Just to clarify that last post ...
If we ignore for a moment the rest of the network, and just think about the distribution cable running along a single street, and lets say there are 15 houses, 5 on each phase. So the DNO sets stuff up so that under no load there's 250V at the feed end of the cable, and under moderate load there's only 230V at the far end - yeah I know, I'm making this up as I go along, it's only to illustrate the issue.
All fine, the users always get something within the allowable voltage range.
Now one or two of the houses on a phase get a 4kW solar PV array installed. Mid-day on a week day when everyone is at work, the sun is blazing - don't laugh, we did have a few weeks of that not long ago and we ended up complaining
So now, instead of there being a power flow FROM the DNO along the cable, there is now going to be a power flow TO the DNO along the cable. Again, ignoring the rest of the network and assuming there's still 250V at the DNO feed end of the cable, the voltage at the houses along that cable will now be higher - potentially going over the 254V upper limit.
So now, the DNO gets complaints from those with the PV arrays that their inverters are cutting out on over-voltage. They have to go back to the nearest tap changer and lower the voltage in the network to allow for this. I don't know if they routinely have taps on the local 11kV/415V transformers - if not then it would go back to the next level at the 33/11kV transformer and affect a wider area.
I could see some interesting situations. Suppose there's a small hamlet at the end of a long distribution cable, and they have night storage heaters because there's no mains gas - I did know someone living in just such a situation. At night, the voltage will drop "quite a bit" due to the heaters which will restrict the DNOs options for lowering the off-load supply voltage. If one (or more) of the residents then gets solar PV installed, the DNO then has to deal with not only the large volt drops under load, but also the potentially large voltage lifts when the sun is shining.
So it's easy to see that embedded generation is far from a "zero cost" to the distribution network.
And that's before you start to think about protection. The distribution cables are protected at the supply end by fuses, put a load of embedded generation onto a branched network and you could potentially significantly increase the fault current in some of those branches - we don't know how fast the PV inverters will react and shut down when there's a fault that drags the voltage down.
