Wang, Feiyue (2021-03). Development of Bridge Deterioration and M&R Decision-Making Models for Use in Network-Level Bridge Management Systems. Doctoral Dissertation.
Thesis
Life cycle planning (LCP) is conducted to ensure optimal maintenance and rehabilitation (M&R) of bridge assets over their life cycle. Furthermore, LCP is required by federal regulations for bridges that are part of the National Highway System (NHS). This study aims to develop Markov chain bridge deterioration prediction models and M&R decision-making model to support conducting network-level LCP and implementing network-level bridge management system, with application to Texas. The specific objectives are to: (a) develop models for predicting the deterioration of bridges prior to applying M&R treatments, (b) develop models for predicting the deterioration of bridges after the application of M&R treatments, and (c) develop a method for prioritizing M&R projects based on benefit-cost analysis. Models have been developed using the National Bridge Inventory (NBI) data that extend from 2001 to 2017 for 43,320 bridges across Texas. The models allow for predicting the deterioration in the NBI bridge condition ratings: deck rating (NBI Item 58), superstructure rating (NBI Item 59), substructure rating (NBI Item 60), and culvert rating (NBI Item 62). The models consider the explanatory factors that affect bridge deterioration, including climate/environment, traffic loading, material type, and M&R type. These deterioration prediction models can be used to forecast the bridges' future conditions, compare the effectiveness of alternative M&R strategies, plan future M&R projects, and estimate future funding needs. Further, a decision-making model was developed for prioritizing M&R projects based on benefit-cost analysis. The M&R project prioritization model has been developed by combining benefit-cost analysis and the bridge deterioration prediction model to help bridge agencies plan M&R activities over multiple years into the future. Finally, the developed models were evaluated based on Texas NBI bridge data.
