = 1.74976e-05 N.s/m^2
Conductivity = 85310.9 W/m^2.K
Turb. Kin. Energy = 7366.16 m^2/s^2
Spec. Dissipation = 4.41291e+07 1/s

Mach Number = 2.089
Reynolds Number = 1.8262e+06



My geometry consists of a 3D structured grid generated via ICEM (.cgns formati) with minimum orthogonality 21.81.

+--------------------------------------------------------------+
| Mesh Quality Metric| Minimum| Maximum|
+--------------------------------------------------------------+
| Orthogonality Angle (deg.)| 21.1814| 89.9999|
| CV Face Area Aspect Ratio| 1.05976| 1.35169e+06|
| CV Sub-Volume Ratio| 1.00003| 2930.34|
+--------------------------------------------------------------+


I only have FAR_FIELD, SYMMETRY and WALL markers. The walls are assumed as adiabatic, and my turbulent properties are: FREESTREAM_TURBULENCEINTENSITY= 0.10 & FREESTREAM_TURB2LAMVISCRATIO= 10.0

I am solving the flow with 1st order upwind schemes, linear solver: BCGSTAB with ILU preconditioner. My CFL is 0.2. I Use 3 multigrid V cycles.


Can anyone help me to figure out what´s going wrong with the setup? I get the following error:

void CSolver::SetResidual_RMS(const CGeometry*, const CConfig*). My residual for omega is always positive during the calculation. So I guess is related to the turbulence initialization...

[pcg]

Try starting at a lower mach number.
It's a bit challenging to start a turbulent simulation Mach 2 from scratch.

[alainislas]

Dear Pedro,

Thank you for your reply. I started a new simulation with Ma = 1. I could run the case for more iterations, this time the Omega Residual reached negative values, but I had the same error.

I checked the flow in paraview, and I found some weird cells with excessive temperature (clipped by >2000 K). Do you think the divergence is attributed to this? Are there any temperature limiters in SU2?

[pcg]

Try turning off the multigrid and using fgmres instead of bcgstab.
Our multigrid does not work very well with the type of mesh you have (high aspect ratio in the Fairfield).