If you do not have the force, you cannot normalize the shapes you obtained. It is then impossible to use them as a model. You will not be able to use the base to synthetize other transfers for example.

In practice you have further steps after your peak picking to obtain modes (operational or not) as they must verify certain conditions like orthogonality.

If you just pick frequencies, you have operational deflection shapes, you may have a coupled response and depending on the sensor your result may be sensitive, so here again you (should) have treatments with SVD analysis in a short frequency band around the picked pick and the sensor batch.

Please could you explain wrt to a practical issue.

Say I want to get the modal parameters and mode shapes of a railway steel truss bridge on which train goes.

I have installed sensors all over my bridge.

Once the train goes over the bridge I can have the free vibrations data of the bridge.

On taking the fft of the bridge I can have the frequency curve which shows some peaks hopefully the modal peaks.

Now I want to get the mode shape of the bridge.

Can I get the mode shape my what every peak picking method or so from the FRFs obtained from all the installed censors.

Or is there is a better way to get the mode shape in this scenario.

Hi

As stated above, Operational Modes are not referenced with respect to the system weight (i.e. its mass & inertia). To reference the system weight in a test, you need a force reference with FRF data at least for the excitation point degrees of freedom as well.

Be aware that Operational Modal Analysis (OMA) also may give you artificial 'modes' that are nothing more than periodic load variation.

If you want to see the steps involved in an OMA, take a look at the enclosed link.

A comment - as long as you have the train on the bridge, vibration would be forced, not free. You have free vibration only once the train has left the bridge.

http://absignal.com/download/OperatingTutorialManual.pdf