Solving the battery swap station location-routing problem with a mixed fleet of electric and conventional vehicles using a heuristic branch-and-price algorithm with an adaptive selection scheme
Introduction
In China, the successful development of electronic commerce leads to the substantial growth of the transportation sector in recent years. From January 2020 to September 2020, the express delivery companies delivered 56.14 billion items, a year-on-year increase of 27.9% in China (State post bureau of the people’s republic of China, 2020). However, increasing delivery with conventional vehicles (CVs) causes environmental problems related to polluting emissions, noise, and congestion. At present, more and more laws and regulations are passed to control the emission of gases from CVs. Developing more environmentally friendly transportation modes, such as electric vehicles (EVs), is becoming necessary. EVs can provide an alternative to CVs. However, EVs face many limitations in real-world applications, such as the limited driving range, significantly considerable recharging time, and the availability of charging infrastructure. To tackle these issues, logistics companies open their battery swap stations (BSSs). The nearly depleted battery is replaced by a fully charged one at a BSS (Yang and Sun, 2015). The process is called battery swapping. The battery swapping process is auspicious. It usually takes less than 10 min and is much faster than the recharge with the fastest technology at a charging station (Kim, 2011). The battery swapping improves vehicles' productivity. Also, the depleted batteries can be recharged together at night at a discounted electricity price, decreasing charging costs (United Nations Environment Programme, 2009).
To render EVs more competitive, we should make simultaneous decisions, including (i) the BSS to be opened from a set of BSSs candidate locations and (ii) EVs' routes to be determined to minimize the total transportation cost and the cost for establishing the BSSs. These decisions combine the strategic level (location of the BSSs) and tactical level (routing of EVs). They integrate locations for electric vehicle battery swap stations and the routing problem (BSS–EV–LRP) introduced by Yang and Sun (2015).
In real-world applications, CVs are mainly used. A complete transition to an electric fleet is a very challenging task. Therefore, most logistics companies gradually introduce EVs into their existing CV fleet (Goeke and Schneider, 2015, Erdelić and Carić, 2019). A mixed fleet of EVs and CVs instead of a complete EV fleet can extend the existing BSS–EV–LRP. EVs and CVs have different characteristics in maximal driving range, operating cost, load capacity, and charging activities. Location and route planning for a mixed fleet should consider the tradeoff between EVs and CVs. Moreover, the existing BSS–EV–LRP models ignore time windows wherein the customer has to be visited. However, it is very common for customers to have their time window requirements in a real-world industrial application.
This research extends the existing work on the BSS–EV–LRP, including several crucial aspects such as a mixed fleet of EVs and CVs and time windows. Specifically, we consider the BBS locations (L), routing problem (RP) with time windows (TW), and a mixed fleet (MF), and that is called the BSS–MF–LRPTW. The proposed BSS–MF–LRPTW determines: (a) the location of the BSSs, (b) the assignment of customers to EVs and CVs, (c) the allocation of EVs to BSSs, and (d) the routes from the depot to customers considering the driving range and the load capacity limitation of EVs and CVs to minimize the total transportation cost and the cost for establishing the BSSs.
The BSS–EV–LRP can be classified as a location-routing problem (LRP), that is proven NP-hard (Hof et al., 2017). By adding the complexity of a mixed fleet and time windows requirements into the BSS–EV–LRP, the resulting BSS–MF–LRPTW is more challenging to be solved than the BSS–EV–LRP. Therefore, a proper and effective method must be developed for the proposed BSS–MF–LRPTW. The existing literature on BSS–EV–LRP proposes heuristic approaches, such as adaptive large neighborhood search (ALNS). However, none of them attempt to obtian the optimal solution based on the exact algorithm. Therefore, this study aims to introduce the exact algorithm, branch-and-price (B&P), to obtain a high-quality solution to the BSS–MF–LRPTW.
Given that the proposed BSS–MF–LRPTW is much more complicated than general LRPs, this research proposes a methodology. It embeds heuristic strategies and an adaptive selection scheme in the B&P framework to radically reduce computational effort while finding a high-quality solution, especially in large-sized instances. Thus, the proposed methodology can obtain the exact solution to small-scale instances with a significantly shorter run time than IBM CPLEX Optimizer. For all the tested cases, it can also find a better solution to large-scale instances more quickly than existing heuristics-ALNS. That makes the solution more practical in real-world logistics applications.
The main contributions of this research are threefold: (a) the BSS–EV–LRP is extended to the BSS–MF–LRPTW, which is motivated by a real-life logistics application. An integer program formulation is developed. It simultaneously determines the routing and locations of BSSs given the time window constraints and a mixed fleet of EVs and CVs; (b) A heuristic branch-and-price algorithm with an adaptive selection scheme (HBP-ASS), which combines exact and heuristic policies, is developed. The HBP-ASS introduces a heuristic version of the dynamic programming algorithm by developing seven heuristic operators in the label extension and dominance procedure. An adaptive selection scheme is also presented to choose the appropriate operator. The proposed heuristic version of dynamic programming can effectively generate new attractive columns and help achieve LP optimality or find a near-optimal solution within a short time. The proposed methodology also extends the elementary shortest path problem with resource constraints (ESPPRC) when solving the subproblem and (c) Effects of the operating cost of the BBS, the size of the mixed fleet, and the battery swapping time on the total cost are examined. This study may help logistics managers gain insight into applications in this problem.
The remainder of this paper is organized as follows. Section 2 provides an overview of the literature on the electric vehicle routing problem, electric vehicle routing problem with a mixed fleet, and electric location-routing problem. Section 3 presents an integer programming model for BSS–MF–LRPTW. Then the model is reformulated to the master problem and pricing problem. Section 4 provides a heuristic branch and price with an adaptive selection scheme (HBP-ASS) for the proposed model. Section 5 presents the performance of HBP-ASS on small-scale and large-scale instances. Managerial insights are also provided. Section 6 concludes the paper with a summary of the main findings.
Section snippets
Literature review
This section positions the BSS–MF–LRPTW in the context of electric vehicle routing problem (EVRP), electric vehicle routing problem with a mixed fleet (E-VRPMF), and electric location-routing problem (ELRP). That highlights some key features with the issues addressed in this research.
Model formulation
The proposed problem is modeled in this section. In Subsection 3.1, we present the problem description for the BSS–MF–LRPTW. In Subsection 3.2, we model the proposed BSS–MF–LRPTW by developing an integer programming formulation. As there are many variables and complicating constraints, the proposed problem is reformulated to have the master problem and the subproblem in Subsection 3.3.
Heuristic branch and price with adaptive selection scheme (HBP-ASS)
The B&P is the hybrid of branch-and-bound and column generation, and it can solve huge integer programming models (Barnhart et al., 1998). In recent years, B&P has established itself as an effective solution technique for some VRPs (e.g., Ozbaygin et al., 2017, Reihaneh and Ghoniem, 2019). However, existing studies have not considered the routing problem for a mixed fleet with EVs and CVs. Also, the B&P is used for several LRPs (e.g., Akca et al., 2009, Ponboon et al., 2016). However, the
Computational results
In this section, the computational performance of the proposed HBP-ASS algorithm is examined based on Python 3.6. All runs were conducted on a Windows 10 Professional 64-bit operating system with an Intel Core i5-8400 CPU and 8G RAM.
The instances are selected randomly from the datasets proposed by Schneider et al. (2014) based on Solomon (1987). According to the geographical distribution of the customer locations, the instances can be divided into three classes: random customer distribution
Conclusion
In this paper, the BSS–MF–LRPTW is presented as a novel approach for strategic planning and operational management of a mixed fleet of EVs and CVs. Compared to existing studies, this problem includes time windows, considers a mixed fleet based on real-world logistics applications, and simultaneously determines BSSs’ locations and vehicle routing plans. This paper also develops an HBP-ASS algorithm combining exact and heuristic policy for the proposed BSS–MF–LRPTW. The HBP-ASS algorithm
CRediT authorship contribution statement
Yanru Chen: Conceptualization, Investigation, Methodology, Writing – original draft, Funding acquisition. Decheng Li: Methodology, Formal analysis. Zongcheng Zhang: Methodology, Software. M.I.M. Wahab: Investigation, Writing – review & editing. Yangsheng Jiang: Investigation, Methodology, Supervision, Writing – review & editing.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
This work was supported by the National Key R&D Program of China [grant number 2018YFB1601402] and the National Nature Science Foundation of China [grant number 71771190].
References (42)
- et al.
Energy consumption estimation integrated into the electric vehicle routing problem
Transportation Research Part D: Transport and Environment
(2019) - et al.
A survey on two-echelon routing problems
Computers & Operations Research
(2015) - et al.
A survey of variants and extensions of the location-routing problem
European Journal of Operational Research
(2015) - et al.
Improved formulations and algorithmic components for the electric vehicle routing problem with nonlinear charging functions
Computers & Operations Research
(2019) - et al.
Routing a mixed fleet of electric and conventional vehicles
European Journal of Operational Research
(2015) - et al.
Routing a mix of conventional, plug-in hybrid, and electric vehicles
European Journal of Operational Research
(2019) - et al.
The electric fleet size and mix vehicle routing problem with time windows and recharging stations
European Journal of Operational Research
(2016) - et al.
Solving the battery swap station location-routing problem with capacitated electric vehicles using an AVNS algorithm for vehicle-routing problems with intermediate stops
Transportation Research Part B: Methodological
(2017) - et al.
The two-echelon capacitated electric vehicle routing problem with battery swapping stations: Formulation and efficient methodology
European Journal of Operational Research
(2019) - et al.
Electric vehicle routing problem with non-linear charging and load-dependent discharging
Expert Systems with Applications
(2020)
Partial recharge strategies for the electric vehicle routing problem with time windows
Transportation Research Part C: Emerging Technologies
The time-dependent electric vehicle routing problem: Model and solution
Expert Systems with Applications
The green mixed fleet vehicle routing problem with partial battery recharging and time windows
Computers & Operations Research
The electric vehicle routing problem with nonlinear charging function
Transportation Research Part B: Methodological
A branch-and-price algorithm for the vehicle routing problem with roaming delivery locations
Transportation Research Part B: Methodological
The electric vehicle routing problem with energy consumption uncertainty
Transportation Research Part B: Methodological
Branch-and-price algorithm for the location-routing problem with time windows
Transportation Research Part E: Logistics and Transportation Review
A survey of recent research on location-routing problems
European Journal of Operational Research
The electric vehicle routing problem with time windows and synchronised mobile battery swapping
Transportation Research Part B: Methodological
A branch-and-price algorithm for a vehicle routing with demand allocation problem
European Journal of Operational Research
The electric location routing problem with time windows and partial recharging
European Journal of Operational Research
Cited by (26)
Dealing with congestion in the optimization of locating single-server battery swapping stations
2024, Information SciencesPlanning decentralized battery-swapping recharging facilities for e-bike sharing systems
2024, Sustainable Cities and SocietyOptimal design of pure battery electric bus system on the grid network
2024, Transportmetrica A: Transport ScienceJoint impacts of ambient temperature, path flexibility, and hybrid fleet on energy-conserving supply chain network
2023, Journal of Cleaner ProductionThe home health care location-routing problem with a mixed fleet and battery swapping stations using a competitive simulated annealing algorithm
2023, Expert Systems with Applications
- 1
ORCID ID: 0000-0003-4137-0721.