Section: New Results

Assessing metrological performances in experimental mechanics

Participant : Frédéric Sur.

Progress was made during this year on several aspects of our collaboration with Institut Pascal on experimental mechanics. As mentioned in Section 4.3, the surface of the specimens under study are marked either by a regular grid, or by a random speckle. Displacement and strain maps are estimated by comparing images taken before and after deformation: through spectral methods (named here "the grid method") in the first case and through digital image correlation (DIC) in the latter.

Our contributions to the grid method are twofolds. First, we carefully analyzed the effect of digital sampling which causes aliasing [17]. We have proposed simple guidelines to minimize the effect of aliasing on strain maps. Second, we have mathematically characterized the properties of the analysis windows commonly used for processing grid images through the grid method [18]. It turns out that a Gaussian window has to be used, mainly because of its good concentration in both spatial and spectral domains in the sense of the Wigner-Ville transform. We eventually published a comprehensive review paper on the use of grid methods in experimental mechanics [15]

We also contributed to DIC-based methods. We have proposed new predictive formulas for the resolution of the displacement maps provided by DIC, which is mainly limited by sensor noise. These formulas take interpolation into account [12]. Indeed, displacement amplitude being often much smaller than one pixel, it is crucial to analyze the effect of the interpolation scheme. We have also proposed an experimental validation of these formula. This requires to take into account the heteroscedastic nature of sensor noise and rigid body motions caused by unavoidable vibrations [13].