Using Data Analytics to Characterize Steel Bridge Deterioration

Abstract

Bridges are a key component of transportation infrastructure. Currently, decisions on what material should be used to build a bridge superstructure are primarily driven by the initial costs in design/construction, but subsequent repair and maintenance costs are less emphasized. Moreover, limited budgets for bridge construction and maintenance require new processes to optimize decision making in choosing appropriate materials for bridges. This study aims to characterize the performance of steel bridge superstructures considering factors such as age, average daily traffic (ADT), design load, and structure length. This paper applies data mining techniques to the 2013 national bridge inventory (NBI) database for bridges whose superstructure material is steel or steel continuous and whose deck type is concrete cast-in-place. This study develops a model to predict the probability of a steel bridge superstructure being deficient given the characteristics of the bridge. The contribution of the paper is that the analyses will help stakeholders better understand what parameters are significant to the superstructure deterioration, and what type of steel bridges are more likely to have deficient superstructures.