An Enhanced Delivery Route Operational Planning Model for Electric Vehicles

Abstract

The recent incorporation of Electric Vehicles (EVs) in the worldwide delivery vehicle fleet has increased the importance of their operation in road networks. In particular, the growth of electronic commerce, while promoting the use of non-polluting energy sources, will make EVs play an important role in cities. Thus, an accurate characterization of the technical aspects of EVs and parcel delivery requirements in delivery route operational planning models is required. To that end, this paper presents an enhanced operational planning model for the route and charging of an EV fleet considering technical and economic real-world constraints, such as battery degradation, acceleration-and speed-dependent power consumption, tolls and penalty for delivery delay and non-fulfillment. Moreover, unlike some previous works, the delivery allocation and the number of EVs used are not predetermined, being rather outcomes of the optimization process. Additionally, the Proposed Approach (PA) allows EVs to pass through intersections more than once. The proposed model is formulated in terms of a series of intersection-and path-related decision variables, characterizing the State of Charge (SOC) of batteries and the navigation time. The resulting optimization problem is cast as an instance of Mixed-Integer Linear Programming (MILP). The model is implemented in the mathematical programming language GAMS and solved using the commercial solvers CPLEX and ODH-CPLEX. The model is tested on two intersection maps, including different types of roads, charging rates and delivery points. Results show the effectiveness of the PA over previously reported approaches in terms of the cost, energy and time associated with the resulting operating strategies. INDEX TERMS Battery degradation, delivery allocation, electric vehicle route planning, penalty costs, acceleration-and speed-dependent power consumption. NOMENCLATURE SETS AND INDICES

Electrical Vehicle Fleet Routing Accounting for Dynamic Battery Degradation

“…Intuitive extensions of the work herein presented include increasing the dimension of the problem, temporally and spatially, including more diverse simulation scenarios. Moreover, accounting for the charging schedules of the fleet of EVs [5] and associating the SoC of EVs across time periods are natural extensions. On top of that, combining predicted vehicle demand forecast and prioritizing the charging of EVs during periods with lower electricity prices are other improvements that have been investigated in the literature, but without considering the dimension of degradation effects in the fleet as herein presented.…”

“…Therefore, new factors come into play, such as the charging strategies [3] and infrastructure 1 , battery range, battery degradation, speed-and acceleration-dependent power consumption [5], and the altered waiting times at recharging stations caused by the longer charging time when compared to fueling internal combustion vehicles (ICVs) [4].…”

“…In parallel, recent research has also focused on how to address such factors, analyzing driving techniques for predictive energy management of EVs [6], or dynamic recommendations based on driving patterns [7]. However, even the most recent works [5] only addresses the degradation modeling within routing problems to a limited extent.…”

“…As such, recent works consider degradation models using approximations. In [5] it is proposed to use an operational planning model for route and charging of an EV fleet, for which the cost of battery degradation is included in the objective function. Its degradation model, however, is simplified in the sense that they use a general degradation cost based on average energy consumption (a parameter set a priori) across vehicles, regardless of their DoD.…”

“…Given the previously described literature gap, the main contribution of this paper is the proposal of a framework that accounts for battery degradation dynamically (i.e., preemptively) when calculating optimal routing strategies for a fleet of EVs. While it still uses a linear approximation of the DoD such as [5], it assigns a cost associated with the difference among DoD across multiple vehicles in the problem, avoiding that any one vehicle has a higher DoD that could result in increased degradation over time, leading to a result similar to [8]. 1 While optimal placing and sizing of charging infrastructure for EVs is another important topic, it is in itself, it constitute yet another research problem.…”