
[Sponsors] 
May 11, 2020, 13:05 
Unsteady Restart Divergence

#1 
New Member
pro
Join Date: Mar 2020
Posts: 25
Rep Power: 5 
Im trying to initialize my unsteady aeroelastic simulation with a steady solution
1. The steady solution uses the Euler solver for an airfoil at 1 degrees. 2. Dual time step method is used, therefore 2 'solution_flow' steady solutions were used. 3. EVERY single time i try to restart the solution always diverges 4. References here, https://arc.aiaa.org/doi/10.2514/1.J058663 and https://www.eccomas2016.org/proceedings/pdf/6597.pdf manage to 'initialize the unsteady solution from a steady state solution'. So im wondering what am i doing wrong? This is the steady cfg Code:
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % STEADY STATE SOLUTION % % SU2 configuration file % % Case description: inviscid NACA64A010 flutter % % Author: Santiago Padron % % Institution: Stanford University % % Date: 070915 % % File Version 7.0.3 "Blackbird" % % % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %  DIRECT, ADJOINT, AND LINEARIZED PROBLEM DEFINITION % % % Physical governing equations (EULER, NAVIER_STOKES, % WAVE_EQUATION, HEAT_EQUATION, FEM_ELASTICITY, % POISSON_EQUATION) SOLVER= EULER % % Mathematical problem (DIRECT, CONTINUOUS_ADJOINT) MATH_PROBLEM= DIRECT % % Read binary restart files (YES, NO) READ_BINARY_RESTART= YES % % %  COMPRESSIBLE FREESTREAM DEFINITION % % % Mach number (nondimensional, based on the freestream values) MACH_NUMBER= 0.87 % % Angle of attack (degrees, only for compressible flows) AOA= 1.0 % % Freestream pressure (101325.0 N/m^2, 2116.216 psf by default) FREESTREAM_PRESSURE= 101325.0 % % Freestream temperature (288.15 K, 518.67 R by default) FREESTREAM_TEMPERATURE= 288.15 %  REFERENCE VALUE DEFINITION % % % Reference origin for moment computation (m or in) REF_ORIGIN_MOMENT_X = 0.5 REF_ORIGIN_MOMENT_Y = 0.00 REF_ORIGIN_MOMENT_Z = 0.00 % % Reference length for pitching, rolling, and yawing nondimensional % moment (m or in) REF_LENGTH= 1.0 % % Reference area for force coefficients (0 implies automatic % calculation) (m^2 or in^2) REF_AREA= 1.0 % % Flow nondimensionalization (DIMENSIONAL, FREESTREAM_PRESS_EQ_ONE, % FREESTREAM_VEL_EQ_MACH, FREESTREAM_VEL_EQ_ONE) REF_DIMENSIONALIZATION= DIMENSIONAL %  BOUNDARY CONDITION DEFINITION % % % Euler wall boundary marker(s) (NONE = no marker) MARKER_EULER= ( airfoil ) % % Farfield boundary marker(s) (NONE = no marker) MARKER_FAR= ( farfield ) %  SURFACES IDENTIFICATION % % % Marker(s) of the surface in the surface flow solution file MARKER_PLOTTING = ( airfoil ) % % Marker(s) of the surface where the nondimensional coefficients are evaluated. MARKER_MONITORING = ( airfoil ) %  COMMON PARAMETERS DEFINING THE NUMERICAL METHOD % % % Numerical method for spatial gradients (GREEN_GAUSS, WEIGHTED_LEAST_SQUARES) NUM_METHOD_GRAD= GREEN_GAUSS % % CFL number (stating value for the adaptive CFL number) CFL_NUMBER= 25.0 % Adaptive CFL number (NO, YES) CFL_ADAPT= YES % % Parameters of the adaptive CFL number (factor down, factor up, CFL min value, % CFL max value ) CFL_ADAPT_PARAM= ( 0.1, 2.0, 1.0, 1e10 ) % %  LINEAR SOLVER DEFINITION % % % Linear solver or smoother for implicit formulations (BCGSTAB, FGMRES, SMOOTHER) LINEAR_SOLVER= FGMRES % % Preconditioner of the Krylov linear solver (ILU, LU_SGS, LINELET, JACOBI) LINEAR_SOLVER_PREC= ILU % % Minimum error of the linear solver for implicit formulations LINEAR_SOLVER_ERROR= 1E4 % % Max number of iterations of the linear solver for the implicit formulation LINEAR_SOLVER_ITER= 2 %  MULTIGRID PARAMETERS % % % Multigrid levels (0 = no multigrid) MGLEVEL= 0 % % Multigrid cycle (V_CYCLE, W_CYCLE, FULLMG_CYCLE) MGCYCLE= W_CYCLE % % Multigrid presmoothing level MG_PRE_SMOOTH= ( 1, 2, 3, 3 ) % % Multigrid postsmoothing level MG_POST_SMOOTH= ( 0, 0, 0, 0 ) % % Jacobi implicit smoothing of the correction MG_CORRECTION_SMOOTH= ( 0, 0, 0, 0 ) % % Damping factor for the residual restriction MG_DAMP_RESTRICTION= 0.75 % % Damping factor for the correction prolongation MG_DAMP_PROLONGATION= 0.75 %  FLOW NUMERICAL METHOD DEFINITION % % % Convective numerical method (JST, LAXFRIEDRICH, CUSP, ROE, AUSM, HLLC, % TURKEL_PREC, MSW) CONV_NUM_METHOD_FLOW= JST % % Spatial numerical order integration (1ST_ORDER, 2ND_ORDER, 2ND_ORDER_LIMITER) MUSCL_FLOW= YES % % Slope limiter (VENKATAKRISHNAN, BARTH_JESPERSEN) SLOPE_LIMITER_FLOW= VENKATAKRISHNAN % % Entropy fix coefficient (0.0 implies no entropy fixing, 1.0 implies scalar % artificial dissipation) ENTROPY_FIX_COEFF= 0.001 % % 2nd and 4th order artificial dissipation coefficients JST_SENSOR_COEFF= ( 0.5, 0.02 ) % %  CONVERGENCE PARAMETERS % % % % Convergence criteria (CAUCHY, RESIDUAL) % CONV_CRITERIA= RESIDUAL % % % Min value of the residual (log10 of the residual) CONV_RESIDUAL_MINVAL= 8 % % Start convergence criteria at iteration number CONV_STARTITER= 0 % % Number of elements to apply the criteria CONV_CAUCHY_ELEMS= 100 % % Epsilon to control the series convergence CONV_CAUCHY_EPS= 1E10 % %  INPUT/OUTPUT INFORMATION % % % Mesh input file MESH_FILENAME= 0_mesh.su2 % % Mesh input file format (SU2, CGNS) MESH_FORMAT= SU2 % % Restart flow input file SOLUTION_FILENAME= solution_flow.dat % % Output file format (TECPLOT, TECPLOT_BINARY, PARAVIEW, % FIELDVIEW, FIELDVIEW_BINARY) TABULAR_FORMAT= CSV % % Output file convergence history (w/o extension) CONV_FILENAME= history % % Output file restart flow RESTART_FILENAME= solution_flow.dat % % Output file flow (w/o extension) variables VOLUME_FILENAME= flow % % Output file surface flow coefficient (w/o extension) SURFACE_FILENAME= surface_flow % % Writing solution file frequency WRT_SOL_FREQ= 1000 % % Writing solution file frequency for physical time steps (dual time) WRT_SOL_FREQ_DUALTIME= 1000 % % Writing convergence history frequency WRT_CON_FREQ= 1 % % Writing convergence history frequency (dual time, only written to screen) WRT_CON_FREQ_DUALTIME= 1 % % Screen output SCREEN_OUTPUT= (TIME_ITER, INNER_ITER, RMS_DENSITY, RMS_ENERGY, LIFT, DRAG_ON_SURFACE, PLUNGE, PITCH) % % VOLUME_OUTPUT= PRIMITIVE, GRID_VELOCITY % HISTORY_OUTPUT= (ITER, TIME_DOMAIN, RMS_RES, AERO_COEFF, AEROELASTIC) % OUTPUT_FILES= (RESTART, CSV, PARAVIEW, SURFACE_PARAVIEW) Code:
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % % SU2 configuration file % % Case description: inviscid NACA64A010 flutter % % Author: Santiago Padron % % Institution: Stanford University % % Date: 070915 % % File Version 7.0.3 "Blackbird" % % % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %  DIRECT, ADJOINT, AND LINEARIZED PROBLEM DEFINITION % % % Physical governing equations (EULER, NAVIER_STOKES, % WAVE_EQUATION, HEAT_EQUATION, FEM_ELASTICITY, % POISSON_EQUATION) SOLVER= EULER % % Mathematical problem (DIRECT, CONTINUOUS_ADJOINT) MATH_PROBLEM= DIRECT % % Read binary restart files (YES, NO) READ_BINARY_RESTART= YES % RESTART_SOL= YES % RESTART_ITER= 2 % RESTART_STEADY_STATE= YES %  COMPRESSIBLE FREESTREAM DEFINITION % % % Mach number (nondimensional, based on the freestream values) MACH_NUMBER= 0.87 % % Angle of attack (degrees, only for compressible flows) AOA= 1.0 % % Freestream pressure (101325.0 N/m^2, 2116.216 psf by default) FREESTREAM_PRESSURE= 101325.0 % % Freestream temperature (288.15 K, 518.67 R by default) FREESTREAM_TEMPERATURE= 288.15 %  REFERENCE VALUE DEFINITION % % % Reference origin for moment computation (m or in) REF_ORIGIN_MOMENT_X = 0.5 REF_ORIGIN_MOMENT_Y = 0.00 REF_ORIGIN_MOMENT_Z = 0.00 % % Reference length for pitching, rolling, and yawing nondimensional % moment (m or in) REF_LENGTH= 1.0 % % Reference area for force coefficients (0 implies automatic % calculation) (m^2 or in^2) REF_AREA= 1.0 % % Flow nondimensionalization (DIMENSIONAL, FREESTREAM_PRESS_EQ_ONE, % FREESTREAM_VEL_EQ_MACH, FREESTREAM_VEL_EQ_ONE) REF_DIMENSIONALIZATION= DIMENSIONAL %  UNSTEADY SIMULATION % % % Time domain TIME_DOMAIN=YES % % Unsteady simulation (NO, TIME_STEPPING, DUAL_TIME_STEPPING1ST_ORDER, % DUAL_TIME_STEPPING2ND_ORDER, SPECTRAL_METHOD) TIME_MARCHING= DUAL_TIME_STEPPING2ND_ORDER % % Time Step for dual time stepping simulations (s) TIME_STEP= 0.0006283185307 % 36 steps per period, based on the pitch natural frequency % % Total Physical Time for dual time stepping simulations (s) MAX_TIME= 0.1884955592 % % Number of internal iterations (dual time method) INNER_ITER= 30 %  DYNAMIC MESH DEFINITION % SURFACE_MOVEMENT= AEROELASTIC % % Motion mach number (nondimensional). Used for initializing a viscous flow % with the Reynolds number and for computing force coeffs. with dynamic meshes. MACH_MOTION= 0.87 % % Moving wall boundary marker(s) (NONE = no marker, ignored for RIGID_MOTION) MARKER_MOVING= ( airfoil ) %  AEROELASTIC SIMULATION (Typical Section Model) % % Activated by GRID_MOVEMENT_KIND option % % The flutter speed index (modifies the freestream condition in the solver) FLUTTER_SPEED_INDEX = 0.65 % % Natural frequency of the spring in the plunging direction (rad/s) PLUNGE_NATURAL_FREQUENCY = 100 % % Natural frequency of the spring in the pitching direction (rad/s) PITCH_NATURAL_FREQUENCY = 100 % % The airfoil mass ratio AIRFOIL_MASS_RATIO = 60 % % Distance in semichords by which the center of gravity lies behind % the elastic axis CG_LOCATION = 1.8 % % The radius of gyration squared (expressed in semichords) % of the typical section about the elastic axis RADIUS_GYRATION_SQUARED = 3.48 % % Solve the aeroelastic equations every given number of internal iterations AEROELASTIC_ITER = 3 %  BOUNDARY CONDITION DEFINITION % % % Euler wall boundary marker(s) (NONE = no marker) MARKER_EULER= ( airfoil ) % % Farfield boundary marker(s) (NONE = no marker) MARKER_FAR= ( farfield ) %  SURFACES IDENTIFICATION % % % Marker(s) of the surface in the surface flow solution file MARKER_PLOTTING = ( airfoil ) % % Marker(s) of the surface where the nondimensional coefficients are evaluated. MARKER_MONITORING = ( airfoil ) %  COMMON PARAMETERS DEFINING THE NUMERICAL METHOD % % % Numerical method for spatial gradients (GREEN_GAUSS, WEIGHTED_LEAST_SQUARES) NUM_METHOD_GRAD= GREEN_GAUSS % % CFL number (stating value for the adaptive CFL number) CFL_NUMBER= 25.0 % Adaptive CFL number (NO, YES) CFL_ADAPT= YES % % Parameters of the adaptive CFL number (factor down, factor up, CFL min value, % CFL max value ) CFL_ADAPT_PARAM= ( 0.1, 2.0, 1.0, 1e10 ) % %  LINEAR SOLVER DEFINITION % % % Linear solver or smoother for implicit formulations (BCGSTAB, FGMRES, SMOOTHER) LINEAR_SOLVER= FGMRES % % Preconditioner of the Krylov linear solver (ILU, LU_SGS, LINELET, JACOBI) LINEAR_SOLVER_PREC= ILU % % Minimum error of the linear solver for implicit formulations LINEAR_SOLVER_ERROR= 1E4 % % Max number of iterations of the linear solver for the implicit formulation LINEAR_SOLVER_ITER= 2 %  MULTIGRID PARAMETERS % % % Multigrid levels (0 = no multigrid) MGLEVEL= 3 % % Multigrid cycle (V_CYCLE, W_CYCLE, FULLMG_CYCLE) MGCYCLE= W_CYCLE % % Multigrid presmoothing level MG_PRE_SMOOTH= ( 1, 2, 3, 3 ) % % Multigrid postsmoothing level MG_POST_SMOOTH= ( 0, 0, 0, 0 ) % % Jacobi implicit smoothing of the correction MG_CORRECTION_SMOOTH= ( 0, 0, 0, 0 ) % % Damping factor for the residual restriction MG_DAMP_RESTRICTION= 0.75 % % Damping factor for the correction prolongation MG_DAMP_PROLONGATION= 0.75 %  FLOW NUMERICAL METHOD DEFINITION % % % Convective numerical method (JST, LAXFRIEDRICH, CUSP, ROE, AUSM, HLLC, % TURKEL_PREC, MSW) CONV_NUM_METHOD_FLOW= JST % % Spatial numerical order integration (1ST_ORDER, 2ND_ORDER, 2ND_ORDER_LIMITER) MUSCL_FLOW= YES % % Slope limiter (VENKATAKRISHNAN, BARTH_JESPERSEN) SLOPE_LIMITER_FLOW= VENKATAKRISHNAN % % Entropy fix coefficient (0.0 implies no entropy fixing, 1.0 implies scalar % artificial dissipation) ENTROPY_FIX_COEFF= 0.001 % % 2nd and 4th order artificial dissipation coefficients JST_SENSOR_COEFF= ( 0.5, 0.02 ) % % Time discretization (RUNGEKUTTA_EXPLICIT, EULER_IMPLICIT, EULER_EXPLICIT) TIME_DISCRE_FLOW= EULER_IMPLICIT %  GRID DEFORMATION PARAMETERS % % % Linear solver or smoother for implicit formulations (FGMRES, RESTARTED_FGMRES, BCGSTAB) DEFORM_LINEAR_SOLVER= FGMRES % % Preconditioner of the Krylov linear solver (ILU, LU_SGS, JACOBI) DEFORM_LINEAR_SOLVER_PREC= LU_SGS % % Number of smoothing iterations for mesh deformation DEFORM_LINEAR_SOLVER_ITER= 500 % % Number of nonlinear deformation iterations (surface deformation increments) DEFORM_NONLINEAR_ITER= 1 % % Print the residuals during mesh deformation to the console (YES, NO) DEFORM_CONSOLE_OUTPUT= NO % % Minimum residual criteria for the linear solver convergence of grid deformation DEFORM_LINEAR_SOLVER_ERROR= 1E14 % % Type of element stiffness imposed for FEA mesh deformation (INVERSE_VOLUME, % WALL_DISTANCE, CONSTANT_STIFFNESS) DEFORM_STIFFNESS_TYPE= INVERSE_VOLUME % % Visualize the surface deformation (NO, YES) VISUALIZE_SURFACE_DEF= NO % % Visualize the volume deformation (NO, YES) VISUALIZE_VOLUME_DEF= NO %  CONVERGENCE PARAMETERS % % % Number of total iterations TIME_ITER= 360 % % Convergence criteria (CAUCHY, RESIDUAL) % CONV_CRITERIA= RESIDUAL % % % Min value of the residual (log10 of the residual) CONV_RESIDUAL_MINVAL= 6 % % Start convergence criteria at iteration number CONV_STARTITER= 0 % % Number of elements to apply the criteria CONV_CAUCHY_ELEMS= 100 % % Epsilon to control the series convergence CONV_CAUCHY_EPS= 1E10 % %  INPUT/OUTPUT INFORMATION % % % Mesh input file MESH_FILENAME= 0_mesh.su2 % % Mesh input file format (SU2, CGNS) MESH_FORMAT= SU2 % % Restart flow input file SOLUTION_FILENAME= solution_flow.dat % % Output file format (TECPLOT, TECPLOT_BINARY, PARAVIEW, % FIELDVIEW, FIELDVIEW_BINARY) TABULAR_FORMAT= CSV % % Output file convergence history (w/o extension) CONV_FILENAME= history % % Output file restart flow RESTART_FILENAME= restart_flow.dat % % Output file flow (w/o extension) variables VOLUME_FILENAME= flow % % Output file surface flow coefficient (w/o extension) SURFACE_FILENAME= surface_flow % % Writing solution file frequency WRT_SOL_FREQ= 1000 % % Writing solution file frequency for physical time steps (dual time) WRT_SOL_FREQ_DUALTIME= 1000 % % Writing convergence history frequency WRT_CON_FREQ= 1 % % Writing convergence history frequency (dual time, only written to screen) WRT_CON_FREQ_DUALTIME= 1 % % Screen output SCREEN_OUTPUT= (TIME_ITER, INNER_ITER, RMS_DENSITY, RMS_ENERGY, LIFT, DRAG_ON_SURFACE, PLUNGE, PITCH) % HISTORY_OUTPUT= (ITER, TIME_DOMAIN, RMS_RES, AERO_COEFF, AEROELASTIC) % OUTPUT_FILES= (RESTART, CSV, PARAVIEW, SURFACE_PARAVIEW) This is the error, Code:
CSysSolve::FGMRES(): system solved by initial guess. Error in "void CSysSolve<ScalarType>::ModGramSchmidt(int, std::vector<std::vector<ScalarType> >&, std::vector<CSysVector<ScalarType> >&) const [with ScalarType = double]":  FGMRES orthogonalization failed, linear solver diverged.  Error Exit  Last edited by pro_; May 12, 2020 at 08:25. 

May 16, 2020, 18:50 

#2 
Senior Member
Heather Kline
Join Date: Jun 2013
Posts: 309
Rep Power: 12 
I'm not sure what the problem is here, but I might be able to offer suggestions on additional information to provide for troubleshooting, such as:
 whether this error code occurs on the 1st iteration, or after multiple, of both the inner and outer iterations.  if you've tried increasing the number of inner iterations  if the steady problem is converged, and how many orders of magnitude the residuals are reduced, prior to restarting  you mention that you change the solver settings, but since the error is listing the linear solver, have you tried changing the linear solver method settings? It looks like you suspect (with reason) that this is an issue with the code  if you continue to have this problem after checking the suggestions above, and/or after someone else is able to weigh in on this question, please go ahead and post this to the github issue tracker: https://github.com/su2code/SU2/issues Please check first whether someone has reported a similar issue, and comment on the existing one if present. Good luck 

May 17, 2020, 06:25 

#3 
Senior Member
Pedro Gomes
Join Date: Dec 2017
Posts: 437
Rep Power: 11 
I fabricated two time domain restart files from the steady results for this problem (manually adding 0 grid velocities), the case also diverges, so the problem is not with steady state restarts.
We have this long running issue https://github.com/su2code/SU2/issues/828 that the Euler wall boundary condition might not be appropriate for time domain (for reasons unknown) I suspect this might be why you cannot replicate those results from years ago, maybe try going back to version 6.0.1... 

May 17, 2020, 08:00 

#4  
New Member
pro
Join Date: Mar 2020
Posts: 25
Rep Power: 5 
Quote:
 v7.0.0 onwards causes divergence  v6.2 leads to divergence as well Will do, Ill have a go with v6.0.1 

May 20, 2020, 15:41 

#5  
New Member
pro
Join Date: Mar 2020
Posts: 25
Rep Power: 5 
Quote:
Essentially, what im trying to say is that, the steady state initialization is only but 'speed up' the solution. Is this right to say? 

May 20, 2020, 15:52 

#6  
New Member
pro
Join Date: Mar 2020
Posts: 25
Rep Power: 5 
Quote:
Quote:
Quote:
Quote:


May 20, 2020, 16:17 

#7 
Senior Member
Pedro Gomes
Join Date: Dec 2017
Posts: 437
Rep Power: 11 
Yes I think that is fair to say (as I mentioned in a previous reply), if you have a repeating cycle it does not matter how you got there (I'm sure there are caveats for special cases, but this should not be one of them).


Thread Tools  Search this Thread 
Display Modes  


Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
PEMFC model with FLUENT  brahimchoice  FLUENT  22  April 19, 2020 16:44 
[ANSYS Meshing] Help with element size  sandri_92  ANSYS Meshing & Geometry  14  November 14, 2018 08:54 
Problem with restart in FSI unsteady simulation  david_mocholi  SU2  0  June 20, 2018 10:44 
fluent divergence for no reason  sufjanst  FLUENT  2  March 23, 2016 17:08 
Divergence problem  Smaras  FLUENT  13  February 21, 2013 06:03 