Over the past few years, there’s been more and more talk about a new class of design optimization tools. Many of these tools create designs directly from the functional requirements that the user sets; others modify existing designs based on finite element simulations. While many of them are both experimental and expensive, they’re worth keeping an eye on. Here’s a rundown of the different approaches that these different tools take, and how they’ll affect your product development process.
Topology and shape
As you begin working on increasingly complex designs, it’s important to know the difference between topology and shape, as most design optimization software will only vary one of them at a time.
The topology of an object refers to the properties that stay constant as it is deformed, twisted, and stretched. If an object (for instance, a coffee cup) can be massaged until it looks like something else (for instance, a donut), then the two are topologically equivalent - regardless of how different they look. For most simple mechanical designs, topology boils down to how many thru holes there are in your part.
Shape, on the other hand, refers to all of the properties that stay constant when an object is translated, moved, and scaled uniformly. Because of this limited definition, it can vary a lot from part to part; while a coffee cup and a donut have identical topologies, their shapes are very different.
Types of optimization software today
Topology optimization
There are a number of software packages that can perform some type of topology optimization. Many of them (like SolidThinking Inspire) are used simply for ideation: they create concept designs that engineers then need to validate and recreate in traditional design software. Others are capable of producing usable designs, though in most cases they’ll be edited for aesthetics and manufacturability. Regardless, these packages require the user to establish a design space (essentially a solid block of material) and apply loading conditions to that design space. Then the software runs FEA on the part and removes material where stress is below a certain threshold - and repeats.
Because topology can be varied, the resulting part might look totally different than the input design space. In general, topology optimization software creates large beams with roughly round cross sections.
Parametric lattice optimization
While topology optimization tools have gotten a lot of attention, I’m not aware of any pure shape optimization software on the market today. Instead, there’s been a focus on parametric lattice optimization. In these tools, a volumetric (3D) or surface (2D, usually non-Euclidian) lattice structure is generated in a design space. The user defines the loading conditions for the part, and usually is allowed to set initial values for the size and shape of the lattice. Then, the parameters of that lattice - beam thickness and cell shape - are varied in order to distribute stress evenly across the part.
Some lattice optimization software allows the lattice density to reach 100%, resulting in totally solid regions of the part. As a result, it actually does produce changes in the part’s topology. Other software doesn’t allow this, and instead sets an upper limit on each beam’s size so that topology can’t be changed during the optimization process.
What it means to you
Cost
In general, optimization software is expensive - often running into the mid five figures. Most of it is fairly immature software, too: the workflow and capabilities are still in flux. Even if you’ve got $50k burning a hole in your pocket, it’s probably a good idea to ask for a demo of the software before you get your hopes up.
Workflow
Let’s just put it out there: With all of these packages, you’re going to get way friendlier with STLs. Pretty much all of the optimization software out there today requires either STLs or OBJs, and in general if you want to play in this realm then you’ll need to accept that mechanical features won’t be easy to manipulate.
There’s also a significant adjustment when it comes to thinking about your part’s design space. You’ll want to give the optimization software as much room as you possibly can, and think critically about how the part will interact with the rest of its environment or assembly.
You’ll also need to know a lot about your part’s loading conditions, and be ready to set up and rely on accurate finite element analyses. For some parts that might not be an issue, but the process of defining complex simulations can be arduous - especially when you’re working with STLs that don’t offer much in the way of feature recognition or editability.
Manufacturability
Unfortunately, I’m not aware of any optimization packages (save maybe Tosca) that take into consideration manufacturability or cost - at all. This can be a huge impediment to your product development process, and it can be frustratingly difficult to overcome.
When evaluating optimization software, think carefully about what your manufacturing constraints are going to be - and whether or not those constraints can be set accurately in your design tool. If a software package is going to make it difficult to reduce overhanging faces, for instance, then see if there are ways to adjust your design space in order to force the results you want.
Additional software
The farther you get into optimization, the more you’ll want to be familiar with the full build prep process. I’ve found it helpful to have a working knowledge of a variety of STL operations, including sculpting and repair. It’s also good to spend some time generating support structures and testing build orientations. Especially given that optimization software tends to produce designs that are extremely hard to build, knowing more about the DFM process can be invaluable.
Conclusion
Design optimization software is at a weird place today. Most of the big CAD providers are thinking a lot about how design will work in the future, and for most of them generative or algorithmic design is a big part of that. But the tools out there today are far from maturity, and the workflow can be a shock to designers who are used to working in conventional ways.
In my next post, I’ll go into more detail about the optimization process, and give examples of how it can be used on real parts. Stay tuned!
Modernize Your Engineering Curriculum
This eBook is aimed at the engineering professor, new or experienced, that is interested in how their peers are thinking about the challenges associated with modernizing.
