Abstract
This recording is from the CDFAM Computational Design (+DfAM) Symposium and features Prof. Tim W. Simpson of Penn State University.
In this insightful presentation, Prof. Tim W. Simpson, a pioneer in the field of Design for Additive Manufacturing (DFAM) at Penn State University, reflects on a decade of applying DFAM principles in various contexts. Prof. Simpson shares his extensive experience, insights, and lessons learned from years of practical application and academic research in DFAM. His talk provides a valuable historical perspective, current state of the field, and his vision for the future of DFAM.
This session is especially beneficial for professionals, researchers, and academics engaged in computational design, additive manufacturing, and particularly those with a focus on DFAM. The in-depth exploration of DFAM’s evolution and its potential trajectory will be of immense value to those invested in this field.
Please note: The CDFAM Computational Design (+DfAM) Symposium is a respected platform for exchanging insights and innovations in computational design and additive manufacturing. This recording is intended for educational and professional development purposes.
Transcript
From YouTube’s automatic captions, lightly cleaned; expect some errors. Each timestamp opens the video at that moment.
Read the full transcript · 4,177 words
Foreign, good morning everybody, see if we can cover 10 years of DfAM history in about 20 minutes, so yeah, apologies for everything that we’re going to leave out, do what I can here. So great to be here, wear many hats in the additive space, for those of you that saw, read my my interview with Duann there, thanks for putting that up. Been at Penn State 25 years now, mechanical engineering and industrial engineering, co-director on Sim 3D, our metal 3D printing lab, that’s been a 10-year anniversary, and then some. Also doing work, we founded a graduate program online, and resin, actually got to meet one of our students, Sam Welch, there, that we worked with and never actually met in person, so it’s been fun connecting with people in real life here. Working fortunately with Added Manufacturing Media, met Pizza Linsky a couple years ago and said hey, do you want to write a column about additive insight, so that’s been fun, I can’t believe I came up with an idea every month for six years, but it happened.
1:09 We did some fun stuff along the way, AMWTF, added manufacturing why the failure, basically is sharing some of our failures in our lab, one of the best ways to learn aren’t out there, but we can’t publish failures, right, everything has to be good when we publish it out there, so good learning opportunities. And now co-hosting some of the episodes on AM Radio, fortunate to work with the Barnes Global Advisors, I’d been doing a lot of training and education on my own, and now really there’s a lot of interest in, actually Chelsea and I have been tag teaming that, geez, every week for the past month now, just about everybody wants training, which is, I guess, is a good thing. And then I’ve been fortunate this past, trying to help connect some of the additive dots there.
1:50 So in terms of the history, we’ve sort of seen a lot of evolution over the years, and for those of you students, formula upright, can’t tell you how many uprights have been designed, developed with additive manufacturing and 3D printing, now you sort of see this as a common component. When I started at Penn State 25 years ago, it was LOM, a layered object modeling, where we’re literally gluing pieces of paper together, that is not a functional part, right, but fast forward 15 years and now we’re using lasers to melt powder, to fuse, you know, titanium together to make a solid working component. And so there actually had a picture, it’s literally butcher paper that’s going back and forth, you’re heating it up, there is a laser on there, you got to pick it out, all those sorts of things.
2:35 In terms of the technology, and now, as I said, lasers are melting titanium powder to make parts for us and whatnot, there’s a part in the in the build chamber there, and if anybody sees that part versus this part, see a lot of support structures and a lot of other things. To the students credit, Vincent Moranis, he was an undergrad, picked up some top op SolidWorks and parametric optimization work with my colleague Todd Palmer on this, Todd’s a material and metallurgist, he’s not a design engineer, so he didn’t think about overhangs and all those sorts of things at the time. So, you know, you have 21st century technology making it, and then, as Corey Dickman says, go caveman on it, throw it in a vice, cut, hack, grind, remove those support structures as part of this.
3:18 And I love telling this tale, and this is now 10 years old now that we did this, and so here’s this, you know, you created this lightweight part, and you spend 54 hours building it, of which less than half of that time is the actual part, two thousand dollars of powder and material in it, of which three quarters is in their support material, that support ends up as scrap and waste on the floor. So now you’ve done this lightweight component, and half of your time, and through the quarters of your cost, ends up as scrap material. Again, 10 years ago our additive tools, our topology optimization tools weren’t where they are today, and what we of course realized is that there’s a lot of trade-offs.
3:57 Now new trade-offs and decisions that designers have to make, you might, from an optimization standpoint, that is the lightest weight structure, right, you give it a objective function and it minimizes it, you’re going to go out and find that particular structure, but there are trade-offs and other considerations, thinking about build orientation, thinking about support structures, overhangs and all that fun stuff that might increase the weight of your component a little bit. But overall, guess what, a lot less support material, which means a lot less print time, which means a lot less cost, which means a lot less prototyping, which means much more likely to make a successful business case out there when it hits industry.
4:36 Fast forward a little bit, one of the features then, and this is again commonplace now, right, previously our top pop tools didn’t take design guidelines and overhangs and things into consideration, this was one of the first that I remember seeing, Cognikad from Paramatters, which is now part of Carbon, we learned about this yesterday, but it was really one of the first to take manufacturing guidelines and be able to integrate them in there, and not only for additive, but let’s do it for casting, let’s do it for machining, let’s do it for forging and everything else there. And so now this is part of this, and tell you what, I was really excited when I saw this yesterday, I’ve been looking for 30 years now for CFD tools to do this as well, and so be able to integrate and do our CFD analysis for thermo fluids and those sorts of things in addition to what we’ve been doing for structures forever.
5:20 Now, as part of that, of course, we now have cool generative tools that are up there, in particular, I was at the the launch at SolidWorks when we saw live parts from Desktop Metal, it was literally like, you know, alien space technology, seeing these parts grow from a seedling and go out and try and connect to the component, and and learn how to basically have enough structure to different vibrations and different loading conditions, sometimes it works, sometimes it didn’t, but it was all part of our generative tools. Now are mimicking nature, think about how much nature has evolved different mechanisms and biology over the years, our tools are now being able to do that as well.
5:58 You know, you’ve seen again, hats off to Autodesk pushing gender design, Fusion 360. Students, that undergrads are loving this stuff, bringing it in, graduate students as well, some great case studies that are up there that we’ve been using and talking about and referencing. People too, I like this one from Airbus, this is that big structural component that separates the first class from the rest of us that are out there, I like to show this one because part of this is now it shows our role, we are defining the problem, this is a big thing, most of engineering, we, you know, we assume the problem is given, here’s the loads, here’s the constraints, here’s the interfaces, you do the math, right, and you can even look up the answer in the back of the book and see that you’re correct.
6:41 Not so in the real world, now you got to go out there and you’ve got to understand what are those loads, what are those constraints, what are those mating interfaces, as part of that, and most of the time when you do that, once you do that, solve the problem, you realize you missed something, our preferences are wrong, there’s an additional constraint I didn’t consider, this, whoa, that’s not what this thing looked like. And so now we need to get engineers and designers to start thinking about formulating and understanding problems, not just solving those, and so the generative tools and all of this great stuff is great at solving problems, problem formulation is important, and of course chat GPT is going to help us with both of those here, well, as it’s continuing to evolve, as we get to those.
7:24 You saw that, of course, can’t leave out lattices, love lattice tools as well and structural components, this was a lightweight in case that we did, circa what was this, about 2016, 2017, in oil and gas component, you throw it in, very painful to generate these lattices, we’ve talked about that for a while, you bring it in, you do your structural analysis. I love showing this slide because this was one of the first where you think about this, when you’re doing a structural analysis, what color do you look for, red, why, because that’s where it’s high stress, constraint, those sort of thing. When you’re doing stress analysis with additive, what color do you look for, blue, why, low stress means your parts overweight, again with additive, your part is overweight, you don’t need material there because there’s no stress in there from the loading conditions.
8:12 Well, why did I formulate this way, why is there so much material, why is this, well, that’s because we’re used to thinking subtractively, as I remove material, I’m creating waste, it create takes time, cost, and all of those sorts of things, so that’s again different way of thinking when it comes to additive manufacturing and design. I won’t beat this one to death, I hate it as a design engineer saying that our CAD tools, our FEA, CFD, was our bottleneck, continues to be the case, I think there’s still some hurdles there, but that free complexity Mark was talking about this as well, you know, took five, six, seven, eight different tools to generate your lattice, analyze your lattice, look at your build plan, do some simulation along the way, repeat, repeat, repeat, thank God for students that have the time and patience to put up with this stuff and make this happen.
8:56 So, but nowadays, again hats off to the folks at anthropology, recently connected with Metaphor, two or three clicks, right, and I can change that lattice on the fly, generate it, spit it out, and do some awesome stuff with that, and so I think this has been game changing, now being able to, the representation issues, being able to generate these structures very easily, very quickly, to be able to explore that space. But end of the day, I love this quote from Bill Brinley, right, we’re not going to fly a 3D printed part just because it’s cool, it’s got to buy its way onto the engine just like anything else.
9:31 And so this was, we were at a Tech Exchange, came and shared this, talked about it, substitute engine with car, with your hip, with an airplane component, whatever, right, end of the day, when we’re out there, just because it’s cool doesn’t mean that, you know, additive is the best way to go. And I think we’re really seeing a shift now to the industrialization, to the deployment and adoption of additive, and so it is a lot about the cost, end of the day, so to what extent do your tools integrate the costs in there, and so we’ll talk about that as a wrap up here, but an example.
10:02 So here is the other big pattern that I’ve seen, not only is our mindset changing, not only are our tools changing, but thinking about these stages then at which we deploy these, and so I started seeing patterns with all of the folks we were working with in industry, as you always start out and you’ve done some part screening activity that said, hey, here’s the part that I’m going to take and now make with additive, it’s machine, it’s cast, it’s forage, to take too long, I can’t get it, you know, there’s a whole list of things, and so there’s this basically a replication of the existing part, and then you realize, oh, that part was designed for casting and machining, it wasn’t designed for additive, so I have to do all this post-processing, I have to do all these support structures, it’s cost expensive, and it’s the same performance, why the hell would I want to use additive manufacturing.
10:46 Well, you got to get to the good stuff, right, you got to get to the chance there where you have the freedom to start changing your design, to start optimizing and tweaking it and adapting it, so just like we design parts for every other process, we need to design parts for additive manufacturing as well. Well, and so what we see in what existing company in the way, and so, you know, startups, you don’t have legacy components and things, right, you’re trying to make a splash, you can start on the far right, every existing firm, you’re going to be on the left here, not only from a performance and cost, but risk, understanding the risk associated with your part with additive and everything else, how’s it going to fail, what’s it going to cost, et cetera, etc.
11:25 So this is the other opportunity for computational tools that are out there in design, is doing part screening, go through your bill of inventory, we worked with Defense Logistics Agency, 4 million, you know, 40 million parts, they coordinate, they wanted to say, what could we make with additive, well, if I can’t change anything, then I have material and size, you take those two as factors and you go quickly and you screen all your parts and say, oh well, I’ve got four and a half million parts, in this example I now only have about forty thousand. So if those two screening criteria, you kick everything out, well, if those 4.5 million parts, how many of them were designed for additive, none of them, it wasn’t around at the time, right.
12:03 And so now we’re starting to get more sophisticated tools, Caster among others, that do the screening and also make redesign recommendations and other things like that, actually Christmas comb, talk a little bit about this later on, what is your repository of data to then be able to train machine learning algorithms to make these recommendations and improvements. So this is going on, these are great tools that are out there, but at the same time engineers like things simple, and so we actually created, there’s been a couple of worksheets now that are available to basically, what are the top nine pain points that Westinghouse was facing when they were trying to use laser powder bed fusion with pen United, student went in, met with everybody, the the two engineers, experts at the time were saying, everybody keeps asking me the same questions, all right, well here’s the checklist, every engineer has Excel on his or her desk, nine things, score them, rank them, and then you can basically say, yep, good, bad, no, somewhere in between.
12:57 And I’ve been amazed how many folks in industry now, just, oh yeah, I got that spreadsheet, it was like, oh yeah, that was something we created in the summer, and now it’s easy to use and put it up there, so that was about three or four years ago. So what those worksheets do is they help you understand, right, what are the limitations, in this case, of additive, laser powder bed fusion, DED, whatever else, and so we’ve now differentiated, and Chelsea’s going to talk more about this, there’s really sort of, we talk about designing for additive of which that adapt, you’re modifying your part for AM, you’re modifying it, where MFAM is really anything you’re doing to save or reduce cost associated, so how do I modify my build orientation and or things to minimize post-processing, minimize build time, etc, etc, out there. And there actually is a structure for this, there’s a sequence that that companies go through that we’ve seen as part of this out there.
13:50 Likewise the other half is the DfAM side, this is the hype, M fans the reality, right, of making that part, DfAM is the hype, this is what gets you on the cover of Metal AM Magazine and other things that are out there, as part of that, this is what everybody wants to do, this is what all the students want to do when they come and work for your companies, but they’re not worried thinking about risk and cost and all of those sorts of things. Turns out there is a sequence of this, we’ve worked with companies over the years and they tend to ask questions in a certain sequence and pattern, there, tools as part of that, and again Chelsea will, we’ll talk a little bit more about that in the next one.
14:22 For those of you that again, there’s so many different opportunities, so many different tools, how do I get organized, they’re actually now playing card decks, we actually had these at, out as giveaways at the DfAM Accelerator, top 24 heuristics for additive, over in Zurich, ETH Zurich, Manuel’s here somewhere and others as part of that, but I like this particular deck of cards because you can go to Thingiverse and download them and print them out, and now you’ve got the handholds to basically share with folks. And so really it’s all about designing for additive, where we’re now combining MFAM and DFAM, right, we’re thinking about addressing those limitations to avoid costs, delays, etc, and then how do we capitalize and take advantage of the benefits.
15:07 And if we think about where are we in that replicate phase, you don’t have the freedom to be able to top up in lattice, etc, you’re stuck on the left hand side dealing with all of those limitations, as you adapt, you start shifting to the right, and then obviously you’d love to be in the optimized case there, where you have the freedom to clean sheet, completely redesign your part and optimize it. So this is something to be thinking about with all of your tools now, because the workflow that you need for replicate versus adapt optimize is different, and so if you think about it, if I’m doing a replicate case, I’ll go 3D scan something, I may do a little cat, I apply my design guidelines because I want to direct apples to apples comparison, that’s where it’s got to start.
15:50 Only after you can get over that hurdle, then you can start tweaking your design, let’s put some lattices in to light weight it, let’s now start doing generative to come up with these cool designs and those sorts of things. I think we are very far along now in terms of the top embodying guidelines, lattice tools, generative etc, etc, are great, I think from our perspective, I think this is the new bottleneck right now, to what extent can your geometries be fed into our analysis tools that the engineers are still required to use and or show or demonstrate, whether that’s an FEA structural analysis, heaven forbid, CFD analysis for a heat exchanger as well, and figuring out what are the best hydraulic diameters etc, you know, there’s a lot of work that needs to be done on this.
16:33 And so I’ve been excited to see several of the presentation here, we’ve been working and talking to Vadim Shapiro intact, trying to now do meshless geometries, and I see geometry without the mesh over there, among other things, but now how are we integrating the interoperability between our geometry generation tools, being able to create all of these, and now put them into FEA, CFD, and whatever other cool things are out there. Again, was excited to see yesterday, we’re now moving into what I would say generative engineering, not only are we generating designs, but we’re able to do that in the engineering analysis simulation tools and integrate those together, and again I’ve been in sort of the FEA, CFD world for 30 plus years now, and have been wanting to see CFD tools that are as easy to use as our FEA and whatnot.
17:21 But finally, end of the day, last thing, cost, cost, cost, so that fancy geometry, whether it’s a lattice, you know, etc, as you have support structures, as you do building, understanding the cost of your AM par, and so they’re very detailed cost models that are out there, we’ve actually developed a simplified cost model. If you take a look here, these are from the Wallers Report, rough breakdown varies by process, varies by material, et cetera, but about 60 percent of your costs are actually in the build, so if you can estimate build time, material cost, etc, then you can use that to then estimate the cost of your additive component and it works pretty well.
18:00 And so actually able to, if you tell me the mass of your part and the weighty and the messier part and the material, I can give you a rough order of a rough order magnitude cost for your additive component, and so that allows you to now start doing cost studies while you’re doing your generative design, while you’re doing your optimization, rather than waiting to get to that detailed cost model that you then go and feed into, you know, Opera and these other tools that are out there. And so the final message here, that cost model, again, I go back to that oil and gas component, you know, basically bottom line, if I took that component, printed it in 718 on laser powder bed, the solid component runs me about three thousand bucks.
18:42 Oh well, hey, I’m going to go to a dual laser, I’m going to go to a quad laser, right, I’m going to speed it up, well, when you go through that it only drops the cost about 20 percent. What about, oh, I’m going to wait till powder costs come down, oh, you know, they’ve dropped 40, 50 percent over the past seven or eight years, well, forty percent reduction in feedstock only gets you about 14 savings from that. Here’s the story, here’s why we’re here, right, design, if I can design it and lightweight it, I reduce my build time, I reduce my material cost, I can reduce post-processing, I can do everything else, this is the kicker, this is the savings, right, design for additive, that is the value multiplier when it comes to AM, when it comes to park costs.
19:21 And so look at that, 40 reduction in weight, I’m over 50 percent of the cost, now I’m within the ballpark of having a viable business case and being able to connect that and make that happen within my company, within industry out there. But final say, as part of this, you know, we got to think beyond design, and so many of us that get so mired and we love our design tools and those sorts of things, right, we actually have a readiness model that we’ve developed, seeing this over the years, I go in and talk to a lot of companies, go from our CAD to our applied design guidelines, eventually we get into customized workflow, but you can’t forget the processes, right, you need the technicians, the operators, the process engineers that you’re working with.
20:00 For me, Corey Dickman, wherever he is out there, if I hadn’t had him all these years, I’d have been lost in terms of everything that I learned from that, Chelsea as well, now with experiences that she’s been through, that’s on the process sign. Don’t forget materials, right, why are we using these materials, let’s pick different materials that are out there, can we do a substitution, you know, what what can I get with aluminum silicon 10 mag versus some of the other allies that are out there. And then finally it’s about the people, and bring it into companies, AM is a team sport, if you don’t have a team working across these three pillars within the organization, supported by upper management that has inspired creative engineers, designers, material scientists, and process engineers working together, then I think you’re going to stumble and trip along the way as part of that.
20:45 So with that, thank you very much, again, if I left you out your software or anything, I apologize, I only had 20 minutes, so thank you, excuse me.
Listen, search and cite
Listen to this talk as a podcast
More from CDFAM NYC 2023

Generative Design & Digital Manufacturing at NASA Goddard
Ryan McClelland · NASA Goddard Space Flight Center

Generate-to-Validate: Optimal Designs and the Need to Deliver Under Uncertainty
André Wilmes · Rafinex

Defining Performance not Just Geometry with Software Process + Material Science
Hardik Kabaria · Carbon

Reinventing the Wheel – Representing & Manufacturing Geometry with a Program
Mark Burhop · Siemens Technology

Interspecies Forms: Robotic Feedback Systems Mycelia Growth & Computational Form
Natalie Alima · Adidas / Biolab Studio

Designing Material Properties with Advanced Toolpath Generation
Alex Roschli · Oak Ridge National Laboratory

Design Automation Using Machine Learning & Computational Design
Rhushik Matroja · Cognitive Design Systems






















