A Practical Approach to Generative Design

The concepts of generative design and topology optimization appeal to us as engineers for obvious reasons, especially in aerospace. Lighter parts that use less material and carry loads most efficiently. What's not to like?

The reality is a bit more challenging though, as we live and we manufacture in a highly constrained world. Most topology optimization that I've seen is almost academic in nature. The algorithms and tools do a great job at optimizing - doing exactly what they are intended to do. But, the results are usable only in a directional sense. The output is highly complex and is rarely directly usable for manufacturing.

Topology optimization outputs tend to be very large files with an immense amount of data generated by the algorithm. An engineer, on receiving this output, more often than not, uses this optimized output as directional. He or she creates a new part with the knowledge gained from topology optimization in combination with other knowledge about the application or manufacturing method that the algorithm doesn't know. This works. It's effective. It's not efficient. Ideally, topology optimization would result in a more usable output.

I've recently become familiar with a startup and their toolset which addresses this challenge. Novineer has created a generative design software which conducts a robust optimization of part geometry based on anisotropic material properties and user-specified load cases. The difference though is in the output, of which, Novineer will give you multiple.

Unlike tools which only provide an ".stl" file which is of large size and cannot be edited in CAD, Novineer can provide a ".stp" file which is a standard format readable by CAD software.

A performance-maximized part, though, still is likely to have many surfaces and elements which make editability a challenge, so Novineer goes a degree further to provide an alternate output based on discrete polygon geometries. The result is a very small, and highly editable ".stp" file which can be quickly and easily modified in CAD to take into account anything the engineer knows, but the algorithm does not.

There is a secondary benefit to this editable file as well, which can be seen on the left side of the image above. Because of the polygon approximation approach, the resultant optimized design has simple, flat surfaces, which make editing and 3D printing easier, but also lend themselves to more traditional technologies which rarely gain topology optimization benefits, like machining.

To go more technical than I can, there's more on this on Novineer 's website: https://novineer.com/case-studies/balancing_performance_with_editability

Novineer also has some very interesting development when it comes to optimizing, not only geometries, but toolpaths for planar and true 3d technologies. Perhaps we'll come back to that topic another day...

Please reach out to Ali Tamijani , CEO at Novineer for more information or a demo of these capabilities. He and his team will also be available at RAPID + TCT in June. Of course, I'm also always available for questions.