Integrated multi-item packaging and vehicle routing with split delivery problem for fresh agri-product emergency supply at large-scale epidemic disease context

Fresh agri-product emergency supply is crucial to secure the basic livelihood of residents at large-scale epidemic disease context. Considering the massive demand and limited transportation resources, this study integrates multi-item packaging and vehicle routing with split delivery to improve the emergency supply capacity. Firstly, three specific objectives of fresh agri-product emergency supply at large-scale epidemic disease context are formulated, i.e., average response time, infectious risk possibility and transportation resource utilization. Then, a multi-item packaging strategy is proposed to consolidate different categories of fresh agri-products according to the food cold chain temperatures.An optimization model integrating multi-item packaging and vehicle routing with split delivery is developed to jointly decide the optimal packaging scheduling, vehicle assignment and delivery routing. Next, an improved genetic algorithm based on solution features (IGA-SF) is designed to solve the integrated model with multiple decision variables. Finally,a case on fresh agri-product emergency supply of Huangpi District, Wuhan in the context of the Corona Virus Disease 2019(COVID-19) is carried out to illustrate the efficiency and feasibility of the proposed model. The numerical results of medium-to-largescale cases demonstrate that the proposed IGA-SF could save 23.91% CPU time and 37.80% iteration number on average than genetic algorithm. This study could satisfy different emergency scenario requirements flexibly, and provide scientific decision support for provincial and national governments on fresh agri-product emergency supply.

Enhanced solution representations for vehicle routing problems with split deliveries

In this study,we investigate a forest-based solution representation for split delivery vehicle routing problems(SDVRPs),which have several practical applications and are among the most difficult vehicle routing problems.The new solution representation fully reflects the nature of split delivery,and can help reduce the search space when used in heuristic algorithms.Based on the forest structure,we devise three neighborhood search operators.To highlight the effectiveness of this solution representation,we integrate these operators into a standard tabu search framework.We conduct extensive experiments on three main SDVRPs addressed in the literature:The basic SDVRP,the multidepot SDVRP,and the SDVRP with time windows.The experimental results show that the new forest-based solution representation is particularly effective in designing and implementing neighborhood operators,and that our new approach outperforms state-of-the-art algorithms on standard datasets.

A two-stage heuristic method for vehicle routing problem with split deliveries and pickups

The vehicle routing problem(VRP) is a well-known combinatorial optimization issue in transportation and logistics network systems. There exist several limitations associated with the traditional VRP. Releasing the restricted conditions of traditional VRP has become a research focus in the past few decades. The vehicle routing problem with split deliveries and pickups(VRPSPDP) is particularly proposed to release the constraints on the visiting times per customer and vehicle capacity, that is, to allow the deliveries and pickups for each customer to be simultaneously split more than once. Few studies have focused on the VRPSPDP problem. In this paper we propose a two-stage heuristic method integrating the initial heuristic algorithm and hybrid heuristic algorithm to study the VRPSPDP problem. To validate the proposed algorithm, Solomon benchmark datasets and extended Solomon benchmark datasets were modified to compare with three other popular algorithms. A total of 18 datasets were used to evaluate the effectiveness of the proposed method. The computational results indicated that the proposed algorithm is superior to these three algorithms for VRPSPDP in terms of total travel cost and average loading rate.