Abstract
Introduction: this work proposes a model, and two heuristic algorithms to assign customers to trucks and visiting days, as a first phase in the solution of a real-world routing problem, which is closely related to the Periodic vehicle routing problem, but a strategic decision of the company imposes the additional constraint that every customer must always be visited by the same truck. Methods: The proposed model aims to group the customers that are visited the same day by the same truck as close as possible. The first proposed heuristic has a constructive stage, and five underlying improvement heuristic, the second one uses an exact linear programming algorithm. Results: The algorithms are evaluated by instances taken from the literature and generated, taking into account the characteristics presented in the real-world case addressed.
Highlights
Introduction: This work proposes a model and two heuristic algorithms to assign customers to trucks and visit days as a first phase in the solution of a real-world routing problem, which is closely related to the PVRP (Periodic Vehicle Routing Problem), but a strategic decision of the company imposes the additional constraint that every customer must always be visited by the same truck
In [9], the assignment of customers to days was made by adapting the GAP of [1] to the PVRP, while in [10] and [11], integer linear models were proposed to assign customers to visit days to minimize the maximum demand in each day, but there is no consideration of the geographical coordinates of the customers
(ii) We show that if the constraint that every customer must always be visited by the same truck in a PVRP is added, the problem can be transformed back to a classical PVRP with one truck, a wider planning horizon, and a greater set of allowable combinations of visit days. (iii) We propose two heuristics for solving overlapping centroid-based clustering
Summary
This work proposes a model and two heuristic algorithms to assign customers to trucks and visit days as a first phase in the solution of a real-world routing problem, which is closely related to the PVRP (Periodic Vehicle Routing Problem), but a strategic decision of the company imposes the additional constraint that every customer must always be visited by the same truck. In the VRP (vehicle routing problem) the “cluster first-route second” strategy is widely used, where each cluster represents the customers assigned to one truck. For solving the MDVRP (multi-depot vehicle routing problem), in [5] and [6], a clustering phase is performed to assign customers to depots. For the PVRP (periodic vehicle routing problem), it is common to solve an allocation phase by assigning visit days to customers and solve a VRP for each day. We bring together the truck and visit day assignment phases of our particular problem, which we will usually refer to as the clustering and allocation phases; each cluster is associated with one truck
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have