Exploratory Investigation of Legacy Bridge Databases in Virginia

This report is an investigation of data mining of legacy bridge databases, focusing on the Pontis and National Bridge Inventory databases maintained by the Virginia Department of Transportation (VDOT). This project is applied in nature as it was sponsored by the Virginia Center for Transportation Innovation and Research (VCTIR). Exploratory data analysis was performed using Microsoft Excel, and resulted in patterns and projections that were unavailable directly from the raw inspection report data.
Data mining techniques were applied to 6 sub-studies conducted in response to VDOT interests. First, district-based trends in various element condition states across Virginia were determined and the results were summarized using graphs created from the Pontis element data. Second, a state-wide identification of potential inspector data-entry errors was conducted. Third, as part of a wider VCTIR study, the effectiveness of zinc-based coating systems was investigated by looking at deterioration of steel girders using National Bridge Inventory (NBI) and Pontis data; deteriorated sections of girder elements on the same set of bridges were also identified. Fourth, joint closures occurring between 1995 and 2010 were discovered and categorized. Fifth, priority bridges were identified around Virginia, and trips to the nine VDOT district offices were taken to investigate which maintenance actions caused improvements to specific elements on those bridges. Lastly, linear least squares regression and Markov chain condition state transition modeling were performed on selected elements to predict deterioration.