Simulation-based design optimization has the potential to dramatically improve the design of aerospace vehicles over the next decade. Many aerospace vehicles operate under conditions which are governed by multiple disciplines that interact in highly nonlinear ways. For instance, flexible wings that deform under aerodynamic loads are governed simultaneously by both aerodynamics and structural dynamics, two disciplines that must be tightly coupled to obtain the true flying wing shape and, in turn, evaluate the performance of the aircraft. Multidisciplinary design optimization (MDO) methods are designed to address concurrently the difficulty of coupled analysis and integrated design using sophisticated numerical methods. First developed over 20 years ago, MDO techniques have matured significantly over the past decade. Low-fidelity MDO methods are now routinely used by industry in preliminary design optimization however, these tools frequently rely on empirical models that have fundamental limitations. High-fidelity, physics-based simulation can overcome these limitations, but significant research challenges remain to reduce the computational cost and increase the robustness of MDO techniques applied to high-fidelity models.
The goal of my research is to develop new MDO methods and techniques for high-fidelity simulations that can be used in a time-critical design cycle in industry applications. My research applications focus on aircraft design, however, I also hope to make contributions to applications of MDO in other aerospace-related fields. My research focuses on three areas in which new developments will have an important impact on the design optimization process, and in turn, on the performance of future generations of aircraft: high-performance computing (HPC) for design optimization, analysis and design optimization of unsteady aeroelastic phenomena, and optimization of composite structures.
Optimization using high-fidelity simulations typically require thousands of times the computational resources of high-fidelity simulations alone. Therefore, developing algorithms that make effective use of HPC resources is essential to obtain results within a 12 to 24 hour time frame that is practical for industrial design applications. Unsteady aeroelastic phenomena are becoming more important in the aircraft design problem as advanced structural materials enable lighter, more flexible aircraft. Developing new design methods to incorporate these design constraints into high-fidelity aircraft optimization problems will maximize the use of both passive and active load alleviation, and avoid costly late-stage design modifications. Composite materials have the potential to enable significant weight savings by giving the design engineer greater freedom to tailor structures to meet stringent design requirements. However, the design of composite structures is often complicated by the addition of manufacturing constraints. These manufacturing constraints place an additional level of complexity on the structural design and are frequently challenging to formulate in a manner compatible with efficient optimization methods. Advances in these areas of high-fidelity simulation-based MDO have the potential to significantly improve the performance of the next generation of aerospace vehicles.