Suggestion - contact the authors and ask them if they have any data on the cases sufficent for you to create a model.
Your best bet to get test data and model is to contact directly universities doing research on model updating methods.
If you need to work data from scratch, the simplest case to use is a straight rectangular cross section beam. Such structures tend to be accuracte as steel shows little variation, cross sections have uniform dimensions and as the beam tend to be straight. A simple case to start with and one where modelling and test consumes very little time.
For what it is worth, you find some ramblings of mine on the model correlation subject here
Hi Nils, what exact data do you need? I have recently made a FE model and calibrated it based on ambien vibration experimental results and modal analysis in the FEM. I dont have the raw data for the ambien noise only the modal frequencies after being processed and some graphs.
Thank you for your response. I will need a finite element model in an appropriate format, e.g. Abaqus (*.inp), Nastran (*.bdf) or Ansys (*.cdb).
In addition, a corresponding universal file (*.unv) with experimental results including Dataset 55 (Data at nodes Analysis Type = Normal Modes), Dataset 15 (Nodes) and optionally Dataset 82(Tracelines) would be helpful. Your support is greatly appreciated.
Hi Nils. I did experimental modal analysis for a simple beam and extracted modal properties. Also, I created FE model and then I did calibration between them. Which data you want?
In SDT, we use the data of the GARTEUR group on ground vibration testing as examples to illustrate our test/correlation tools (see http://www.sdtools.com/pdf/IMAC97gart.pdf for details on the structure, the measurements and the project)
In the linked archive file, you will find
- the .unv file containing the test definition (wireframe + sensors), the measured transfers and an identification result.
- the .inp file containing a very simple model of the structure.
The interest of such simple data is that low computation time is required to test and develop.
You can download a full example of a model validation and calibration case using MATLAB (and CALFEM and ABRAVIBE, both free toolboxes) of a simple Plexiglas plate from my website http://www.abravibe.com. The example includes data (FRFs) from an impact test. There is a paper describing it all here: http://www.sandv.com/downloads/1406bran.pdf.
A useful reference is - MORASSI A. & TONON S., “Dynamic Testing for Structural Identification of a Bridge”, J. Bridge Engineering, ASCE, 13(6), 573-585, 2008. The technique I've used is to "calibrate" your FE model with the data from the EMA testing. When you set up your FE model, there are certain assumptions made and these usually involve the boundary conditions. When you perform EMA, you actually measure the boundary conditions.