Hello @Anonymous ,
This is a common observation with users of random part nesting. While random part nesting is great at finding solutions to general problems with lots of variety in parts, there are special cases like the one you have here where it does not work as well. You have essentially found a special case where a completely different nesting algorithm is more ideal than the random part nesting algorithm we're using in the Inventor Nesting product. Random part nesting would be a much better solution if there were several different shapes rather than just one being repeated. There are many scenarios of one shape other than just a circle where random part nesting will not give you the optimal result, and that actually isn't unusual for random part nesting algorithms.
As you may know, Inventor Nesting includes some of the technology from the TruNest product, and TruNest supports the following nesting algorithms:
- True-Shape Nesting (multiple random shapes; punching/chiseling, profile cutting/torching)
- Rectangular Nesting (rectangular shapes; Linear saw cutting, Shearing)
- Linear Nesting (lengths for bars, extrusions, rods; chop saws)
- Part Packing Algorithm or High-Volume Nesting (one or two shapes nested in high volume, stamping, profile cutting/torching with multiple heads)
While we have included the True-Shape nesting algorithm into Inventor Nesting, we have not included the other three (yet), which are designed for more niche manufacturing processes or are designed for more specific geometry scenarios.
In this particular case, the Part Packing algorithm would be more suitable, as it is designed to find repeatable patterns that are optimally designed for high volume of one or two parts. It would come up with a solution similar to what you have shown in your solution, which random part nesting will not. Here's an example of part packing on something a little more complex than a circle:
This solution is great for die or stamping applications for high volume production, but it also solves this specific scenario better.
I don't want to go into tons of detail about the math and why it is the way it is, but there is a good reason why you see the behavior you do with random part nesting. I hope this assures that if you were using more than one part, random part nesting will do a better job than you can by hand, and if you were using part packing, it would likely outperform anything you can do by hand in the single part scenario as well.
Phillip Doup Architect - Fusion Platform UI
