Data-Driven Vehicle Specification of Conventional and Electric Heavy-Duty Commercial Trucks for Improved Energy Consumption

In the heavy-duty commercial trucks sector, selecting the most energy-efficient vehicle can enable great reductions of the fleet operating costs associated with energy consumption and emissions. Customization and selection of the vehicle design among all possible options, also known as “vehicle specification,” can be formulated as a design space exploration problem where the objective is to find the optimal vehicle configuration in terms of minimum energy consumption for an intended application. A vehicle configuration includes both vehicle characteristics and powertrain components. The design space is the set of all possible vehicle configurations that can be obtained by combining the different powertrain components and vehicle characteristics. This work considers Class 8 heavy-duty trucks (gross combined weight up to 36,000 kg). The driving characteristics, such as the desired speed profile and the road elevation along the route, define the intended application. The objective of the optimization is to minimize the energy consumption. Therefore, a means to evaluate the energy consumption of the vehicles over the desired route is required. This work leverages real-world truck data and additional simulated data to train machine learning models (neural network and random forest). The models can capture the relationship between vehicle specification and driving characteristics with the vehicle’s energy consumption, with an R² correlation of 0.9986 for conventional trucks and R² of 0.9998 for electric trucks. The result of this work is the development of a framework that allows to explore a large design space and find the most efficient vehicle configuration for a specific route.