Joint Design for Electric Fleet Operator an Charging Service Provider: Understanding the Non-Cooperative Nature

TL;DR

This paper introduces a bilevel optimization framework modeling the strategic interaction between a charging service provider and a fleet operator in electric vehicle logistics, highlighting the noncooperative dynamics and proposing an iterative solution approach.

Contribution

It presents a novel bilevel mixed integer programming model capturing the noncooperative behavior of two entities in EV fleet planning, with a double loop solution method.

Findings

Numerical studies demonstrate the model's ability to reveal strategic interactions.

The framework can be generalized to other two-agent network planning problems.

The iterative solution approach effectively finds optimal decisions.

Abstract

This work proposes a new modeling framework for jointly optimizing the charging network design and the logistic mobility planning for an electric vehicle fleet. Existing literature commonly assumes the existence of a single entity, the social planner, as a powerful decision maker who manages all resources. However, this is often not the case in practice. Instead of making this assumption, we specifically examine the innate noncooperative nature of two different entities involved in the planning problem. Namely, they are the charging service provider (CSP) and the fleet operator (FO). To address the strategic interaction between entities, a bilevel mixed integer program is formulated, with the CSP and FO problems expressed in the upper and lower levels respectively, in a joint decision making process. These decisions involve the CSP infrastructure siting, sizing, substation capacity upgrades, the FO fleet composition, vehicle routing, charging, and delivery assignment. To relieve computational burdens, we utilize a double loop solution architecture to iteratively reach optimality. We conduct detailed numerical studies on a synthesized small network and the simulation results reveal the unique aspects of this two entity framework. This modeling perspective can be generalized to other system design problems with two interacting agents planning and operating resources across networks.