In case of generator with brushless excitation, voltage decay takes very long time after generator is tripped. Under fault condition, this characteristic will keep on feeding fault current for long duration, thus damaging the equipment. This appears to be inherent phenomenon.
Views are invited on what sort of protection scheme can be employed to arrest long-feeding fault current so as to prevent such damage.
To a cursory reader. this post makes no sense. First of all, you do not say whether you are talking a dc machine or an ac machine. You do not say if the prime mover remains connected. why is the machine still connected to a fault if tripped.
AC alternator is likely.A DC generator would inherently have a DC supply to the field How big -size matters. What is the "long-feeding" time? These questions are at the crux of the problem.
Attempts to stop the prime mover quickly can be disastrous. Even so, the mechanical time constants are much longer than the electrical time constants- so speed effects can conservatively be ignored. When a fault is external(on the load side of the breakers-which may well be on the load side of a transformer)it is not "usually" be a problem, but for ground or phase faults on the generator side it is a major problem. A line to ground fault inside a winding can involve a second winding and eat holes in the stator core- expensive.
So- protection against internal faults has to deal with reduction of field ASAP. Now one is looking at high current, low R/L situations In the situation where DC exciters were used field breakers also introduced a resistive element in parallel with the field winding to balance fast field current decay with allowable overvoltage in the field and supply. With brushless excitation, the same considerations are involved. It is possible, on detection of an internal fault, to not only disconnect from the world at large but to disconnect the field taking into account these factors. Certainly I would expect that proper design of a given excitation system would account for this.
The brushless exciter is DC type, where its dc field current is fed from PMG through AVR. Its rotating armature produces AC voltage which is rectified by rotating diode wheel (mointed on same shaft) and its dc output is fed to generator field throgh connector. Generator rotor, exciter armature, diode wheel, PMG rotor are coupled to each other and rotating at same speed. The voltage decay time is of the order of 30 sec. As such, no arrangement is possible to disconect the dc supply to generator field winding.
Unfortunately. the OP gave almost no information as to just what the situations were. It is vert difficult to protect against any possible mode of failure under any circumstance. I think that is what the OP was after.
I was speaking of large machines -but in that case the excitation system should be able to discharge the field within a reasonable time- 30 seconds is not reasonable even considering that there are two fields (generator and exciter )that are of concern. Is there no field discharge resistor /diode combination that will, in case of a shutdown of the field supply, allow the main field to discharge at an optimum rate? Most of the schematics that I looked at,for brushless excitation systems have such a resistor. I assume the AVR is not on the rotating shaft or, if rotating, there is some form of internal fault sensing or external control? How big a machine is this? That has a bearing on protection- it may not be economic to repair rather than replace.
I am assuming the main output is a.c. If so, what you have described is an alternator. This is a fairly standard arrangement for the excitation of alternators used on diesel generators for locations where a good quality a.c. supply with low electrical noise is required. When I was working, these were used for our telecoms sites.
The long voltage decay is due to the run down of the rotation of the rotor shaft. Short of installing brakes on the shaft there is nothing you can do about this.
As you have correctly identified, you can't get to the output of the diodes on the rotor. To kill the output as quickly as possible you could disconnect the output of the AVR (assuming the AVR is a separate unit and not part of the alternator) - but it won't kill the output entirely due to the residual magnetism within the iron.
The only way to remove the output asap is to use some form of disconnect switch on the output - probably a frequency sensing switch to detect the change of output frequency as the alternator runs down would be quickest (assuming that your prime move is stable enough so that the disconnect switch doesn't disconnect during normal operation).
Otherwise an under-voltage switch could be OK, particularly if you arrange to disconnect the output of the AVR from the alternator to stop the AVR from trying to maintain the correct output voltage as the machine runs down.
If people post questions here, the least they can do is to describe their situation well. If they leave so many loose ends, as this OP did, they do not deserve thought out answers. I enjoy helping out but not to those who carelessly or intentionally make it difficult to answer.
In many cases, particularly with large machines with high inertia, the speed and the frequency can remain nearly constant over what is an electrically long time- and the time to drop from 60Hz to 59Hz can be significant- but normal variations in this range that do not merit disconnection are not uncommon- particularly for isolated machines.
While this is not directly related: I do recall dealing with a situation in a hospital where the switch over from grid to diesel was about 10 seconds (this did not affect absolutely needed supply) and there was a slow down of large induction motors - with power factor capacitors. The frequency held well but there was a phase shift that was of more concern- so that the effective induction generator (capacitor excited) was out of phase with the incoming generation- Bloody high inrush currents and torques caused major mechanical failures. It was better to drop out the capacitors in this case.
I am making some assumptions with respect to the original query BUT!! Salmon Egg, as astute as ever, recognizes that lack of information leads to guesses based on assumptions which may not be valid. He is right in this regard.