Dynamic decision model for pavement management using mechanistic pavement performance submodel

Abstract

A dynamic decision model for pavement management has been developed on the basis of a dynamic programming formulation. The transition probabilities are determined by a mechanistic pavement performance model formulated within a stochastic framework. In this way, the individual distress modes may be modeled in the pavement condition states, which can be helpful in identifying the proper rehabilitation treatment. Furthermore, the Markovian assumption that the transition probabilities are time-invariant is no longer necessary with the proposed methodology. A numerical illustration demonstrates that the impact of variations in excess user and highway agency costs and other management decisions on the optimal rehabilitation policy can be evaluated explicitly.