Ant Colony System for the Periodic Capacitated Arc Routing Problem

In this paper, we present a new Ant Colony System (ACS) approach for the Periodic Capacitated Arc Routing Problem (PCARP) on a mixed network. In the CARP problem the objective is to find routes starting from the depot such that each required arc/edge (called task) is serviced, capacity constraints are satisfied and total travel cost is minimized. The PCARP extends the classical CARP problem to a planning horizon of several days. Each task requires a certain number of visits within this time horizon. Hence, the objective of the problem is to choose the visit days for each task and to solve a CARP for each day in order to find efficient routes over the whole horizon. In our approach, we suggest a new technique for solving the PCARP problem. We have combined the ACS algorithm with a heuristic for insertion of tasks in a current solution. Hence, the ACS is used to optimize the order of tasks for insertion and the heuristic is used for insertion the tasks. The results of computational experiments carried on problem taken from literature indicate that the proposed approach outperforms existing algorithms in most cases. A comparison with the best memetic methods is presented and the results prove the robustness, the rapidity and the performance of our method.