Energy Consumption Assessment of Fuel Cell Heavy-Duty Vehicles for Long-Haul Applications: A Comparative Study Using VECTO and GSP & AI-Assisted Regulatory Modelling

Abstract

The decarbonisation of long-haul freight transport has increased interest in fuelcell heavy-duty vehicles, while simultaneously placing greater emphasis on reliable and comparable energy consumption assessment methods. Within the European regulatory framework, the VECTO simulation tool is used to evaluate energy consumption and emissions of heavy-duty vehicles. However, its level of abstraction and regulatory focus differ from those of engineering-oriented simulation environments commonly used during vehicle development. This thesis presents a vehicle-level energy consumption assessment of a fuel-cell heavy-duty vehicle for long-haul applications using VECTO and compares the results with simulations performed in the Global Simulation Platform (GSP), an internal MATLAB-based simulation platform. A VECTO-compliant fuel-cell heavy-duty vehicle model was developed based on an internal prototype and evaluated under a representative long-haul mission profile. The results demonstrate that the VECTO declaration-mode model can reproduce the prescribed mission with stable longitudinal dynamics and physically reasonable energy flows. While VECTO and GSP exhibit similar mission-level trends in battery power and state-of-charge behaviour, systematic differences are observed in transient response and absolute operating levels. These differences are shown to be consistent with the backward regulatory simulation approach employed by VECTO and the forward, engineering-oriented simulation philosophy of GSP. In addition to the simulation study, this thesis investigates the use of an AI-assisted Agentic Retrieval-Augmented Generation (RAG) framework to support the VECTO modelling workflow. The framework facilitates interpretation of regulatory documentation, parameter constraints, and XML schema definitions, reducing ambiguity during model setup while preserving engineering judgement. The results indicate that AI-assisted methods can effectively support complex regulatory modelling tasks and improve efficiency and robustness in simulation-based energy assessment workflows.