Decision-dependent Robust Charging Infrastructure Planning for Light-duty Truck Electrification at Industrial Sites: Scheduling and Abandonment

TL;DR

This work addresses electrifying industrial-site fleets of light-duty trucks by integrating a two-stage robust charging infrastructure planning framework with scheduling under abandonment. The approach combines a mixed-integer linear program for charger installation and a scheduling layer that accounts for waiting, abandonment, overnight charging, and range anxiety, with a decision-dependent uncertainty set for parking durations that is affine in charging decisions and linearizable to a MILP. A fix-and-optimize heuristic accelerates long-horizon planning, enabling near-optimal solutions (gap < 0.1%) on year-long data. A case study at an open-pit mining site with around 200 trucks demonstrates how overnight zones, uncertainty in parking duration, and charger mix (slow/fast) shape robust infrastructure deployments and operating schedules, highlighting the practical benefits of a year-round robust planning approach for decarbonizing industrial fleets.

Abstract

Many industrial sites rely on diesel-powered light-duty trucks to transport workers and small-scale facilities, which has resulted in a significant amount of greenhouse emissions (GHGs). To address this, we developed a two-stage robust charging infrastructure planning model for electrifying light-duty trucks at industrial sites. The model is formulated as a mixed-integer linear programming (MILP) that optimizes the charging infrastructure, selected from multiple charger types and potential locations, and determines opportunity charging schedules for each truck based on the chosen infrastructure. Given the strict stopping points and schedules at industrial sites, we introduced a scheduling problem with abandonment, where trucks forgo charging if their waiting times exceed a maximum threshold. We also further incorporated the impacts of overnight charging and range anxiety on waiting and abandonment behaviors. To represent the stochastic and heterogeneous parking durations of trucks, we constructed a decision-dependent robust uncertainty set in which parking time variability flexibly depends on charging choices. We applied the model in a case study of an open-pit mining site, which plans charger installations in eight zones and schedules a fleet of around 200 trucks. By decomposing the problem into monthly subproblems and using heuristic approaches, for the whole-year dataset, the model achieves an optimality gap of less than 0.1 % within a reasonable computation time under diverse uncertainty scenarios.

Figures (30)

• Figure 1: The studied multi-zone charging scheduling problem for the light-duty truck fleet; At any half-hour parking time slot $\mathcal{S}$ shown in light green, a truck can either choose to wait, abandon charging, or charge. The effective parking durations $pp$ are shown in dark green.
• Figure 2: Day-to-day carryover of waiting time in the same zone. For Truck 2 on Day 1, the first admissible slot initializes $w=0$. The next slot in the same zone accumulates to $w=\Delta t$, and carryover to Day 2 in the same zone yields $w=2\Delta t$.
• Figure 3: Zone-change resets of waiting time. For Truck 2, waiting grows from $w=0$ to $w=\Delta t$ across two slots in Zone 1. Transition to Zone 2 resets $w=0$, after which accumulation resumes in subsequent uncharged slots.
• Figure 4: The fix-and-optimize heuristic where fast-charging scheduling is pre-assigned and fixed based on the forecast high charging demands
• Figure 5: Identified eight charging zones and roads connected to eight charging zones (indicated as gray dashed lines)