This paper proposes a hybrid methodology that integrates Model-Based Systems Engineering (MBSE), discrete-event simulation, and simulation-based optimization to support the design and analysis of complex systems. Grounded in the Arcadia method and documented through UML diagrams, the approach offers a structured framework to define operational needs, mission objectives, and architectural alternatives before implementation. To validate the methodology, it is applied to a last-mile delivery system involving autonomous aerial and ground vehicles operating in a gated residential environment. The case study progresses through operational modeling, logical architecture, simulation, and optimization phases. A discrete-event simulation model is developed in Python using SimPy, incorporating a non-homogeneous Poisson process to capture realistic demand patterns. Key performance indicators, such as delivery delay and energy consumption, are measured and used to compare dispatching strategies. In the final phase, a Simulated Annealing algorithm is used to explore optimal fleet compositions. Results demonstrate that the proposed approach can reduce average wait time by over 99% and energy consumption by 12.6% when compared to a baseline configuration. The methodology ensures traceability between stakeholder needs, design models, and performance outcomes, supporting informed decision-making in early design stages. The proposed framework is generalizable and can be extended to other domains characterized by operational uncertainty and complex architectural trade-offs. It contributes to bridging the gap between high-level system architecture and quantitative performance evaluation through a unified, model-driven workflow.