Evolving an Aircraft Using a Parametric Design System
Jonathan Byrne
Natural Computing Research and Applications Group
University College Dublin
Ireland
The Importance of Mapping
Direct Mapping (Voxels)
Buildable Objects by Pollack and Funes (1999)
Issues with Direct Mappings
Shape not smooth or contains small holes
Scales poorly
Parents being valid doesn't guarantee valid children
Qualities of a Good Mapping:
Generates valid designs
High locality
Hierarchy, modularity and regularity
Parametric Design
Rapid prototyping technique
Define the relationships between components
Originally in Sketchpad
Why Parametric?
Adoption in Aeronautics and Architecture
Powerful applications in 3D printing
Programmers get it for free
Generates valid designs
Easy to evolve
Grasshopper
Turning Torso (Calatrava)
Messe Halle (Herzog & De Meuron)
Messe Halle (Herzog & De Meuron)
Rolex Learning Center (SANAA)
Open-Vehicle Sketch Pad (openVSP)
link
OpenVSP Hangar
OpenFOAM
Field operations and manipulation
Industrial Standard
Built on openMPI
OpenFOAM Toolbox
Meshing utilities
Incompressible (hydrodynamic) solvers
Transonic solvers
Heat transfer
Electromagnetic solvers
Financial solvers
SIMPLE Algorithm
Semi-implicit method for pressure linked equations
Iterative Navier-Stokes solver
Initial state set using inviscid solver
Aircraft Optimisation
Airfoil Representation
Wing Representation
Best Lift Maximisation
Best Drag Minimisation
Multi-Objective Optimisation
Conflicting design objectives
Lift versus drag of a wing
No perfect solution
Initial Generation
Evolving the Pareto Front
Final Generation
Blended Wing Body Optimisation
Pareto Front for BWB
Optimised Blended Wing Body Designs
Cessna 182 Optimisation
Final Pareto Front
Optimised Cessna 182 Designs
MIG 21 Optimisation
Final Pareto Front
Optimised MIG 21 Designs:
Conclusions
Parametric representations are easily optimised
Designer allowed to specify the search space
EAs capable of actively suggesting design improvements
OpenFOAM is a toolbox of fitness functions