Network-Level Bridge Deterioration Prediction Models That Consider the Effect of Maintenance and Rehabilitation

Abstract

With the increasing number of aging bridges that need maintenance and rehabilitation (M&R), it is important to plan and prioritize M&R projects proactively. This process requires mathematical models capable of predicting bridge condition over multiyear planning horizons taking into account the long-term performance effects of M&R treatments. This paper describes the development, validation, and application of Markov chain–based bridge deterioration prediction models. Bridge condition is measured in terms of the National Bridge Inventory (NBI) deck rating, superstructure rating, substructure rating, and culvert rating. The models consider explanatory variables that affect bridge deterioration, including climate/environment, traffic loading, material type, and M&R type. Because data on M&R work history are often disconnected from bridge condition data, the study used a novel approach that allows for inferring past M&R type and timing based on changes in bridge condition ratings. The models were developed using condition data for 43,320 bridges across Texas extending from 2001 to 2017. The developed prediction models can be used as a tool to support bridge asset management planning in Texas. The modeling approach, however, can be used by transportation agencies across the world to develop bridge deterioration prediction models using their local empirical data.