Adjoint Solver Diverges for "MAXIMUM_HEATFLUX"

[aero35]

Hi,

I am using SU2 version 7.3.0 to obtain the direct and adjoint solutions of the blunt body at M = 7.1. I use RANS equation with SA turbulence model. Direct and discrete adjoint solutions (for drag and total heatflux) are converged successfully. However, the adjoint solver for maximum heatflux diverges in the first iteration (see the .cfg file settings below). Also, I have the same issue with SU2 version 7.2.1.

I would appreciate it if anyone can assist me with this matter.
Thanks,

---

% ------------- DIRECT, ADJOINT, AND LINEARIZED PROBLEM DEFINITION ------------%
%
% Physical governing equations (EULER, NAVIER_STOKES,
% WAVE_EQUATION, HEAT_EQUATION, FEM_ELASTICITY,
% POISSON_EQUATION)
SOLVER= RANS
%
% Specify turbulent model (NONE, SA, SA_NEG, SST)
KIND_TURB_MODEL= SA
%
% Mathematical problem (DIRECT, CONTINUOUS_ADJOINT)
MATH_PROBLEM= DISCRETE_ADJOINT
%
% Restart solution (NO, YES)
RESTART_SOL= NO
%
SYSTEM_MEASUREMENTS = SI
AXISYMMETRIC = YES

% ----------- COMPRESSIBLE AND INCOMPRESSIBLE FREE-STREAM DEFINITION ----------%
%
% Mach number (non-dimensional, based on the free-stream values)
MACH_NUMBER= 7.1
%
% Angle of attack (degrees)
AOA= 0.0
%
% Free-stream pressure (101325.0 N/m^2 by default, only Euler flows)
%FREESTREAM_PRESSURE= 663
%
% Free-stream temperature (288.15 K by default)
FREESTREAM_TEMPERATURE= 293
%
% Free-stream viscosity (1.853E-5 N s/m^2, 3.87E-7 lbf s/ft^2 by default)
%FREESTREAM_VISCOSITY= 7.96E-5
%
% Reynolds number (non-dimensional, based on the free-stream values)
REYNOLDS_NUMBER= 5200
%
% Reynolds length (1 m by default)
REYNOLDS_LENGTH= 0.00635
%REYNOLDS_LENGTH= 1.0

% ---------------------- REFERENCE VALUE DEFINITION ---------------------------%
%
% Reference origin for moment computation
REF_ORIGIN_MOMENT_X = 0.00
REF_ORIGIN_MOMENT_Y = 0.00
REF_ORIGIN_MOMENT_Z = 0.00
%
% Reference length for pitching, rolling, and yawing non-dimensional moment
REF_LENGTH= 0.00635
%REF_LENGTH= 1.0
%
% Reference area for force coefficients (0 implies automatic calculation)
REF_AREA= 0.000709287

% ----------------------- BOUNDARY CONDITION DEFINITION -----------------------%
%
% Navier-Stokes wall boundary marker(s) (NONE = no marker)
MARKER_ISOTHERMAL= ( WALL, 1000.0 )
%
% Marker of the far field (NONE = no marker)
MARKER_FAR= ( FARFIELD)
%MARKER_OUTLET= ( OUTLET, 663)
%
MARKER_SYM = (SYMMETRY)

% ------------------------ SURFACES IDENTIFICATION ----------------------------%
%
% Marker(s) of the surface in the surface flow solution file
MARKER_PLOTTING = ( WALL )
%
% Marker(s) of the surface where the non-dimensional coefficients are evaluated.
MARKER_MONITORING = ( WALL )

% ------------- COMMON PARAMETERS TO DEFINE THE NUMERICAL METHOD --------------%
%
% Numerical method for spatial gradients (GREEN_GAUSS, WEIGHTED_LEAST_SQUARES)
NUM_METHOD_GRAD= GREEN_GAUSS
%
% Courant-Friedrichs-Lewy condition of the finest grid
CFL_NUMBER= 1.0
%
% Adaptive CFL number (NO, YES)
CFL_ADAPT= NO
%
% Parameters of the adaptive CFL number (factor down, factor up, CFL min value,
% CFL max value )
CFL_ADAPT_PARAM= ( 1.1, 0.9, 5.0, 50.0 )
%
% Number of total iterations
ITER= 5000

% ------------------------ LINEAR SOLVER DEFINITION ---------------------------%
%
% Linear solver for implicit formulations (BCGSTAB, FGMRES)
LINEAR_SOLVER= FGMRES
%
% Preconditioner of the Krylov linear solver (JACOBI, LINELET, LU_SGS)
LINEAR_SOLVER_PREC= LU_SGS
%
% Minimum error of the linear solver for implicit formulations
LINEAR_SOLVER_ERROR= 1E-6
%
% Max number of iterations of the linear solver for the implicit formulation
LINEAR_SOLVER_ITER= 5

% -------------------------- MULTIGRID PARAMETERS -----------------------------%
%
% Multi-Grid Levels (0 = no multi-grid)
MGLEVEL= 0
%
% Multi-grid cycle (V_CYCLE, W_CYCLE, FULLMG_CYCLE)
MGCYCLE= V_CYCLE
%
% Multi-Grid PreSmoothing Level
MG_PRE_SMOOTH= ( 1, 2, 3, 3 )
%
% Multi-Grid 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= 0.95
%
% Damping factor for the correction prolongation
MG_DAMP_PROLONGATION= 0.95

% -------------------- FLOW NUMERICAL METHOD DEFINITION -----------------------%
%
% Convective numerical method (JST, LAX-FRIEDRICH, CUSP, ROE, AUSM, SLAU, SLAU2,
% AUSMPLUSUP, AUSMPLUSUP2 HLLC, TURKEL_PREC, MSW)
CONV_NUM_METHOD_FLOW= JST
%
% Monotonic Upwind Scheme for Conservation Laws (TVD) in the flow equations.
% Required for 2nd order upwind schemes (NO, YES)
MUSCL_FLOW= YES
%
% Slope limiter (VENKATAKRISHNAN, MINMOD)
SLOPE_LIMITER_FLOW= VENKATAKRISHNAN
%
% Coefficient for the limiter
VENKAT_LIMITER_COEFF= 0.03
%
% 2nd and 4th order artificial dissipation coefficients
JST_SENSOR_COEFF= ( 0.5, 0.04 )
%
% Time discretization (RUNGE-KUTTA_EXPLICIT, EULER_IMPLICIT, EULER_EXPLICIT)
TIME_DISCRE_FLOW= EULER_IMPLICIT

% -------------------- TURBULENT NUMERICAL METHOD DEFINITION ------------------%
%
% Convective numerical method (SCALAR_UPWIND)
CONV_NUM_METHOD_TURB= SCALAR_UPWIND
%
% Monotonic Upwind Scheme for Conservation Laws (TVD) in the turbulence equations.
% Required for 2nd order upwind schemes (NO, YES)
MUSCL_TURB= YES
%
% Time discretization (EULER_IMPLICIT)
TIME_DISCRE_TURB= EULER_IMPLICIT

% ---------------- ADJOINT-FLOW NUMERICAL METHOD DEFINITION -------------------%
% Objective function in optimization problem (DRAG, LIFT, SIDEFORCE, MOMENT_X,
% MOMENT_Y, MOMENT_Z, EFFICIENCY,
% EQUIVALENT_AREA, NEARFIELD_PRESSURE,
% FORCE_X, FORCE_Y, FORCE_Z, THRUST,
% TORQUE, FREE_SURFACE, TOTAL_HEATFLUX,
% MAXIMUM_HEATFLUX, INVERSE_DESIGN_PRESSURE,
% INVERSE_DESIGN_HEATFLUX)
OBJECTIVE_FUNCTION= MAXIMUM_HEATFLUX
%
% Convective numerical method (JST, LAX-FRIEDRICH, ROE)
CONV_NUM_METHOD_ADJFLOW= JST
%
% Monotonic Upwind Scheme for Conservation Laws (TVD) in the adjoint flow equations.
% Required for 2nd order upwind schemes (NO, YES)
MUSCL_ADJFLOW= NO
%
% Slope limiter (NONE, VENKATAKRISHNAN, BARTH_JESPERSEN, VAN_ALBADA_EDGE,
% SHARP_EDGES, WALL_DISTANCE)
SLOPE_LIMITER_ADJFLOW= VENKATAKRISHNAN
%
% Coefficient for the sharp edges limiter
ADJ_SHARP_LIMITER_COEFF= 3.0
%
% 2nd, and 4th order artificial dissipation coefficients
ADJ_JST_SENSOR_COEFF= ( 0.0, 0.01 )
%
% Reduction factor of the CFL coefficient in the adjoint problem
CFL_REDUCTION_ADJFLOW= 0.8
%
% Time discretization (RUNGE-KUTTA_EXPLICIT, EULER_IMPLICIT)
TIME_DISCRE_ADJFLOW= EULER_IMPLICIT
%

% --------------------------- CONVERGENCE PARAMETERS --------------------------%
%
% Min value of the residual (log10 of the residual)
CONV_RESIDUAL_MINVAL= -6
%
% Start Cauchy criteria at iteration number
CONV_STARTITER= 10
%
% Number of elements to apply the criteria
CONV_CAUCHY_ELEMS= 100
%
% Epsilon to control the series convergence
CONV_CAUCHY_EPS= 1E-6
%

% ------------------------- INPUT/OUTPUT INFORMATION --------------------------%
READ_BINARY_RESTART= NO
% Mesh input file
MESH_FILENAME= 140x200_new.su2
%
% Mesh input file format (SU2, CGNS, NETCDF_ASCII)
MESH_FORMAT= SU2
%
% Mesh output file
MESH_OUT_FILENAME= mesh_out.su2
%
% Restart flow input file
SOLUTION_FILENAME= solution_flow.dat
%
% Restart adjoint input file
SOLUTION_ADJ_FILENAME= solution_adj.dat
%
% Output tabular format (CSV, TECPLOT)
TABULAR_FORMAT= TECPLOT
OUTPUT_FILES = (RESTART, RESTART_ASCII, TECPLOT_ASCII, SURFACE_TECPLOT_ASCII, SURFACE_CSV)
%
% Output file convergence history (w/o extension)
CONV_FILENAME= history
%
% Output file restart flow
RESTART_FILENAME= restart_flow.dat
%
% Output file restart adjoint
RESTART_ADJ_FILENAME= restart_adj.dat
%
% Output file flow (w/o extension) variables
VOLUME_FILENAME= flow
%
% Output file adjoint (w/o extension) variables
VOLUME_ADJ_FILENAME= adjoint
%
% Output Objective function gradient (using continuous adjoint)
GRAD_OBJFUNC_FILENAME= of_grad.dat
%
% Output file surface flow coefficient (w/o extension)
SURFACE_FILENAME= surface_flow
%
% Output file surface adjoint coefficient (w/o extension)
SURFACE_ADJ_FILENAME= surface_adjoint
%
% Writing solution file frequency
OUTPUT_WRT_FREQ= 250
%
%
% Screen output
%SCREEN_OUTPUT= (INNER_ITER, RMS_DENSITY, RMS_ENERGY, LIFT, DRAG)
%HISTORY_OUTPUT= (ITER, RMS_RES, AERO_COEFF, HEAT)

SCREEN_OUTPUT= (INNER_ITER, RMS_ADJ_DENSITY, RMS_ADJ_ENERGY, SENS_PRESS, SENS_AOA)
HISTORY_OUTPUT= (ITER, RMS_RES, SENSITIVITY)

[Aslihan35]

Hi all,

It would be greatly appreciated if anybody can help with adjoint solver divergence for maximum heatflux.

Thank you

[Aslihan35]

Hi all,

I resolved the issue by having a minor change in the settings. I defined the maximum heat flux as a custom objective function and added the following lines to the .cfg file.

OBJECTIVE_FUNCTION = CUSTOM_OBJFUNC
CUSTOM_OBJFUNC = 'MAXIMUM_HEATFLUX'

The adjoint solver converged smoothly after these changes.

Thank you,

[pcarruscag]

Hmmm, there's a bug then. Those two methods are not referring to the same value.

[pcarruscag]

Please click the "Create issue from discussion" button

[Aslihan35]

I also scaled the custom objective function to make a comparison and the results were promising.