Carrying on from my #13 and thinking about LN's post. I return to the effect caused by block wide bulk time-switched overnight space and water heating.
From what PC1966 said in response to my query fuse rupture does not then seem likely to be due to the simultaneous switch on surge of 50 flats worth of 3kW load each.
Noting that all three fuses rupture and the flats are all single phase it may well still be something all flats have in common like their space and water heating. For the first period after switch on (first hour at least) it may be a storage heater (3kW) and immersion heater (3kW) are all energised and thermostats closed; after a few hours thermostat starts cycling which will/would reduce average load per phase of incomer. Until this happens current per phase could be......
Worst case: 66 x 6kW = 396kW or 132kW per phase. 132000/230 = 575 A per phase.
With background load of 40A per phase as estimated before, current per phase could be 575 +40 = 615A
315A BS88 type gG at 615A ruptures after about 700-800 seconds or 13 minutes. see page 5:
https://espm.co.uk/BS88 fuse info.pdf
We do not know trip time delay for 400A mccb set to trip at 220A. It may not be set up correctly either which seems likely if 3 fuses rupture before mccb trips.
So apartment block has a problem with overcurrent in the guise of overload.
Replacement of the fuses shortly afterwards do not then suffer rupture until following night because new fuses are cold and thermostats are now cycling so mean phase current is less than there rated current. The once nightly epoch of overload has passed.