Introduction
Realising green electrification in road freight is pivotal for net-zero, yet most evidence and tools were built for passenger EVs rather than heavy-duty commercial fleets whose choices hinge on route structure, depot topology, payload–range trade-offs, driver hours, and strict service windows.
This project uses an AI-enabled, Smart Predict-then-Optimize (SPO) pipeline to model and ultimately understand how electrification alters effective (revealed) logistics demand: origin–destination flows, shipment size and timing, reliability requirements, and carrier/mode choice, under technology and policy scenarios, then differentiates through fleet sizing and electric vehicle routing with charging to produce prescriptions that minimise cost, meet service levels, and reduce emissions.
AI improves in three ways that matter for electrification:
it fuses high-dimensional, nonstationary data (ELD/telematics, TMS orders, energy prices, congestion, weather, charger availability) to learn nonlinear thresholds and interactions that drive shipper behaviour
it addresses endogeneity between price, service quality, and demand with causal ML rather than assuming fixed coefficients
via SPO, it trains forecasts for the downstream objective (cost, reliability, emissions) instead of generic statistical fit, delivering decisions that remain robust under uncertainty and distribution shift (Elmachtoub & Grigas, 2022; Bertsimas & Kallus, 2020).
The work concentrates on three questions.
First, the decision-focused value:
By how much does SPO-based, AI demand modelling improve total logistics cost, service reliability, and emissions versus parametric baselines (Elmachtoub & Grigas, 2022)?
Second, the optimal fleet–charging mix under uncertainty:
What combination of battery-electric trucks and charger types (depot, opportunity, megawatt), with what siting and schedules, minimises system cost while meeting service levels when demand is stochastic and temperature/grade affect usable range, benchmarked to current total-cost-of-ownership and adoption evidence for zero-emission HDVs (Basma et al., 2021–2023)?
Third, policy efficiency:
Which instruments, capex subsidies, LCFS-style credits, carbon fees, access/toll exemptions, deliver the largest electrification gains per public pound once operational responses are optimised with SPO and demand reallocation is taken into account, and how sensitive are results to regional energy prices and corridor densities (International Energy Agency, 2023 & 2024)?
Methodology and Validation
Methodologically, we will learn probabilistic, lane-level demand and service-elasticity models from telematics and order data; generate counterfactual demand under technology and policy scenarios; and embed these forecasts in multi-period fleet composition and electric VRP with time windows and charging, capturing dwell, degradation, payload penalties, and ambient conditions.
Strategic analysis will evaluate competition and cooperation among carriers and infrastructure providers—shared depots, megawatt-charging hubs, and data pooling—linking micro-level routing costs to macro TCO and adoption timing. Validation will combine held-out corridors, event studies (e.g., depot openings, low-emission zone expansions), and reconciliation to ICCT benchmarks and IEA outlooks for external validity. The results will be converted into better routing and infrastructure choices, and identify the most cost-effective policies to accelerate heavy-duty fleet decarbonisation.
References:
Basma, H., Saboori, A., & Rodríguez, F. (2021, November). Total cost of ownership for tractor-trailers in Europe: Battery electric versus diesel. International Council on Clean Transportation.
Basma, H., Buysse, C., Zhou, Y., & Rodríguez, F. (2023, April). Total cost of ownership of alternative powertrain technologies for Class 8 long-haul trucks in the United States. International Council on Clean Transportation.
Bertsimas, D., & Kallus, N. (2020). From predictive to prescriptive analytics. Management Science, 66(3), 1025–1044.
Elmachtoub, A. N., & Grigas, P. (2022). Smart “Predict, then Optimize”. Management Science, 68(1), 9–26.
International Energy Agency. (2023, April). Global EV Outlook 2023: Catching up with climate ambitions. IEA.
International Energy Agency. (2024, April). Global EV Outlook 2024: Moving towards increased affordability. IEA.