New formulations for the traveling repairman problem with time windows

Abstract

The Traveling Repairman Problem (TRP) is one of the most important variants of the Traveling Salesman Problem (TSP). The objective function of TRP is to find a Hamiltonian path or tour starting from the origin while minimizing the total latency (waiting or delay time) for all customers. The latency of a customer is defined as the time passed from the beginning of a tour (or path) until a customer’s service is completed. TRP with time windows (TRPTW) is the case where the earliest and latest times for visiting each customer are restricted by prescribed time windows. The literature on TRPTW is scarce. We only found one formulation for TRPTW and one formulation for its variant. In this paper, we propose four new mathematical models for TRPTW with O(n2) binary variables and O(n2) constraints. We computationally analyze the performance of existing and new formulations by solving symmetric and asymmetric benchmark instances with CPLEX 12.5.0.1 and compare the results in terms of CPU times and optimality gap. We observed that our two formulations were extremely faster than existing formulations, and they could optimally solve symmetric instances up to 150 nodes and asymmetric instances up to 131 nodes within seconds.