Abstract
The commercial aviation industry generates approximately 2.5% of worldwide CO2 emissions annually, resulting in up to 1,800 megatonnes of CO2 emitted per year by 2050. Electrified propulsion systems provide the most promising short-term path to reduce CO2 emissions because of its ability to be integrated in existing tube-and-wing aircraft designs while also maintaining compatibility with existing airport infrastructures. Many possible electrified propulsion system architectures exist, requiring multidisciplinary design optimization to identify ones that maximize system-level performance and meet the necessary safety regulations. However, the increased architectural complexity of electrified propulsion systems requires configuration-agnostic design methods to be developed, enhancing traditional conventional aircraft design methods. In this thesis, a unified propulsion system architecture optimization and synthesis environment was developed to account for nominal aircraft performance, off-nominal propulsion system failure scenarios, and system-level safety constraints in a configuration-agnostic manner, and consisted of three overarching contributions.
First, a graph-based propulsion system analysis framework was developed, providing a formal method to represent and analyze advanced propulsion systems in a configuration-agnostic manner. The framework abstracts a propulsion system as a graph with component constitutive models at each vertex and power transfer relationships along each edge, overcoming topological restrictions in prior state-of-the-art propulsion system analysis methods. Uni- and bi-directional power flows are simulated using a fixed point iteration, enabling power to be propagated across branches of a propulsion system with different path lengths. Integrating the GPSA Framework with a parametric aircraft sizing tool provides a configuration-agnostic aircraft design environment, enabling system-level tradeoffs between aircraft with disparate propulsion system architectures to be quantified.
Second, currently published FAA flight performance requirements for aircraft with discrete, homogeneously sized propulsors were generalized to accommodate advanced propulsion systems with heterogeneously sized or distributed propulsors and their corresponding critical failure modes. A monotonic, bounded mapping between the specific excess power loss and FAA-published flight performance requirements was constructed to reproduce the climb gradient regulations for conventional aircraft while extending the same flight performance logic to accommodate advanced aircraft concepts and their propulsion systems. To enable seamless integration into the aircraft sizing process, traditional n/(n - 1) engine inoperative correction factors were also generalized as a function of the failure mode severity. An advanced aircraft sizing study revealed that, for aircraft with a propulsion system failure less severe than a legacy one-engine inoperative condition on a conventional aircraft, sizing under the generalized flight performance requirements will reduce propulsion system over-sizing. System-level performance may improve, but depends on the operational constraints imposed.
Third, a bi-level optimization framework was constructed to assemble redundant, safety-feasible propulsion system architecture topologies utilizing embedded, problem-specific heuristics, while also identifying the optimal power management strategy. The optimization framework decomposes a mixed integer optimization problem into its discrete variables, finding the best system architecture, and its continuous variables, finding the optimal component sizes and power management strategy. Depending on the failure mode, the bi-level optimization framework assembles propulsion system architectures with no worse system-level performance and possibly improved system-level safety relative to fixed architecture baselines and those synthesized using stochastic optimization methods with problem-agnostic heuristics. In some cases, system-level safety is improved with minimal system-level performance penalty via asymmetric cross-connection schemes generated by the optimization framework.
Together, the computational frameworks developed in this thesis support the design and optimization of future aircraft concepts and their advanced propulsion system architectures in a configuration-agnostic manner while accounting for nominal aircraft performance, off-nominal propulsion system failure scenarios, and system-level safety constraints.
Paul Mokotoff
PhD Student and Graduate Research Assistant
Paul Mokotoff is a graduate student research assistant in the IDEAS Lab at the University of Michigan.