Logistic Optimization for Multi Depots Loading Capacitated Electric Vehicle Routing Problem From Low Carbon Perspective

Abstract

In this paper, a multi depots capacitated electric vehicle routing problem where client demand is composed of two-dimensional weighted items (2L-MDEVRP) is addressed. This problem calls for the minimization of the transportation distance required for the delivery of the items which are demanded by the clients, carried out by a fleet of electric vehicles in several depots. Since the 2L-MDEVRP is an NP-hard problem, a heuristic algorithm combined variable neighborhood search algorithm (VNS) and space saving heuristic algorithm (SSH) is proposed. The VNS algorithm is used to solve the vehicle routing problem (VRP) sub-problem, and the SSH algorithm is used to solve the bin packing problem (BPP) subproblem. We propose the space saving heuristic to find the best matching solution between the next loading item and the feasible loading position. The SSH-VNS algorithm is tested by using benchmark instances available from the literature. The results show that the SSH-VNS algorithm has better performance compared with other published results for solving capacity vehicle routing problem (CVRP) and two-dimensional capacity vehicle routing problem (2L-CVRP). Some new best-known solutions of the benchmark problem are also found by SSH-VNS. Moreover, the effectiveness of the proposed algorithm on 2L-MDEVRP is demonstrated through numerical experiments and a practical logistic distribution case. In the last section, the managerial implications and suggestions for future research are also discussed.INDEX TERMS Vehicle routing problem, two-dimensional loading, electric vehicle, variable neighborhood search, space saving heuristic.

An Enhanced Delivery Route Operational Planning Model for Electric Vehicles

“…The EV charging scheduling problem, including the charging amount of energy and the charging option, and using a Mixed-Variable Differentiate Evolution algorithm, is presented [29]. A multi depots capacitated EV routing problem is proposed in [30], which is solved using a heuristic algorithm that combines Variable Neighborhood Search and Space Saving. In [31], a new sustainable parcel delivery model that compares EVs with conventional vehicles is studied, considering enconomic aspects and emissions.…”

“…On the contrary, in [22], the allocation of deliveries is predetermined, but the number of EVs is an optimization output. Furthermore, in [12]- [15], [18]- [20], [27], [30], the allocation and order of deliveries as well as the number of EVs used is determined including all or some aspects such as using simplified maps (Euclidean distances), partially preallocating deliveries or applying heuristic techniques. In contrast, in the PA, the aforementioned modeling features are explicitly represented by constraints and decision variables of the optimization process.…”

“…Additionally, unlike [13], [17], [19]- [27], [29], [30], [32], the PA considers the battery degradation costs by explicitly modeling the operation of EV batteries depending on their life cycle, efficiency and technology. In [12], [14]- [16], [18], [28], [31], the battery degradation cost is included either as a fixed cost or not taking into account life cycle, efficiency and technology [7].…”

“…Finally, from a methodological viewpoint, rather than using heuristic and metaheuristic techniques [12], [14]- [16], [18], [20], [22], [30], [31], the proposed model is formulated as a MILP model, which guarantees finite convergence to optimality [33]. Nevertheless, the PA is considered an NPhard problem due to the high number of binary variables and possible combinations.…”

Modelo De Programação Linear Inteira Mista Para a Otimização Do Roteamento De Frotas De Veículos Elétricos Com Serviços De Entrega / Mixed Whole Linear Programming Model for the Optimization of Electric Vehicle Fleet Routing With Delivery Services

“…alternativa de solução, a localização das estações de troca de baterias, cujo objetivo é determinar a estratégia de localização das estações de troca, usando heurísticas hibridas baseadas em Busca Tabu[Yang e Sun, 2015] e; em alguns casos, modelar as perdas no sistema de distribuição de energia elétrica mediante problemas de otimização não linear[Arango et al 2018]. Também, existem esforços em analisar problemas de tempo de entrega de mercadorias e de otimização de espaço dentro do veículo.Os primeiros são resolvidos por algoritmos genéticos em conjunto com ferramentas de lógica difusa, devido aos tempos não determinísticos que apresentam os percursos dos veículos[Mehlawat et al 2019], e os segundos são resolvidos por algoritmos de Busca em Vizinhança Variável em conjunto com heurísticas de economia de espaço[Zhu et al 2020].Todos esses modelos existentes na literatura têm como motivação resolver problemas específicos, alguns focados na redução do tempo computacional e outros na melhoria da qualidade da solução. Como se observa, ainda são escassos os trabalhos de métodos de programação exata para direcionar os diferentes PRV relacionados com aplicações reais e acadêmicas.…”