A joint optimization-simulation framework for decision-making in complex networks

Shared Automated Electric Vehicles (SAEVs) represent a promising travel mode that integrates automation, electrification, and vehicle sharing, offering potential benefits in service efficiency and sustainability. Effective deployment of SAEV systems requires the coordinated optimization of long-term strategic planning decisions and short-term operational controls under uncertain travel demand, which remains a significant challenge. This paper proposes a joint optimization-simulation framework for SAEV system planning and operation under stochastic demand. A Mixed-integer Nonlinear Programming (MINLP) model is first formulated to maximize the operator’s profit while capturing charging dynamics and demand uncertainty. The model is decomposed into two interconnected levels: an upper-level optimization model that determines strategic decisions such as fleet size and vehicle distribution, and a lower-level simulation model that represents operational decisions, including vehicle relocation and movements. To explicitly account for demand uncertainty, a two-stage stochastic optimization formulation is developed, supported by a set of tailored solution algorithms. A case study based on the Austin metropolitan network, covering six counties and 2,210 traffic analysis zones, is conducted to evaluate the proposed framework. Results demonstrate that the integrated approach effectively improves system profitability and service performance while maintaining operational stability across different demand realizations, highlighting its suitability for large-scale SAEV system planning and operation.