Date of Award

May 2014

Degree Type

Thesis

Degree Name

Master of Science

Department

Engineering

First Advisor

Anoop Dhingra

Second Advisor

Sudhir Kaul

Keywords

Damage Detection, Mixed-Mode Crack, Modal Properties, Truss, Wavelet Transform

Abstract

The use of dynamic response in damage identification has been gaining considerable attention over the last two decades. The aim of these methods is to detect the presence of a defect or a crack in components or structures. This study focuses on using modal properties for the damage detection of mixed-mode cracks in truss structures. The behavior of a mixed-mode crack is simulated by developing a macroscopic model that is integrated with the finite element model of a truss structure. The modal properties obtained from the model of the damaged structure are found to be comparable to the results of the continuous system model. The direct use of modal properties such as natural frequencies and mode shapes is investigated for simple and large truss structures. It is observed that the traditional approach of using modal properties in damage detection is limited to simple structures with relatively large cracks. Therefore, a damage detection algorithm that uses the wavelet transform is developed in this study. Multiple analyzing wavelets are investigated to enhance the capability of using mode shapes for extracting salient information related to specific damage characteristics. The proposed algorithm is found to be effective and reliable in detecting relatively small mixed-mode cracks even in the presence of noise. The influence of multiple parameters such as number of truss members, truss member orientation, crack size, crack orientation, etc. is investigated through the application of the proposed algorithm to the Warren truss and the Howe truss structures. The amplitude of wavelet coefficients at a predefined damage location is found to be related to crack size, therefore allowing an evaluation of damage severity. The parameters associated with damage characteristics and geometrical properties are found to be very influential in damage detection, especially when the structure is large and complex.

Share

COinS