The failure here wasn't overcurrent, which is what fuses (or breakers) mitigate. Here it was poor connection, no fuse or breaker would help. And in fact didn't since the author mentioned he had breakers on these lines.
Also, and particularly in high current circuits like these, fuses generate significant heat on their own, at the connection points.
The better bet is to look to conservative standards (like ABYC for the marine industry) and follow their rules for wire termination. In this case that would be properly crimped ring terminals over simply wrapping stranded cable around a screw. That said I am sure the NEC has something to say about domestic solar installations.
The idea is your inline fuse overheats and melts instead of the entire controller turning into fire due to the heat cooking the plastic at the rail or terminal. It's a passive device for both overcurrent and overheating. Or just step into 2026 and use mag breakers, there are lots of DC breakers that do this: https://www.amazon.com/Miniature-Circuit-Isolator-Magnetic-D...
The issue wasn't heat generated by excessive current, but heat generated by excessive resistance and normal current.
Mag breakers and fuses won't help. A thermal fuse could work,provided it was thermally bonded to the bad connection. So you'll need a lot of thermal fuses - one at every potential bad connection.
This only happens when there is an overcurrent condition. As OP,
scrumper, BenjiWiebe all told you there was no overcurrent condition. The system was rated for a 100 amp, the current was under that.
The thermal overload breaker won't help here, because it only trips when the current is above breaker rating. And one would install at least a 100A breaker into a circuit rated for 100A, or it'll just trip all the time.
(You can see the detailed curve on datasheet, example page 2 here [0]. It is surprisingly slow - for 1.5X overload, it takes >10 minutes to trip).
The problem was because of bad connection on 100A line. We can talk about mistakes made here, but lack of breaker was not one of them.
Also, btw, the 'mag' part of the breaker is irrelevant. The magnetic component is for sudden large overcurrent events, to provide a quicker break than what a bimetallic strip can do.
Also, and particularly in high current circuits like these, fuses generate significant heat on their own, at the connection points.
The better bet is to look to conservative standards (like ABYC for the marine industry) and follow their rules for wire termination. In this case that would be properly crimped ring terminals over simply wrapping stranded cable around a screw. That said I am sure the NEC has something to say about domestic solar installations.