[descot]

Hello, I have been trying to perform high-speed simulations using grid_movement for some time. However, after passing a certain speed (in my case 500 m/s), the calculation crashes when I try with a ramp or does not start with the following error:

Error in "void CSolver::SetResidual_RMS(const CGeometry*, const CConfig*)":
-------------------------------------------------------------------------
SU2 has diverged (NaN detected).
------------------------------ Error Exit -------------------------------


Here is a simple example with a cylinder:


% ------------- DIRECT, ADJOINT, AND LINEARIZED PROBLEM DEFINITION ------------%
%
SOLVER= RANS %NAVIER_STOKES %EULER
KIND_TURB_MODEL= SA %NONE
MATH_PROBLEM= DIRECT
REF_DIMENSIONALIZATION= DIMENSIONAL


% ----------- COMPRESSIBLE AND INCOMPRESSIBLE FREE-STREAM DEFINITION ----------%
%
MACH_NUMBER= 0.0
AOA= 0.0
SIDESLIP_ANGLE= 0.0
INIT_OPTION= TD_CONDITIONS
FREESTREAM_TEMPERATURE= 250.0
FREESTREAM_PRESSURE= 280.0


% ------------------------- TIME-DEPENDENT SIMULATION -------------------------------%

RESTART_SOL= NO



TIME_DOMAIN= YES
TIME_MARCHING=DUAL_TIME_STEPPING-2ND_ORDER %DUAL_TIME_STEPPING-1ST_ORDER %DUAL_TIME_STEPPING-2ND_ORDER %DUAL_TIME_STEPPING-1ST_ORDER
TIME_STEP= 0.001
MAX_TIME= 50.0
INNER_ITER= 10
RESTART_ITER= 0
TIME_ITER= 500



GRID_MOVEMENT= ROTATING_FRAME
TRANSLATION_RATE = -633.941 0.0 0.0
MACH_MOTION= 2

%MARKER_SHROUD= (wall)
%RAMP_ROTATING_FRAME= YES
%RAMP_ROTATING_FRAME_COEFF= (-100.0, 1.0, 1000)

% ---------------------- REFERENCE VALUE DEFINITION ---------------------------%
%
REF_ORIGIN_MOMENT_X = 0.00
REF_ORIGIN_MOMENT_Y = 0.00
REF_ORIGIN_MOMENT_Z = 0.00
REF_LENGTH= 10.0
REF_AREA= 20.0

% -------------------- BOUNDARY CONDITION DEFINITION --------------------------%
%
%MARKER_EULER= ( wall )
MARKER_ISOTHERMAL= (cylinder,250)
MARKER_WALL_FUNCTIONS= ( cylinder, STANDARD_WALL_FUNCTION )


MARKER_FAR= ( farfield )
MARKER_PLOTTING= ( cylinder )
MARKER_MONITORING= ( cylinder )

% ------------- COMMON PARAMETERS DEFINING THE NUMERICAL METHOD ---------------%
%
NUM_METHOD_GRAD= WEIGHTED_LEAST_SQUARES %GREEN_GAUSS
CFL_NUMBER= 10
CFL_ADAPT= NO
CFL_ADAPT_PARAM= ( 1.5, 0.5, 1.0, 100.0 )
RK_ALPHA_COEFF= ( 0.66667, 0.66667, 1.000000 )

LINEAR_SOLVER= FGMRES %BCGSTAB %FGMRES
%LINEAR_SOLVER_SMOOTHER_RELAXATION= 10
LINEAR_SOLVER_PREC= LU_SGS %(ILU, LU_SGS, JACOBI)
%LINEAR_SOLVER_ITER= 5


% ----------------------- SLOPE LIMITER DEFINITION ----------------------------%
%
%VENKAT_LIMITER_COEFF= 0.1
%REF_SHARP_EDGES= 3.0
%SENS_REMOVE_SHARP= NO

% -------------------------- MULTIGRID PARAMETERS -----------------------------%
%
MGLEVEL= 0
MGCYCLE= V_CYCLE
MG_PRE_SMOOTH= ( 1, 2, 3, 3 )
MG_POST_SMOOTH= ( 2, 2, 2, 2)
MG_CORRECTION_SMOOTH= ( 0, 0, 0, 0 )
MG_DAMP_RESTRICTION= 0.8
MG_DAMP_PROLONGATION= 0.8

% -------------------- FLOW NUMERICAL METHOD DEFINITION -----------------------%
%
CONV_NUM_METHOD_FLOW= JST
%ENTROPY_FIX_COEFF= 1.0
MUSCL_FLOW= NO
SLOPE_LIMITER_FLOW= BARTH_JESPERSEN %BARTH_JESPERSEN
JST_SENSOR_COEFF= ( 0.5, 0.02 )
TIME_DISCRE_FLOW= EULER_IMPLICIT

% -------------------- TURBULENT NUMERICAL METHOD DEFINITION ------------------%
%
CONV_NUM_METHOD_TURB= SCALAR_UPWIND
MUSCL_TURB= NO
SLOPE_LIMITER_TURB= BARTH_JESPERSEN
TIME_DISCRE_TURB= EULER_IMPLICIT

% --------------------------- CONVERGENCE PARAMETERS --------------------------%
%
CONV_RESIDUAL_MINVAL= -15
CONV_STARTITER= 10
CONV_CAUCHY_ELEMS= 100
CONV_CAUCHY_EPS= 1E-6

% ------------------------- INPUT/OUTPUT INFORMATION --------------------------%
%
MESH_FILENAME= mesh_cylinder_lam.su2 %mesh
MESH_FORMAT= SU2
MESH_OUT_FILENAME= mesh_out.su2
SOLUTION_FILENAME= restart_flow.dat
READ_BINARY_RESTART= YES
TABULAR_FORMAT= CSV
CONV_FILENAME= history
RESTART_FILENAME= restart_flow.dat
VOLUME_FILENAME= flow


SURFACE_FILENAME= surface_flow
OUTPUT_WRT_FREQ= 5
SCREEN_OUTPUT= (ITER, RMS_RES, LIFT, DRAG)
OUTPUT_FILES= (PARAVIEW)
BREAKDOWN_FILENAME= forces_breakdown.dat
WRT_FORCES_BREAKDOWN = YES

mesh -> Tutorials/compressible_flow/Laminar_Cylinder
/mesh_cylinder_lam.su2

I used the cylinder mesh provided in the SU2 tutorials for my example. However, the problem does not seem to be related to the mesh since I tried with a BL well-resolved mesh for the problematic speed.

Thanks for your help!

[bigfootedrockmidget]

The numerical stability of solvers often depends on the velocity, so it is not guaranteed that your setup works when you change the velocity.You could try to start a simulation with a lower velocity that converges and then restart from that solution.
How do you initialize the flow? A better initial condition might help in getting the solution started properly.



You also have to check what the actual Reynolds number and the Mach number is and choose the appropriate solver. Probably RANS is OK, but just check to be sure. If Ma<0.3, you might be better off using the incompressible solver.