You are on page 1of 30

STRUCTURAL HEALTH MONITORING FOR CIVIL 

INFRASTRUCTURE –
INFRASTRUCTURE 
FROM INSTRUMENTATION TO DECISION SUPPORT

Anne S. Kiremidjian
Dept. of Civil and Environmental Engineering
p g g
Stanford University

IWSHM 2011
S t b 13 15 2011
September 13‐15, 2011

This research is supported by the NSF CMMI Research Grant No. 0800932, the John A. Blume 
Fellowship, and the Samsung Scholarship.
Diversity of Civil Infrastructure
Diversity of Civil Infrastructure
Why Monitor Civil Infrastructure?
Gradual Long‐Term Deterioration
(from Dr. Hae Young Noh)

3/17
Why Monitor Civil Infrastructure? ‐
Extreme Event Damage
Extreme Event Damage
Outline
• Wireless/Wired Monitoring System Design
Wireless/Wired Monitoring System Design

• Algorithmic Development

• Example Applications

• Decision Support System

• Conclusion

5
Structural Monitoring System Sensors & 
Structural Monitoring System Sensors &
Network
Control Center
Data 
management 
and archiving 

Data Storage
Manager on 
Manager on
site
Base 
Station

Decision Support &  Data Analysis and Post‐
Emergency Response  processing System
S t
System
Synchronization Signal Processing

Spectral Analysis
Structure ‐level 
analysis
l i Feature Extraction & 
Feature Extraction &
Damage Classification

Decision making
The Smart Structural Monitoring 
System
Decision • structure 
Sensors
Support
• model updating specific
• decision making • Damage 
• warning specific
id
• guidance • Data 
Data
• active control collection/
storage

Data
Analysis Sensor
& N t
Network
k
Archiving
• Wired
• physical modeling 
• Wireless
• signal processing
• Combined
• statistical analysis
• Communication
7/17
Damage‐Specific Sensors
g p
Currently Available Still needed
Wired sensors
Wired sensors Wireless/Wired 
Wireless/Wired Crack
sensors
Fiber optic sensors Accelerometers Corrosion
High definition  Strain gages Displacement
digital cameras
Laser
Laser  GPS Materials specific
Materials specific
interferometers
Laser scanner Tilt meters
Deep penetration  Temperature 
radar
Humidityy
Corrosion
Anemometers
The Smart Structural Monitoring 
System
Decision • structure 
Sensors
Support
• model updating specific
• decision making • Damage 
• warning specific
id
• guidance • Data 
Data
• active control collection/
storage

Data
Analysis Sensor
& N t
Network
k
Archiving
• Wired
• physical modeling 
• Wireless
• signal processing
• Combined
• statistical analysis
• Communication
9/17
Sensor Networks
Sensor Networks
• Wireless vs. wired 
Wireless vs. wired
• Advantages of wireless systems
– Scalable
Scalable 
– Ease of installation
– Portability
– Lower cost
• Challenges
– Potential signal loss
– Communication barriers
Communication barriers
The Smart Structural Monitoring 
System
Decision Multiple • structure 
Support Sensor
• model updating specific
• decision making • Damage 
• warning specific
id
• guidance • Data 
Data
• active control collection/
storage

Analysis
Sensor
& Data
N t
Network
k
Archiving
• Wired
• physical modeling 
• Wireless
• signal processing
• Combined
• statistical analysis
• Communication
11/17
Types of Algorithm
yp g
• Damage diagnosis– statistical 
• Device control pattern recognition methods
pattern recognition methods
– Wake up 
– AR/ARMA/ARX
– Database structure • Hypothesis testing
• Gaussian Mixture Modeling‐ GMM
– Baseline collection – Wavelet Based – Haar and 
• By time Morelet wavelets
• By season • Comparison of wavelet energies at 
• By temperature high scales
• By humidity • Gaussian Mixture Modeling
– Synchronization – Rotation ‐ to drift ‐to damage
• Hardware – internal 
clock
• Software – pre ‐& post 
synchronization
Damage Detection Using Statistical 
g g
Signal Processing
• Main approach
– Use single sensor pre‐ and post‐ damage measurements
– Combine information from multiple sensors
– Computationally efficient 
efficient ‐ local micro
local micro‐processing
processing
– Independent of the sensor – can be used with acceleration, 
,
strain, etc.
– Scalable with increased sensor density
– Reduces amount of transmitted data – power saving
Steps in Damage Diagnosis
Statistics Based
ARX
Collect data AR
GMM
Wavelet
Extract Change point detection
features

Classify damage

14/17
Examples
• Using stationary vibration signals 
Using stationary vibration signals –
– AR with information criteria testing
– AR with Gaussian Mixture Model
AR with Gaussian Mixture Model
• Using non‐stationary vibration signals – e.g. 
earthquake motions
th k ti
NEES – UNR Project‐¼ Scale Bridge Test
AR & G
AR & Gaussian Mixture Algorithm 
i Mi t Al ith
(Nair & Kiremidjian, 2006)
Decision Support System
Decision Support System
Test Schedule – 4‐span Bridge Test at UNR

Baseline Signals
g

Minor cracks at base of 
column

Major cracks at base of 
col. & spalling
l & lli

Concrete spalling and 
exposure or rebar

Rebar buckling/ 
breaking; concrete 
pouring out of core
i t f
Final Test – White Noise

Test Damage Measure


Reno - Setup Day DC1 and DC2 baseline
Mild Shaking Day 1, White Noise 21.05
Mild Shaking Day 2, White Noise 21 36.79
Mild Shaking Day 3, White Noise 41 56.97
Final Test Day,
Day White Noise Run 51 59 80
59.80
C l
Column Device
i

DM

Test Damage
Measure
Reno - Setup Day DC1 baseline
and DC2
Mild Shaking Day 1,
1 21 05
21.05
White Noise
Mild Shaking Day 2, 36.79
White Noise 21
Mild Shaking Day 3, 56.97
White Noise 41
Final Test Day, White 59.80
Noise Run 51
Example of Wavelet Damage Diagnosis
(Noh et al., 2011)                
• Feature
Feature from 
from
wavelet energies of 
signal
• Used with non‐
stationary signals –
e.g. earthquake 
response motions
• Develop fragilities 
for rapid damage 
i di i
indication
NEES 4 Story Steel Frame
Wavelet Based  Algorithm 
g

• Scaled structural system 
tests Stanford/SUNY 
tests – Stanford/SUNY
Buffalo
• Development of fragility 
Development of fragility
functions in terms of 
structural response 
parameters obtainable 
t bt i bl
from real time 
measurements – wavelet 
based fragilities 
(Noh, Lignos, Nair and Kiremidjian 
2011)
The Smart Structural Monitoring 
System
Decision Multiple • structure 
Support Sensor
• model updating specific
• decision making • Damage 
• warning specific
id
• guidance • Data 
Data
• active control collection/
storage

Analysis
Sensor
& Data
N t
Network
k
Archiving
• Wired
• physical modeling 
• Wireless
• signal processing
• Combined
• statistical analysis
• Communication
23/17
Decision Support System
Decision Support System
• Visual
Visual representation of the 
representation of the
structure
• Visual representation locations 
p
of the wireless system
• Interface to wireless network
• System command and control 
center  
• Display results of monitoring 
Display results of monitoring
analyses
• Issue alerts
Components of a Decision Support 
System
• Monitor the sensor
• Monitor the sensor 
communications system
• Serve as the 
communications 
environment between 
i tb t
manager and system
– Initial set‐up
– Modifications of system 
Modifications of system
parameters
– Modifications of 
monitoring settings
• Serve as the information 
delivery environment
– Periodic queries
– Following a major event
Following a major event
Decision Support System
• Provide support for decision making for follow‐on 
actions
• Enable web services for 
– wide distribution of alerts and other information
– remote access by operators and other users.
C l i
Conclusion
• Wireless monitoring systems – inevitable part of 
the future
h f
• Few applications in Europe, Asia and the US 
• Must combine structural with other monitoring 
systems, e.g.
–BBuilding environmental/energy/lighting/security 
ildi i t l/ /li hti / it
monitoring
– Bridge, highway, tunnel, pipeline, transmission line, 
g , g y, ,pp , ,
etc. management systems 
• Key to success is providing information and 
d
decision support, not just data
d
27
Need for Full‐scale and Field Testing

• Objectives
– Systematic damage at 
different levels
different levels
– Different damage 
patterns
– Different damage 
sequences
q

NIED – E‐Defense Test 
y g,
5‐Story Full Scale RC Building, 
August, 2011
Acknowledgement
• National Science Founcdation
• National Institute of Standards and 
Technology
gy

• Students
Dr. E. Straser ‘98 Dr. K. Nair ‘07
A Kotapalli ‘99
A.Kotapalli 99 Allen Cheung
Allen Cheung
N. Mastroleon ‘00 Dr. H‐ Y Noh ‘11
C Ch i i ‘03
C. Charistis
THANK YOU!

You might also like