End windings in large 2-poles turbogenerators tend to vibrate at 2XLF due to elm forces caused by stray fields from the field winding (from retaining rings) and from the phase currents.
It can excite radial as well as tangential basket modes if these exhibits 2 lobes at a natural frequency (resonance) close to 2XLF.
I may understand that radial modes are always excited as soon as phase currents depart from zero after synchronization. Now I think that for basket tangential modes to be excited, one must produce some active power. Is that right?
With only reactive power being produced, I think that radial basket modes would be much more excited than tangential modes. Is that right?

I don’t work with generators and I’m sure the question is way over my head. But that never stopped me from trying to talk through it.

Regardless of the phase relationships to the other fields in the end region of the machine (which are more complex than I understand), there is tangential attraction between adjacent coils carrying current. The attraction is strongest where the current is highest and weakest where the current is 0. So the two-pole current pattern creates a two-lobe wave where there is compression at the lobes of the wave and decompression 90 degrees away. And the pattern rotates at syncronous speed with the wave. The compression seen at any given location is a frequency of 2*LF.

So this would seem to be A tangential 2*LF two-lobe mode that is excited for both real and reactive power (not just for active power). But perhaps not THE tangential basket mode that you are interested in? If so can you describe it a little more?

I have used normal charge mode accels to monitor inside the endwindings of large generators with good results, you see when the support need to be tightened. You can even do shaker test with the largest air driven shaker you can find and a diesel compressor during operation. You should avoid resonances in this structure at 1 and 2xLF as levels get impressive. I can´t say that any active/ reactive load dependency was recorded in these cases. When there was a undamped resonance you got hi levels, possibly depending on load generally. Other indication of problem was the production of massive amounts of saw dust from a brand new 600MW generator stator isolation material.... after a few weeks operation. To improve this it was a massive 6-8 month research of a 5-8 man highly educated taskforce so I can´t help much on the theory side either. Modifying circular devices w/o radial support is difficult. Olov

Well, end winding vibrations are no easy stuff to understand from the vibration viewpoint. It is yet well mastered by large manufacturers.
Bump tests in cold conditions do not always prove useful to assess the resonance frequencies in actual operating conditions, not to mention aging problems (resins stability like shrinkage, electric transients, etc.) and the utter complexity of the geometry.
If anyone has results about actual measurements with accelerometers on windings in operation, I would gladly discuss them with that person.

There are 12-16 charge mode accels, happen to be B&K as they been there since start 25+ years ago on most nuke turbo gens here. Latest we have been involved in was to supply racks with charge amps for them to be continuously monitored with the turbine vibe analysis systems of a couple of brands. I used to work in one of those plants so I have changed them when they got bad and used them for shaker test during full temp and operation IRL. At least 3 nuke blocks have them like that and by that 6 turbines and a smaller prototype generator in a heatplant. Special with these gens part from the transducers are that they are watercooled also the rotor.... at 3000RPM.
There are a couple of brands with optical accels but they cost a lot more and have less frequency range.
If I can be of any help, you are welcome. Olov

Oli,
You probably think of such optical accelerometers as Vibrosystm and others (EDF uses Japanese brands as far as I know). True, their bandwidth is limited.
We had practically no problem with nuke 4-poles endwindings in Belgium.
Trouble is more with peak load units like combined cycles where thermal transients are sharp and generators are two-poles units.
In order to accommodate large and frequent excursions of differential Cu/core differential expansions, manufacturers increasingly introduces endwinding supports with some axial decoupling with mixed results. Bracing endwindings in all circumstances is very complicated.
As far as the dependence of vibrations on MW MVAR, I try to qualitatively understand what happens with the pulling forces generated on endwindings as a result of stray mag fields generated by the rotor endwindings (beneath the retaining ring) and the stator endwindings, mimicking the reasoning that leads to core vibrations, without delving into F.E. intricacies. We could pursue the discussion directly if you wish so.
I focus on global basket modes (mainly two-lobes).