Hi Curious, for Q1 notice how the fillet is split up into multiple sections - that center zone is not a piece of a fillet surface it's a patch put in after the filleter has tried to slice out a self intersecting area of the regular fillet.

I'm not entirely sure if it's because of not enough space or if it's because of low surface quality, if you turn on surface control points you can see they are quite dense and it's possible the surface has a kind of micro-ripple type structure in it which confuses the fillet solvers.

I was thinking maybe not be enough space, I took it into Rhino to test a bit there and Rhino's FilletSrf command starts to go crazy with some bunching up at around radius = 120.

Ok, I think I see why the space is even more limited than it initially appears, it's because there is some significant twisting/skewing of the surfaces.





That kind of skew has an impact on surface shaping and general behavior which can be hard to notice visually. You can kind of think of it sort of as if the surface is dragging one side of it behind the other just as it is also going into a bend.

A better surface would look something more like this:



Rather than this:



How was that surface constructed exactly? I see some curves in there which do not look super dense like the surface. But there seem to be some duplicates stacked up on top of each other.

- Michael