Thought Leadership

The Mechanical Side of Generative Design Part One – Transcript – Talking Aerospace Today

In this episode, Todd Tuthill and John O’Connor reintroduce generative design in the context of digital transformation maturity, but this time with how it relates to mechanical engineering.

Patty Russo: Greetings and welcome to another episode of Talking Aerospace Today. I’m Patty Russo. If you are a regular listener, you now know me as the host of this podcast, but today I’d like to introduce our guest host, Maddie Goetze. Maddie is a talented writer on our A&D global marketing team. Welcome, Maddie to the Talking Aerospace Today circle. And for our listeners, I’m sure you’ll enjoy this discussion on generative design for mechanical engineering.

Maddie Goetze: Hi everyone and welcome to another episode of Talking Aerospace today from Siemens Digital Industries Software. I’m Maddie Goetze, and I’m the content writer for our aerospace and defense industry marketing team here at Siemens, and I’ll be your host for this episode. Thank you for joining us today. In our previous episodes, we took our first step into the fourth stage of digital transformation maturity, generative design. We were joined by Tony Nicoli, Senior Director for Aerospace here at Siemens, who gave us many insights about generative design in the context of digital transformation maturity specifically, electrical systems.

Maddie Goetze: Today we’ll continue to explore generative design, this time and how it pertains to the domain of mechanical engineering. So without further ado, please welcome back our VP of A&D here at Siemens, Todd Tuthill. Welcome back, Todd.

Todd Tuthill: Hey, Maddie. Thank you and welcome to you. It’s great to have a new voice on the podcast. Thank you for taking time out of your day. I’m really looking forward to this conversation with you and John today. So thank you, Maddie.

Maddie Goetze: Yeah, absolutely. Thank you. And joining us today to talk about the mechanical side of the conversation is John O’Connor, Director of Aerospace Product and Management at Siemens. Welcome, John.

John O’Connor: Thank you. Great to be here.

Maddie Goetze: So Todd, last time we introduced the concept of generative design and discussed at length its relation to electrical systems. Can you start with a recap on what generative design is in broad terms before we delve into the specifics of mechanical systems?

Todd Tuthill: Absolutely, Maddie. For our listeners maybe who are joining this podcast out of sequence, let me kind of set the stage. We’re talking about digital transformation maturity and the five steps that are involved in that and we’re step four today. Of course, digital transformation maturity is configuration, connection, automation, generative design, and closed loop optimization. And as you said earlier, we’re going to focus today on generative design, specifically generative design in the mechanical area. And unless you’ve been living under a rock in the tech world in the past year, you’ve heard all these terms about artificial intelligence and very specifically terms around generative AI and generative design.

Todd Tuthill: So I kind of want to start with talking about those two terms because they’re used interchangeably a lot in the media and a lot with companies, but they’re really not the same thing. They’re related, but not the same thing. If we talk about generative AI now, that’s what a lot of people will know is things like ChatGPT. It’s the ability to provide a question or a little bit of input and have the AI, through a large language model, go out and search the interwebs and all kinds of data and this large learning model it has and it will create some text and maybe write a term paper for you, or maybe even an image, or maybe even music. That’s generative AI.

Todd Tuthill: When we talk generative design, that’s a much more narrow kind of feel that we’re talking about. It’s really the realm of people like US, engineers and industrial design companies. People designing products. Maybe architects designing buildings and homes. That’s the generative design aspect and it really starts with that concept, the third part of digital transformation, when we talk about automation. It begins to automate things that people used to do by hand in designing things. It’s the automation of design, and if you think about what the benefits that brings, the kind of the goal of that is to be able to do design iterations, single design iteration orders of magnitude faster than a human being could design things.

Todd Tuthill: And you say, “Okay, well, gee, is that a bad thing?” I mean, what do you think John? Is that what we can you see mechanical designers saying, “Oh, I don’t know if I want a machine designing for me.” That sound like a bad thing to you, John? What do you think?

John O’Connor: No, I think there’s always looking for a way to increase speed and the speed of the development process apps to make a ultimately like better choices in the end for any design.

Todd Tuthill: Yeah, but I don’t want to say that- Okay. We’re- this isn’t Skynet. We’re not taking over and eliminating the need for engineers. We’re actually making engineers lives and the lives of product designers and design teams much better because what it does, it’s, we’re trying to take away some of the mundane aspects of design and leave the higher level creating thinking for the engineers so that you can do multiple design iterations very, very quickly and began to optimize all of those design iterations. That’s what we’re talking about. So do you think that would sound attractive to engineers, John? Or do you think, you know, would that be something you’d be concerned about?

John O’Connor: No, absolutely. I think there’s the ability to do increasing levels of automation in the design process probably for as long as there’s been any sort of, you know, CAD modeling and all people been trying to find ways to automate the design process using software. And so, you’ll never be without the need for a human to be making decisions about the design and kind of having that kind of real world input from a human experience. But there are lots of things that happen in the design process that you could do much more efficiently if you had a generative approach to doing the development.

Todd Tuthill: So Maddie, that’s it. The, in a nutshell, you know, the idea of automating design processes in this narrower field, than just generative AI. And you say, “Okay, is AI related to the generative design?” It can be, but generative design as John’s going to talk about later existed before there was AI, and there’s aspects of generative design that are related and aspects that aren’t. So it’s a really interesting field, but I think that’s enough for the first answer to the first question, so.

Maddie Goetze: Thanks, John and Todd for your insights there. So before we move forward, last time we talked with Tony about how generative design can be applied to electrical systems. So what are the differences between the electrical domain and the mechanical domain, which is the focus of today’s discussion?

Todd Tuthill: Maddie, I’ll give you two differences as it relates to this discussion. First, we’ll talk about digital transformation in general in the mechanical domain versus the electrical domain. And maybe the EEs the crowd, and I’m an EE, won’t like this answer, but the gearheads, the MEs are winning. There’s more maturity. There’s been more digital transformation, more model based design I think happening in the mechanical domain than the electrical domain. Certainly, you know, we’ve been using simulation and things in electrical domain for quite a while, but when we talk about model-based transformation, I think I see the mechanical domain ahead of us, which kind of gives them an edge.

Todd Tuthill: So, so that’s one thing. And the other thing really is with respect to the kinds of outputs and the kinds of things that the generative design would produce. When we talked with Tony on the previous episode, and by the way, if you haven’t listened to that episode, you have my permission, hit the pause button right now. Go back. There’s two episodes in the feed. They’re fabulous episodes. Listen to those and kind of get an idea of the electrical domain. But once you’ve done that, come back and listen to this episode. But Tony talked about the idea of kind of a two-dimensional kind of wiring design, wiring connections, and then there’s an enormous amount of documentation that goes around interfaces and connections and architecture inside something as complex as an aircraft.

Todd Tuthill: So that’s really the focus of the generative design efforts right now in that particular field. When we come to mechanical design, it’s a whole different thing. We’re talking about this three dimensional space. We’re talking about wrapping PMI around all these kinds of things and really working in this 3D model. And as I talked earlier about the automation of all the PMI and all the things that wrap around that model. So it’s, at an 1000 foot level, it’s kind of the same thing, creating engineering artifacts, but when you get into the details, it really gets into the differences between the mechanical domain and electrical domain. So that’s kind of where we’re adding, and really looking forward to all the insights John has from his career about the mechanical domain and where we’re going in this exciting field.

Maddie Goetze: It is very exciting. So John, would you like to share some of your experience with the mechanical design from the perspective of traditional design techniques and contrasting them with some examples of generative design?

John O’Connor: Sure. And, you know, just to take a moment and kind of level set on where this field has started and where it has now grown to, you know, generative design, kind of its original sort of implementation was kind of more of an iterative inspirational type of design process where you’re making really just pure topology. So you’re optimizing for pure geometry without any real concern for manufacturability or the kind of geometry quality in terms of surface quality and things like that that you’d be able to get. So you would be able to create very exciting designs, but they would be effectively conceptual designs that you would then have to manually rework to create something that be a manufacturing design.

John O’Connor: And then from there, the generative modeling capabilities developed so that you could actually do a type of interactive design which would allow you to be able to drive the design to a result that created geometry that was manufacturable and created, you know, excellent services that you could machine to etcetera. And from there it then moved into a partially generative sort of design, where now it’s not really, you know, an iterative process or an interactive process, but it’s based upon certain conditions and initially the conditions were very simple, but then evolved to include multiphysics, different types of algorithms to drive the design based upon what the ultimate requirement was. You know, stiffness, weight what other mechanical considerations there would have been for the design. And the key part there is that not only were you were able to be creating a geometric representation, but it could also help you to create the associated product and manufacturing information.

John O’Connor: So you could create the design within certain tolerances driving to certain dimensions and be able to incorporate into the generative process more intelligence than would allow you to create a design that you know is going to meet the finalrequirements without having to have a significant amount of rework into it. And today, you’re beginning to see a lot of activity in machine learning as a component of AI and being able to drive other types of other types of design criteria. So cost, other types of design for manufacturing goals, etcetera. And so the ability now for a general design to be able to learn off of other previously generated designs based upon similar constraints or similar types of design conditions makes it so that you know create this higher level of generative design in the overall process.

John O’Connor: And key to that is being able to create that final geometry in a way such that it actually allows you to combine capabilities with traditional CAD systems. So you can do a generative process that creates, say, a mesh-based topology representation with ancillary PMI or other types of information based upon a multiphysics simulation, but then be able to merge that into a workflow that’s more typical for a traditional CAD modeling process. And being able to do both of those things together is what gets you to the highest level of general design that we have right now.

Todd Tuthill: Hey, John, can we step back to something you said something really interesting I’d like to dive down. You talked about manufacturability and you talked about how when generative started, it really wasn’t manufacturable. Can you comment a bit about additive and maybe what impact additives had on the ability to optimize and create generative designs for manufacturing?

John O’Connor: Yeah, that’s a great observation. Where a certain number of highly optimized designs, so the highly optimized topologies that you can create would not have been really manufacturable. You would not have been able to use typical manufacturing processes that would have involved machining a mold or being able to do any type of machining of metal to create the design. It’d be very hard to do that. So additive really has opened the door to a much wider range of geometries that allow you to be able to create something that is really fully optimized, and you can see that in a lot of areas. Anything that’s related to fluid flow or thermal problems, heat transfer issues for example in heat exchangers, can really benefit by having very, very complex topologies in their design, and additive provides you with the ability to do that.

John O’Connor: But there’s a trade off as well, where you want to be able to choose sort of how highly optimized that topology is because you might not want to do everything via an additive process. You might want to do a certain amount of optimization based upon the best topology you can get while still being able to manufacture it using conventional manufacturing methods. So there’s always a tradeoff there, and it additive itself has really opened the door to being able to fully explore this area and sort of fully exploit that capability, but many people find that you want to be able to choose when and where you apply it.

Patty Russo: Thank you for listening to this conversation. And, Maddie, thank you for standing in in my absence. Maddie, Todd, and John will be back to conclude this discussion on generative design in the next episode. I’m Patty Russo, and we look forward to connecting again on the next episode of Talking Aerospace Today.


Siemens Digital Industries Software helps organizations of all sizes digitally transform using software, hardware and services from the Siemens Xcelerator business platform. Siemens’ software and the comprehensive digital twin enable companies to optimize their design, engineering and manufacturing processes to turn today’s ideas into the sustainable products of the future. From chips to entire systems, from product to process, across all industries. Siemens Digital Industries Software – Accelerating transformation.

Leave a Reply

This article first appeared on the Siemens Digital Industries Software blog at https://blogs.sw.siemens.com/thought-leadership/2024/02/02/the-mechanical-side-of-generative-design-part-one-transcript/