Generative Design: The Road to Production

Generative design is promising to revolutionize the way we design—by using algorithms inspired by the way in which bones grow in animals—to place material where it needs to be and removing it from where it doesn’t. Generative design is promising to shatter the old philosophy of form follows function and replace it with a new philosophy of form follows force.

Here at engineering.com we have covered generative design in quite a lot of detail—it is definitely something of a ubiquitous topic on this website.

But given the amount of coverage we devote to this topic and how it is promising to change the very world we live in, we have to ask this question: Where are all the products that have been promised?

We spoke to some of the companies involved in various aspects of generative design to ask those questions and more.

First up, we spoke to ANSYS, which is very much involved in generative design and topology optimization, particularly with its ANSYS AIM software.

According to ANSYS, end-use products are indeed being manufactured as a result of generative design, but they’re still out of reach for the general public.

“The aerospace and defense community has rapidly adopted the ANSYS topology optimization for tasks such as lightweighting,” said Steve Pilz, Additive Manufacturing Engineering, ANSYS.

“While much of this work is classified and the production scales are relatively low compared to industries such as automotive and consumer products, we can say that in applications such as satellites, rockets and weapons, the adoption rate has been high.”

So, the products are out there, but we just don’t have the security clearance to know exactly what they are. At the moment, the costs associated with generative design products at low scales of production are prohibitive for mainstream use.

However, it’s not just cost that prohibits the technology’s widespread adoption in consumer goods. There are cultural considerations, too.

“In some industries, such as those mentioned above,” continued Pilz, “they have not been reluctant to move into production, but in other markets, factors such as production scale, unit part cost, lack of production capabilities and fear of consumer acceptance have caused restraint.”

But is that fear of consumer acceptance equal across the board? Or are there some industries that are less risk averse when it comes to new design paradigms?

“Sporting goods manufacturers may be the first to adopt generative design in mainstream commercial products,” said Pilz, “particularly those for extreme athletes that are looking for weight savings and are not primarily concerned with nontraditional aesthetics.”

“Physics-driven design can be tailored to produce designs that can be made with traditional manufacturing techniques, such as extrusion, forging, casting, etc.,” said Pilz. “While often not as optimal as designs driven without traditional manufacturing constraints, the unit costs can make optimization viable for lower cost/higher production run components.

“We are seeing this already and have been driven by customers to enhance our software capabilities so that techniques like casting, extrusion, etc., are viable alternatives for a physics-driven optimal design.”

So, it’s clearly not a case of traditional manufacturing not being up to the job. It seems to be a case of how you adapt the generative design tools to be used with traditional manufacturing techniques. After all, if you use less material and still manage to reduce the number of tooling operations, the unit cost will still be reduced. Generative design is providing designers with new ways of looking at old problems.

But when will we see these products hit the main stage? When will we see these goods in our living rooms?

“Predictions are hard, particularly if they involve the future, but a number of factors will decide the velocity by which newer design and manufacturing technologies will overtake traditional techniques such as subtractive machining,” concluded Pilz.  “Some of these factors will be the rate at which consumer perception changes about aesthetics, that is, what is beautiful in a structural or multiphysical sense, as well as the rate at which unit part costs come down as newer technologies, such as additive manufacturing, continue to overtake more wasteful or environmentally harmful production techniques.”

Next up from the CAE realm, we spoke to Ravi Akella, director of Product Management - Generative Design and Digital Manufacturing Simulation at Autodesk.

“Autodesk generative design is being incorporated by our customers into their product development process.” said Akella. “While this is leading to productivity gains and higher-performing designs, the technology only became commercially available in April, so we haven’t yet seen examples of commercial products. However, we believe customers are well underway in adding and benefiting from generative design in their product development workflows.”

Of course, the Autodesk software that Akella is referring to is the Autodesk Generative Design platform, which first appeared commercially in Autodesk Fusion 360 in April of this year. We covered that story here if you’d like to see how it works.

Anyone who has seen the Autodesk Generative Design platform (or its predecessor, Dreamcatcher) in action will agree that these tools can create some fairly mind-bending designs. But what is it going to take for designers to migrate these designs from prototypes into actual large-scale production?

According to Autodesk, there are still cultural and technological hurdles to overcome first. 

“The outputs of Autodesk generative design, at first glance, can be unusual and nonintuitive to engineers accustomed to more traditional-looking designs,” said Akella.“But our customers are beginning to appreciate that the technology allows them to explore a plethora of practical, manufacturable solutions by simply specifying the engineering problem they aim to solve. This is unlike traditional tools such as topology optimization that merely enable incremental improvements to traditional designs. We are observing that once engineers have performed comprehensive validation using traditional simulation tools, they build trust in the outputs of generative design. This growing trust along with an increasing rate of adoption and ever-decreasing costs of advanced manufacturing techniques means that generatively designed parts will soon be in production. We’ll also see an increase in production parts created generatively once our technology starts creating designs suited to traditional, lower-cost manufacturing processes.”

So, we are seeing some sort of consensus among the CAD people—lowered costs in manufacturing will enable the move from expensive prototypes into cheap mass production. And slowly but surely, engineers and consumers are becoming accustomed to the strange new forms that generative design allows. 

Therefore, which industries will be the first to bring these generative-designed parts to market? 

“The adoption of Autodesk generative design is happening most quickly in industry segments where a premium is put on high-performance and uniqueness of designs,” said Akella. “We have seen interest from customers ranging from makers of home appliances to hand tools. In addition to these characteristics, the most practical cost benefits are realized when multiple parts are consolidated into one unique and original component, as highlighted in our recent success with General Motors. Part consolidation simplifies the customers’ supply chain and reduces downstream assembly costs.”

We speak to General Motors later in this article, so we will elaborate more on what Autodesk means by part consolidation later on. But generally speaking, when we think of parts being consolidated like this, we tend to be speaking of parts manufactured with additive manufacturing. The printing of multiple parts at one time means a reduction in assembly time, inventory and fastenings.

But is the growth in generative design driven exclusively by additive manufacturing? This was definitely the case in the beginning, but Autodesk agrees that this is no longer necessarily the case.

“Autodesk generative design produces geometrically complex designs that are manufacturable and practical with additive manufacturing,” confirmed Akella. “With every passing year, additive manufacturing is becoming more mainstream and accessible as costs come down and more materials are supported. And, as we expand to producing designs that are suited to processes like machining, casting, etc., we will enable more and more customers to mass produce their generatively designed products.

“Generative design is inherently about understanding requirements, from performance to materials to manufacturing processes, and providing engineers with a variety of practical, manufacturable options to make informed trade-offs between time to market, cost and performance,” he continued. “Ultimately, our goal is to enable designers and engineers to produce more, better, with less—more new ideas, products that better meet the needs of users, in less time and with less negative impact on the environment. Autodesk generative design will not be tied to one manufacturing method in the future,” said Akella.

That’s good news for traditional manufacturers. Generative design is not just for those with high-end metal printers, but for anybody with a CNC-enabled workshop and access to generative-designed software. 

So, it appears that Autodesk is gearing up for the generative design. When will we start to see products designed with generative design become mainstream? Like ANSYS, Autodesk is optimistic that we will see them soon.

“As soon as generative design starts creating designs that are suited to multiple manufacturing methods, we expect mainstream adoption to grow rapidly,” observed Akella. “We also see large potential in industries where high-value, low-volume products are produced.”

That comment is a reflection of what we heard from ANSYS. High-value, low-volume products, such as those in aerospace, automotive sports, and medical devices would fit that description.

That’s all from the CAD/CAE folks who are leading the charge into generative design. Now let’s see what the product designers and manufacturers have to say.

When we think of a utopian future of generative design, one image that is conjured up is that of architecture. We envision gleaming, sinewy spires taking loads where they need to be in an efficient, organic fashion. There is a lot of room for generative design disruption in the built environment, ranging from mass-produced fixtures and brackets, to concrete structures 3D printed on the spot.

Construction consultancy firm Arup Group has been looking at generative design for quite some time.

We spoke to Salomé Galjaard, senior product and interaction designer in Arup's Amsterdam office to get the lowdown on how the built environment is utilizing generative design. As you can see from the image of tensegrity nodes that follows, Arup has been involved in designing fixtures for use in architectural projects.

The image shows the evolution of so-called tensegrity nodes, from a traditional design on the left, to a fully generative design-based product on the right.

“This didn’t sound very practical,” continued Galjaard. “So, my reasoning at the time was, instead of adapting our designs to available materials and traditional production techniques, isn’t there something out there that we can actually use to help us solve this challenge? And that’s what we researched, and we took that tensegrity project as an example because you have all the boundary conditions and you know exactly what you want to design and tests that also can be done and applied with 3D printing in mind.”

It’s worth noting that Galjaard and Arup designed these tensegrity structures a few years ago and announced them to the world back in June 2014. The generative design landscape was still very much in its infancy and there weren’t as many software options available to perform generative design back then. It was therefore still very much a case of trial and error for Arup to arrive at a solution that it was satisfied with.

“We tried various software to get to the final iteration,” said Galjaard. “You can see in the picture we show two design iterations—the middle one that is still quite similar visually to the original node was our first iteration. And you can see version 2.0, the second iteration on the far right—this is where we really made the big step with the freedom of form with 3D printing in mind.

“For the final one in the picture, we used OptiStruct by Altair for the topology optimization and Rhino and other support tools for modeling and for integration of the other functional elements.”

As with the other companies we spoke to, Arup is aware of the cultural hurdles associated with bringing generative design to the masses.

“If you look at our models, especially the 2.0 version, this is not something that the human mind can conceive—well, not me at least!” noted Galjaard.

“I’m working on multiple topics that feature technology that is considered new and we consider being essential to the building sector, or to society as a whole. And the real threat that is common to this thinking is that it’s not really the technology or ideas that are holding us back; it’s really people management and people unwilling to accept change—complete change, value change, all happening at the same time.

“What you’re seeing with 3D printing is they prefer to take traditional manufacturing out and put 3D printing in without changing anything around that, and this approach isn’t going to work. These new technologies will only be allowed to work if you allow the context around it to also adapt, and it’s not as simple as people want it to be sometimes. It may take a little bit of time [to reach mass adoption],” said Galjaard.

Of course, architectural projects operate on different time scales compared to consumer product development lifecycles, and this is another reason why civil engineering and the built environment are slow to adopt the technology.

“I think there are multiple reasons [for low production adoption] working at the same time. One is the time frame we are working at with the development of designs and the construction of buildings. If we have an idea now, you will only see that built on-site in say five years from now. We work in a complex value chain where no one has complete control over the final results, so it is a lot of different people and a lot of different companies that have to move at the same time or at least following in a series, and what you see in these processes is that they all start with wonderful ideas and throughout the years and the phases they become victim to planning pressure and costs and worries,etc., so a lot of new ideas and challenges are just designed out,” explained Galjaard.

Again, it is the cultural reluctance to embrace new ideas that is preventing the widespread acceptance and usage of products and processes related to generative design. Galjaard is confident, however, that technology is up to the task.

“So, I don’t think it’s to do with 3D printing not manifesting itself—more it is to do with change being difficult to establish because change is scary, change might be costly, and change has to always be better in every single way than what we have now—and that is something that is very difficult to foresee and to guarantee, so it’s little steps.”

So, there you have it—the built environment sector is also optimistic about the applications of generative design, but like everyone else, they are wary of the cultural hurdles associated with the uptake of new technologies, especially in a safety conscious industry such as construction. As Galjaard mentioned, it’s just a matter of proving that the technology works, and demonstrating the products until they reach a point of maturity where everyone realizes that the benefits of the technology outweigh the costs—be they financial or cultural.

We mentioned earlier that Autodesk has been working with automanufacturing giant General Motors on a variety of generative design-related items.

From GM, we spoke to Kevin Quinn, director of additive manufacturing, to see exactly how the vehicle manufacturer has been bringing generative design to the masses—or has it?

“Currently, we have not made a production part,” said Quinn. “We have created a proof-of-concept seat bracket that combines eight components into one using new generative design software from Autodesk,” continued Quinn. “The concept part is about 40 percent lighter and 20 percent stronger than the bracket we use today. We are actively working on several other part design projects using this technology that we expect to go into production on low-volume vehicle applications soon.”

But what’s the holdup? Generative design has been around for a few years already. Why are companies reluctant to move from prototype into production?

“I am unable speak to what other companies think, but I can tell you that at GM we explore many options to lightweight our vehicles,” said Quinn.

“This has less to do with how the part(s) look and more to do with how they perform and at what investment. For each application, we will evaluate cost and performance benefits as compared to a more traditional manufacturing process. We may choose to invest more to 3D print a part if the added value in terms of performance merits the additional cost. If a traditional manufacturing process is less expensive than 3D printing and there is not a substantial performance benefit, then traditional manufacturing processes will be used.”

From that statement, we can see that generative design is indeed synonymous with additive manufacturing for the most part. And Quinn confirms that.

“At GM, we see great potential for generative design combined with additive manufacturing processes to enable part designs that are lightweight with the performance criteria we expect,” he continued. “Metal 3D printing costs are still relatively high in comparison to traditional manufacturing methods. GM has taken a leadership role working with the 3D print industry to advance capabilities,” said Quinn.

Although there is a kind of symbiotic relationship between additive manufacturing and generative design, GM does see generative design branching out into other, more traditional forms of manufacturing.

“Generative design can be used to design parts for some traditional processes,” Quinn continued. “For example, casting processes could be used to manufacture some of the generatively designed parts.”

So, GM is all in with additive manufacturing (as long as the performance benefits of the part being manufactured outweigh the other costs, but how long does the company think it will take for generative-designed products to reach the main stage?

“It’s often difficult to predict the adoption curve of new technology,” said Quinn. “We see great potential in generatively design parts coupled with additive manufacturing to give us options that quite frankly we haven’t seen before in the auto industry.”

We have had some insights from a wide range of industries and from different roles within those industries. We can all agree that additive manufacturing is the key enabler of this technology.

So, to round things off we spoke to Scott Kraemer, production development engineer at 3D printer manufacturer Carbon, to see how the actual additive manufacturing companies view the future of generative design.

First up, we asked Carbon if it had heard of any production-level parts being manufactured with its system.

“Carbon’s proprietary Digital Light Synthesis (DLS) technology and programmable liquid resins drove the production and iteration process of Adidas Futurecraft 4D, a high-performance athletic shoe that is currently slated for 100,000 pairs in 2018 and millions following that,” Kraemer revealed.

Boom! We have caught our snipe! Yes, Carbon is totally onboard the mass production train!

“Additionally,” continued Kraemer, “many other companies are using Carbon’s digital 3D manufacturing solution to produce a variety of products, such as commercial blender rinse nozzles for Vitamix, mobile device protection for Incase, automotive parts for companies like Ford and BMW, and medical devices such as 3D-printed dentures, pediatric airway stents, and tuberculosis diagnosis tests.”

“The $12 trillion manufacturing industry is slow to change because it can be expensive to modify manufacturing processes, and many companies are set in their existing supply chains,” replied Kraemer (echoing Arup’s sentiments).

”They’ve largely been reluctant because of the lack of scalability, cost-effectiveness, materials and technology needed for generatively designed production parts. However, with technologies like DLS and Carbon’s wide range of programmable resins, it’s much quicker and cost-effective to design, engineer, iterate and make production parts.”

Now we have found our unicorn. The Adidas project provides a fine example of how generative design is coming into our homes enmasse. Let’s hear some more about that product from Carbon and how generative design has enabled that product to come to fruition.

“Carbon’s rapid product development process enabled Adidas to iterate over 50 different lattices for the midsole before landing on the current design,” continued Kraemer.

“Traditionally, product development has followed four steps: design, prototyping, tooling and production. This process is so slow that many product development programs can only afford three to five redesign cycles before finalizing the design, constraining the product team’s innovation. The Carbon-Adidas collaboration made 10 times as many iterations possible. Further, each iteration was produced with the same process and material as the ultimate product. These were not just prototypes for assessing a design’s visual appeal: Adidas could actually test midsole performance in the design stage. In other words, Carbon’s technology makes prototyping obsolete and allows for scaling up.”

For anyone who has seen the Carbon 3D printer at work, it is very fast indeed, so this does contribute to the fast iterative cycle and, consequently, the rapid production process that is allowing millions of pairs of sneakers to hit the streets. Other 3D printer systems are still a little on the slow side, so that would explain why Carbon seems to havethe edge. You can see the Carbon printer in action in the video below.

And Carbon is not planning to rest on its laurels now that it has the edge.

“The 300 pairs of Futurecraft 4D released in April 2017 were just a start,” continued Kraemer. “More than 5,000 pairs were released for retail in fall/winter 2017, ramping up to 100,000 pairs by the end of 2018. No other company has stated plans to scale a 3D-printed product to the hundreds of thousands. Another more recent example of large-scale production would be our partnership with Vitamix, where Carbon’s DLS technology is being used to make tens of thousands of rinse nozzles per year and getting deployed in stores. These are proofs that 3D printing can be used at scale for manufacturing, with the right tech and materials to change the way companies create, test and sell customized and mass-produced products.”

Carbon is optimistic that the future of generative designed consumer goods is indeed just around the corner, and it does see it self as a leader in this field.

“Transforming the traditional manufacturing sector will involve people, processes and technology to all come together for generative-designed products to become mainstream,” said Kraemer. “Along with Carbon there are already a handful of companies out there simplifying generative-designed products with additive manufacturing (like Desktop Metal for the metal additive manufacturing space).

“The complicated and unique lattice structures of Carbon and Adidas’ Futurecraft 4D sole is also one of the industry’s leading examples of generative design applied to a mass-produced product that captured mainstream interest upon its announcement. Carbon’s partnerships with Adidas and Vitamix are examples of the vote of confidence from industry giants that additive manufacturing has the very real potential to bring generatively designed products to the masses.”

You can also see the Carbon video of their collaboration with Vitamix below.

Whew! What a journey.

What can we conclude from speaking to these companies?

Well, the consumer products created with generative design are already hitting our shelves in the form of the Adidas sneakers and Vitamix blender nozzles, and we should expect more products to follow in the form of sporting goods, cell phone cases and home appliances.

Of course, there will be more products in the defense and aerospace sector, but they are super-secret and don’t really count as consumer goods,which is the focusof this article.

From GM, we can expect to see lightweighted parts appearing on vehicles in the near future, but as GM hassaid, this will depend on the performance benefits of the parts.

Regarding the cultural obstacles preventing mass adoption, we can say that the experts largely agree that there needs to be a holistic change in the product development workflow that is bespoke to the generative design (and additive manufacturing) process. Companies should not think that they can reap the benefits of these technologies as long as they have a mind-set that is rooted in legacy processes.

But that said, the experts all seem to agree that while generative design is driven by additive manufacturing, it is not exclusive to 3D printing, and companies involved in casting could stand to benefit from generative design (as long as they modify their mind-set and workflow accordingly).

And, of course, there is the issue of social acceptance of these weird organic forms. How long will it take for the public to embrace these products? Who knows? If Carbon’s order book is anything to go by, the public is on its way to accepting them even as we speak.

But like the cultural changes inside companies, societal changes will also require an incremental approach before gaining acceptance.

Conclusively, we are going to see a lot more companies adopting this approach to design and manufacturingover the next couple of years, and you can be sure that we will let you know of the latest product releases that are made with generative design here on engineering.com when and as they happen.

That just leaves one question: Are they generative designed or generatively designed products?

We’re probably going to need some consensus on the exact terminology to use.

Semantics experts: let us know what you think in the comments!