An exact algorithm for vehicle routing problems with temporal dependency constraints

TL;DR

This paper introduces an exact algorithm for vehicle routing problems with complex temporal dependencies, capable of handling all studied types and outperforming existing methods.

Contribution

It proposes a novel fragment-based formulation and a price-cut-and-enumerate algorithm that effectively solves a broad class of temporal dependency constraints.

Findings

Significantly outperforms state-of-the-art benchmark methods.

Able to solve previously intractable instances.

Handles all types of temporal dependencies studied in the literature.

Abstract

Temporal dependencies between customer visits, such as synchronization constraints, pose a fundamental challenge in vehicle routing. These dependencies, which arise in applications such as home healthcare routing, aircraft scheduling, and technician routing, introduce inter-route constraints that make the resulting problems significantly harder to solve. We present an exact solution method for vehicle routing problems with temporal dependencies capable of handling all types of temporal dependencies studied in the literature, unlike most existing approaches that target specific subclasses. Our approach is based on a fragment-based formulation in which routes are represented as sequences of a new type of fragment, designed to handle temporal dependency constraints. This formulation is solved via a price-cut-and-enumerate algorithm that computes a lower bound using alternating column-and-row generation, obtains an initial upper bound, and iteratively refines both bounds through fragment enumeration and branch-and-cut, supported by several new classes of valid inequalities. Computational experiments show that our method significantly outperforms state-of-the-art benchmark methods and is able to solve previously intractable instances while covering a wider range of temporal dependencies.