Acoustic resonance testing for nondestructive detection of forged or casted serial parts

Acoustic resonance testing (ART) is an integral nondestructive testing method that is used to inspect components and assemblies with regard to different properties or variations in quality by evaluating the test object’s eigenfrequencies or other parameters determined from its natural vibration behavior.

As a comparative method, ART mainly focusses on the examination of serial parts which are produced in large quantities and with low cycle times, e.g. forged or casted metallic parts. Besides the possibility for complete automation, a big advantage of ART compared to other volume-oriented methods is the fast in-line quality assessment of an entire specimen within a matter of seconds.

The principle of ART is based on the fact that a specimen’s natural vibration behavior and its eigenfrequencies mainly depend on geometry and material properties, but also on structural defects, e.g. cracks.

One objective of ART is to detect defective parts by evaluating the test object’s measured eigenfrequencies. Generally, the exact geometric dimensions and the exact material properties of single parts in a serial production vary randomly within acceptable ranges, for example because of manufacture-related effects, entailing variations in the eigenfrequencies of the good parts.

These effects are superposed by changes in the eigenfrequencies caused by intolerable component variations. This impedes a reliable classification of the components with the help of ART. To solve this problem a new compensation method of those random perturbations respectively a method to differ between eigenfrequency shifts caused by acceptable as well as intolerable variations is required.

A current research project focusses on detecting forged or casted metallic parts with intolerable geometric variations by evaluating the component’s eigenfrequencies, whereat such parts are also characterized by large acceptable component variations.

Previous investigations using simulated data showed that the exact dimensions of components can be estimated from their eigenfrequencies after describing those correlations with the help of linear regression analyses. This contribution presents the latest results of the project, especially the adjustment of this procedure to real parts and the associated difficulties.

http://www.izfp.fraunhofer.de