Performance Prediction of Bridge Deck Systems Using Markov Chains

Abstract

Bridge management systems have adopted Markov-chain models for predicting the future condition of bridge components, systems, and networks. These models are developed based on two assumptions. First, bridge inspections are performed at predetermined and fixed time intervals (i.e., constant inspection period). Second, the future bridge condition depends only on the present condition and not on the past condition (i.e., state independence). This paper evaluates the impact of these assumptions on the performance prediction of bridge deck systems using field data obtained from the Ministère des Transports du Québec. Transition probability matrices are developed for the different elements of the deck system and adjusted for the variation in the inspection period using Bayes’ rule. This investigation indicated that the variation in the inspection period may result in a 22% error in predicting the service life of a bridge deck system. Also, the statistical tests used to assess the validity of the state independence assumption of Markov chains showed that this assumption is acceptable with a 95% level of confidence, which is reasonable for network level analysis.