> they'll push so hard that the grid itself will change phase
That's a thundering herd problem: where all invertors have the same set-points and they all are synchonised to push in the same direction at once.
In networks, thundering herd problems are fixed by given each sender different random delays.
For power networks we could choose statistical methods to get individual solar generators to lead or lag so that the frequency becomes an aggregate vote.
Given that part of the blackout was due to large amounts of solar going offline at the same time, it's possible that all invertors with common software were tripped at the same set-point.
So trip conditions also need to be randomly fuzzy. E.g. if frequency drops below 49Hz +/- random spread of 0.5Hz.
Although it's difficult to match financial incentives against random variations (individual generators are incentivised to power outside of boundaries to keep getting paid, and a trip event can be expensive - due to restart costs).
Electricity market design is hard because the design needs to be resilient to perverse incentives.
Conceptually, your idea seems reasonable, but because the goal is to put the power from the solar micro-inverters onto the grid, they must lead the grid phase by a few degrees. If they lag instead of leading, they are draining power from the grid and not supplying power into the grid.
People with PV arrays want to make money by selling power into the grid. Perhaps if they were a little less greedy, they could back off the phase difference if they detect the grid phase drifting from too much "pushing." After all, they can't sell any power at all while the grid is down.
That's a thundering herd problem: where all invertors have the same set-points and they all are synchonised to push in the same direction at once.
In networks, thundering herd problems are fixed by given each sender different random delays.
For power networks we could choose statistical methods to get individual solar generators to lead or lag so that the frequency becomes an aggregate vote.
Given that part of the blackout was due to large amounts of solar going offline at the same time, it's possible that all invertors with common software were tripped at the same set-point.
So trip conditions also need to be randomly fuzzy. E.g. if frequency drops below 49Hz +/- random spread of 0.5Hz.
Although it's difficult to match financial incentives against random variations (individual generators are incentivised to power outside of boundaries to keep getting paid, and a trip event can be expensive - due to restart costs).
Electricity market design is hard because the design needs to be resilient to perverse incentives.