MCB Tripping - Overheating

The heat off a wire won't cause a MCB to trip BTW, the MCB is supposed to be sized to protect the wire but it only does that by detecting an overload condition due to high amps, rather than the temp of the wire.

What did the CT on the incoming wire (ac grid supply) read when the power dropped? Do you have that logged?

MCB itself shouldn't get hot.

Maybe it should be replaced.

Let me get this right...

Sorry if this seems remedial but I like to be sure: Is "Deye grid CT power" a combination of "heatpump" and "myenergi car" as well as the other house loads, or do we need to add "heatpump" and "myenergi car" to "Deye grid CT power" to calculate the total house load? So was the TOTAL house load 17636W at the time of the second trip or was it 17,636 + 5,000 (heatpump) + 5,000 (EV) = 27,636w?

Also confirm this for me... Does this brown wire loop like I drew?

So different now from the photos? Previously the 40A and one of the 63A's were fed from the 10mm2 coming off the 80a MCB?

I presume you don't have the HA graphs for the periods where the MCB tripped? I'm thinking that you probably had an additional unexpected load on the MCB and it tripped on that basis. As to how to troubleshoot that... I guess that given that it's still only March that isolation of potential loads isn't going to be a popular suggestion...? 😉

Which CT is the shelly and where is it physically placed and what is it's data being used to load-control? Is that white one on the bottom of the 100A AC supply not the "Deye grid CT power"? Also you said this: "...Zappi CT both of which are connected directly to the tails in the Meter box", that's what's confusing me. Do you have any connections to any devices within the meter box that aren't being seen by the last CT in the consumer unit (fuse box)?

If you feel like uploading a line drawing of the meter box, consumer unit (showing major circuits/loads only), Deye, Myenergi and EV charger and then draw in where the individual CT's are, that might help.

In my mind you have too many import limiting devices which are independent loads and not enough overhead built into the draw off the grid. None of these devices are interlinked, each one works to manage the import independently of the rest and part of the problem is the resolution at which many of these CT based devices operate as some only see the grid-load only every 5 to 10 seconds. I know that the shelly is normally faster than that, maybe every second, but the asynchronous nature of independent import load management is what's probably causing the issue here as one device is possibly seeing load gaps and increasing it's consumption on a slow resolution while other devices are also making the same change within that same resolution period. 💥

Yeah the Zappi should dial down based on grid current (I've never exceeded on my own supply to test the theory though), and that would obviously require the Zappi grid limit is set properly and that the Grid CT is 'seeing' all current as per poster above. Have you checked settings/CT location?

The resolution/real time of the CT readings another important point made above.

It's an interesting dilemma alright, as this is the way domestic electrical supplies are heading in the future.

If all those cables are now 16mm2 and you're seeing a 56 degree C hotspot only around the screw-terminal of that breaker, then I'd suggest you might have a bad breaker and/or cable. But it's still not going to be the trigger for the main 80A MCB to trip.