Here is a nice chart from my favorite site to ponder on:) Let me be informed the difficulty of the AVR in dealing with leading p.f .What is the issue with this situation? Why is such a unstability pops up with leading p.f loads?
I'm not one for clicking on unknown links so I'm going to assume that you're looking at a generator capability curve for a small diesel generator set. Since I don't want to get hung-up on naming conventions I'm also going to assume that when you say "leading p.f. loads" you're referring to operating the generator in the underexcited region, i.e., the excitation/field current is below what it would be at unity p.f., all other parameters being equal.
The main problem is that the field current is at its weakest and therefore the magnetic field that couples the power from the prime mover through the air gap to the stator is also weaker. If it gets too weak then additional loading may cause the rotor to pullout and slip poles. The problem is further exacerbated by the negative feedback action of the AVR. Unlike inductive loads that tend to depress the voltage and make the AVR increase the voltage thereby strengthening the air gap magnetic field, capacitive (leading pf) loads tend to raise the terminal voltage, which in turn causes the AVR to lower the field current in an effort to maintain the desired voltage. This is the dangerous situation that your favorite site (I suspect it begins with a C) wants you to avoid by never going below 0.95 pf underexcited. The Minimum Excitation Limiter and Stability Controller are supposed to help you not get into this operation region, assuming that they're set properly.
Whenever you connect a capacitor to a bus, the bus voltage tends to increase marginally. (There are precise formulae to calculate the incrase in volatge depending upon the kVAr rating of the capacitor). This being so, the feedback signal for the generator AVR is tapped from the bus. When you have a leading power factor, this means that the connected load is more capacitive and as such, the bus voltage is increased. This "pseudo-increase" is incorrectly sensed by the AVR as increase in the generated voltage and thus it reduces the excitation thus causing voltage hunting. This becomes sort of a cumulative process, until the AVR reduces the excitation to almost zero.
"Strength does not come from physical capacity. It comes from indomitable will." -- Mahatma Gandhi