Reefers are refrigerated containers commonly used for transporting perishable goods such as meat, fish, vegetables, and fruit. Due to the growth in reefer usage, energy consumption of reefers at container terminals grows significantly to 40% of the total energy consumption at terminals. When reefers are connected to the electricity grid, peaks in energy consumption can lead to high costs. A specific grid capacity is reserved for the container terminal, exceeding the reserved capacity will result in high energy costs. Therefore, reducing peak energy consumptions of reefers at container terminals will reduce energy costs and reduce the carbon footprint of reefer transport.

Previous research has taken highly technical views on the problem. Refrigeration and insulation techniques research result in efficient reefers, but innovations in this area seem to reduce. However, Lukasse et al. (2013); Barzin et al. (2015. 2016) have shown that advanced control systems can reduce the energy consumption of individual reefers significantly. More recent and promising studies of van Duin (2016) focus on the broader picture of multiple reefers connected simultaneously at the container terminal. Filina and Filin (2008) have taken a process-based view by investigating factors that lead to power-out moments within the supply chain. Meanwhile, the root-cause of energy consumptions remains un-researched.

A sequential multiple regression analysis with backwards feature selection is performed. A model is found in which the number of arriving reefers, dwell time, plug-in temperature, insulation value, and cargo type are found to be significant. The developed model shows that the number of arriving reefers explains 76,6% of the variance, dwell time 4,6%, cargo type 1,1%, thermal insulation 0,3% and the delta plug-in temperature 0,4%.

Finally, an improvement for the dwell time is developed as this produces the highest yield. It is suggested that long stay reefers can only be targeted if the revenue of the container terminal does not reduce. Therefore, two Revenue Management schemes are proposed: a complex dynamic pricing scheme and peak pricing. Dynamic pricing requires fixed capacity, perfect knowledge of demand, and price sensitivity. Peak pricing scheme is less fine-tuned and does not require perfect demand knowledge. Additionally, price sensitivity is not a requirement when using peak pricing as the peak price an incentive for fast collection of the reefer. Considering the perfect demand knowledge requirement, it is attempted to predict the energy consumption using only data known to the terminal prior to the ships arrival. Using a neural network, it is attempted to predict the dwell time with the purpose of using this in the energy consumption model. The neural network has shown that the dwell time cannot be predicted with the available data. This reduces the prediction accuracy of the model. Additionally, Dutch importers of perishable goods indicate that demand is not price-sensitive. Therefore, a peak pricing scheme is advised to provide an incentive reduce the dwell time. An effective implementation of peak pricing provides 5,5% to 11,6% reduction in energy consumption. Estimations show that the energy reduction is achieved without decreasing revenue of the terminal.