Advisor Information
Dr. Robin Gandhi, Dr. Brian Ricks, Dr. Jong-Hoon Youn
Location
Room 231
Presentation Type
Oral Presentation
Start Date
1-3-2019 10:30 AM
End Date
1-3-2019 11:45 AM
Abstract
Bridges in the US scored a C+ on the infrastructure report card in 2017. There is a need for substantial improvement in conditions of these bridges as they are structurally deficient and can become unsafe in the near future. The most recent estimate puts the nation’s backlog of bridge rehabilitation needs at $123 billion. Throughout the country, many state departments of transportation (DOT) have limited resources, leaving them with difficult decisions about where to invest and allocate their limited resources. To make cost-effective decisions, these bridge stakeholders need clean data and studies to estimate the future condition of bridges. This will allow them to have data-driven accurate life-cycle models and improved inspections intervals.
Previous researchers have identified potential independent variables that may cause deterioration in bridges using a variety of deterioration models. Unfortunately, these researchers limit their data to specific regions and type of bridges. This severely limits the general applicability of their results.
In this research, we approach bridge health-related decision making challenges using a novel data science perspective. This study of bridge health deterioration provides new insights into making bridge rehabilitation and reconstruction decisions. In this research, we demonstrate the use of large datasets, including records of all the bridges maintained by Federal Highway Agency, known as the National Bridge Inventory or simply NBI and precipitation data from Center of Disease Control (CDC) and Prevention to perform sound statistical analysis. We specifically contribute to this domain by 1) providing a reference implementation of a big data pipeline for bridge health-related datasets and compute scores for evaluating bridges; 2) demonstrating the feasibility of using data science to study the deterioration of the bridges
Further, the curated datasets and methods developed through this research are used to analyze the statistical significance of independent variables related to bridge deterioration as identified in the literature and from subject matter experts at the Nebraska State DOT. The data used in our research spans all available inspection records in the NBI and precipitation rates from all US states.
Understanding the effects of Precipitation on Bridge Health in the US
Room 231
Bridges in the US scored a C+ on the infrastructure report card in 2017. There is a need for substantial improvement in conditions of these bridges as they are structurally deficient and can become unsafe in the near future. The most recent estimate puts the nation’s backlog of bridge rehabilitation needs at $123 billion. Throughout the country, many state departments of transportation (DOT) have limited resources, leaving them with difficult decisions about where to invest and allocate their limited resources. To make cost-effective decisions, these bridge stakeholders need clean data and studies to estimate the future condition of bridges. This will allow them to have data-driven accurate life-cycle models and improved inspections intervals.
Previous researchers have identified potential independent variables that may cause deterioration in bridges using a variety of deterioration models. Unfortunately, these researchers limit their data to specific regions and type of bridges. This severely limits the general applicability of their results.
In this research, we approach bridge health-related decision making challenges using a novel data science perspective. This study of bridge health deterioration provides new insights into making bridge rehabilitation and reconstruction decisions. In this research, we demonstrate the use of large datasets, including records of all the bridges maintained by Federal Highway Agency, known as the National Bridge Inventory or simply NBI and precipitation data from Center of Disease Control (CDC) and Prevention to perform sound statistical analysis. We specifically contribute to this domain by 1) providing a reference implementation of a big data pipeline for bridge health-related datasets and compute scores for evaluating bridges; 2) demonstrating the feasibility of using data science to study the deterioration of the bridges
Further, the curated datasets and methods developed through this research are used to analyze the statistical significance of independent variables related to bridge deterioration as identified in the literature and from subject matter experts at the Nebraska State DOT. The data used in our research spans all available inspection records in the NBI and precipitation rates from all US states.