J. Eberhardsteiner, M. Gingerl, L. Ondris:
"Some Accuracy Aspects of 3D Electronic Speckle Pattern Inferometry";
Poster: 13th Danubia-Adria Symposium on Experimental Methods in Solid Mechanics, Rajecké Teplice, Slowakei; 26.09.1996 - 28.09.1996.
In the scope of a comprehensive experimental investigation the stiffness and strength behavior of biaxially loaded wood has to be studied. The testing equipment developed for that purpose consists of a biaxial servo-hydraulic loading machine and a 3D electronic speckle pattern interferometer (ESPI). The latter is used for contactless measuring of the spatial displacement field at the surface of the flat wooden specimen. In the context of a quantitative evaluation of the state of deformation and the corresponding state of strain, the knowledge of real accuracy of the used measuring configuration is of great importance.Basically, the accuracy problem has two aspects: a) the accuracy of the measuring method itself, which is a fraction of the wave length of the used light in case of interferometric methods and b) the measuring accuracy of the ESPI system which is influenced by a lot of disturbances. Some of these faults are inevitable, but most of the disturbing influences can be minimized or taken into account during the quantitative evaluation of the deformation components.
The usage of an ESPI system in the framework of material testing requires strict avoidance of vibrations (from the hydraulic loading system or external vibrations sources like road and railway traffic) of the measuring object and the optical components of the interferometric set-up. Erroneous measuring results are also caused by temperature and air turbulences due to thermal radiation of the hydraulic circuits. An other source for the loss of accuracy arises from the recording and digitalization of the measuring information by means of a CCD camera. Reduction or elimination of optical and electronic noise can be done by applying suitable filter algorithms. However, a careless use of such filter procedures may cause a significant loss of information.
In order to minimize errors during a quantitative full-field evaluation particular attention should be paid to an accurate determination of object dimensions and of illumination and observation geometry. For a uniaxial interferometric set-up with known illumination geometry the sensitivity of the measuring system can be expressed by the sensitivity vector S_k. In case of a three-dimensional deformation analysis a so-called geometry matrix G_{jk} representing three sensitivity vectors is defined as
\begin{displaymath} N_j \cdot \la = G_{jk} \cdot d_k\ , \end{displaymath}
where N_j denotes the fringe order, \la is the wave length and d_k represents a component of deformation. In an accurate evaluation procedure the variation of G_{jk} due to divergent illumination within the measuring area has to be taken into account. Inaccurate determination of distances and angles for illumination and observation geometry influences the geometry matrix in different amount. Distance errors are mostly neglectible. An erroneous determination of angles, however, may yield to significant errors of measuring results. A change of the illumination angle by 2^\circ results in a relative error of sensitivity up to 20 % depending on the illumination angle.