Problem with solid union using fillet curves

For a course I am tryin to design a column, based on multiple lofts through curves. I want to base it on filletted rectangles, but for some reason when I finsihed the lofts, and capped them, solid union only works when the rectangles are not filletted. I need it to be one large solid element, any idea how I can fix this?column (34.3 KB) Column problem.3dm (166.9 KB)

I think I would do this all a little differently… but kept my changes to only a few.

I got radius rectangles to work by making three changes:

  • I used Polygon instead of the single curve in the .3dm file. (it works for 3, 5, 8, 9…)
  • I used the same ‘Plane’ input as the Rectangle component with zero radius.
  • I changed loft options to ‘Type’ = 1 (Loose). (24.2 KB)

Thanks. I will look into it. The base curve cant be done with polygon, since that one is part of the overal design. I will see if the rest works later!

It doesn’t work using the original curve, not sure why. Perhaps Flip Curve? (Nope!) or maybe it’s not a regular pentagon? P.S. That’s it, the pentagon in the .3dm file is slightly irregular.

The radius and height of the five circles looks arbitrary and irregular, which results in wobbly lofts of the rectangles. I would smooth those out.

What would you do differently?

This is just a guess, but I found that when I designed something for 3D printing that had filleted corners I had to use this Rhino command to end up with properly shaped corner fillets:

_DivideAlongCreases _SplitAtTangents=_Yes

I have no idea why this worked, or if it might help your situation, but you might try baking your GH geometry, use the above command in Rhino, and then see if you can use the Rhino Solid/Union command to make a single object.

This uses your irregular pentagon, which ideally I would replace with a standard pentagon (polygon). (19.7 KB)

P.S. A rotation increment of 72 instead of 36 degrees looks cool.

You can remove the “-x” expression on the Z input to build the tower up instead of down, then either modify the Graph Mapper or just reverse the Scale ‘F’ input. (19.5 KB)

The Rectangle ‘R’ (Radius) input can be increased to 27 before SUnion breaks.

Here’s another approach for this type of geometry:

It’s not quite as flexible as Joseph’s more elegant method - with mine you have to use an even number of base curves (the Count slider) to ensure the bottom and top horizontal surfaces line up. (13.3 KB)

Twist is a curious form of morphing but there are a number of differences between this model and the original objectives.

  • Brep Join does nothing.

  • SUnion instead fails.

  • Orient isn’t necessary, you can get the same effect by wiring the Divide ‘P’ to the Polygon ‘P’ input.

  • The square sections all have the same orientation instead of a radial orientation aimed at the center.

  • You aren’t using the irregular pentagon from the .3dm file.

  • It fails badly with 5 instead of 6 (as you mentioned).

  • The Twist angle doesn’t make sense; it appears to be way more than a few degrees.

Thanks for the feedback - I reckon I’ll have to find a way to be more careful.

It is so weird that with some the union does work, and with others it doens’t…

Small changes can affect whether or not SUnion works. It has to do with intersections, coplanar surfaces, tolerances, well formed curves and surfaces. Get used to it, Grasshopper is full of weird quirks and points of failure that are not intuitive.