Detection of structural damage phenomena such as cracks, delaminations, or loose fasteners is important because it increases safety and may allow significant reduction in operational costs. Many structural health monitoring techniques are based on detecting changes in vibration features obtained from the measured response. The technique presented in this work is based on the random decrement signatures calculated from the random response. No knowledge of input signal is required. Random decrement signatures have been used for damage detection purposes in the past. The significance of this work is that it explores the possibility of identifying the type of damage and focuses on detection of opening and closing cracks. Statistical features of the signatures are used to detect the presence of nonlinearity that often occurs due to onset of damage.
A numerical study was performed where simulated measurements were obtained from a finite element model of a frame containing a damaged member. The results of this numerical study showed that the new technique is able to detect crack type damage in a complex structure. It was able to highlight the damaged spar, but it was not able to precisely locate the damaged member in the spar. The results suggest that reliability of damage detection depends on the amount of noise in the measurements.
Experimental validation was performed using a cantilever beam experiment. The damaged beam used in the experiment contained a real fatigue crack instead of saw cuts that are often used to simulate damage. The technique was shown to be able to detect damage when excitation level was greater than 1.0g RMS. The exact location of damage is not always detected reliably, as sometimes a segment of the structure adjacent to the actual damaged segment is indicated. The advantage of the new technique is that it is model-free and could be used on structures excited by ambient forces that are difficult or sometimes impossible to measure.