Simulation-Based Optimization and Evaluation of Urban Air Mobility

Current mobility systems face increasing strain due to rapid urbanization and growing mobility demands, resulting in transportation bottlenecks, longer travel duration, and exacerbated ecological externalities. Urban Air Mobility (UAM), utilizing electric vertical takeoff and landing (eVTOL) aircraft operating in low-altitude airspace, addresses these limitations by leveraging underutilized vertical airspace and cleaner energy. However, UAM implementation faces substantial challenges in integrated system design, with current research predominantly focusing on isolated aspects examining either demand-side factors or supply-side considerations without adequate integration. This dissertation aims to develop comprehensive methods, algorithms, and frameworks to address these gaps through simulation-based optimization and evaluation approaches. An integrated methodological framework is constructed to consider the interdependence of essential components for UAM system design. User preference modeling, strategic infrastructure planning, and operational optimization under uncertainty are three prerequisites for successful UAM deployment. The dissertation contributes to these research domains across behavioral modeling methods, network planning methodologies, operational strategies, and empirical insights. In terms of user preference analysis, discrete choice models reveal significant heterogeneity in UAM acceptance patterns driven by complex interactions between service attributes, sociodemographic factors, and latent attitudinal constructs. Environmental awareness and price sensitivity serve as fundamental mediating factors, with shared UAM services offering substantial potential for reducing per-passenger costs while enhancing service acceptance and market accessibility beyond traditional high-income segments. The research demonstrates that UAM operators must develop differentiated service strategies rather than uniform approaches to capture diverse market segments effectively. To address demand uncertainty in UAM network design, the dissertation develops a sequential two-layer stochastic optimization framework that embeds discrete choice models within Monte Carlo simulations to generate demand endogenously from each network design strategy. This approach addresses real-world constraints where infrastructure decisions must precede demand realization, providing robust network planning methods for transportation agencies when high-quality passenger preference data is unavailable. Further, novel infrastructure planning approaches are developed through a multi-objective optimization framework that integrate real land use data, capacity constraints, cost considerations, and incremental deployment strategies. The multi-phased siting framework facilitates gradual evolution from initial eVTOL operations to advanced modular flying vehicles, demonstrating that strategic four-phased deployment yields approximately 13000 euros higher daily generalized travel cost savings. The framework identifies optimal infrastructure cost subsidy ranges of 20-40%, balancing performance gains with economic sustainability. To address operational optimization under demand uncertainty, the dissertation develops stochastic programming frameworks for UAM ride-pooling operations. The hybrid optimization approach combining NSGA-II with Targeted Ruin-and-Recreate local search algorithm demonstrates superior performance, achieving fleet size reductions of approximately 30% and flight distance savings of up to 40,000km. The framework successfully addresses critical limitations of existing deterministic approaches by ensuring operational strategies remain viable when confronted with heterogeneous passenger acceptance behaviors in real-world operations. The proposed integrated UAM system design framework incorporates innovative models, methodologies, and algorithms. We evaluated this framework through comprehensive experiments in large-scale real-world networks. The results demonstrate strong practical applicability and robust performance under various operational conditions.