Is this actually how it is done? I would assume that the complete ENTSO-E grid cannot be perfectly phase-synchronous anyway, simply due to the physical dimension of the grid and the speed of light.
I would have guessed that frequency-synchronicity is sufficient for coupling two grids, and then you could use something like a phase shifting transformer to control the power flow?
As far as I'm aware this is how grid operations are done in the overwhelming majority of the world. You spin a generator up to speed, match phases and close the switch to connect it to the rest of the network. Due to how electricity works, if any connected generator is slightly out of phase with the rest that will generate currents causing it to pulled back into phase in a closed feedback loop. So while it is technically possible for a generator to be out of phase with the network, it can never be outside of phase for very long because literally all the other generators on the network are pulling it back into phase.
If the grids are particularly far from each other and/or with differing frequencies, a DC high voltage line can be used instead but that is certainly not the most used method.
You only need to match phase at the connection point, so the phase of the grid in France doesn't really matter (as long as ENTSO-E isn't destroying itself before you do anything).
This does get much trickier if you have multiple connection points and it is impossible to be synchronized at all of them because the in-grid distance between the two points is different on the two grids. Then you definitely do need fancier tricks.
True, but you need to have a very large unbalanced grid for that to occur: at the grid frequency (50 Hz), the corresponding wavelength is 6000 kilometer. So assuming that math is correct, you would need a path length difference of 3000km between two transformers to get full phase inversion between the paths.
That's more than twice the max diameter of the entire European grid (Tunesia-Gibraltar-Turkey seems the longest path). So as long as there's transformers every few hundred kilometer, and multiple paths between most of the connection points, the overall phase synchronization should be adequate to not need those fancier tricks.
Some country interconnections, in particular those to UK and Nordic countries, have the ability to connect even unmatched phases, by for example converting to and from DC (AC -> DC -> AC).
That ability to connect unmatched phases is secondary to the primary function of those interconnects: over those distances it is more efficient to transmit the power in DC instead of AC. [0]
Sometimes connecting unsynchronized grids is the primary reason. The Outaouais HVDC Intertie is only about 35km long, but it allows the import and export of electricity between Ontario and Quebec.
Even more fun is that Denmark despite its' very tiny size is split in two when it comes to electricity. West Denmark is connected with East Denmark through HVDC.
This was how the Tres Amigas Superstation was supposed to work.
Unfortunately, ERCOT decided they didn't want to be subject to the same regulations as the Eastern and Western interconnects, so the project collapsed.
I looked into this once, and they .... just don't worry about it.
Each generator just matches whatever it sees. If you are trying to match two other generators (like in a triangle, and due to the speed of light it's impossible for all to be perfectly in sync), then it matches to an average of the two, and apparently that works.
When comparing Sigfox to LoRa, they are not even that bad in terms of openness:
The Sigfox protocol specifications are public: https://build.sigfox.com/sigfox-device-radio-specifications
This is in contrast to LoRa, which remains a closed protocol (I am not talking about LoRaWAN, that part of the spec is open, but LoRa, i.e. the physical layer stuff).
So there really is a tradeoff here:
If you want an open wireless protocol compatible with many standard RF transceivers, but a proprietary backend and device library, go for Sigfox.
If you want a closed, proprietary wireless protocol that only works with Semtech chips, but an open backend, go for LoRa/LoRaWAN.
Since several projects have done that one might ask them if they have considered this. For all we know, they may have discussed this with Semtech, in which case the status may be knowable.
I would have guessed that frequency-synchronicity is sufficient for coupling two grids, and then you could use something like a phase shifting transformer to control the power flow?