Aero-Structural-Trim Design Optimization Using SU2, Nastran, and OpenMDAO

2^nd Annual SU2 Conference

Harsh C. Patel, Juan J. Alonso

13 July 2021

0

Presentation Outline

• Research objectives
• Aero-Structural-Trim sensitivities
• Design application
• MDAO framework
• Aircraft weight optimization
• SU2 Python API developments

​

​

1

1

Research Objectives

• Can coupled, multi-physics, sensitivity analyses be useful in industrial settings?

​

• Can such workflows be implemented in a flexible and modular framework? 

​

• Objectives
• Incorporate non-linear, coupled-physics, sensitivities for industry-relevant aeroelastic considerations in conceptual design phase
• Calculate coupled aero-structural-trim sensitivities for all necessary cost and constraint functions
• Incorporate trim and metrics for stability & control into optimization process

​

2

2

Aero-Structural-Trim Coupled Sensitivities

3

3

Aero-Structural-Trim Coupled Sensitivities

• Coupled adjoint equation

​

​

​

​

​

​

• Total sensitivity equation in terms of adjoint variables

4

4

Aeroelastic Design with Sensitivity Analysis

• Quantities of interest
• Aerodynamic: lift, drag, moment, range, endurance
• Structural: weight, stresses, stiffness, thickness
• Trim: angle of attack, control surface deflections, roll rates

​

• Design considerations
• Sensitivity of C_D and drag divergence with respect to wing shape
• Wing static stability and flutter considerations

​

​

​

5

5

Demonstration Test Case

• Efficient Supersonic Air Vehicle (ESAV)
• Objectives for supercruise, supersonic dash, maneuverability, and extended range
• A tailless configuration with conventional and non-conventional control surfaces for stability

​

​

​

​

6

6

Demonstration Test Case

• ESAV model utilizes finite element and aerodynamic models consistent with typical aerospace industry methodologies

​

7

7

Computational Framework

• SU2
• Flow calculations, mesh deformation, and shape parametrization

​

• Nastran
• Structural modeling and evaluation of trim and aeroelastic responses

​

• MELD
• Load and displacement transfer schemes

​

• SNOPT
• Interfaced through pyOptSparse

​

• OpenMDAO

​

8

8

Nastran Aerodynamic Correction Factors

3D FEM mesh

(node-based)

2D VLM mesh

(centroid-based)

9

9

Nastran VLM Pressure Distributions

10

10

SU2 CFD Corrections via FA2J

n = the number of structural nodes linked to each aerodynamic node

11

11

Derivatives in OpenMDAO

• Totals: derivatives of model outputs w.r.t model inputs
• Partials: derivatives of component outputs w.r.t component inputs

​

12

13

13

ESAV Weight Minimization

14

14

SU2 Flow Contours 

Distribution A. Approved for public release: distribution unlimited. 88ABW-2020-3798

15

15

SU2 Surface Sensitivities

16

16

Aeroelastic Analysis Convergence

17

17

18

18

PYSU2 API Development

• Extensions to pysu2 and pysu2ad for use in aeroelastic MDO
• With external structural solvers and externally defined design variables (see GitHub issue #1262 and Pull Request draft #1300)

​

• Exposing mesh deformation routines and adjoint counterparts to Python
• Refactorization introduced CDeformationDriver and CDiscAdjDeformationDriver
• Legacy SU2_DEF and newer CMeshSolver can both be used

​

• Added new “get” and “set” methods to pysu2
• Geometry information, flow states, AoA/AoS updates, etc.

​

​

19

19

Summary

• Enable routine aerodynamic CFD, structural FEM, trim analysis and design optimization for industrial use in conceptual design
• Use tested solvers like SU2 and Nastran for physical modelling

​

• Flexibility in problem setup and solver interfacing
• OpenMDAO facilitates this orchestration
• Arbitrary flight conditions in multipoint optimizations

​

• Sensitivity analysis may provide insight and a path to a solution
• Multi-disciplinary coupling is too complex to characterize
• Traditional engineering and design approaches may be inadequate

​

​

20

20

Thank you!

21