Bent packing

A circular puzzle by Dr. Arne Köhn

Ever got angry at a puzzle and bent the pieces? And then asked yourself whether you can make a new puzzle out these pieces? no? Well, me neither. I designed this puzzle in my head while bringing my kid to bed and then did some minor adjustments on paper.

The goal of this puzzle is to put all pieces into the cylindrical box such that there are no unfilled holes in the box (in other words: make an "apparent cylinder"). Are rotations required? Depends!

license

This puzzle is distributed under a Creative Commons attribution license (CC-by 4.0). You may print, change, distribute, and sell the puzzle as long as you give credit to me.

As I am not selling this puzzle, consider making a donation to a good cause if you like the puzzle.

printing

If you just want to print the puzzle, you can download the model files from printables.

There are three vertical pieces (v1 needs to be printed twice) and three horizontal pieces.

The vertical pieces are best printed on the side (this is obvious for v1) so the layers glide more smoothly with the other pieces and the box.

This puzzle is quite forgiving with tolerances and nozzle sizes, as there are no sharp corners needed. I print it with a 0.8 nozzle.

While not needed for a solve, the piece slide is much more satisfactory when printed with high infill to add weight. I use 50%.

rendering and making changes

This design requires the use of The Belfry OpenScad Library, v2.

To render the pieces without an interactive openscad session, run:

openscad-nightly --backend=manifold -D torender=\"all\" bent_packing.scad -o bent_packing_all.3mf

you can also render the pieces individually:

for p in box lid h1 h2 h3 v1 v2; do
  openscad-nightly --backend=manifold -D torender=\"$p\" bent_packing.scad -o bent_packing_$p.3mf
done

There are some variables at the top of the scad file that change the dimensions of the puzzle.

You can print the threadtest object if you want to test which slop value works best for you. It prints three different threads and you can change the slop values for them.

--backend=manifold is not necessary but cuts rendering time drastically.

random things

The code is structured in a way that it should be fairly easy to design new puzzles with the same geometry. If you do so, forward it to me if you want, i am interested in new designs.

I used minkowski for beveling. Derek Bosch made me aware of this technique and adapted his approach by using bicones for the operation. I think that worked out quite well.

There is a solver (only the assembly part) for this puzzle over at https://github.com/akoehn/circular_assembler Do check it out, the code is straight-forward to read and fun.