Data-Driven Analysis of Transportation Infrastructure Systems using Embedded Wireless Sensing and Cloud-Based Data Architectures

Abstract

In 2021, America’s infrastructure received an overall score of ‘C-’ from the American Society of Civil Engineers suggesting a national crisis may be looming due to aging infrastructure systems combined with chronic underinvestment in system renewal. Given the importance of transportation infrastructure systems in ensuring societal prosperity and quality of life, there is an urgent need to explore new approaches to managing these systems. While this is a universal problem, it is especially acute in cities and towns with extremely limited financial resources (i.e., under-resourced communities). While sensing technologies have been proposed to monitor such systems, current solutions have a number of challenges including: monitoring solutions (wired and wireless) remain expensive with proprietary architectures limiting versatility; there is a lack of data processing tools that extract information impacting decision making; solutions are not always tailored to the needs of the system end-user. This dissertation explores the creation of an end-to-end sensing and data management systems for the monitoring of transportation infrastructure systems. The work focuses on wireless telemetry as a primary approach to ease system installation in stationary assets and to support asset mobility for moving assets. Specifically, the work highlights efforts to develop wireless sensors that use cellular networks to seamlessly move their data to cloud-based data repositories. The second major contribution of the work is the development of automated data processing tools that can process raw sensor data to extract information specific to assessing, often in real-time, the performance of the system under study. The third is the mapping of information extracted to drive stakeholder decision-making processes for each of the transportation systems studied. The first transportation system considered are retaining wall systems. Asset managers are reliant on visual inspection information to perform risk assessments; due to the nature of visual inspections these assessments are qualitatively done. This work explores the deployment of a cellular-based wireless monitoring solution that monitors the long-term behavior of reinforced concrete cantilevered retaining walls. The work focuses on automated algorithms that extract metrics associated with wall performance including lateral earth pressures and wall deflections before performing a quantitative risk assessment for asset manager decision-making. The same end-to-end data architecture is also shown to be versatile as a rapid-to-deploy monitoring system for under-resourced communities seeking affordable monitoring solutions to identify low-pressure zones in their drinking water distribution systems. This solution is demonstrated in Benton Harbor, Michigan with pressure measured sparsely across the city to identify locations where water pressure is low. The third transportation system considered in this study are highway assets (e.g., bridges) undergoing heavy truck loading. The thesis adopts the use of weigh-in-motion weight data from a state-wide network of weigh-in-motion stations to infer the travel trajectories of heavy trucks based on Bayesian inference methods. The last system explored by the work is the tracking of a transit system’s fleet of buses with an end-to-end data architecture developed to visualize system performance, especially the on-time arrival of buses servicing fixed route service. The work engages community stakeholders to assess their data visualization needs in order to inform the design of a user dashboard that can empower stakeholder insight to the performance of the public transit system.