Decarbonizing cold chain transport: Predictive energy management for refrigerated generator-axle trailers

Refrigerated semi-trailers are crucial for transporting perishable goods, but their conventional diesel-powered transport refrigeration units (TRUs) are a notable source of CO 2 emissions. A promising alternative involves a battery-electric TRU powered by an energy-harvesting generator axle. However, this configuration introduces a control trade-off: actively generating electrical energy via drag torque imposes an additional fuel penalty on the tractor, yet the battery must remain sufficiently charged at all times to guarantee uninterrupted cooling and prevent cargo spoilage. Resolving this trade-off is challenging because limited route foresight, uncertain traffic flow, varying cargo cooling demands, and fluctuating ambient temperatures all introduce substantial uncertainty into the optimization problem. This paper presents a novel, data-driven predictive energy management system (PEMS) to address it. The core of the strategy is a receding-horizon Forward Dynamic Programming (FDP) algorithm. Control decisions are based on a ”drag efficiency” metric, the ratio of generated electrical energy to additional fuel consumed, which is estimated in real-time via Sparse Gaussian Processes (SGP). To safely constrain the planner without relying on macroscopic route data, individually initialized Kalman Filters continuously forecast the microscopic recuperation potential and TRU cooling demand. Evaluations across standardized National Renewable Energy Laboratory (NREL) cycles and over 57 h of real-world transport drives in two European countries, providing different temperature and traffic circumstances, confirm the system’s robust performance under predictive uncertainty and unscheduled trip disruptions. Results demonstrate that the predictive controller reduces CO 2 emissions by an average of 11.6% compared to a conventional rule-based two-point controller on real-world routes.