A Surrogate-Based Multi-Disciplinary Design Optimization of Movables for a Modern Wide-Body Aircraft Wing Integration internship

Abstract

The aerospace industry faces increasing pressure to enhance aircraft efficiency, driven by economic and environmental imperatives. The aircraft wing is central to this effort, but its design is governed by a complex, multi-disciplinary trade-off between high-speed cruise performance and low-speed take-off and landing requirements. This thesis addresses this challenge by developing and applying a surrogate-model-based methodology to perform a multi-objective design optimization (MDO) of a modern wide-body aircraft wing.

A parametric model featuring twelve design variables—including span, sweep, high-lift device types (e.g., Slats, Kruegers, SSF, DSF, ADHF), and control surface geometry—was created. A low-fidelity, semi-empirical simulation toolchain was used to evaluate over 1600 design configurations, generating a comprehensive dataset. A high-fidelity surrogate model was then trained and validated on this data, achieving an R-Squared value greater than 0.9 for all key performance metrics.

The analysis of the design space confirmed that span extension is the most dominant variable for improving aerodynamic efficiency, at the cost of structural weight and handling qualities. Subsequent multi-objective optimization studies were performed, culminating in an analysis based on Net Present Value (NPV). The key finding of this work is that aerodynamic efficiency, driven by low block fuel, is the overwhelming contributor to the aircraft's lifecycle economic value. The performance-optimized wing achieved an NPV just 7.4\% shy of the final, fully-optimized design, indicating that design choices aimed at reducing production cost provide only a marginal benefit compared to those that minimize cruise drag.

This thesis successfully demonstrates a robust MDO framework, delivers quantitative insights into the key design drivers of modern wings, and provides a data-driven foundation for future higher-fidelity design cycles.