Generative designs don’t have to be 3D printed. Other manufacturing methods enter the fray.
by Jean Thilmany, Senior Editor
There are those that say generative design ensures the best product design possible. Engineers enter the specifications a product needs to meet, and the tool responds by creating a number of potential designs that fit the bill. It even recommends potential materials to be used.
Now, the resulting designs can be manufactured in a number of ways.
Increasingly, CAD software makers, such as Autodesk, are building manufacturing preferences into their topology optimization and generative design tools, meaning generative design can take into account a particular manufacturing method. The engineer enters the manufacturing method as one of the design constraints. The resulting generatively created designs depict products that can be made according to that method and that meet other desired specifications, says Kim Losey, Autodesk’s head of marketing for Fusion 360.
Two years ago, Autodesk released generative design to subscribers of its Fusion 360 Ultimate product development software. Autodesk moved to marry generative design with manufacturing methods when it announced late last year that PowerMill, PowerShape, and PowerInspect would become part of the company’s Fusion 360 solution.
Now, Autodesk says, CAD and CAM reside on the same Fusion 360 platform.
The inclusion of CAM brings manufacturing considerations to the front of the generative engineering process, saving engineers and manufacturers time and money.
The generative technique is a completely different approach to design. The design concept allows engineers to define design parameters: such as material, size, weight, strength, and cost constraints–before they begin to design. Then, using artificial-intelligence-based algorithms, the software presents an array of design options that meet the predetermined criteria, Losey says.
By calling upon advanced software and computing power the generative system runs through many possibilities, each meeting the engineers’ design specifications. It presents the best designs to engineers, allowing them to pick the optimal design. Engineers are involved at the beginning and at the end of the design process. If they don’t feel any of the returned designs meet their standards, they can tinker with inputs and cue the generative-design system to start again.
The computer-generated (“generative”) designs might be unorthodox, new, and unexpected, with geometries that wouldn’t naturally occur to the designer. No matter how different, if the design is shown to work, it can be created by additive manufacturing, also called 3D printing, Losey says.
In the past, the design software often created shapes that could be produced only by 3D printing because the resulting organic, nature-like forms that were commonly created did not lend themselves well to processes such as machining, casting or fabrication, she adds.
But that led many people to assume the designs could be produced only through additive-manufacturing techniques.
Now, the scope of generative-design automation is expanding to include traditional manufacturing processes, such as milling, die casting, and even water milling.
Today, you can input traditional manufacturing constraints into Autodesk Fusion 360 and use its generative-design functionality to produce optimal design solutions that can be manufactured according to one of several predetermined methods, she adds.
Autodesk’s generative design software, part of its Fusion 360 platform, includes capabilities to sort through the design possibilities reflected by particular manufacturability considerations. This enables engineers to compare how easily, quickly and cost-effectively components can be produced via different manufacturing methods, Losey says.
By applying constraints that sort design possibilities according to selected processes, design options can be evaluated in a new light. Both engineers and manufacturers are often surprised at the results, she says. Multi-axis machining might be the best choice, as it turns out, even for a very elaborate part, when certain constraints are in place.
And bringing manufacturing to the front of the decision-making process underscores the role generative design has to play in the choice of product, Losey adds.
Estimating manufacturing costs
PowerMill 5-axis CAM software provides expert CNC programming strategies for complex 3- and 5-axis subtractive, high-rate additive, and hybrid manufacturing. Meanwhile, PowerShape is CAD software that creates complex 3D geometry to better control CAM software such as PowerMill. It works with any combination of surface, solid, or mesh data and quickly creates damage and collision-free toolpaths for large, complex parts, Losey says.
A recent video from Autodesk, in reference to questions to keep in mind while populating a prospective generative design, asks: “Will you deviate from the traditional design scope to include manufacturing constraints? Will you use 5-axis, 3-axis or even 2.5-axis milling? What about die casting? Can your design be water milled? Or does additive manufacturing provide the best value and performance?”
Clearly, the manufacturing process is a big influence on the type of geometry produced, Losey says.
While generative design can provide results optimized for your manufacturing method of choice, how do you evaluate the tradeoffs between performance and the cost to make your part?
Autodesk includes software—the aPriori Cost Insight Engine—as part of its generative design package. It generates a manufacturing cost estimate for each design alternative created.
“It interrogates a model for features or tolerances that would be costly based on certain manufacturing methods,” she says. “It allows engineers to fully explore costs and to strike balance between performance and cost.”
Another way generative design reduces manufacturing costs: the generative process can result in a single, solid-mesh body that can replace multibody components, such as a welded assembly. This reduces a manufacturer’s need to spend money on jigs, fixtures, welders, and welding material. And single-body parts can be made via additive manufacturing, which also reduces the need to design and manufacture tooling.
Also, generative design allows engineers to replace small assemblies with a single component to reduce manufacturing costs and bill-of-materials complexity.
Milled motorcycle parts
MJK Performance had shied away from the generative design method because Phil Butterworth, MJK designer and co-owner, felt the visual style evoked through generative design wouldn’t fit his company’s style. The company, of Calgary, Canada, produces aftermarket parts for Harley Davidson motorcycles.
“It always made me think of those spiderlike, organic, alien models,” he says. “I just thought it’d give us the same style only blockier, covered in intricate little pieces and looking silly.”
But MJK recently turned to generative because weight, strength, and style were all necessary considerations for the triple clamps it makes for motorcycles and Butterworth knew the method could help with weight reduction.
On motorcycles, stanchions from the telescopic fork are attached to the triple clamp, and sliders at the other end are attached to the front-wheel spindle. The clamps are part of the fork that connects the motorcycle’s handlebars, steering stem, and shock absorbers.
“Our clients want the look of a 200-pound race bike for their 1,000-pound Harley,” Butterworth says. “As a result, we need to make our parts as light and strong as possible but also stylish. Every part has to look like it belongs on a hundred-thousand-dollar bike.”
But the triple clamps are large and bulky, leading Butterworth and his fellow designers to attempt to reduce their weight. With an additional factor: The clamps had to be fully machinable on a 2.5-axis mill because the company makes all its parts at its small, four-machine shop.
Butterworth says he was surprised to learn that generative design might be the answer to his needs. He’d assumed generative design was mostly limited to additive manufacturing.
He soon discovered that Autodesk’s generative design technology could develop parts for conventional 2.5- and 3-axis milling and set Fusion 360 to design parts that could be made exclusively through those methods. Using that input, the CAD and CAM product quickly returned a range of designs that are lightweight and fully machinable, Butterworth says.
Not only that, “when the model came back, it looked like something I would buy immediately. I was blown away. It looks racy; it’s got all the cool geometry. It has all the cool stuff generative does but in a style people understand and recognize,” he says.
After studying the potential designs, Butterworth selected one and then spent approximately 20 minutes editing it to suit the characteristic MJK Parts style.
Through use of generative design, engineers went from computer model to a prototype within a few hours. And the weight of the triple clamp was reduced by 23% compared to a similar triple clamp designed by an engineer rather than the generative program. Nor did the generative design sacrifice safety, Butterworth says.
Normally, this process would take one or two days to complete. But with generative design MJK was able to accelerate this process. They went from a computer model to a prototype in a few hours. Fusion 360 can easily be set to produce designs for 2.5-axis milling, he says.
Within JPL, its Atelier division is the team charged with trying new approaches and processes, and its recommendations are passed on to teams working on specific missions. This division is collaborating with Autodesk to evaluate generative design for the proposed lander.
Topology optimization joins in
Capabilities are also changing fast in topology optimization, which differs from generative design. When using topological optimization, an engineer defines loads within a 3-D space and the program removes material to attain a shape that uses the least amount of material while still retaining required stiffness and density. The method also creates parts with odd shapes, though it typically gives fewer results than does generative design.
Additive manufacturing, with its capability to create never-before-seen shapes, seemed a natural for both.
But, to give you an example of the move to find methods beyond additive has quickened, take the example of the following two engineering journal submissions.
An October 2016 paper in the journal “Advances of Engineering Software” stated that: “Despite being an effective and a general method to obtain optimal solutions, topology optimization generates solutions with complex geometries, which are neither cost-effective nor practical from a manufacturing perspective.” That paper was written by Sandro Vatanabe and his coauthors.
In it, Vatanabe, now an engineering professor at the FEI University Center in São Paulo, Brazil, and his colleagues proposed techniques to restrict the range of solutions for the optimization problem. The paper was entitled Topology optimization with manufacturing constraints: A unified projection-based approach.
Three years later, a paper entitled “Topology optimization for multi-axis machining” presented a topology-optimization approach that incorporates restrictions of multi-axis machining processes. In it, author Matthijs Langelaar posits that particular cutting tool shapes and maximum insertion lengths can be included in topology optimization without much additional computational effort. He then shows examples of how to generate optimized, machinable, three-dimensional parts. Langelaar is an associate professor of structural optimization and mechanics at Delft University of Technology in the Netherlands.
Langelaar set out a formulation developed for 5-axis processes, though the technique also covers other multi-axis milling configurations, 2.5-axis milling and 4-axis machining by including the appropriate machining directions. In addition to various tool orientations, user-specified tool length and tool shape constraints can also be incorporated in the filter.
So, while a generative design or topology optimization study can result in interesting new designs, perhaps previously unfathomable designs, engineers needn’t limit themselves to 3-D printing to manufacturing the design of their choosing.
Butterworth says generative design has opened up countless new avenues at his company.
“Before I even start programming a mill, I know my part is going to pass any performance test,” he says. “Whatever the weight optimization in the simulation does, the 2.5-axis generative design just gave us a solid model right away.”