Thesis Defence: Data-Driven Robust Optimization for the Operational Design of Electric Vehicle Battery Swapping Stations Under Multi-Source Uncertainty

Quanyu Long, supervised by  Dr. Babak Tosarkani, will defend their thesis titled “Data-Driven Robust Optimization for the Operational Design of Electric Vehicle Battery Swapping Stations Under Multi-Source Uncertainty” in partial fulfillment of the requirements for the degree of Master of Applied Science in Mechanical Engineering.

An abstract for Quanyu Long’s thesis is included below.

Defences are open to all members of the campus community as well as the general public. Please email babak.tosarkani@ubc.ca to receive the Zoom link for this defence.

Abstract

The rapid growth of electric vehicles has increased interest in battery swapping stations (BSSs) as a practical alternative to conventional charging, particularly for reducing charging time and improving service convenience. However, effective BSS operation requires coordinated decisions on battery swapping, charging, inventory management, energy sourcing, and battery recovery under uncertain operating conditions. This thesis develops a decision-support framework for the operation of an electric vehicle BSS that integrates seller-supported battery supply, refurbishment-center returns, and hybrid energy sourcing from the grid, solar power, and wind energy. The research is conducted in two stages. First, a deterministic MILP model is developed to maximize station profit while satisfying operational, technical, and inventory constraints. Second, the deterministic model is extended into a data-driven robust optimization framework that captures multi-source uncertainty using principal component analysis (PCA) and robust kernel density estimation (RKDE) to construct a budgeted polyhedral uncertainty set from historical data.

The proposed framework is applied to a case study of a battery swapping station in Tongliao, China. Results from the deterministic model show that greater seller participation reduces unmet demand by improving battery inflow stability, while higher residual energy in returned batteries improves profitability and lowers emissions-related charging costs by reducing dependence on additional grid-based charging. The data-driven robust optimization results further show that the proposed PCA and RKDE framework provides a practical way to represent uncertainty while preserving computational tractability. The robust model supports more reliable charging, inventory, and energy-allocation decisions under adverse conditions, and the sensitivity analysis of the uncertainty-set parameters highlights the trade-off between conservatism and operational flexibility. This thesis provides an integrated and data-driven optimization framework for battery swapping station operation and provides managerial insights as well as practical implications for improving the reliability, economic performance, and sustainability of future electric vehicle energy service systems.