How does generative design unlock engineering innovation?

For those who aren’t aware of the term, what exactly is ‘generative design’?

---

Generative design is a method of achieving results by sharing your goals, objectives and constraints with the computer, and partnering with it to explore the solution space before you commit to just one and take it all the way through to manufacturing and delivery.

When CAD and CAM tools were first introduced in the 1960s, computing was scarce, with analysis and simulation used incredibly sparingly. Humans did all the hard work and used the computer to simply preserve their thoughts or designs.

In a nutshell, computers were used for documentation, not for ideation and exploration. As a result, we ended up committing to the first design that sort of worked, as opposed to iterating and finding the best design. Generative design helps you find that best design by showing you the entire solution space and enabling you to navigate and discover what some of the trade-offs are before you commit to one prematurely.

How far has ‘generative design’ progressed since we last spoke about the concept at Autodesk University 2014?

I’ve been intent on trying to overcome this frustration that I’ve got with the computer, where I must tell it everything I want it to do, quite literally. I also get frustrated in that, day by day, the computer gets no smarter, it’s as dumb the next day as it was the day prior; it never learns from any of my interactions. In both of those two areas, that’s where we’re making the most significant progress.

Through Project Dreamcatcher [Autodesk’s generative design system], we’re now able to describe what it is that we want the computer to accomplish without specifically stating the pathway to go there. The user may not know which manufacturing method they want to use, be that casting, milling or 3D printing.

They may not know what material they should use or the different trade-offs between their objectives. They want to understand the solution space across cost, manufacturing time or, for the part itself, strength, mass or light-weightedness of the result.

Getting these descriptions initially into the computer, we’ve made tremendous progress. We’ve also seen an acceleration in terms of formulating these criteria in direct parallel to the compute power offered by the cloud. Our next challenge is taking the raft of solutions that come from the explorations that the computer can do and recommunicating them back to the user.

A big challenge is that users define what they’d like to see, but what they get back doesn’t precisely match that. Generative design represents a step change in terms of that journey.

Our research has mostly been focused on the user interface, more so than on the simulation analysis or form generation because those problems, for the most part, have been addressed. Designs that look ‘engineering’ minded, we’re well on the path; more creative, human-empathetic, computer-to-user designs, that’s where we’re applying the most work.

What have been the reactions from designers and engineers who’ve been introduced to generative design?

There’s two camps, one which says, “Oh no you don’t, that creative process, insight and wisdom is what I as an engineer or designer would bring, how dare the computer encroach upon that expertise!”

The other says, “Thank god you’re allowing the computer to shoulder the drudgery of simple drawing, drafting and form creation and allowing me to partner better with the computer to understand what the real goals and constraints around the problem are.”

Fundamentally, why do designer or engineers of any description get excited by their careers? Because they like exploring how goals map to user’s concerns, learning how their customers truly conceive of the constraints. That’s the conversational level that generative design enables, as opposed to going all the way down to the minutiae and modelling every bolt and thread.

How reliant is generative design on the fundamental technologies which support it, such as cloud compute power and additive manufacturing? 

Generative design is the first design paradigm designed specifically for the scale of cloud computing, in fact it would be almost impossible to accomplish generative design on anything other than the infinite scalability that we have in elastic computing in the cloud.

Desktops, that form of computing which occurred in the box underneath your desk, often resulted in non-out of the box thinking because it constrained your work to whatever you could accomplish on limited computing power. That never changed, even as machines got better and could accommodate more advanced processes, i.e. simulations.

The process of accomplishing a goal started off with a design, moved to simulation and analysis and any imperfections meant going literally back to the drawing board again. That whole process of modify, analyse, repeat was based on constraints which weren’t engineering-based, the limitations were always time, money and patience.

Within that system, there’s no guarantee that you’d come up with a great, breakthrough results. However, if you switch it and make the computing so available and fluid that you could compute all the solutions in the time that it would have taken to do just one, that creates an entirely new, very exciting relationship with design tools. Now, so long as you can express what it is that’s truly driving your design, you’re going to get a tremendous result.

When are manufacturers going to start taking full advantage of new, advanced digital technologies?

Take additive manufacturing, that technology has been largely blocked to date; blocked because using additive to reproduce the same part that had been formally milled or cast is a complete waste of time. It just means that you’re going to have a bigger fault in the manufacturing process as it’s not’s quite as reliable as those other systems just yet, and it’s certainly going to be far more expensive.

Generative design completely unlocks the power of additive manufacturing because the forms it can create don’t constrain themselves to those solutions that humans would have come up with. In fact, the shapes that generative design creates often look like a lot more like nature.

Some of the topologies that would be challenging to cast or mill, are just as trivial to 3D print as a solid block. Where we find the greatest adoption so far is in those manufacturers who already have some expertise with the most astute optimisation tools and have also been exploring additive manufacturing applications, and they see that these two things married together can take them to the next level.

We are talking here about industrial innovation, on both the design and manufacture side. How do you see industrial innovation evolving?

We’re going to reimagine industrial processes. One of the most exciting things about generative design, even combined with additive manufacturing, is you don’t have to adopt a radically new design and manufacturing process all at once to start driving benefits.

Being able to explore a space, even if you don’t intend to reproduce any objects that have been derived in generative design, gives you an understanding of what the actual trade-offs are. I used to hand around this example of a motorcycle swing-arm which was produced in Dreamcatcher and came out asymmetric – the left and right side were totally different.

Designers naturally gravitate towards symmetry as a balance and design principle. That’s not at all necessary for the solutions that the computer’s able to come up with, it doesn’t have that constraint that it brings along from prior designs. It’s able to show you that symmetry isn’t necessarily a requirement. You don’t have to pick that solution, but it’s informed you of a flexibility in what you might have imagined to be a fixed limitation.