Application of simple and hybrid local search heuristics for the long-term optimization of pavement maintenance strategies at the network level

Abstract

Infrastructure maintenance (considering preservation, maintenance and rehabilitation activities) is one of the major issues of public agencies. Either insufficient maintenance investment or inefficient maintenance strategies lead to high economic expenses in the long term. Recent figures illustrate these facts: Spanish pavements are currently presenting poor conditions and the investment needed in road maintenance is estimated in more than US$7 billion, while annual budget was reduced 20% in 2012. Moreover, users demand higher quality, comfort, and safety conditions. Infrastructure and pavement managers need tools to help them in the decision making process in order to maximize the long-term effectiveness of maintenance strategies subject to budgetary and technical restrictions. When dealing with large pavement networks, genetic algorithms have extensively been applied to solve this problem. However, little attention has been paid to local search heuristics, which have been proved to perform more robust results than genetic algorithms in similar combinatorial optimization problems.This paper explores the suitability of applying local search heuristic algorithms to optimize the long-term effectiveness of maintenance treatments at the network level. For this purpose, two local search heuristic methods (Simulated Annealing and a hybrid Simulated Annealing-Greedy Search) were applied in an illustrative case study. Mean values of long-term effectiveness obtained with the hybrid method are statistically higher than those of the simple method with a reduction of standard deviation of 44%. In broad terms, the hybrid method leads to more efficient and robust results than simple Simulated Annealing, resulting in a suitable method for the optimization of maintenance strategies at the network level.