In our process, we use to make measurements in our power generation for a large scale of active power.

But in our synchronous compensator We use to make the same thing, but for different reactive power. I don't know the effective theses measurements, because if we have some problem in the field, is possible detect for one reactive power only, for example the rated reactive power lagging, in our case -18 MVAr.
The fact is: In our procedure, we have to make vibrations's measurements in -18, -10, 0, 5, 10 and 30 MVAr. I think that one measurement in 18 MVar for example is enough!

I would like to know what the worst condition for vibration in synchronous condenser (compensator)!?

Depending on the number of poles, the above comment can be valid.
In particular for 2P machines this would apply.
In order to determine worst case, you should measure the frequency of vibrations. The most likely frequency should be in step with the RPM. This type would be caused by shaft unbalance, while vibrations at line frequency (F) and higher (2F, 3F, and so on) is more likely to be caused by the magnetic conditions.