The question has a flawed premise. If there is a non zero voltage drop across the armature resistance, then the generator cannot have zero voltage regulation, since the no load voltage at the terminals will exceed the full load voltage by the aforementioned voltage drop.
Quick Summary: Check your reactive droop control setpoints on all three machines and make sure they match.
Forgive me; I don't know how much you've worked with this, so I'm going to write this for the uninitiated.
When machines are in parallel, they have to share load. If their controllers are set to maintain one, specific voltage, they don't share very well, shown in the next figure.
Here, the reactive load sharing pretty much depends on tiny little differences between the machines. In general, my experience has been that the machine that starts taking on the load will try to take all of it, and there isn't a great way to tell which one will do it.
The VR6 voltage regulator you mentioned has something called reactive droop control. The figure below shows how that cleans things up a bit.
The red line represents the total amount of reactive load in the system between the machines.
If both machines are set with the same droop percentage, then they will share the reactive loading in proportion to their KVAR ratings. The no-load set point can be adjusted up or down to increase or decrease the proportion of the reactive load taken - just imagine the red load lines sliding left or right when the black droop lines move up or down.
Looking at the VR6 manual that I googled, It doesn't seem to actually make you set the droop rheostat, so much as tell you that you can if you want to. Your three parallel machines all probably have slightly different droop characteristics. In this case, the one with the shortest droop compensation will eat up the most load, demonstrated below.
You can probably get pretty close to matched droop by just making sure all three machines droop rheostats have been turned by the same amount.
Best Answer
Of course you can connect a load that's larger than full load.
The data sheet tells you what happens at full load. Unfortunately, the data sheet probably doesn't tell you what happens at more than full load. You would have have to do the experiment to see whether it failed to self-excite into it, exploded, blew a fuse, tried then shut down safely, or how long it took to overheat.
There is a school of hobby engineering that assesses maximum ratings by ignoring the data sheet and turning it up until it explodes, then backing off a bit. Sounds like you adhere to this teaching. Professional engineers tend to stick to the data sheet. Often the manufacturer has spotted something in the months or years of development and qualification that the user hasn't yet.