AI Designed This Organic Flying Wing… I Had to Build It (+16% Efficiency)

Exploring Optimal Wing Designs with AI

Introduction to Wing Design

• The video begins by reflecting on a previous design inspired by bird wings, which demonstrated stability and efficiency through a dihedral shape.

• The focus shifts to exploring the vast possibilities of wing shapes, aiming to find an optimal design using artificial intelligence (AI).

• The presenter introduces the concept of creating a fully adjustable parametric wing that allows for extensive design freedom.

AI Optimization Process

• The initial expectation was that AI would make minor adjustments to the existing design; however, it produced a significantly more complex shape.

• Previous attempts at optimization were limited in scope, using only simple trapezoidal designs and five variables for performance prediction.

• Acknowledges the computational challenges faced during earlier simulations, which took considerable time due to reliance on Computational Fluid Dynamics (CFD).

Defining Design Variables

• To enhance flexibility in design, 16 parameters are defined for the new wing model compared to just five previously.

• Key parameters include leading edge sweep angle and chord distribution along the span, allowing for smooth tapering of the wing.

• Twist is controlled with four variables across different sections of the wing, crucial for tuning stall behavior and lift distribution.

Structural Considerations in Design

• Five variables define dihedral shaping along the span, enabling exploration of various structural benefits versus straight-wing designs.

• Emphasizes that real aircraft wings are complex structures requiring simplification into equivalent models for effective analysis.

Scoring Function Development

• Introduces multidisciplinary optimization (MDO), considering not just aerodynamic efficiency but also stability and structural integrity.

• Each wing design is graded based on an objective function: aerodynamic efficiency minus penalties related to structure and stability.

Evaluation Methods

• Utilizes vortex lattice method (VLM), which runs simulations quickly compared to CFD, allowing testing of thousands of designs efficiently.

Evolutionary Algorithm Implementation

• Describes differential evolution as a method where random designs evolve over generations through mixing successful parameters from previous iterations.

Finalizing Design Parameters

• After extensive testing and refinement through simulation, an optimized wing design emerges as a winner from thousands of candidates.

Transitioning from Simulation to Real Fabrication

Creating 3D Models

• A detailed 3D model is developed based on optimized parameters before moving towards physical fabrication.

Building Process Overview

• The transition from simulation results to actual building marks an exciting phase in bringing theoretical designs into reality.

Maiden Flight Experience

Initial Flight Testing

• Expresses nervousness about maiden flights despite thorough preparation; acknowledges potential risks involved.

Performance Assessment

• Reports excellent flight performance with stable gliding capabilities even under challenging weather conditions.

Evaluating Success Post Flight

Efficiency Gains Observed

• Concludes that AI optimization resulted in approximately 16% increased efficiency compared to prior designs.

Future Directions

• Plans future videos focusing on further flight testing and understanding why specific geometries were chosen by AI.