I've written about this is several prior threads.
Pass current through a resistive element and it gets hot. The internal fuse link is a calibrated resistance that melts when a certain current is passed through it. Below that current, it will remain intact and conductive but will run warm and it will run warmer as the current increases toward its "blow" rating.
In a perfect world all the heat would come from the fuse link itself. But, there is also resistance between the fuse ends and the fuse clips. This resistance gets worse as the materials oxidize, and they will, with time and exposure to the atmosphere. Passing current through this resistance generates heat, too. The heat can change the temper of the metal which in this case also provides a spring for the clip in order to retain the fuse. The spring pressure can also lower the contact resistance. Lastly, the contact area between fuse clip will become resistive as it reduces in area (usually due to incorrect length fuses).
The main fuse usually runs at about 2/3rds of its rating. Which is warm to touch, but not hot. If clips are making high resistive contact, the heat from that will add to the fuse heat and then melt the fuse element near the fuse ends, where the added heat is introduced through conduction.
So, the correct length fuse is important.
The fuse and the clip contact area must be clean/clear of oxidation (regardless of whether the corrosion is transparent or not).
And, the clips should hold on to the fuse so that a tool is required to remove it. Bare fingers shouldn't do it, a clear sign that there are clip issues.
If you have added extra electrical load to the bike like higher wattage headlights, three ohm coils, and a Dyna-S, you have increased the amperage flowing through the fuse and it will run warmer and closer to it's "blow" temp simply because of that. And, of course, more amperage through any fuse clip oxidation present, will also generate more heat.
Normally, the SOHC4 draws about 10 amps through a 15 amp fuse. High watt headlights, dyna and low ohm coils can add more than 3 amps to this norm.
Still surprised at "fuse" failures?
Cheers,