A cradle-to-grave life cycle assessment focused on end-of-life (EoL) was conducted in this study for three configurations of a light-duty commercial vehicle (LDCV): diesel, compressed natural gas (CNG), and battery electric vehicle (BEV). The aim is to investigate the impact of recycling under two EoL scenarios with different allocation methods. The first is based on the traditional avoided burden method, while the second is based on the circular footprint formula (CFF) developed by the European Commission. For each configuration, a detailed multilevel waste management scheme was developed in compliance with the 2000/53/CE directive and ISO22628 standard. The results showed that the global warming potential (GWP) impact under the CFF method is significantly greater when compared to the avoided burden method because of the A-parameter, which allocates the burdens and benefits between the two connected product systems. Furthermore, in all configurations and scenarios, the benefits due to the avoided production of virgin materials compensate for the recycling burdens within GWP impact. The main drivers of GWP reduction are steel recycling for all of the considered LDCVs, platinum, palladium, and rhodium recycling for the diesel and CNG configurations, and Li-ion battery recycling for the BEV configuration. Finally, the EoL stage significantly reduces the environmental impact of those categories other than GWP.

This paper introduces a novel physics-informed exploration strategy for a deep reinforcement learning (DRL)-based energy management system (EMS), specifically targeting the challenge of dealing with constrained action sets. RL-based controllers for electrified vehicle energy management systems have faced obstacles stemming from the selection of infeasible actions, obstructing their practical deployment. The absence of a mechanism for assessing control action feasibility prior to application has compounded this issue, primarily due to the model-free nature of RL-based controllers. Adding a safety layer to the RL-based controller addresses the abovementioned issue, but this often results in suboptimal policies and necessitates an in-depth understanding of the powertrain. Alternatively, theoretical remedies incorporate penalty terms into the immediate reward function to manage infeasible conditions. However, this approach can slow down the training process as the agent learns to avoid infeasible actions. To surmount these challenges, this paper introduces a novel physics-informed exploration strategy, coupled with prioritized experience replay, enabling the agent to swiftly learn to avoid selecting infeasible control actions without the need for a separate safety layer. Real-time simulation results highlight the superior performance of the proposed DRL-based controller over the baseline DRL-based controller with a safety layer, particularly in terms of overall fuel consumption.

In recent decades, the automotive industry has moved towards the development of advanced driver assistance systems to enhance the comfort, safety, and energy saving of road vehicles. The increasing connection and communication between vehicles (V2V) and infrastructure (V2I) enables further opportunities for their optimisation and allows for additional features. Among others, vehicle platooning is the coordinated control of a set of vehicles moving at a short distance, one behind the other, to minimise aerodynamic losses, and it represents a viable solution to reduce the energy consumption of freight transport. To achieve this aim, a new generation of adaptive cruise control is required, namely, cooperative adaptive cruise control (CACC). The present work aims to compare two CACC controllers applied to a platoon of heavy-duty electric trucks sharing the same linear spacing policy. A control technique based on reinforcement learning (RL) algorithm, with a deep deterministic policy gradient, and a classic linear quadratic control (LQC) are investigated. The comparative analysis of the two controllers evaluates the ability to track inter-vehicle distance and vehicle speed references during a standard driving cycle, the string stability, and the transient response when an unexpected obstacle occurs. Several performance indices (i.e., acceleration and jerk, battery state of charge, and energy consumption) are introduced as metrics to highlight the differences. By appropriately selecting the reward function of the RL algorithm, the analysed controllers achieve similar goals in terms of platoon dynamics, energy consumption, and string stability.