Skip navigation
Please use this identifier to cite or link to this item:
Title: Physics-Based and Data-Driven Methods for Structural Health Monitoring At Fine Spatial Resolution
Authors: Kumar, Vivek
Advisors: Glisic, Branko
Contributors: Civil and Environmental Engineering Department
Keywords: Attribute Analysis
Crack Identification
Ground Penetrating Radar
Machine Learning
Structural Health Monitoring
Two-Dimensional Sensing
Subjects: Civil engineering
Issue Date: 2021
Publisher: Princeton, NJ : Princeton University
Abstract: Structural health monitoring (SHM) process involves several elements: selection of monitoring strategy, installation \& management of monitoring system, data management, and closing activities. In the last four decades, each of these processes has attracted considerable attention and we have reached the stage of commercial deployment for long-term monitoring of civil infrastructure. Furthermore, monitoring technologies are consistently being adapted from and exported to domains of mechanical engineering, aerospace engineering, archaeology, and geotechnical engineering. Still, one challenging area for civil structure scale of monitoring is design and implementation of sensing technologies to cover large areas while maintaining fine spatial resolution at an affordable cost. Densely instrumenting the structure with point sensors would result in the SHM system cost a significant fraction of the structure's budget which would not be approved by the stakeholders. Distributed fiber-optic sensors provide a relatively affordable solution yet only provide one-dimensional information about the structure. In this dissertation, this gap of affordable sensing technologies for large area monitoring at fine spatial resolution is addressed using traditional sensing technologies such as electrical strain gages (ESG) and ground penetrating radar (GPR). These technologies are mature for assessment of parameters over large areas of structures with fine spatial resolution. Using novel large-area electronics, affordable two-dimensional sensors are demonstrated using ESG while GPR is used as three-dimensional sensor. They are combined with data-driven and physics-based approaches to achieve the goal of large-scale monitoring at finer scales. The data-driven approach includes use of supervised machine learning (ML) models to determine the concrete material properties using GPR scans. ML models are developed using laboratory curated concrete beams and laboratory tested cylinders. These models are used for estimating concrete properties such as density, porosity and compressive strength on the deck of a pedestrian bridge. Physics-based approach includes analytical modeling for a dense array of resistive strain gage sensors for crack identification, i.e., detection, localization, quantification. Algorithms are developed based on these analytical models for crack identification, and verified using laboratory tests and employed on real-life structure. The results pave way for truly affordable and distributed sensing technologies for comprehensive monitoring of structures at finer scales.
Alternate format: The Mudd Manuscript Library retains one bound copy of each dissertation. Search for these copies in the library's main catalog:
Type of Material: Academic dissertations (Ph.D.)
Language: en
Appears in Collections:Civil and Environmental Engineering

Files in This Item:
File Description SizeFormat 
Kumar_princeton_0181D_13761.pdf28.19 MBAdobe PDFView/Download

Items in Dataspace are protected by copyright, with all rights reserved, unless otherwise indicated.