Health assessment of beams–experimental verification

Abstract

A novel nondestructive structural health assessment procedure now under development at the University of Arizona and presented in a companion paper, is experimentally verified. The experimental verifications of the procedure to identify defect-free and defective fixed ended and simply supported beams are presented in this paper. In this approach, acceleration and rotational time histories are measured at pre-selected node points. They are then post-processed to remove several sources of error including noise, high frequency content, slope, and DC bias. The post-processed response information is then successively integrated to obtain the corresponding velocity and displacement time histories. Even when these sources of error were removed from the response information, the proposed method failed to identify the beams. Several factors including noise, data latency, scale factor and cross coupling error were investigated. Amplitude and phase errors in the accelerometer's measurements were found to be the root cause. Alternative approaches are proposed to mitigate them. Following the suggested procedures, defect-free and defective fixed ended and simply supported beams are correctly identified. The proposed NDE procedure is accurate and robust, and can identify defects at the local element level in the context of the finite element representation. The laboratory experiments clearly and conclusively verified the proposed algorithm, i.e. a beam can be identified without using input excitation information and using only noise-contaminated response information and established its application potential.