Optimal maintenance decisions for pavement management

The object of this investigation was to develop a procedure for making optimal maintenance decisions for a deteriorating system. The particular system chosen for study is a pavement, and a methodology is developed to ensure that pavements meet certain performance criteria while minimizing the expected maintenance cost. A cumulative damage model based upon a Markov process was developed to model pavement deterioration. The optimal repair action for each possible pavement state in the planning horizon was found by means of probabilistic dynamic programming. Sample sequences of repair actions were generated during a simulation in which the optimal repair policy was applied to sample pavement condition histories. Several sensitivity studies were performed to study the variation in expected cost, including the effect of delaying the optimal program.