Smarter EV Fleet Routes Designed to Harness Local Solar Surplus

A research team in Japan has created an innovative planning model that aligns electric delivery vehicle (EDV) operations with local solar power generation, offering logistics companies a roadmap to reduce emissions and make use of surplus renewable energy.

Image: Waseda University, Journal of Energy Storage

A new approach to EV routing

The study, conducted at Waseda University, introduces an advanced electric vehicle routing problem (EVRP) framework. At its core is a machine learning tool — a random forest regression model — which forecasts when and where excess solar power will be available near charging stations on the following day.

“By integrating this prediction into delivery routes, we can actively lower CO₂ emissions while supporting the local use of renewable energy,” explained lead researcher Ryoji Miyabe.

Unlike conventional models that mainly focus on distance or depot-based charging, the new system incorporates both private and public charging stations. This transforms EVs into flexible, mobile units that can absorb solar surplus throughout the city.

How the model works

The approach is tailored for mid-mile delivery networks where goods are transported from distribution depots to retail outlets. Electric trucks can recharge either at the depot or at strategically chosen fast-charging stations (50 kW output) along their routes.

Key inputs include customer locations, service time windows, and vehicle constraints. The model then combines solar forecasts and smart meter data to estimate available PV surplus. A mixed-integer linear programming (MILP) optimisation step follows, balancing service requirements with the goal of minimising carbon emissions.

The final output is a complete daily mission plan: routes, customer stops, charging times, and charging locations.

Predicted surplus PV for May 17, 2023

Image: Waseda University, Journal of Energy Storage

Case study in Utsunomiya City

To test the concept, the team simulated deliveries in Utsunomiya City, central Japan. The scenario involved three Mitsubishi FUSO eCanter trucks, each with a 41 kWh battery and an 80 km range, one logistics depot, 14–16 customer sites, and five charging stations.

Over a simulated week in each season, with two shifts per day, 56 separate delivery cases were analysed.

Results showed that the proposed low-carbon EVRP significantly cut CO₂ emissions compared to conventional approaches:

• 16.6% lower emissions on average than a standard VRP (25.7 vs. 30.8 kg-CO₂ per case)

• 21.4% lower emissions than a typical EVRP without carbon considerations (25.7 vs. 32.7 kg-CO₂ per case)

Emission reductions varied seasonally: 22.7% in spring, 17.9% in summer, 19.6% in autumn, and 7.2% in winter. The most dramatic saving came during one morning shift in May 2023, where emissions were slashed by 66.1% compared to a standard VRP.

Charging strategy over distance

An important insight was that cutting emissions depends more on when and where vehicles charge than on minimising driving distance. In some cases, longer routes were chosen, but charging at the right locations with abundant solar surplus outweighed the extra travel energy use.

“Our research shows that charging strategy is the real key to decarbonisation,” Miyabe concluded.

The findings were published in the Journal of Energy Storage under the title “Low-carbon routing and charging planning for electric freight trucks utilising local surplus solar power.”