Pardon my ignorance, the potential being this is a couple years out. But I am not sure the answers you received on this are that robust. My understanding of Kirchoff's Law , Ohm's Law, PFC and energy quadrants, switching quadrants etc., tells me that overdriving the grid in terms of voltage is not how a proper utility-grade GTI should operate, if it is to contribute useful power to the distributed loads; not to mention being absorbed effortlessly by the utility power provider's own multi-megawatt active and/or passive in-line PFC (reactive power management) hardware.
First of all, unless you are in the multi-megaWatt class of power plant, then your attempt to stick your V-neck out one tick more than the grid such that you force-feed the system with all of your available power, that will happen -- all your available power can be put to go sparring with Hoover Dam generators for example. You will lose. Those generators will see you as an irritating reactive load -- assuming a direct connection, at close to 1MVolts. That isn't the case.
What you are dealing with is an extremely complex network which requires big compute power, in real time, to be able to know what the power available, real versus apparent power and distributed loading, plus alternative active subsystems they incorporate throughout their grid, and what would happen, what limits are,and finally, real-time because the consumers create a monster of unpredictable loading, particularly reactive loads.
Reactive loads are effectively inductive loads, meaning they kick-back opposite currents into the grid. So that is what you would do if your GTI algorithm wa as I understood you: Impedance of the line is 100mOhm, I have 2000Amperes available, so I will just bump up the voltage by 20V above the grid's.
That makes no sense to me -- please feel free to correct me if I am somehow missing something. All that will do is heat your inverter to it's maximum output fighting the billion dollar million Watt grid to no avail. You would fair better to store that energy for night-time use --the utes won't even pay you for it (unless you are megawatt class).. Here's how I think it works.
The proper GTI approach is to compensate for the grid's activity, whether is be reacting to a reactive load, or buckling to a sudden passive load, or some PV farm down the road using your algorithm (backfiring onto you). It might be clearer to picture two small local PV farms on the same 3-phase distribution, or even single phase local line, fighting each-other with reversing currents, only using the grid's cables to battle-it out, meanwhile, nobody benefits, and the grid is adjusting by effortlessly driving around your phase noise.
SO if you simply look at fundamentals; use a circuit simulator. Have IDENTICALLY SET voltage sources, don't worry about reactive loads, use resistors. Tie two sources together on a single-phase simulation. There is series resistance per distance, etc. you put an ammeter on those sources, each isolated by THEIR SOURCE IMPEDANCE, which you can adjust and see how Kirch's Law works -- because the bottom line is this -- your load is not the generators in the dam. Nor is your farm supposed o be a load to those generators.
So you absolutely positively MUST be aligned, synchronized, perfectly sinusoidal in phase, frequency and amplitude -- moreover, the PFC issue is all about being a CURRENT SOURCE, synchronized with the grid's real-time voltage phase. Because you need to INJECT POWER, which means your current times their voltage, which must be your voltage too.
Any THD (harmonics, or deviation from the grid's nearly perfect sinusoid) is wasted power, because trust me, the grid wins, and it will not even feel you as it destroys your inverter (of course, you have circuit protection).
So in reality, it has everything to do with the grid's impedance, your source impedance, synchronization, PFC and, THD.
The impedances, BTW, are dependent upon the transmission cable, as well as the intended grid voltage. If you are anywhere near 480VAC 3-phase output, you are probably doing this through a distribution transformer at 220 single phase with a center tap for two legs of 110VAC single phase. So the impedance of your transformer and electrical wiring is probably a lot lower than it would be if you were interfacing directly to a 12kV power line. So the impedances of the grid change depending upon distribution system switches, faults incurring pother switching and parallel lines becoming twice the impedance they were a minute prior, etc. Add to that the substation nearest you, and what it is kicking in and out.
Lastly, the safety systems are critically important and private or personal PV system owners are not happy with this, but the truth is, unless you store you own power in cells, or some fancy kinetic or magnetic storage system, you need a lot of PV cells to even think about powering yourself wen the grid is down. And, if you have excess PV power, you cannot feed it to the system if the grid is down. When it comes back up, there are waiting periods that vary from region to region etc.
So GTI is not trivial, and for any appreciable power, it is a partnership with the grid owner/operator/utility company. It is not just a clever inverter when you get to high power sources. So this is a large reason why utility companies do not pay the same rate for power sourced from PV homes. It is more of a nuisance than anything, but for the home owner, it certainly helps cut the bill -- but not by force-feeding the utility grid they won't even pay for. It is solely by not demanding as much from the grid, and GTI, simply means it is transparently switched into your house wiring (seamlessly), not supplying other homes in the neighborhood seamlessly.
For any small PV farm, there is not much sense to having GTI for purposes of selling excess energy. GTI means a completely different thing, primarily a safety thing, and again, for the safety of the grid, its workers, its equipment and even its customers, not to benefit you or be fair about paying for your energy when more than likely,it was dissipated by improperly interfacing with the grid.
Pardon the dissertation.