Mixed multi-echelon location routing problem with differentiated intermediate depots
Introduction
Urban logistics covers the circulation of goods related to national economy and people's livelihood within or between cities. It is defined as finding efficient and effective ways to transport goods in urban areas and considering the adverse effects on congestion, safety and environments (Savelsbergh & Van Woensel, 2016). Express industry, which supports parcel delivery, plays an important role in urban logistics. Since 2000, the fast-growing express industry has spawned many express companies, such as SF Express and YTO, and these companies have built their own logistics networks. However, as differences in the regional distribution of e-commerce packages have increased in recent years, express companies have severe misallocation in logistics facilities, causing high costs. For instance, in some high-demand areas, express companies opened some small-scale intermediate depots (IDs), which can be replaced by a large-scale ID, and in some low-demand areas, some oversize IDs can scale down. It is urgent to redesign their logistic networks.
One important feature of the express industry is the significant difference in the volume of E-commerce parcels in different function areas of the metropolises. For example, the number of parcels in residential areas, shopping malls, and office buildings are greater than that in recreation areas and some factory districts. The high-demand areas usually are intensively populated, leading to high housing price and less available space, while the opposite is true in the low-demand areas. Thus, in high-demand areas, express companies prefer to open large-scale IDs outside instead of small-scale IDs. Low-demand areas are on the contrary. This leads an application prospect of mixed multi-echelon (MME) network. The MME network consists of limited urban distribution centers (UDCs), several differentiated IDs, i.e., large intermediate depots (LIDs), medium intermediate depots (MIDs) and small intermediate depots (SIDs), and a large number of terminals. Different types of IDs function differently. In high-demand areas, SIDs are preferred as transshipping parcels to terminals and several LIDs are built in outer layer as sorting parcels from UDCs and transshipping them. In low-demand areas, express companies usually use MIDs to transship parcels. Therefore, parcels are first transshipped from UDCs to LIDs, then transferred to SIDs, and finally delivered to terminals in high-population-density areas, while they are transshipped from UDCs to MIDs and then delivered to terminals in low-demand areas.
A mixed multi-echelon location routing problem (MME-LRP) is proposed to redesign a high-quality logistics network where the location and the number of UDC and terminals are given. Most giant express companies possess one UDC, the location of which is known. Locations of terminals depend on characteristics of end-users, such as distance to the nearest terminal, population densities, parcel demands (Zhao et al., 2018). MME-LRP has two kinds of decisions: location and routing. Three types of IDs, i.e., LIDs, MIDs, and SIDs, should be chosen from ID candidates, where LIDs and MIDs have a joint candidate set, which means every location in the set can be used as either LID or MID, and SIDs have a separate candidate set. The problem involves three levels of vehicle routes: first-level routes from the UDC to LIDs and MIDs, second-level routes from LIDs to SIDs, and third-level routes from all types of IDs to terminals. As some terminals are close to LIDs, parcels can flow between LIDs and terminals directly. For example, in Fig. 1, square and circles are UDC and terminals, and they are predetermined. Rhombuses are joint candidate set and each one can be used as LID or MID. Triangles are seen as separate candidate set that they can only be used as SIDs. First-, second-, and third-level routes are named as primary, secondary and tertiary routes respectively. They are represented by three kinds of lines.
The main contributions of this paper are summarized as follows:
- (I)
We extend MME-LRP presented by Winkenbach et al. (2016). Winkenbach et al. (2016) put forward an MME network combining single-echelon and two-echelon, whereas our MME network is a two-echelon and three-echelon mixed network considering differentiated IDs. Three types of IDs in the MME urban logistics network perform distinct functions: LIDs undertake the sorting and transshipment functions between the UDCs and SIDs. MIDs have the same functions but work between the UDCs and terminals. SIDs are located between the LIDs and terminals as transshipment points. In the problem, LIDs and MIDs are chosen from a joint candidate set while SIDs have a separate candidate set. Several terminals are allowed to be allocated to LIDs directly.
- (II)
We put forward a hybrid heuristic algorithm based on iterated local search and intensified by path relinking (HH-ILS-PR) to solve the MME-LRP. The HH-ILS-PR extends and improves the MS-ILS + PR (Nguyen et al., 2012a) that is originally proposed to solve 2E-LRP. We design two characteristic mechanisms to exchange status between two open-type for each used joint candidate and allow direct assignment of terminals to LIDs. Moreover, we embed a new perturbation in our algorithm.
- (III)
We put forward some suggestions for express companies to redesign and optimize their urban logistics networks. In areas with large-volume parcels, it is more cost efficient to transship parcels through LIDs and SIDs successively, while in areas with small volume parcels, transshipment through MIDs is more suitable. Additionally, LIDs can cover terminals directly when these terminals are close to LIDs.
The paper is organized as follows. Section 2 briefly surveys literature on relative problem. In Section 3, the problem is described and the MILP formulation is presented. A hybrid heuristic is proposed in Section 4. The computational results are reported and analyzed in Section 5. Finally, concluding remarks are given in Section 6.
Section snippets
Literature review
The MME-LRP is closely related to capacitated location routing problem (CLRP) and two-echelon location routing problem (2E-LRP). CLRP was proposed sixty years ago (von Boventer, 1961). It consists of opening one or more depots from a given set of candidate locations predefined, allocating customers to opened depots considering capacity, and designing a number of routes to suffice demands of customers. Several papers presented exact algorithms to solve the problem. Baldacci et al. (2011)
Problem description and mathematical formulation
The following assumptions apply to the MME-LRP:
- (I)
This paper optimizes the MME network based on planning horizon, which maybe one day, half day or several hours. In addition, planning horizon of express companies is given.
- (II)
Parcel demand from each terminal can be predicted by applying machine-learning tools or other advanced methodologies. This paper only considers parcels to be delivered (not the ones to be picked up) because the volume of the former is significantly larger than that of the latter
A hybrid heuristic algorithm for MME-LRP
MME-LRP optimization is apparently NP-hard. Exact algorithms are not feasible when instances reach a certain size. Therefore, we provide a hybrid heuristic algorithm based on iterated local search complemented with path relinking (HH-ILS-PR), which extends and improves MS-ILS + PR, to address the presented problem. Our HH-ILS-PR consists of iterated local search (ILS) and path relinking (PR). ILS contains three processes: generating initial solutions by heuristics, improving solutions by local
Computational experiments
This section is the computational experiments that are conducted to evaluate our HH-ILS-PR on MME-LRP on three sets of instances. The algorithm is coded in java, and all the experiments are conducted on a personal computer with an Intel(R) Core (TM) i5-7500 3.40 GHz CPU, 16 GB RAM, and Windows 10 operating system.
Conclusions
In this paper, we consider a mixed multi-echelon location routing problem with limited capacities on differentiated intermediate depots and vehicles. In the MME network, differentiated ID candidates are not predetermined, and therefore we have to decide the type of each ID. Algorithm HH-ILS-PR is put forward to solve the problem. The algorithm brings several improvements to Nguyen’s MS-ILS + PR to fit the problem: some empty routes are configured to scale up our neighbors in local search, a new
CRediT authorship contribution statement
Yunkai Chen: Investigation, Writing – original draft. Quanwu Zhao: Funding acquisition, Project administration, Resources, Investigation, Supervision. Wei Wang: Writing – review & editing. Shu Zhang: Investigation.
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.
Acknowledgment
This research was supported by the Humanities and Social Sciences Foundation Program of the Ministry of Education (Project No. 19YJA630122), the Fundamental Research Funds for the Central Universities of Chongqing University (Project No. 2019 CDJSK 02 XK 12, 2019CDCGJG331), and the Technological Innovation and Application Program of Chongqing (Project No. cstc2019jscx-mbdxX0008).
References (35)
Route first—Cluster second methods for vehicle routing
Omega
(1983)- et al.
A computational comparison of flow formulations for the capacitated location-routing problem
Discrete Optimization
(2013) - et al.
Lower and upper bounds for the two-echelon capacitated location-routing problem
Computers and Operations Research
(2012) - et al.
A two-phase method for multi-echelon location-routing problems in supply chains
Expert Systems with Applications
(2019) - et al.
Flexible two-echelon location routing problem
European Journal of Operational Research
(2019) - et al.
A granular variable tabu neighborhood search for the capacitated location-routing problem
Transportation Research Part B: Methodological
(2014) - et al.
A comparative study of heuristics for a two-level routing-location problem
European Journal of Operational Research
(1980) - et al.
The fleet size and mix location-routing problem with time windows: Formulations and a heuristic algorithm
European Journal of Operational Research
(2016) - et al.
A simple and effective evolutionary algorithm for the capacitated location-routing problem
Computers and Operations Research
(2016) Methods for solving combined two level location-routing problems of realistic dimensions
European Journal of Operational Research
(1983)
A multi-start iterated local search with tabu list and path relinking for the two-echelon location-routing problem
Engineering Applications of Artificial Intelligence
Solving the two-echelon location routing problem by a GRASP reinforced by a learning process and path relinking
European Journal of Operational Research
The two echelon open location routing problem: Mathematical model and hybrid heuristic
Computers and Industrial Engineering
A simple and effective evolutionary algorithm for the vehicle routing problem
Computers and Operations Research
Two-echelon location-routing optimization with time windows based on customer clustering
Expert Systems with Applications
Multi-start simulated annealing heuristic for the location routing problem with simultaneous pickup and delivery
Applied Soft Computing Journal
An exact method for the capacitated location-routing problem
Operations Research
Cited by (1)
Smart multi-commodity location-routing model for perishable goods with an emphasis on big data under uncertainty and congestion
2024, Interdisciplinary Journal of Management Studies