Method for Estimating Transition Probability in Bridge Deterioration Models

Abstract

A method for estimating transition probability is applied to the deterioration models of a bridge management system. Deterioration is modeled as a time-dependent function of the condition changes of critical elements in bridge systems. The methodology applies “life data” analysis techniques employed in reliability studies of engineering systems and incorporates the inherent censorship of the utilized bridge data. It involves a two-phase process: one concerned with the condition of a set of similar bridge components composing a system, and the other concerned with the condition of a specific component within the system. The information required for modeling purposes is derived from historical data typically available in bridge inventory and inspection systems. The methodology focuses on deterioration models previously created for use in a bridge management system. It is illustrated by a case study involving deterioration processes of a structural deck system. The methodology has been evaluated through comparison of predicted and realized condition measures over a 10-year period. On the basis of the analysis conducted and its results, it is concluded that the developed methodology significantly improves the accuracy of bridge deterioration models and the network-level condition prediction process.