Experimental modal analysis on vibration data measured by digital image correlation

This paper discusses the application of 3D vision-based techniques to recover the Operational Deflection Shapes of a reference specimen consisting in a rectangular section cantilever beam. The research work focuses on the analysis of the accuracy of this measuring approach and discusses the strong points and the limitation of the photogrammetric approaches to experimental modal analysis. Two different types of area-based matching techniques are considered here: the former is based on incremental displacement estimation of subsets of subsequent images. This technique works the approach of Particle Image Velocimetry (PIV), with an Eulerian approach. The second technique considered here is on the contrary based on the estimation of the absolute displacement, considering the first acquired image as the reference for all the subsequent ones. In this last case a Lagrangian approach is used and the applied algorithm is the Digital Image Correlation (DIC). The main advantage of Eulerian (i.e. PIV) approach is that it is possible to analyse also the vibration of a moving object (like in belt dynamic characterization), where the reference region imaged in the first acquired picture, comes out of the field of view much before the end of the test. On the other hand, the Lagrangian approach shows some accuracy problems in the applications where the estimation of the displacement is request instead of velocity. In these cases, the incremental displacement estimated at each new image has to be composed with all the previously estimated ones to obtain the overall motion. Experimental tests were done using a cantilever beam as a case study. The excitation of the beam is done by means of an electromagnetic shaker, while the reference transducer used to validate the output of vision-based techniques is a Scanning Laser Doppler Vibrometry. Results show that the two image-based methodologies have similar performances when estimating natural frequencies, loss factors and ODSs.