BaseLevel Ambient Vibrations in Structural Health Monitoring

1
Base-Level Ambient Vibrations in Structural
Health Monitoring

• Anthony Mizumori
• Washington University in St. Louis
• Department of Civil Engineering
• 2002 REU Program
• 8/2/02

2
Overview

• Structural Health Monitoring
• Ambient Vibrations
• Data Collection
• Model Development and Simulation
• Verify Accuracy of Simulation
• Modal Identification Results
• Conclusions

3
Structural Health Monitoring

• Determination of stiffness and damage in a
  structure
• Utilizes only dynamic properties of the structure
• More efficient than current damage identification
  techniques

4
Ambient Vibrations

• Origins
• Ground/Base motion
• Wind loading
• Intra-structure motion
• Necessary as dynamic stimulus in SHM, but
  knowledge of actual ambient forces not necessary
• How does damage detection vary w/ types of
  ambient motion?

5
SHM Methodology

• Equations of Motion
• NExT finds free response
• ERA finds modal parameters
• Least-square analysis finds E

6
Data Collection

• Dual channel collection near busy 4-lane road
• PCB 3513 accelerometer
• 6 acquisition periods at 5 minutes each
• Horizontal and Vertical

7
Simulation of Ground Acceleration

• Collection of data near structure
• Analyze frequency- and time-domain
  characteristics
• Is separation valid?

Ground Acceleration
Spectral Density
8
Time-Domain Characterization

• Characteristics
• Maximum amplitude
• Peak duration
• Frequency of Occurrence
• Peak location
• Generate PDFs

9
Frequency-Domain Characterization

• Find Mean Power Spectral Density
• Interpolate spectrum
• Assume random phase
• Inverse Fourier Transform

PSD
10
Generation

• MATLAB script generates variables under
  determined PDFs

Time Domain Frequency Domain
11
Verify Accuracy of Simulation
Simulated Record
Simulated Spectrum
Experimental Record
Experimental Spectrum
12
Simulation Performance

• Load a 4-storey finite element model
• Experimental records
• Simulated records
• Identify lateral modes
• 15, 90, 300s loadings

13
Study of Modal Identification

• What effects do inputs have on modal
  identification?
• Input spectrum
• Non-stationary properties
• 4 cases

14
Results

• Less than 1 error for all cases
• Artificial mode identification
• High density in input
• Distinguished by large damping ratios

15
Conclusions

• Effective method for modeling non-stationary
  signals
• False modes can be distinguished in
  identification scheme
• Future considerations in modal identification
• Wind loadings
• Spatial variations
• Multi- directionality
• Extend to actual damage localization
• Require mode-shape identification

16
QUESTIONS ??