InfinitForm’s shapes look like they are made with CAD for CAM, not by Nature for 3D printers
Michael Bogomolny isn’t new to us. We saw him at his previous venture, ParaMatters, which was sold to Carbon, a 3D printer company, in August 2022. He has since gathered his ParaMatters team (six of them Ph.D.s) and secured a $2.5M investment to launch InfinitForm. It may very well produce the first generative software that designs usable and manufacturable parts.
Big CAD vendors may scoff, claiming that their generative design software has options to create manufacturable parts, and indeed, they do. However, whether they produce machinable part designs is a matter of debate. Such software has yet to spur widespread industry acceptance of generative design. The machinable shapes offered as proof are made by carving away non-machinable details and smoothing out otherwise gnarly, organic shapes resulting from a bone-growth algorithm on which they are based. It’s a roundabout approach at best and, at worst, incomplete — leaving non-machinable details behind.
Bogomolny, once the creator of shape optimization tools (ParaMatters CogniCAD) now has a better way. His InfinitForm software, still to be released, will optimize with cutting tools very much in mind.
We asked Bogomolny if his shape optimization is now an optimization of parameters because it appears to work with a solid model, but he doesn’t reveal the secret sauce. Are his methods patentable? Bogomolny will not apply for a patent because then he’d have to expose the algorithm.
No bones about it
From the conversation and demo, we learned how the user can specify the number of axes and size of the cutting tools. Varying cutting-tool size allows a tradeoff between the time it takes to make the part and its level of detail. Large tools can quickly remove material, but smaller tools can yield more detail .. though take more time, Bogomolny explains.
The size of the cutting tool is seen at the onset of optimization. This sets InfinitForm apart from conventional shape optimizers, which work by adding elements along stress paths or removing elements where no stress is present, all without consideration of the CNC machines that make 80% of aerospace and automotive parts.
With InfinitForm, fillets assume the radius of the cutting tool. Surfaces are smooth, straight, and look “like a human made them on CAD” for CAM. There are no little lumps one would see on a twig or bone.
A generatively designed part looking like it’s made of bone is no accident. Shape optimization software is based on bone growth. 19th-century doctor Julius Wolff noticed the increase in bone thickness when the bone was stressed, an effect that came to be known as Wolff’s Law. Altair was the first to implement bone-growth algorithms in a commercial software product when it launched OptiStruct in 1994. Bone growth continues to be the basis for structural generative design to this day — until now. Until InfinitForm.
A change of course
Starting with cutting tools in mind is a more direct approach to attaining parts that can be made with cutting tools. InfinitForm also distinguishes itself by outputting a solid model, while other shape optimizers output a triangulated mesh.
The solid modeling kernel in use currently is OpenCASCADE, an open-source kernel. But as Bogomolny will be wooing SOLIDWORKS users, he expects to switch to a Parasolid kernel at some point.
For those subjected to years of hype that generative design produces the lightest of parts, InfinitForm’s results will stand out as having obviously excess material. For example, the part shown above has sharp edges that could be removed and straight surfaces that could be gouged out to remove weight with zero effect on maximum stress. But Bogomolny doesn’t claim InfinitForm is strictly a shape optimization program. InfinitForm represents the best compromise of the least material and most manufacturability.
It is a compromise Bogomolny provides after much soul-searching. His whole career and higher education have been dedicated to pure optimization. How hard must it have been to consider low-tech manufacturing operations like cutting metal when everyone else was discussing 3D printing? But Bogomolny had endured a cathartic experience with ParaMatters, which had blazed a trail, duplicating how Nature optimizes shapes, with an internal structure filled with irregularly shaped cells, not lattices. Like the inside of a femur, for example.
But the price of Nature’s perfection proved too high. A perfectly optimized part, such as an organic shape, can only be made on a 3D printer and takes a lot of time. It can’t be mass-produced, and if it’s to be made of metal, it will be costly.
What company would create a part that costs $2,000 when it could make the same part for $100 with CNC? Bogomolny asks rhetorically.
One reason that part optimization will catch on now (besides being able to make machinable parts) is the advances in hardware, says Bogomolny. With the processing power available in GPUs, optimization, which involves solving FEA repeatedly on models with millions of degrees of freedom, can happen in the blink of an eye. He tells of one 700,000-element model with 2.4 million DOF solved in 0.7 seconds on a cloud server.
Bogomolny says the example shown was optimized in about five minutes. It’s almost finished — you must add fillets to CAD program, which will take a few more minutes.
InfinitForm plans its first release to be cloud-based, but a local solution is next, as local workstations loaded with GPUs are becoming faster.
A local solution is a must-have for military and defense contractors, who, for security reasons, can not use the cloud, says Bogomolny.
About InfinitForm
InfinitForm will have a product ready for “limited access for a few carefully chosen design engineers” in June but will take the summer, when everyone else is on vacation, to suitable the program, so it plans for a September release.
The company is headquartered in Orange County, California and has eight employees.
“You can do a lot with a few people,” says Bogomolny.
IfinitForm received initial funding of $2.5M from San Francisco-based Schematic Ventures, which may be one of the few investment firms that understands the technology engineers need and appreciate. Bogomolny’s background and education must have helped. He has a PhD in Structural Engineering (Technion — Israel Institute of Technology) with a dissertation on optimizing structures subjected to dynamic loading. He did post-doctoral research in the optimization field at the Technical University of Denmark, including structural and fluid, material, and shape optimization.
Welcome, InfinitForm
InfinitForm demonstrates a valid and practical approach to shape optimization. We have long criticized topology optimization algorithms for producing super lightweight parts, as if weight mattered without considering how manufacturing companies make their products. However, established CAD vendors paid no attention, insisting that their routines can help engineers think outside the box they find themselves trapped in and, therefore, their routines are able to conjure up shapes engineers would never conceive of. They wave off protests of “unmanufacturable” with “you can 3D print it.” Never mind the conceit of software developers who think they can design a product better than design engineers. Such developers ignore how reality is a world of both design and manufacturing … and choose instead fantasy world in which design exists all by itself.
Here comes InfinitForm, with consideration of metal-cutting tools built in. It’s the most sensible and most welcome approach to generative design we have seen to date.
