Scholarly Communications Project


Locating Structural Damage in Real Time Using Contrast Maximization

by

Vikas Juneja

Dissertation submitted to the Faculty of the Virginia Polytechnic Institute and State University in partial fulfillment of the requirements for the degree of

DOCTOR OF PHILOSOPHY

Approved

Raphael T. Haftka, Co-Chair
Harley H. Cudney, Co-Chair
Robert A. Heller
Daniel J. Inman
Scott L. Hendricks

April, 1996
Blacksburg, Virginia


Abstract

A technique to identify structural damage in real time using limited instrumentation is presented, Contrast maximization is used to find the excitation forces that maximize the difference in the response of the damaged structure and the analytical response of the undamaged structure. The optimal excitations are then matched against a database of optimal excitations to locate the damage. To increase the reliability of the approach when modeling and measurement errors are present, the contrast maximization approach is then combined with an approach based on changes in frequency signature to develop the average angle technique. A damage detectability measure is defined which, for a given level and location of damage and a given amount of modeling and measurement errors, compares the magnitude of the damage to the magnitude of errors in a single number. The success of the average angle technique in damage detection is quantitatively defined by a success factor. The technique is first tested analytically on a 132 degree of freedom truss. The structure can be either equipped with active members or collocated shakers/sensors. The technique has a high success rate in damage detection. The technique is then tested numerically on a 36 degree of freedom truss equipped with 3 collocated shakers/sensors. To simulate experimental conditions, an extensive study is carried out in the presence of numerical noise. It is seen that the success factor in the presence of noise depends upon the success factor in the absence of noise and the damage detectability measure. The members are classified into three groups based on the success factor in the presence of noise. A mathematical relationship between the damage detectability measure and the success factors with and without noise is developed using linear regression. Using the results of numerical simulations in the presence of noise and this mathematical relationship, we find members where we expect to locate damage experimentally, for a given amount of damage and given amount of noise. A similar truss is built and the FEM model of the structure is corrected using experimental data. The average angle technique is applied to locate damage in a member when the member has a low level of damage (25%). The damage detectability measure indicated that the measurement errors are large compared to the damage and it is difficult to detect damage in most detectable locations. The steps taken to rectify that are described and after each step we show the improvement in damage detectability. The average angle technique is used to locate damage in 5 members. The experimental results indicate that the technique can robustly identify the damaged member with limited instrumentation.

Full text (PDF)


The author grants to Virginia Polytechnic Institute and State University or its agents the right to archive and display their thesis or dissertation in whole or in part in the University Libraries in all forms of media, now or hereafter known. The author retains all proprietary rights, such as patent rights. The author also retains the right to use in future works (such as articles or books) all or part of this thesis or dissertation.
[ETD main page] [Search ETDs][etd.vt.edu] [SCP home page] [library home page]

Send Suggestions or Comments to webmaster@scholar.lib.vt.edu