Event Details

Dr Ramin Raeesi will present a seminar to the Management Science Department

Abstract: Energy, fuel or emissions estimation have been increasingly incorporated into the recent variants of the vehicle routing problems (VRPs) to plan energy-efficient or emissions-aware routes for conventional and alternative fuel freight vehicles. In Electric VRPs with Time Windows (EVRPTW), energy consumption estimation is yet more crucial and deemed an integral part of routing with a significant impact on the practical optimality (if not feasibility) of the solutions due to the limited driving range of Electric Commercial Vehicles (ECVs). The real impact of energy estimation inaccuracies on routing decisions is, however, unclear, as is the trade-off between improving estimation accuracy and increasing model complexity and intractability. This paper quantifies the real cost and impact of energy estimation inaccuracy on EVRPTWs by analysing the influence of various factors, including variable speed, payload, acceleration/deceleration rates, rolling resistance, road slope, and weather conditions. Through extensive realistic testing, the expected inaccuracies from ignoring these variables—both independently and collectively—are evaluated, and their effects on routing feasibility and optimality are examined. Furthermore, a new approach to robustness is proposed based on some intuitive results used within a bi-objective integer programming optimisation framework to mitigate the impact of estimation inaccuracies. Unlike traditional robust optimisation methods, which often hedge against worst-case scenarios, this approach provides a flexible decision-making framework. It allows decision-makers to select solutions from an efficient frontier with guaranteed optimality or an optimality gap, without the need for an overly complex energy estimation model. By balancing robustness and computational efficiency, this methodology enables more practical and adaptable routing decisions for electric commercial vehicles.