Inverse Design Optimization

khavart · June 20, 2019, 04:37

Hello, I'm currently running an inverse design optimization on a nozzle.

My goal pressure distribution come from said to be optimized nozzle. (Nozzle_Opti.png)

I copy the pressure distribution at the outlet. Then I create a new nozzle which is not optimized (Nozzle_Base.png) and set the TargetCp.dat as following.

Code:

"Global_Index", "x_coord", "y_coord","Pressure","Pressure_Coefficient","Mach_Number"
3, 0.123, 0, 15000,-2.14844, 2.58
2, 0.123, 0.0143,  15000,  -2.14844, 2.58
223, 0.123, 0.000595833,  15000, -2.14844, 2.58
224, 0.123, 0.00119167,  15000,  -2.14844, 2.58
225, 0.123, 0.0017875,  15000,  -2.14844, 2.58
226, 0.123, 0.00238333,  15000,  -2.14844, 2.58
227, 0.123, 0.00297917,  15000,  -2.14844, 2.58
228, 0.123, 0.003575, 15000,  -2.14844, 2.58
229, 0.123, 0.00417083,  15000,  -2.14844, 2.58
230, 0.123, 0.00476667,  15000,  -2.14844, 2.58
231, 0.123, 0.0053625,  15000,  -2.14844, 2.58
232, 0.123, 0.00595833,  15000,  -2.14844, 2.58
233, 0.123, 0.00655417,  15000,  -2.14844, 2.58
234, 0.123, 0.00715,  15000, -2.14844, 2.58
235, 0.123, 0.00774583,  15000,  -2.14844, 2.58
236, 0.123, 0.00834167,  15000, -2.14844, 2.58
237, 0.123, 0.0089375, 15000,   -2.14844, 2.58
238, 0.123, 0.00953333,  15000,  -2.14844, 2.58
239, 0.123, 0.0101292, 15000,   -2.14844, 2.58
240, 0.123, 0.010725,  15000,  -2.14844, 2.58
241, 0.123, 0.0113208,  15000,  -2.14844, 2.58
242, 0.123, 0.119167, 15000,  -2.14844, 2.58
243, 0.123, 0.0125125,  15000, -2.14844, 2.58
244, 0.123, 0.0131083, 15000,  -2.14844, 2.58
245, 0.123, 0.0137042, 15000, -2.14844, 2.58

Then my config files is :

Code:

% ------------- DIRECT, ADJOINT, AND LINEARIZED PROBLEM DEFINITION ------------%

PHYSICAL_PROBLEM= EULER
MATH_PROBLEM= DIRECT
REGIME_TYPE= COMPRESSIBLE
AXISYMMETRIC= NO
RESTART_SOL= NO
SYSTEM_MEASUREMENTS= SI

% -------------------- COMPRESSIBLE FREE-STREAM DEFINITION --------------------%

MACH_NUMBER= 0.8
AOA= 0
SIDESLIP_ANGLE= 0.0
INIT_OPTION= TD_CONDITIONS
FREESTREAM_OPTION= TEMPERATURE_FS
FREESTREAM_PRESSURE= 400000
FREESTREAM_TEMPERATURE= 288
REYNOLDS_LENGTH= 0.005
%

% ---------------------- REFERENCE VALUE DEFINITION ---------------------------%

REF_ORIGIN_MOMENT_X = 0.25
REF_ORIGIN_MOMENT_Y = 0.00
REF_ORIGIN_MOMENT_Z = 0.00
REF_LENGTH= 1.0
REF_AREA= 1.0
REF_DIMENSIONALIZATION= DIMENSIONAL

% ---- IDEAL GAS, POLYTROPIC, VAN DER WAALS AND PENG ROBINSON CONSTANTS -------%

FLUID_MODEL= IDEAL_GAS
GAMMA_VALUE= 1.4
GAS_CONSTANT= 287.058
CRITICAL_TEMPERATURE= 131.00
CRITICAL_PRESSURE= 3588550.0
ACENTRIC_FACTOR= 0.035

% -------------------- BOUNDARY CONDITION DEFINITION --------------------------%

MARKER_EULER= (Nozzle)
MARKER_RIEMANN= (Inlet, TOTAL_CONDITIONS_PT, 400000.000000, 288, 1, 0.0, 0.0, Outlet, STATIC_PRESSURE, 20000.000000, 0.0, 0.0, 0.0, 0.0)
MARKER_SYM= (Symmetry)

% ------------------------ SURFACES IDENTIFICATION ----------------------------%

MARKER_PLOTTING= ( Nozzle )
MARKER_MONITORING= ( Outlet )

% ------------- COMMON PARAMETERS DEFINING THE NUMERICAL METHOD ---------------%

NUM_METHOD_GRAD= GREEN_GAUSS
CFL_NUMBER= 2.0
CFL_ADAPT= YES
CFL_ADAPT_PARAM= ( 1.5, 0.5, 0.5, 1000)
MAX_DELTA_TIME= 1E6
RK_ALPHA_COEFF= ( 0.66667, 0.66667, 1.000000 )

% ----------- SLOPE LIMITER AND DISSIPATION SENSOR DEFINITION -----------------%

MUSCL_FLOW= YES
SLOPE_LIMITER_FLOW= NONE
MUSCL_ADJFLOW= YES
SLOPE_LIMITER_ADJFLOW= VENKATAKRISHNAN
ADJ_SHARP_LIMITER_COEFF= 3.0
LIMITER_ITER= 999999

% --------------------- INVERSE DESIGN SIMULATION -----------------------------%

INV_DESIGN_CP= YES
INV_DESIGN_HEATFLUX= NO

% ------------------------ LINEAR SOLVER DEFINITION ---------------------------%

LINEAR_SOLVER= FGMRES
LINEAR_SOLVER_PREC= LU_SGS
LINEAR_SOLVER_ERROR= 1E-4
LINEAR_SOLVER_ITER= 4

% -------------------------- MULTIGRID PARAMETERS -----------------------------%

MGLEVEL= 2
MGCYCLE= V_CYCLE
MG_PRE_SMOOTH= ( 1, 2, 3, 3 )
MG_POST_SMOOTH= ( 0, 0, 0, 0 )
MG_CORRECTION_SMOOTH= ( 0, 0, 0, 0 )
MG_DAMP_RESTRICTION= 1.0
MG_DAMP_PROLONGATION= 1.0

% -------------------- FLOW NUMERICAL METHOD DEFINITION -----------------------%

CONV_NUM_METHOD_FLOW= JST
ENTROPY_FIX_COEFF= 0.0
TIME_DISCRE_FLOW= EULER_IMPLICIT
RELAXATION_FACTOR_FLOW= 0.95
JST_SENSOR_COEFF= ( 0.5, 0.02 )

% ---------------- ADJOINT-FLOW NUMERICAL METHOD DEFINITION -------------------%

FROZEN_LIMITER_DISC= NO
FROZEN_VISC_DISC= NO
INCONSISTENT_DISC= NO
CONV_NUM_METHOD_ADJFLOW= JST
TIME_DISCRE_ADJFLOW= EULER_IMPLICIT
RELAXATION_FACTOR_ADJFLOW= 1.0
CFL_REDUCTION_ADJFLOW= 0.8
LIMIT_ADJFLOW= 1E6
MG_ADJFLOW= YES

% --------------------------- CONVERGENCE PARAMETERS --------------------------%

EXT_ITER= 10000
CONV_CRITERIA= RESIDUAL
RESIDUAL_REDUCTION= 7
RESIDUAL_MINVAL= -20
STARTCONV_ITER= 10

% ------------------------- INPUT/OUTPUT INFORMATION --------------------------%

MESH_FILENAME= Mesh_FFD.su2
MESH_FORMAT= SU2
MESH_OUT_FILENAME= mesh_out.su2
SOLUTION_FLOW_FILENAME= solution_flow.dat
SOLUTION_ADJ_FILENAME= solution_adj.dat
OUTPUT_FORMAT= PARAVIEW
CONV_FILENAME= history
BREAKDOWN_FILENAME= forces_breakdown.dat
RESTART_FLOW_FILENAME= restart_flow.dat
RESTART_ADJ_FILENAME= restart_adj.dat
VOLUME_FLOW_FILENAME= flow
VOLUME_ADJ_FILENAME= adjoint
VALUE_OBJFUNC_FILENAME= of_eval.dat
GRAD_OBJFUNC_FILENAME= of_grad.dat
SURFACE_FLOW_FILENAME= surface_flow
SURFACE_ADJ_FILENAME= surface_adjoint
WRT_SOL_FREQ= 250
WRT_SOL_FREQ_DUALTIME= 1
WRT_CON_FREQ= 1
WRT_CON_FREQ_DUALTIME= 10
WRT_RESIDUALS= NO
WRT_LIMITERS= NO
WRT_SHARPEDGES= NO
WRT_SURFACE= NO
LOW_MEMORY_OUTPUT= NO
CONSOLE_OUTPUT_VERBOSITY= HIGH
WRT_BINARY_RESTART= YES
READ_BINARY_RESTART= YES
REORIENT_ELEMENTS= YES

% ----------------------- DESIGN VARIABLE PARAMETERS --------------------------%

DV_MARKER= (Nozzle)

%DV_KIND= FFD_SETTING
%DV_PARAM=(NOZZLE,1,1,0.0,1.0)
%DV_VALUE= 0.1

DV_KIND=FFD_CONTROL_POINT_2D,FFD_CONTROL_POINT_2D,FFD_CONTROL_POINT_2D,FFD_CONTROL_POINT_2D,FFD_CONTROL_POINT_2D,FFD_CONTROL_POINT_2D,FFD_CONTROL_POINT_2D,FFD_CONTROL_POINT_2D,FFD_CONTROL_POINT_2D,FFD_CONTROL_POINT_2D
DV_PARAM=(NOZZLE,1,1,0.0,1.0);(NOZZLE,2,1,0.0,1.0);(NOZZLE,3,1,0.0,1.0);(NOZZLE,4,1,0.0,1.0);(NOZZLE,5,1,0.0,1.0);(NOZZLE,6,1,0.0,1.0);(NOZZLE,7,1,0.0,1.0);(NOZZLE,8,1,0.0,1.0);(NOZZLE,9,1,0.0,1.0);(NOZZLE,10,1,0.0,1.0)
DV_VALUE= 0.001,0.001,0.001,0.001,0.001,0.001,0.001,0.001,0.001,0.001

% -------------------- FREE-FORM DEFORMATION PARAMETERS -----------------------%

FFD_TOLERANCE= 1E-10
FFD_ITERATIONS= 1000
FFD_DEFINITION= (NOZZLE, 0.0559, 0.0, 0.0, 0.125, 0.0, 0.0, 0.125, 0.05, 0.0, 0.0559, 0.05, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0)
FFD_DEGREE= (10, 1, 0)
FFD_CONTINUITY= NO_DERIVATIVE
FFD_SYMMETRY_PLANE= YES

% ------------------------ GRID DEFORMATION PARAMETERS ------------------------%

DEFORM_LINEAR_ITER= 1000
DEFORM_NONLINEAR_ITER= 2
DEFORM_CONSOLE_OUTPUT= YES
DEFORM_LINEAR_SOLVER_ERROR= 1E-14
DEFORM_STIFFNESS_TYPE= CONSTANT_STIFFNESS
DEFORM_LINEAR_SOLVER= FGMRES
DEFORM_LINEAR_SOLVER_PREC= LU_SGS
DEFORM_TOL_FACTOR = 0.000001
DEFORM_COEFF = 1E6
VISUALIZE_DEFORMATION= YES

% --------------------- OPTIMAL SHAPE DESIGN DEFINITION -----------------------%

OPT_OBJECTIVE=INVERSE_DESIGN_PRESSURE
OPT_CONSTRAINT= NONE

OPT_GRADIENT_FACTOR= 1E-6
OPT_RELAX_FACTOR= 1E3
OPT_ITERATIONS= 40
OPT_ACCURACY= 1E-20
OPT_BOUND_UPPER= 0.1
OPT_BOUND_LOWER= 0
FIN_DIFF_STEP=0.0001


DEFINITION_DV= ( 15, 1 | Nozzle | NOZZLE, 1, 1, 0.0, 1.0); ( 15, 1 | Nozzle | NOZZLE, 2, 1, 0.0, 1.0); ( 15, 1 | Nozzle | NOZZLE, 3, 1, 0.0, 1.0 ); ( 15, 1 | Nozzle | NOZZLE, 4, 1, 0.0, 1.0 );( 15, 1 | Nozzle | NOZZLE, 5, 1, 0.0, 1.0); ( 15, 0.1 | Nozzle | NOZZLE, 6, 1, 0.0, 1.0); ( 15, 0.1 | Nozzle | NOZZLE, 7, 1, 0.0, 1.0 ); ( 15, 1 | Nozzle | NOZZLE, 8, 1, 0.0, 1.0 );( 15, 1 | Nozzle | NOZZLE, 9, 1, 0.0, 1.0 );( 15, 1 | Nozzle | NOZZLE, 10, 1, 0.0, 1.0 )

I'm using FFD boxes for the deformation and I have already observed the deformations.

The problem is that when I run the optimization I can't recover the optimized design and it stops after only 1 iteration, saying :

Code:

Found: Mesh_FFD.su2
Found: TargetCp.dat
New Project: ./
Removing old designs in 10s. Done!

Sequential Least SQuares Programming (SLSQP) parameters:
Number of design variables: 10 ( 10 ) 
Objective function scaling factor: [1.0]
Maximum number of iterations: 40
Requested accuracy: 1e-26
Initial guess for the independent variable(s): [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]
Lower and upper bound for each independent variable: [(0.0, 0.0001), (0.0, 0.0001), (0.0, 0.0001), (0.0, 0.0001), (0.0, 0.0001), (0.0, 0.0001), (0.0, 0.0001), (0.0, 0.0001), (0.0, 0.0001), (0.0, 0.0001)]

  NIT    FC           OBJFUN            GNORM
    5     1     2.557944E-07     7.675535E-04
Positive directional derivative for linesearch    (Exit mode 8)
            Current function value: 2.55794353e-07
            Iterations: 5
            Function evaluations: 1

Have one of you ever had this problem ?

Thanks for your help,

Kevin

Nozzle_Opti.png

Nozzle_Base.jpg

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
FFD 2D for inverse shape design using target CP	ericclaro	SU2 Shape Design	1	May 30, 2019 19:53
Difficulties running 3D inverse design optimisations	dominico	SU2 Shape Design	1	September 19, 2017 09:56
Problem with Inverse design of a 2D cascade	Bartosz.Ziegler	SU2 Shape Design	10	February 17, 2016 13:50
Inverse design and Optimization of Aircraft Wings	Simon Nava	Main CFD Forum	8	May 22, 2007 15:55
Info: Short Course On Thermal Design of Electronic Equipment	Arnold Free	Main CFD Forum	0	August 10, 1999 11:18