Structural Health Monitoring Studies on a Girder Bridge

1
Structural Health Monitoring Studies on a Girder
Bridge

• Ashley Lucas
• Washington University in St. Louis
• REUJAT 2003

Japan Advisor Dr. Xia YongUniversity of Tokyo
U.S. Advisor Dr. Shirley DykeWashington
University
2
Overview

• Introduction
• Girder bridge example
• Methods studied
• Background on the girder bridge in Japan and
  summary of provided data
• Results and Conclusions
• Further exploration
• Acknowledgements

3
Introduction

• Structural Health Monitoring using statistical
  methods

4
Girder Bridge Example
5
(No Transcript)
6
(No Transcript)
7
(No Transcript)
8
Data Collected

• Vibration data
• Excitation large vehicle crossing
• trials run on both lanes of the bridge varying
  speeds both directions
• Monitoring data
• Hourly and daily peak stress values(ideally over
  the course of one year)

9
Methods Studied

• Vibration data
• System Identification
• Time History Analysis
• Monitoring data
• Fatigue Analysis
• Extreme Value Analysis

10
Methods Studied

• Vibration data
• System Identification

• Eigensystem Realization Algorithm (ERA)

• Power Spectral Density estimate

11
Methods Studied

• Vibration data
• System Identification
• Time History Analysis

12
Methods Studied

• Monitoring data
• Fatigue Analysis
• Equivalent Constant Amplitude
• reference stress range
• sp smax smin/2
• must fulfill fatigue assessment criterion
• this determines years of service

13
Methods Studied

• Monitoring data
• Fatigue Analysis
• Linear Damage Accumulation
• Miners rule to determine Fatigue Life
• Di ni/Ni ? 1
• Life 1/Dyear
• Ni values also determined from fatigue assessment
  criterion

14
Methods Studied

• Monitoring data
• Extreme Value Analysis
• Gumbel model vs. GEV model

15
SHM challenges

• Insufficient measurements
• Unknown structural or material properties
• Unknown excitation

These all became limitations!
16
The Girder Bridge Bridge A
17
The Girder Bridge Bridge A

• Skew 44
• Sampling frequency 250 Hz
• Excitation ? 40 km/hr, lanes 1 and 2 60
  km/hr, lanes 1 and 2 (both
  directions)Duration 6-10 seconds
• Vehicle was a large truck
• Vibration and Monitoring data

18
Vibration Data
Time History Analysis
19
Monitoring Data

• Given 1 month of data
• Extreme Value Analysis
• Gumbel model
• General Extreme Value model
• Fatigue Analysis
• Equivalent constant amplitude
• Linear damage accumulation (Miners rule)

20
Results and Conclusions

• Using the Monitoring Data
• Compared the Gumbel and GEV models

• Using the Vibration data
• Determined peak stress values
• Generated histograms for those stress sensors

21
Time History Analysis for the max 6 stress sensors
Recurring sensors 6, 13, 18
22
Gumbel model for Sensors 6 and 13
23
General Extreme Value model for Sensors 6 and 13
24
Comparison of Gumbel and GEV models
25
Further Exploration

• Perform the fatigue analysis when more
  parameters/codes are determined
• Try other statistical models for the extreme
  value analysis

26
Acknowledgements

• Dr. Shirley Dyke and Dr. Xia Yong
• Diego Giraldo
• University of Tokyo Bridge and Structures
  Laboratory
• National Science Foundation

27
Questions?