0
Research Papers

# Structural Health Monitoring of Glass/Epoxy Composite Plates Using PZT and PMN-PT Transducers

[+] Author and Article Information
Valeria La Saponara1

Department of Mechanical and Aerospace Engineering, University of California, Davis, CA 95616vlasaponara@ucdavis

David A. Horsley

Department of Mechanical and Aerospace Engineering, University of California, Davis, CA 95616

Wahyu Lestari

Embry-Riddle Aeronautical University, Prescott, AZ 86301

1

Corresponding author.

J. Eng. Mater. Technol 133(1), 011011 (Dec 02, 2010) (8 pages) doi:10.1115/1.4002644 History: Received March 04, 2010; Revised June 10, 2010; Published December 02, 2010; Online December 02, 2010

## Abstract

The structural health monitoring of composite structures presents many challenges, ranging from sensors’ reliability and sensitivity to signal processing and a robust assessment of life to failure. In this research project, sensors constructed with both PZT-4 ceramic and single-crystal PMN-PT, i.e., $Pb(Mg1/3Nb2/3)O3−PbTiO3$, were investigated for structural health monitoring of composite plates. Fiberglass/epoxy specimens were manufactured with a delamination starter located in the middle of the plate, and were subjected to axial tensile fatigue at a high stress ratio. A surface-mounted PMN-PT pair and a surface-mounted PZT-4 pair were positioned on each side of the delamination starter and excited in turns at set intervals during fatigue loading. This project had two goals: (1) assess the performance of the two piezoelectric materials and (2) develop a signal processing technique based on wavelet transforms capable of detecting damage features that are independent of the transducers (being damaged concurrently to the host composite specimens) and thus can estimate life to failure of the composite specimens. Results indicate that the PMN-PT transducers may be more resilient to fatigue damage of the host structure and possibly generate less dispersive Lamb waves. However, these aspects are compounded with higher costs and manufacturing difficulties. Moreover, the proposed signal processing method shows promise in estimating impending failure in composites: It could, in principle, capture and quantify the complex wave propagation problem of dispersion, scattering, and mode conversion across a delamination front, and it will be further investigated.

<>

## Figures

Figure 1

Left: two typical glass/epoxy specimens of this project, with a 50.8 mm central delamination. Right: surface-mounted transducers.

Figure 2

Views of fractured specimen C. Arrows indicate the transducer location. All transducers debonded at fracture. Width=25.4 mm.

Figure 3

Dispersion curves in 0–90 kHz center frequency range, from experimental data (circle and square symbols), compared with Ao estimate from classical lamination theory (1) (solid line). Top plots: specimen B with PMN-PT pair (left) and PZT pair (right). Bottom plots: specimen C with PMN-PT pair (left) and PZT pair (right).

Figure 4

Top: acquired waveform (original and denoised) for specimen D, PMN-PT pair, at 24,000 cycles. Bottom: absolute value of Gabor wavelet transform of denoised signal, which is a (32,768×132) matrix. Axes show, respectively, the ith row component in the 32,768×132 matrix and the log2 of a function of the scale (as calculated by WAVELAB (22)), which is related to the jth column component. Software: MATLAB and WAVELAB .

Figure 5

Top: close-up of contour plot of absolute value of Gabor wavelet transform of denoised signal for specimen D, PMN-PT pair, at 24,000 cycles. Note different intensity levels (from a maximum of 0.0283 to a minimum of 0.00707). Software: MATLAB and WAVELAB (22). Bottom: contour area calculated for lowest intensity level (0.00707). Only the solid areas were added for the rest of the analysis. Software: MATLAB and CONTOUR2AREA.M (29).

Figure 6

Top: cycles versus residual strain. Center and bottom plots: trends of log10 of the contour area with the lowest intensity, selected from the absolute value of Gabor wavelet transform. Specimen D failed 11,110 cycles after the last signal acquisition. A, B, and C failed 2,942, 846, and 600 cycles, respectively, after the last signal measurement.

## Errata

Some tools below are only available to our subscribers or users with an online account.

### Related Content

Customize your page view by dragging and repositioning the boxes below.

Related Journal Articles
Related Proceedings Articles
Related eBook Content
Topic Collections