CDFAM NYC 2023 · New York · 14–15 June 2023

Computational Disruption: Design Engines & Additive’s Untapped Potential

Listen as a podcast

Abstract

This recording features a presentation delivered at the CDFAM Computational Design (+DfAM) Symposium held in New York City on June 14-15, 2023. The featured speaker is Nathan Shirley, Experience & Design Lead on a cross-functional team at HP, composed of designers, programmers, and engineers.

In this presentation, Nathan Shirley provides an in-depth examination of computational design and its intersection with additive manufacturing. Drawing from his extensive experience at HP, Shirley illuminates the untapped potential of integrating computational design engines with additive manufacturing processes. Through the presentation, attendees gain insights into the complex methodologies and software tools leveraged by HP’s cross-functional team to devise innovative solutions for their clients. Moreover, the talk encapsulates case studies and practical applications, demonstrating the efficacy and scalability of these strategies in an engineering context.

This presentation is aimed at professionals and scholars in the fields of engineering, additive manufacturing, and computational design. It is particularly relevant for those seeking to understand the cutting-edge applications and integration of computational design in an advanced manufacturing landscape.

Transcript

From YouTube’s automatic captions, lightly cleaned; expect some errors. Each timestamp opens the video at that moment.

Read the full transcript · 3,133 words

0:00 Foreign, good morning, presentation this second here, but I want to show you something real quick, but I’m going to talk for just a minute about my background, a couple minutes about the team I’m part of, and then going to some case studies about how we’ve applied this in the real world, real business applications for computational design to have an impact, in this case, and sustainability and in the medical industry. Just by way of introduction, sorry, so I’m an industrial designer by trade, you know, we all come from different parts, and I think everybody brings something interesting, computational design is the common language, but we all have very unique perspectives looking into it.

0:55 I got my start and, of all things, Tupperware, you know, in the homeware, housewares industry, but very quickly I found myself trying to parameterize entire ranges of containers and not design the same thing 20 times in different sizes, and got very deep into parametrics, even though we weren’t doing, you know, food wear and stuff like that, so that kind of got my brain ticking, we were doing a lot of hole patterns, a lot of things like that. And then very quickly I realized that I was a tech geek and I wanted to make a transition into the the industry, so HP was kind enough to kind of take me on, and I started working on things like this, which is our one of our printers for social media, and kept kind of developing in the background my skills with computational design, starting out in Grasshopper.

1:44 And as HP was also going, you know, full tilt into 3D printing and building, you know, 3D printers for production, very quickly those skills became kind of crucial to new businesses that we were starting, and I was able to be part of developing the team’s capabilities inside of HP. So I’m going to talk about two of these, about how we enabled, through computational design, revolution and how molded fiber is made, and applying, you know, creating a an orthotics brand with customized orthotics. So that’s that’s me real quick, and then as far as the team I’m a part of.

2:22 Actually I’ll throw a few things up just to show the experiments that I was doing within HP before I was able to, you know, have this officially be my role, I was showing our managers things like this, where I was saying, you know, we spend all day doing design languages, but, you know, we could, we could codify those languages and expand upon them, and also the the various sizes and form factors. And so I’m starting to think about things in a in a way where instead of talking about whether something should be organic or rectilinear, we just play with dials and try it and see what really works, and get get very deep into that.

3:01 And I was also trying to understand how, how form could be solved in a computational way, kind of pushing the limits of like something that kangaroo, trying to figure out what would this mean for design if we could leverage the forces within, you know, with, and what I’m already thinking about, like tension and, you know, with within a surface or something like that, but have it actually solved from a set of curves. And so these are the things I’m playing with, and I’m trying to push my skills into something like, you know, how would you 3D print a Wii for a custom fit geometry for like a headset, and this is me kind of on my own pushing, pushing capabilities.

3:34 But then along comes, and I was kind of presenting this to the design teams internally, and then they came in with the need and they said, how do we do this, and I I was able to engage. But real quick, now pivoting to the team that I’m a part of internally, this is kind of a an interesting slide, if you look, you know, within the polymer space, this maps kind of the accuracy you need for a part, and the level of kind of the number of parts, of the quantities, and 3D printing, and specifically MJF kind of lives in the middle sweet spot, but mostly in the past it’s been prototyping, you know, one-offs or iterative prototypes, so you can eventually go into another mode of production.

4:20 But ultimately, you know, there’s millions of individual files and applications in here, and they’re changing constantly, that’s like a service bureau context, but we were very quickly realized that within a, you know, what we have, maybe a dozen applications, or a handful of applications, with slight differences between each file, you can have millions and millions of parts, and we’re interested in driving production of parts, not not prototyping specifically. So to be able to do that you have to leverage computational geometries, you know, whether it’s a dental aligner that’s being thermally formed onto a 3D printed part, or you can, in a ways that a human might not be able to understand intuitively.

4:56 And we’ve kind of, we we saw that there was just a few categories that are really well positioned for this, and so we look, HP had three different internal verticals that were started up to kind of build our capabilities and start businesses in these contexts. We had a footwear, we have a footwear business, and basically that’s leveraging engineered lattices, metamaterials, with our specific materials, so that we can create, you know, kind of a better than foam context with, you know, the right level of energy return, and get, you know, really get performance improvements for for athletes, for example.

5:35 On the other hand, we have body fitment, which, you know, comes into footwear, but really plays well in the medical space, so anything from a scan for orthotics, prosthetics, things like that, HP started a business called Arise Orthotics, and I was part of that endeavor, which I’ll get into a little bit. And then here there’s kind of this more industrial context of automated high mix, where you might be making, you know, a batch of 100 or 20, it might be tools, or or grabbers, or something on assembly line, you need to be able to to work very quickly, but you can enable the production of something else that wouldn’t be possible or wouldn’t be efficient, or might not have level control you need.

6:10 And then mass customization, I think that that will come more into play as the the price point kind of matures around 3D printing, but once that’s unlocked, you know, the world is our playground, right, as we have materials that can really do whatever we can envision. So the team that I’m a part of is now those, those teams that were all developed for those individual, those internal startups, we all came together at the end and say, hey, we just all fell into computational design, let’s make a team, and let’s provide a service to our customers, so that we can help them accelerate and get to the market faster leveraging our capabilities and our experience.

6:47 But a lot of people think, you know, all you need is an idea and a printer, and design often gets, you know, kind of as an afterthought. But first you need some kind of concept or aesthetic design of how this thing’s going to work, and then you need to, you know, really make a geometry workflow, or design engine, so much goes into this, and the really hard part is integration, to make all that work together seamlessly, you know, in the cloud, or, you know, to be more than just the demo. So this team I’m a part of, we leverage all this deep additive experience and computational design experience, the assets we produced to unlock applications, and we’re going and finding, you know, different industries where we can apply this.

7:25 It’s really fun as a computational designer, because one day I’m talking about prosthetics, the next thing I’m talking about eyewear, the next day I’m talking about, you know, an industrial context or automotive, it’s it’s very fulfilling in that regard. Essentially what we’re trying to do is take all the work we put in over years to get over this learning curve, and help other companies that are not as well funded, you know, accelerate into their step into the market using all that experience. So jumping into some case studies, molded fiber, excuse me.

8:01 So just just so you can kind of see it in context, this is one of our tool sets, but traditionally this would be done with a brass and steel tool set going through a paper slurry, and you, you suck the slurry through, it’s like a filter, it’s like an egg carton, or any kind of consumer packaging is made this way, it’s been done in a, you know, in the same way for basically for 100 years, where you’re seeing, seeing, or carving the steel, and then you’re making a custom wiring mesh, which is often cut into small pieces and welded back together, you know, in incredibly long lead times, and a lot of highly skilled labor involved.

8:40 So here’s the opportunity that we had, you know, metal molded fiber tooling, it’s expensive, it’s heavy, you might need a crane or multiple people to move it around, highly skilled labor, 12 leaks of weeks of lead time, but we knew from printing some small coupons that we could make something 3D printed that was robust enough and had some potential, so we could replace that wire mesh and and serve the industry in a fraction of the time, with more control at a local level. So just just to understand the complexity involved, you know, on the one side you see a wire mesh that’s welded together just to make a square, a box, right, and we need to be able to reproduce that level of detail in 3D printing, which, at the time, this is about, you know, 20 20 pandemics just starting at this point, and there was no, millions of 300 micron holes and a part.

9:31 So we had to figure out how to do that programmatically, you know, that we had engineers who were trying to do this in SolidWorks and then Grasshopper, and just, yeah, that’s not going to work, you kind of peter out around a thousand holes if you have a really good computer, in a, you know, a reps contest, a con B reps context, so that’s problem number one, we just can’t, the complexity, the rock complexity, is is more than we we can handle with off-the-shelf solutions. And then if we want to do this quickly, instead of, you know, the CAD is unreliable, if you’ve ever tried thickening, offsetting a surface that has a small radius on it, usually have to delete the radius and then thicken and add it back in, all these steps, these workarounds, resurfacing, it’s just a complete mess, and it’s not really, you know, you might spend a day on a file, or you might spend three weeks wrestling with a file that a customer gave you just just to patch it back together, and that’s not a way to start a business.

10:21 So we had to solve this problem too, and then once you can do it, you know, are you just going to hire more and more CAD designers to scale the business, ideally no, you need to automate all of that to have an efficient business. So, so quick shout out to Mode Lab, because at this point I had done a, I had kind of showed it was possible through Dendro and through Grasshopper with a tiny patch to say this is how we’d approach it, they said great, that looks promising, and I said let’s call on the calvary, because the time our team was still developing, I’ve gotten got engaged with Mode Lab with Ronnie, which I’m pretty sure he’s here, yeah, there you go, and and we were off, you know, into this very deep, very specific context, highly engineered, you know, my background from industrial design, I was kind of shifting sides of my brain into this.

11:13 But what we ended up building is this geometry pipeline that leverages voxel algorithms and an automatically, you know, generates the entire tool stack, so we’re doing CAD in a voxel context so that it’s kind of never fails to solve the problem, because it’s just, you know, it’s a simpler space to work in, and we digitally engineer millions of holes, there might be three or four of these parts strung together to make them a larger part because the build volume is so big, that we’ve made five foot long parts through this, and all this is happening through an interface that we developed in in minutes instead of, you know, three or four weeks to design, 12 weeks to to produce, but we can really change the paradigm there.

11:54 And there was a team inside of HP, we had, you know, a whole site working on, or part of a site working on this within HP, that we, four or five other people that I was part of, and we were collaborating with Modelab as well, and what we built, we built this. So essentially, so we didn’t need to create a UI, this was, you know, because we had our own people internally, but, you know, on the left, the right hand side we have our parameters, there’s, you know, over three or four hundred parameters you can go in and control, after you input the initial surface, and once you get everything set up the way you want to and customize for the machine you’re going to put it in, then you press, you know, compute, and outcome all the whole locations to do the booleans, with all of the complexities for managing the supporting the screen, because we need to print it on its own because it’s so, you know, so detailed, and then supporting it with this Gladys, you know, this is on par the complexity of like an injection molded tooling solution, and automating that whole process.

12:58 And then, you know, just in the way of working, if you get, some of you might recognize that we’re in Houdini in this context, but it’s really just like a, we think of it as like an operating system for geometry, where we can just go and build whatever we want in a sandbox together, without limitations from, you know, from wherever a particular solution might be at that time. So, you know, we’re building nodes inside of nodes inside of nodes that Rhythm leveraging into new contexts and new business offerings as well, which is great for someone like me in an industrial design context to work with somebody with a computer science background, someone with an Emmy background, and we’re all in here just riffing off each other’s work, it’s very efficient.

13:37 So the result is we now have a fully automated perforation and tool stack design to locally vary the pulp thickness, which wasn’t possible before, you didn’t have the control to say I want more pulp here or here or here, and now we can basically make parts that are, you know, 20 or 30 percent lighter but have the same structural value, and that’s huge for an industry that produces, you know, millions a week, you know, the the kind of volumes we’re talking about there, this is a, this is a big deal. And they’re also ultra light and fast to produce, and we’ve produced thousands of these production tool sets and are engaged with some of the largest molded fiber companies out there to to start off with.

14:20 So so that’s the first one, and then I’ll just kind of flip through the context of the Arise Orthotics that this project, we came in, and they already had like an engineering solution, kind of a bare bones functional thing they were prototyping with, but they didn’t have a design language to to make it cohesive to represent the brand out to to the doctors and patients. And so we came in as a team, also pulled in some team members from Modelab, and a number of internal team members, went through the process of design, you know, like we typically do, but we had to keep in mind that it should be extensible across any patient condition.

14:56 And so here you have like the doctors interface where they’ve taken the scan, they’re making their their prescription and their treatment plan, to be able to do those things, all based on specifically on the customer, but at the same time, you know, it has to look like a product, it has to have value in a product context, and so we’ve added these elements to communicate the support that you’re getting from it in a visual way, so when they unbox it they feel confident, not embarrassed to show, you know, if someone were to see this they could actually tell someone else, hey, these are a great product, and recommend them without feeling self-conscious.

15:31 And that might need to incorporate a detail like this heel post which is very exaggerated, you know, to and still be able to fit all that. Here is kind of the parametric thing, the way that the design language actually scales across the part, and how fluid that can be, and, you know, this level of consideration just to be able to get the design to be cohesive across, you know, any shape that we might have to put it on. And then just one little tidbit, I’ll let this play out, where we’re leveraging this now into the same learnings that we’ve had into something like eyewear, and then this is, I’ll be done after this, but you can see, you know, you have the workflow, you have to derive the landmarks, you create the glasses procedurally, parametrically, but they’re already custom fit to your face, you can choose your styles.

16:19 But also, since this is basically a blend shape, you can, you can blend between all the styles, you can do all kinds of customization of the thickness of the of the tilt, and also for the the patient, the customers specific, you know, facial geometry and eyesight. So yeah, I’m hopefully that was interesting for you guys, and I’ll leave it at that. Foreign.

More from CDFAM NYC 2023

Design Computation Human

Design Computation Human

Onur Yüce Gün · New Balance

Mud y Robots: Keynote Presentation

Mud y Robots: Keynote Presentation

Ronald Rael · UC Berkeley / Emerging Objects

Roadmap to Replicators

Roadmap to Replicators

Neil Gershenfeld · MIT Center for Bits and Atoms

Scaling & Standardizing Patient Specific Design

Scaling & Standardizing Patient Specific Design

Adam Wentworth · Mayo Clinic

Computational Design Ops

Computational Design Ops

Ronnie Parsons · Mode Lab

Overcoming the Design Dilemma

Overcoming the Design Dilemma

Ryan Carter · Johns Hopkins APL

Generative Design & Digital Manufacturing at NASA Goddard

Generative Design & Digital Manufacturing at NASA Goddard

Ryan McClelland · NASA Goddard Space Flight Center

Digital Materials – Geometry and Data Combined

Digital Materials – Geometry and Data Combined

Marek Moffett · General Lattice

Ten Years Teaching Additive

Ten Years Teaching Additive

John Hart · MIT

10 Years of Applying DFAM

10 Years of Applying DFAM

Tim Simpson · Penn State University

MfAM Walks so DfAM Can Run!

MfAM Walks so DfAM Can Run!

Chelsea Cummings · The Barnes Global Advisors

Informatics and AI for Materials Development

Informatics and AI for Materials Development

Wes John Alder · Citrine Informatics

Predicting Fatigue Life in AM Metal Parts Using DigitalClone

Predicting Fatigue Life in AM Metal Parts Using DigitalClone

Behrooz Jalalahmadi · Sentient Science

Designing Material Properties with Advanced Toolpath Generation

Designing Material Properties with Advanced Toolpath Generation

Alex Roschli · Oak Ridge National Laboratory

Design Automation Using Machine Learning & Computational Design

Design Automation Using Machine Learning & Computational Design

Rhushik Matroja · Cognitive Design Systems

Learning to Generate Shapes

Learning to Generate Shapes

Karl Willis · Autodesk Research

Design for Artificial Intelligence

Design for Artificial Intelligence

Chris McComb · Carnegie Mellon University

Register for Updates and Discounts on CDFAM events.