Contents     Functions              PDF Index

3. Test/analysis correlation tutorial

Modal testing differs from system identification in the fact that responses are measured at a number of sensors which have a spatial distribution which allows the visualization of the measured motion. Visualization is key for a proper assessment of the quality of an experimental result. One typically considers three levels of models.

• Input/output models are defined at sensors. In the figure, one represents these sensors as arrows corresponding to the line of sight measurements of a laser vibrometer. Input/output models are the direct result of the identification procedure described in chapter 2.
• Wire frame models are used to visualize test results. They are an essential verification tool for the experimentalist. Designing a test well, includes making sure that the wire frame representation is sufficiently detailed to give the experimentalist a good understanding of the measured motion. With non-triaxial measurements, a significant difficulty is to handle the perception of motion assumed to be zero.
• Finite element models are used for test/analysis correlation. In most industrial applications, test and FEM nodes are not coincident so that special care must be taken when predicting FEM motion at test nodes/sensors (shape observation) or estimating test motion at FEM DOFs (shape expansion).

The tools for the declaration of the wire-frame model and of sensor setups are detailed in section 2.2. Topology correlation and sensor/shaker placement tools are details in section 3.1. A summary of general tools used to compare sets of shapes is made in section 3.2. Shape expansion, which deals with the transformations between the wire-frame and FE models, is introduced in section 3.3. The results of correlation can be used for hybrid models combining experimental and analytical results (see section 3.4) or for finite element model updating (see section 6.5).