The 737 MAX, Digital Thread, and What PLM Could Have Prevented

Episode Summary

The episode titled "The 737 MAX, Digital Thread, and What PLM Could Have Prevented" delves into critical aspects of Product Lifecycle Management (PLM) through a panel discussion featuring experts in the field. Host Michael Finocchiaro introduces a diverse group of seasoned professionals including Kenn Hartman from DSA, Rob Ferrone as a product data plumber, Oleg Shilovitsky from OpenBOM, Jos Voskuil with his PLM coaching background, Martin Eigner with 40 years of experience in consultancy and blogging, Patrick Hillberg teaching at Oakland University on managing product lifecycles for sustainable futures, and Brion Carroll contributing insights from a practical perspective. The discussion centers around the complexities of digital thread implementation, organizational dysfunction leading to catastrophic failures like those seen in the Boeing 737 MAX, and the broader implications of PLM beyond design and manufacturing into service and recycling.

Key technical insights include the necessity of connecting all aspects of product life cycles—from ideation through use and disposal—to ensure safety and compliance. The panelists emphasize that while PLM systems are crucial for managing data throughout a product’s lifecycle, effective implementation requires not just technology but also robust organizational practices to prevent failures. They highlight the importance of continuous feedback loops between users and designers, as exemplified by Brion Carroll's example of Toyota using customer input to improve vehicle designs.

The key takeaway for PLM and engineering professionals is that successful PLM strategies must encompass a holistic view of product lifecycle management, integrating data from various stakeholders and ensuring seamless connectivity across all stages. This approach not only enhances safety but also drives innovation and sustainability in manufacturing processes.

---

Full Transcript

Michael Finocchiaro

And we are live. Welcome to the Future of PLM podcast. This is your host, Michael Finocchiaro joined by a panel, a large panel of people far smarter than me, including Rob and Patrick and Oleg and Jos and Martin and hopefully Brian will make it. Who you know, because he's been on other ones. And we have a new member of our panel, Kenn Hartman. So he'll introduce himself in a moment. So I guess we do.

Jos Voskuil

you

Michael Finocchiaro

Very quick intro because everybody sort of knows this already, but we'll just start ⁓ on the top for me is Kenn He's the new guy. So go ahead, Kenn.

Kenn Hartman

Yes, I'm Kenn Hartman. I'm the managing director of DSA. We are a PLM systems integrator. We're in our 32nd year of operation here. ⁓ I'm in my 41st year now of doing PDM and PLM implementation work, and it's really nice to be part of the team.

Michael Finocchiaro

Awesome. Thanks Kenn. Rob?

Rob Ferrone

Yeah, short intro for me. So I'm a product data plumber. I care about getting information to the right people at the right time in the right format so that they can do the right thing.

Michael Finocchiaro

Oleg.

Oleg Shilovitsky

Good morning. Good afternoon. I'm Oleg Shilovitsky I'm CEO of OpenBOM and writing Beyond BLM for 15 years.

Michael Finocchiaro

Awesome. Yes.

Jos Voskuil

I'm also writing blogs. I'm the virtual Dutchman and I have a profession as a PLM coach.

Oleg Shilovitsky

you

Michael Finocchiaro

haha and the legendary Martin.

Martin Eigner

No legendary, Martin. Martin Eigner, 40 years PLM experience, still ⁓ busy right now with lot of consultancy, blogging. PLM is my DNA. So anyway, thanks. Thanks.

Michael Finocchiaro

Hahaha. ⁓

Oleg Shilovitsky

you

Michael Finocchiaro

I think we've got a good 200 years of PLM on the cast this time, once again. So Patrick is going to introduce him in a second, and he's actually going lead the discussion today because he actually does a class, so we're actually doing a virtual class today of experts. so Patrick, you want to introduce yourself and then introduce the subject?

Patrick Hillberg

Yes, happy to thank you. Thank you guys i'm noticing that we can get to 200 years with fewer and fewer people Have you guys noticed this too? So,

Michael Finocchiaro

That's very nice.

Patrick Hillberg

let's see, I've got a few slides and then we'll get into things, but I just want to set things up. And ⁓ this is an intro slide I gave a talk a couple of weeks ago, and this is an intro slide from that. So my name is Pat Hilberg. ⁓ I teach a course at Oakland University. ⁓ Oakland University is near Detroit. So we're deep involved in the US ⁓ auto industry. And the name of the course is Managing Product Lifecycles for sustainable future. So, and you can see some things about the class and about my background and such like as I do a shameless plug to try to get people to come join my class. So much of what we discuss in class in our catastrophic product failures. My students work for companies which build trains, planes, and automobiles. And what we discuss is organizational dysfunction, which may lead to catastrophe. And I'm just going to set this up for the next few minutes and then we'll come back and talk about it for rest of the hour. ⁓ I'm sure everybody on this call is familiar with the V model. ⁓ And this is the systems engineering V. Stakeholders at the left have a set of needs. The corporation develops high level specifications to meet these needs. These are decomposed into manageable pieces and the pieces are integrated and tested. But the challenging complexity is that no one person can keep the entire product in their head. So we decompose the work into pieces, implement those pieces at the bottom of the V and integrate and test the the pieces as they fit into ever larger assemblies moving up the right hand side. Hopefully what we do at the top left of the V will get us to the product that we want at the top right. But when this fails, it does so catastrophically, costing many lives and billions of dollars. So while the V model that I showed in the previous slide is conceptually simple, I think this actually provides a more representative model of how product development occurs. Decomposition does not narrow to a point, but instead it broadens as work is divided across disparate groups. While each subgroup exists within the larger system, each subgroup is also incentivized based on its own goals. And this creates a systemic dysfunction in which low-level implementation decisions are not tested until they become a small piece of the larger whole. So will isolated decisions in the bowels of development be fully understood later in the process? And systemic dysfunction impacts more than just product design. Can we ensure that after a door plug is removed for maintenance, all of its retaining bolts are properly inserted when the plug is reassembled? So what first set me off on this was a fatal issue known as the GM ignition switch recall. And in class, we discuss how the US government required airbags beginning in 1998 and in 2002 introduced a new regulation to reduce unintended airbag deployments. So note that we've got a new requirement into an existing ⁓ set of functionality. We read that a software team implemented this requirement by only powering the airbag circuitry when the ignition switch was in the run position. So the only way that the airbag could go off is if the ignition switch was in the run position. But knowledge of that decision remained siloed within their group. At the time, the switch engineer who eventually made an unethical decision was consumed by problems in an unrelated set of electronics as he tried to prepare the switch for production. but very late in his development process, so late that it would delay the launch and cost GM tens of millions of dollars, and probably the engineer would lose his job. The engineer learned that the switch was not holding tension. And still only the software team knew that the ignition switch was a critical component of occupant safety. The Chevy Cruze and Saturn I in vehicles were released to market after 10 years of dysfunctional decision-making and 124 deaths, GM addressed the problem through an $8 billion recall. So I question I asked my class and we'll kind of get to this in a couple of minutes. Who should go to jail? If a nefarious actor introduced a virus into the airbag software, which eventually led to 124 deaths, they would be prosecuted as a serial killer. It's easy to say that GM is at fault, but who should go to jail? And I frequently have students employed by the automakers in the same role as this engineer who behaved unethically. They'll be facing these decisions. I heard some laughter too. A few years later, amidst teaching the ignition switch, the second Boeing 737 Max crashed. I literally we were running through it right the week that it crashed. I wrote an article about the first crash, a researcher wrote an article about the first class and we discussed that article in class.

Kenn Hartman

.

Patrick Hillberg

On October 29th, 2018, Lion Air Flight 610 lost control and crashed into the sea. A few minutes before it crashed, the first officer was given control of the aircraft, but in training, this officer did not show good air handling skills. He had been handed control a few minutes earlier. The pilot had been fighting the control since takeoff and needed some time to think through the problems that he had been having. A day prior to this flight and a previous flight on this same aircraft, another pilot had to fight through similar problems, including what's called a stick shaker warning. And flight safety rules indicate that the pilot should have returned immediately to his departure airport. However, he flew the plane onto his destination and did not even report the stick shaker event. Just prior to this flight, a maintenance technician replaced an angle of attack sensor but failed to properly calibrate it. The indicator gave a false indication that what was actually a level aircraft was about to stall. And in that case, it forced the nose down. And one year prior to this, the angle of attack sensor was refurbished at a supplier facility in Florida and calibrated on a machine which had a switch set in the wrong position. The miscalibrated sensor sat in an inventory for a year waiting to be used. So let's look at this from the Boeing product development perspective. In 2011, they lost a large order which they had expected to sell to American Airlines. This created a business need for a modification to the 737 model to reduce the use of fuel but require no pilot retraining. Designers chose a larger engine, but being larger, it needed to be placed further forward on the wings. And this might cause a rare instability through changing flight dynamics.

Martin Eigner

you

Patrick Hillberg

The following year, the new model's first test pilot was asked to allow for a piece of software controlled autonomy called the MCAS to address this occasional instability. He resisted the request, but was eventually convinced by engineers to include it only in rare circumstances, which were triggered by both high angle of attack and high acceleration turn. This test pilot felt that most pilots would never see this occur in their careers.

Martin Eigner

Thank

Patrick Hillberg

But four years later in January 2016, the original test pilot had retired. The new pilot advocated for increased use of the MCAS. It was now enabled based strictly on the angle of attack without the need for high acceleration. Engineers in this meeting described it as a quote unquote thoroughly run of the mill decision that generated little or no debate. By November 2016, and just five months prior to its planned product launch, which is an FAA certification, a third test pilot was res-

Martin Eigner

GAH!

Patrick Hillberg

who was responsible for defining pilot training was seeing strange behavior. This pilot had previously lobbied the FAA to remove mention of the MCAS from the flight manual. Removing it from the manual also removed the need to train pilots on what to do if the MCAS failed. Thus, no pilots were trained in how to manage an MCAS failure, and they were not even aware of its existence, which could take control of the aircraft away from them, as happened in the Lion Air crash and later in the Ethiopian Air crash. The 737 Max product was launched, hundreds died, and Boeing has faced tens of billions in losses. They've had negative earnings ever since the first crash, which was seven years ago now. So in class, I asked this question again, who should go to jail? I'll wrap up and then we can get to our conversation. So let's go back to the system V and note that there's a simultaneous V which wraps which impacts the organization, namely the transition from negative revenue during the development phase to positive revenue after launch. The organization is revenue negative until verification and validation are complete on the systems V. But when products become scandals, it is often years after the launch and revenue turns deeply negative. In the resulting legal investigations, we often find an ethical dilemma and a developer who chose a timely product launch over product safety. The complexities which lead to scandal are due to a lack of shared vision between these two Vs. Developers who are testing against a set of requirements and the financial organization which is pursuing positive revenue. Both are needed, but what if they conflict? This becomes worse as we see the growth in autonomous products. The burden of operating the ship, the plane, or the automobile is placed on product developers such as my students, rather than on the captain, the pilot, or the driver. As seen at GM, Boeing, and others, it is these product developers, students in my class, who face criminal prosecution. I developed this slide specifically based on mentoring a former student. She's one of the top safety experts in the world regarding autonomous vehicles. There's a qualification you can get and there's 120 people in the world who have this qualification, including her, but know what she says. In the past few years, the past few years have been very rough. No one cares about safety, security or quality. They only care about product launch. And literally two hours ago, I saw something from ⁓ somebody dealing with at-home robots was saying very much the same thing. So in our world, there's a concept of a digital twin, which is that ⁓ there can be a virtual representation of anything that exists in real life. And also a digital thread of information, which could extend all the way from engineering decision-making through service operations and into pilot training in the case of the 737 MAX. Such a thread might protect against future failures, but the necessary data stores for this thread would need to extend across all of these activities at multiple organizations.

Kenn Hartman

.

Patrick Hillberg

for each of the million parts in every airplane.

Kenn Hartman

Yeah.

Patrick Hillberg

What is the business model which can manage this? I won't take you through this today, but I take my classes in this. What's the ethical model needed to manage this? I'll tell you, we spent a lot of time on ethics in my class. As our systems become more complex, how do we address the very small chances something very bad will happen? Finally, let's get to our debate. So three questions.

Martin Eigner

Good.

Patrick Hillberg

In the case of catastrophic product failures, does PLM have a role in preventing these? If we think that digital threads could prevent these failures, what are the business models needed to manage these threads? And finally, in that dual V scenario, must product developers speak truth to power? So if you're a product developer on the system V,

Oleg Shilovitsky

Thank

Patrick Hillberg

And you're having an ethical conflict with a revenue person or a financial person on the revenue V. Must you tell the person on the revenue V we cannot do this and our business model is flawed. We'll start with the first one and I'm done with that. Thank you very much for listening to that. So, ⁓ on the first one, does PLM have a role in preventing catastrophic failures like the ignition switch or the 737 max. And frankly, I think we're seeing similar things on autonomous vehicles.

Michael Finocchiaro

you

Patrick Hillberg

Anybody have any thoughts?

Jos Voskuil

You want the thumbs up or a?

Michael Finocchiaro

Ha ha ha!

Oleg Shilovitsky

Hahaha

Patrick Hillberg

Yeah, right. Yeah. I'm gonna stop sharing my screen so I can see more video actually.

Jos Voskuil

Thank you.

Michael Finocchiaro

⁓ so who do, who wants to go first? think I saw Martin saying go up first and Yosses. Brian's back. Hey Brian. All working now?

Jos Voskuil

Yeah.

Martin Eigner

I think we have to

Brion Carroll

Hey.

Martin Eigner

keep in mind that PLM system, which should improve administration. We have a lot of tools along the product lifecycle, which are able to give us a better insight in design, for example. I think if you have good requirements, you have proven requirements, we can use system engineering in the early phase to check interdisciplinary design. And we can do a lot of simulation right now, co-simulation and signalized simulation. But in this case, the 7.3, I'm pilot, so I read all the material from this crash. And I tried to avoid flying 7.3 7-Bucks, and I was very successful. I think that is not a result of a single mistake. It's a systemic failure coming from pushing your old design to its limits.

Brion Carroll

Thank

Martin Eigner

corporate and competitive pressure against Boeing with their A320neo. ⁓ Float incentive structures and partial self-certification. That is, for me, unbelievable that the FAA, this re-delegated certification process to Boeing itself, and communication failure inside Boeing and with regulators, cultural problems within the company. and regulatory blind spots and misconducted by the FAE. I it is fully in this. I can say clear no. PLM will not help to protect it. I think I have some experience to improve requirements with artificial intelligence to make requirements better and compliance suited. That is possible. We are doing this in automotive airbag seat belts and It looks like we will be successful. But in this case, where the management, I think the mistake was on the upper management. even, well, the best information of PLM will not protect us against this type of failure. That's my opinion.

Patrick Hillberg

Well, we have hands raised. Brian, then Rob.

Brion Carroll

Yeah, so just quickly PLM should be able to ⁓ protect against these kind of things happening. But in my view, automation has to have guardrails, right? So if there's something that's saying that the plane is ⁓ going up and it wants to bring it down and the guy's trying to hold back as it's going like a land dart into the earth, there should be a way to say shut off. the automation because it's screwed up, right? So that's a design thing, right? But the way I look at PLM is PLM is pre-product, product design development, product and use, feeding back to product design, right? So the key is PLM is not just the engineering content. It's all the things prior to and subsequent of the manufacturing. So if the manufacturing

Martin Eigner

you

Oleg Shilovitsky

.

Brion Carroll

PLM continued in through the in use or in service like it would an industrial plant and it knew that there was a failure in the ⁓ thing saying, hey, you're going up, bring it down. No, bring it down. All that conflict and strive should never happen twice. If it happened once, that's because, you you didn't have guardrails that said shut off automation. It's like instrument flight rules versus visual flight rules, right? When you're landing. So If you put anything in automation, when you're engineering something, you should also say, how do I turn that damn thing off? That's number one. And the fact that it didn't, ⁓ it didn't allow the pilots to correct its misconception the second time should have been fixed because that feedback, whether it's through the box or whatever hell instrumentation connects in through to PLM to say things are weird. change the design, things are weird, change the design. If that loop does not exist, then you won't fix the second time, but it should always have been fixed the second time. So I'm gonna stop babbling.

Patrick Hillberg

Mm-mm.

Oleg Shilovitsky

Thank

Patrick Hillberg

Yeah, pin. I gotta put a, let's put a pin in that thought, because I think it gets to the digital thread part of the conversation, which is next. But, but yeah, okay. Rob, go ahead.

Brion Carroll

Right, it does, yes. ⁓

Rob Ferrone

Yes. So for me, it's a question of what's in scope of PLM and product data management across the life cycle. You know, so does that we talk about, you know, as used, but then are we really talking about as used for every single product made? Right. And in the case of ⁓ airplanes, et cetera, perhaps you could do that. But really, is that part of the life cycle that you need to be able to say, for example, okay, we've got the part there or the parts not there and how you do that if it's not electronic, I don't know, like if a plug's missing, for example, or a bolt's missing, I don't know how you detect that. And then, you know, even going back to the tests, you know, is testing in scope of product lifecycle management, and you could argue that that's actually more of a quality thing, but surely that has to be linked to, you know, part validation and part quality validation. So... Yeah, are we just looking at, all the parts there or are we actually looking at, all the parts functional and do we need to have some smartness in the products that are thinking about the part quality for themselves?

Patrick Hillberg

Yeah, I'll get a fee. Fino, although I'll say, think Rob got a little bit into what is the scope of PLM. but anyway, if you don't go ahead.

Oleg Shilovitsky

it

Michael Finocchiaro

Well, I had a great interview with Michael Rossum of Quicks, which is a startup that does ⁓ manage his test data. And I was a bit shocked to learn that in Formula One, when they're spending literally millions of dollars on each race and millions and millions of dollars on each car, the test data is just printed out Excel sitting on desks. I mean, he was just astounded. That's why he left Formula One to create Quicks because he's like, it's not. So I'm imagining in the aerospace world, must be the same. The test data is just manuals printed out. It's absolutely not, you know, scanned and then read into a system, into an expert system using an AI agent to read it. then ultimately, I mean, I've been talking, writing a lot on LinkedIn about it. think that ultimately there would need to be an agent exclusion and you need to have multiple agents, an agent looking at the test data and saying, wait a second, flagged this particular, wait, wait, wait. This test here is really critical for safety because it was flagged as a requirement and we're seeing, starting to see errors so that we really have to be careful on this. And then there should be an agent and the MRO system saying, hey, know, that part was serviced. ⁓ Is there, what are the impacts of that? And then this test agent would say, hey, I've got an impact. And then there should be a requirements agents that these two can talk to and say, hey dude, what was the deal with that requirement? So I think that there's like an agentic discussion because humans will never remember. That was like two years ago. I forgot that already. And that was a sub sub sub sub contractor and he quit. So you can't really rely on the humans to do it necessarily, particularly in something as complex and spread out as an airplane across hundreds of some contractors. So I think today could appeal to solve it. No, but two, three years with an agent exclusion with a, ⁓ a really persistent method and ⁓ an obsession of trying to make sure that all the test data and all these systems are online, then I think it could be possible. So that's it. I guess no that.

Patrick Hillberg

Oh, oh, like, what do you think?

Brion Carroll

Ha ha ha ha!

Patrick Hillberg

I'm ⁓ going in the order you guys raise your hands.

Martin Eigner

.

Michael Finocchiaro

You

Jos Voskuil

Okay.

Oleg Shilovitsky

I raised my hand. So, well, I think I'm coming with Rob. And from my perspective, the real question is how much data we can collect and how much insight and analytics we can extract from the data. So eventually we will be able to extract more information. Eventually we will be able to connect more pieces of information.

Jos Voskuil

Okay, go.

Oleg Shilovitsky

Eventually, reasoning of models will become stronger. Eventually, connections will be built using different graph and semantic models. So we will be able to get to the point when we can provide an insight that says there is something here that potentially has a higher ⁓ failure and that's something that Martin feels that he doesn't want to fly 737.

Brion Carroll

You

Oleg Shilovitsky

Without analytics, I'm just saying, without analytics. So we can get analytical way and we can get to this point. If this data already there, yes. Is it connected? Not completely. Will we come to the point when the systems will be able to predict it better than today? Absolutely. Will it solve the problem? Unlikely completely because it will be a human factor in everything. So we will increase the... chances to identify those situations. Will it be 100 %? I'm not sure. be, then finally for companies, will it be economically important to do it or not? It depends because Patrick, we had this conversation preliminary. There are more people dying in car crashes than in airline crashes, but people are more impacted by the airline, one airline. And I look at the data, was 240 people died in airline crashes and 40,000 in car crashes. So people are completely, like only in the US and people are quite silent. And it's a very, very different perspective, but then the money is impacted by how a human perceive it. Because when one self-driving car, and you mentioned autonomous vehicle, when one self-driving car is hitting a cat,

Patrick Hillberg

That's just you us, right.

Oleg Shilovitsky

All newspapers are suddenly writing about this. And how many people are hit by a normal car? A lot. So no one is like, it's not in the news, right? So, I mean, the answer is somewhere in between. That's my take.

Patrick Hillberg

yes?

Jos Voskuil

Yeah, I think I was an optimist, but I think I'm the pessimist now in the team. mean, I hear things about things happening in two, three years. And because if we rephrase the question about PLM, we probably would say a modern PLM data-driven infrastructure would avoid this type of accidents. Well, two weeks ago, we heard Boeing in the PLM conference speaking about their long-term journey towards a model-based enterprise.

Patrick Hillberg

You

Jos Voskuil

They are far from that. Nobody is there on that level yet that they control the data. And therefore, yeah, we are very much already also your introduction, Patrick. It's a lot of human activity and integrity, probably that is still where we depend on because we cannot master the complexity anymore. And I personally think we will never master the complexity.

Patrick Hillberg

So, okay, I've got a fourth question, I'll ask it fourth, but you guys have kind of brought it up a little bit, is which do you trust more, the human in the system or the automation in the system? But we won't go there yet. Let's see, I'm move on to the next thing. Kenn, do you have any comments?

Michael Finocchiaro

I think Kenn hasn't talked yet either. ⁓

Kenn Hartman

Well, I'm just wondering if we're in a perpetual state of catch up. The complexity of technology is growing exponentially. And how do you as an engineer stay caught up, especially in the light of the Boeing situation in 97, they merged with McDonnell Douglas. Prior to that, Boeing is ⁓ a culture of engineering excellence. They shift focus to cost cutting. and financial targets, they moved their entire executive staff to Chicago away from their R &D centers. So they're not having the water cooler conversations anymore with the engineering community. And they lost touch of that. ⁓ After that for the next 20 years, we've seen a massive explosion of technical capability in aerospace that has to be extraordinarily difficult to keep up with.

Oleg Shilovitsky

Thank

Kenn Hartman

When you're in the middle of a cost cutting ⁓ era ⁓ and meeting financial targets, you're not going to fund your R &D engine at a level that would allow you to understand all of the things that are critical, safety and reliability issues and test accurately 100 % of the time for all of them. It must have been an enormous mountain decline.

Patrick Hillberg

Yeah, I want to move on to the digital freight question, but I've got hands up from both Brian and Fino. Do you guys have?

Brion Carroll

Yeah, I just wanted to, I just want to touch on one thing. ⁓ and Kenn, Kenn kind of brought it up that the humans that are using PLM, if there's a distraction like cost cutting and they don't put in the necessary information, like I was talking about overriding some automation in the system saying we're, we're, we're level. Don't, don't, don't go down. Don't go down. We're level to turn that off. If you don't have the engineers, the human, thinking and the guardrails. then what you design and what the PLM system represents will be a loser, right? Because you're all going to, people are going to be flying out of the sky like, like, you know, darts at a family gathering. ⁓ To Rob's point, he said that ⁓ you got to kind of look at what is something that you would want to maintain the operational use of. And at some point it becomes dizzying in regards to the content you have to manage. ⁓ Above my head is a fan, right?

Oleg Shilovitsky

you

Brion Carroll

People have fans in their house. Now, to Rob's point, you may not want the fan talking to the backend system all day, all night saying, I think I'm still good. think I'm still good. Oops, I got a problem. And have those problems manifested in a quote design change. However, if you're going to get in an airplane or an automobile, it would seem that if you serialize the product, like an automobile with the VIN or an airplane with whatever the hell they use, that it would make sense.

Patrick Hillberg

tail numbers.

Brion Carroll

that if people are entrusting their lives in that object called the plane, that's probably why running over cats with a automated car is, you know, gets all the air cover because, you know, the cat had no choice and the car just ran over versus a human ran it over, right? Cause that means a human was involved. So if you're going to automate the playback from an airplane and what's happening, IOT, know, floating this content back or for a car, you may not do it for a fan. There's too many of them. It's not worth it. So there's ways to be able to ⁓ include, and this gets to the digital thread that you're about to get into the way in which content can be radioed back into the system. using what Michael talks about of AI agents that go, whoop, that's crap. Whoop, that's crap. Ooh, interesting. Let me, let me dialogue. You know, it depends on what comes back as to what happens. That whole flow and that whole, what I call PLM, which starts, you know, from mind to market, right? It's not just engineering.

Patrick Hillberg

So let me get in, thank you, thank you very much. And we're all kind of bumping on this, so let's get into the digital thread question. So, and note that the pilot on the previous flight had a stick shaker event and he did not report it. So I don't need AI for that, right? I simply need, hey, somehow I'm trying, I mean, there was some sensor that gave a stick shaker warning.

Brion Carroll

Right. Right.

Patrick Hillberg

Record the information on that stick shaker warning. Do not allow, I mean, there's a very simple green red checkbox on this, right? Do not allow this to fly. As a matter of fact, we need to talk to that pilot, it's why he didn't go back. But in any event, let's imagine that without, there's probably, there's this 80-20 rule of we can get 80 % of the way there with 20 % of the technology. Let's just imagine that a digital,

Brion Carroll

Right, that.

Patrick Hillberg

a relatively uncomplicated digital thread with minimal amount, which is just looking at easily accessible data. But we need to get that data is coming from multiple different, I mean, it's coming from Boeing and it's coming from, in this case, Lionair, and it's coming from Lionair subcontracted out their maintenance. And there was another subcontractor involved in maintenance.

Brion Carroll

Right.

Patrick Hillberg

And there's pilot training and all of those failures. So how, what, what is, what, what's, what's the data infrastructure that supports that and more to the point, what is the business model that supports that? Yeah. If you don't go ahead.

Michael Finocchiaro

⁓ Well, I think it actually does talk to digital thread. But in terms of the cat thing, think what was interesting about that was not, well, it was a cat, but the same height of a cat would be a baby. And that was the scary part is that there wasn't a camera there to check for objects that were street level. And that was a design flaw that should have been picked up in the requirements and wasn't even tested for, which shows it, which to the point of digital thread, there should have been a digital thread there to say, Why isn't there a camera looking at the ground? Because that's just as important as looking straight on and looking to the sides, right? ⁓

Patrick Hillberg

What did Waymo learn from the cat?

Michael Finocchiaro

They need a camera pointing down, ⁓ down in front.

Patrick Hillberg

Yeah, but without without all the news stories would Waymo have learned that?

Michael Finocchiaro

unless the ba-bump after the cat was run over had been picked up by another sensor,

Brion Carroll

Probably not. I don't think they would have learned it because that it cost.

Patrick Hillberg

All right, Brian, we got to give other people some opportunity. go ahead.

Kenn Hartman

I guess I have more of a question. ⁓ the black box is a recording device. Are the black boxes today, IOT enabled in real time? Cause none of those failures by the pilots to report those incidents would have gone by. Had the black box been IOT enabled in real time, talking to satellite, dumping data and being at least periodically correlated to see if there were any glitches.

Brion Carroll

Right.

Kenn Hartman

So, know, that's.

Patrick Hillberg

Martin, you know? Martin is a pilot. I'm wondering.

Martin Eigner

Yeah, I would like to mention that I have projects with ALM, and I think the ALM concept is much better than what I see in the data structure, data model, the PLM system. In ALM, it's very typical that the requirements are connected to the test conditions and to simulation things. And I think the software is much easier. So we have the one, typically one-to-one relationship between a requirement and test condition. When we do the same, we did it with VW. There was a project, FUSE, was called Functional System Engineering. We have to combine systems engineering, cameo. So we connected requirements to function behavior, and then we can connect to the test condition. But that's not typical for a PLM environment. But anyway, I'm a little bit optimistic that we can stabilize and improve processes from requirements. system engineering design and testing. But that will not help when I think what Boeing did is that they sell the MCAS to the FAA. It looks like a paid optional feature. It was not mentioned as a major adjustment. Major adjustment has strong conditions and requirements for some redundancy. The very important angle attack sensor

Kenn Hartman

.

Martin Eigner

was only installed once. And no PLM system is able to best installation, ALM, PLM, whatever. The best digital thread will not help when management try to misinform the regulation. That is my problem. But I think when we take over some functions from ALM and PLM and we improve the connection between the requirements and simulation and test, we can improve processes, but not in this case.

Kenn Hartman

That's right.

Brion Carroll

So, yeah, so sorry, I had my hand up for a second. I want to just address two things. Kenn brought up the movement of the executives to Chicago for Boeing as an example. Therefore, you couldn't, as a lead engineer, walk into quote, the business person's office and say, listen, Bob, we got an issue. We got to do something so that they willful decision like Martin brought up to avoid visibility to something ⁓ was.

Patrick Hillberg

So I've got a shirt.

Brion Carroll

owned by the human that made that decision. Because PLM cannot fix what isn't in it, right? You can't imagine things that don't get put into it. When it comes to the digital thread, ALM, PLM, I don't give a crap what LM, they should all be considered product lifecycle management. Application lifecycle management.

Patrick Hillberg

So are we dealing with, I mean, both you guys, are we dealing with PLM systems, which is technology that manages CAD and a little bit else, or are we managing the product of the full extent of the product lifecycle? We always come back to this debate, what is PLM?

Brion Carroll

Right, well, I'm saying, right, so Patrick, Patrick, that's what I'm saying. You cannot solve the problem with what is currently, know, having been in this industry for 40 years, what is currently the core of CAD engineering, bomb management. You can't fix this problem. You have to go and you know, Michael brought up, is the test data part of it? It should be, is, you know, in use. content and connecting as Kenn said, the black box should be all of it is product life cycle management, right? It can't just be engineered.

Patrick Hillberg

Okay, so Rob, Brian, I gotta give other people a chance to talk.

Jos Voskuil

Yeah.

Patrick Hillberg

Rob, what are you thinking?

Rob Ferrone

I mean, so one big problem is that the data belongs to the operator and the product creator can't get their hands on the data legally and contractually. So that's one big problem. And secondly, for me, the digital twin, sorry, digital thread is good for post-mortem, but the it's not alive. And actually, if you want to have live insight into what's going on in all of the products, that for me is then ⁓ digital twin.

Patrick Hillberg

That's the, how would you get, to me, that you need a digital thread in order to provide live data back. Is that your understanding?

Brion Carroll

Right.

Michael Finocchiaro

Yeah.

Rob Ferrone

Whether you call it digital credit, it's just kind of connected data ecosystem. But what my point is, is that the way ⁓ I see digital thread is it's almost static and you can trace information flow through, but it's not live. You don't have live checking against all of the products, etc. All it is saying is, did we build this thing to spec? You know what parts went on it?

Michael Finocchiaro

A digital thread is a prereq for a digital twin.

Kenn Hartman

Yeah.

Oleg Shilovitsky

.

Rob Ferrone

And if you swap things out, you've got a history, it's very static, it's not live. And therefore, for me, a digital twin is something that is living and breathing and live and constantly cycling so that you can be simulating and checking and validating and yeah, which is incredibly complicated.

Patrick Hillberg

I mean, I wanna move on to all, like I agree with that, but just in terminology, I think that my sense is that the threat is needed in order to keep the twin alive. If I wanna know how much gasoline is in my gas tank, I can bring up a lightweight digital twin of my car on my cell phone, which will tell me what's the air in my tires, what's the gas in my tank, and how long do I have in my life. Right. But I need, I need that communicated to me on a daily basis, which I, in my view is, the thread. I'll let you, your hands up. Your hand has been up.

Oleg Shilovitsky

Yes, I was listening and what comes to my mind is that I think there are potentially different problems that we are trying to solve. It's one problem that I deeply care about is how to make data that can be collected and can be ⁓ used. It's what was, I don't remember who said the closed loop. I mean, we need to be able to use the data and to use analysis. and we can use it for the future prediction and not only for future prediction, but for creation of the manuals and creation of the maintenance procedures and creating like the things that analyze the design, the things that analyze procedures. So that's the one thing. On the other side, there are complete stupidity of the ⁓ different reality of the human communication. and organizational communication. I can give you story. I had a visitor from Germany coming to me and he looked at my ceiling and I speaking about this propeller that you have right on top of it. And he said such a... This is the airline conversation, okay? So let me use it. Wait a second.

Brion Carroll

It's called a fan, actually, not a propeller.

Michael Finocchiaro

Hahaha!

Brion Carroll

That's a plug for Hunter fan. That's Hunter fan fans. Yes.

Kenn Hartman

Is it turbocharged?

Oleg Shilovitsky

So, and the person is looking on this ceiling and it was in the room with the bed and he said like, it's a horrible American decision. Like you put this fan in the ceiling and what did it will fail on me? Like I will die. So, and he said to completely you American and unsafe people, you like everything is built not according to the code. It's completely horrible things. Your airplanes like it was like. going all over the place. And I took ask and I asked for statistic, how many people get head injury from this fan falling down. And like even there are no US statistics, I think like two injuries reported in the hospitals in the United States from this injury. So it sounds very horrible on the ceiling for someone. That's a human perception. Like it's the same story for the cat. It's like a human perception is horrible. So, and some people are operating. using this perception and they believe it's dangerous and this is not dangerous and we will not be able to fix it. Some people will be protecting themselves from the things that are completely safe and not protecting themselves from everything that is completely dangerous. no PLM system will be able to fix it.

Patrick Hillberg

So let me, I want to bring us back into the topic though. Boeing has been earnings negative ever since the first crash, every single year. They've had negative earnings totaling $38 billion. I looked it up sometime recently. So my belief, and let me test my belief is theoretically there's a data infrastructure called a digital thread, which could have recognized these problems in advance and therefore saved Boeing $38 billion. And in the immediate aftermath of the Max crashes, two CEOs lost their jobs. So, and we frequently talk about the fact that, the CEO needs to get engaged in PLM. Well, let's save the CEO's job. Let's build on Let's build a lifecycle manager that can trap all this. I'll get to Yasen in just a second. ⁓ But how would we set that up? Yes, go ahead.

Jos Voskuil

Okay, so you don't fix legacy by firing the CEO. think, I mean, it's you have an infrastructure in place and maybe two definitions for me. Digital twin is real time. It's a digital twin. Don't run on documents. Digital threats. There we have two situations. We have the connected digital threat where we have the real time information and we have the documented digital threat would coordinate it where people work with documents. And I think The last one, digital threat, is what still is in place in many places and the connected digital threat, we are still far from that. So we need to rely still on humans and human knowledge until we fix this digital connection.

Patrick Hillberg

Sure. Martin, your hand was up a while ago, I think. I don't think I got you. Did you have anything, Martin?

Martin Eigner

Sorry, you mean me? ⁓ But once again, ⁓ I feel I repeat myself. I really can imagine the digital threat is possible, not yet, but in the future. And I can imagine when we have good organization, good management that we establish well.

Patrick Hillberg

Yes, you, Martin, Dr.

Martin Eigner

to eat it processes for change management, design management, test management, requirement management. I really believe that we are maybe three, four years we have the technology. once again, if management fail, if the culture is bad, then the best technology will not save us. I think that's a mixture between culture, management, behavior of the humans and technology. I think that is a real problem. So I think Boeing is a bad example to say PLM will not help us. I think it is such a systemic failure that PLM is a niche to help in this case.

Patrick Hillberg

Yeah. And if you don't get to you in just a second, let me say the reason I make so much of point, both of the GM ignition switch and Boeing, but Boeing in particular is it's a very well documented example. There's, mean, there's television shows on Netflix about this, right? Um, and it's also it, it's a, it's a pretty concise example. can, you can look at specific things and you can see where the issues are. Anyway, that's an aside. you know, please go ahead.

Michael Finocchiaro

I just want to say that I don't think it's an easy problem by any means. ⁓ agree that we can from this discussion see there's some ambiguity in some of the terminology, right? We don't quite, all of us agree on digital twin digital thread. ⁓ The way I look at it is I agree with Rob in the sense that a digital thread is more you take a ⁓ part number, which becomes a line number and a skew number and later becomes a serial number. and you hope that your CO number responds to some kind of requirement and physical model and 3D model that was done ⁓ proceedings for having this running airplane. The fact that this operational data, which would define that digital twin of the entire airplane is difficult is very evident in the fact that PTC couldn't deal with Thingworks. I mean, they just sold the whole thing. And that's a PLM company. that just got rid of this business that they bought for about what they bought it for 10 years later. It just shows that it's a difficult problem, right? I think the only PLM company today that still is really doing IoT is Siemens with MindSpheres, which is even called MindSpheres, now it's Insights Hub, but they're the only one. mean, DS does a little bit with the Prisa, but it's still not a turnkey IoT solution. And now TSG will hopefully do thing works or they'll just use Kepware as Walker Reynolds thinks. ⁓

Jos Voskuil

But you want to be vendor independent as a company.

Michael Finocchiaro

You do, but I'm just saying that it's a difficult problem to solve. If a PDC couldn't figure it out after 10 years, that means it's not an easy one. It's not a low hanging fruit. The real time world and the PLM world, they're hard to connect. So that's why my opinion is it should be more of an agentic thing. You have agents that can sit in the real world and throw up alerts. And then there's somebody in that more static development world that's saying, wait, there was a problem over there. The agents wake up and says, Hey, we better look into this a little further. Something like that. But I don't know if you, I don't think you'll ever get to one system that'll do it all because it's different kinds of data. have transactional data, time series data, collaboration data, CAD data. They're just all so heterogeneous. I don't see one system managing all of it. That's all.

Patrick Hillberg

Brian, I want to get to you in just a second, ⁓ briefly, ⁓ building management systems are doing IOT. So it's not coming out of, ⁓ don't get me wrong about what Siemens is doing or anything else. I consulted as an expert witness on a patent lawsuit on managing buildings and they're using all the same technology as is going on in factories, but they're way ahead in terms of actually monitoring the building as it occurs.

Brion Carroll

Right.

Patrick Hillberg

Brian, you've been very patient. But believe me, enough time for the last question.

Brion Carroll

Yeah, that's fine. So two things. One is ⁓ while you guys were talking, ⁓ I'm driving to Boston today and my wife said, check the tires. And so I went onto my Toyota app and I found that all my tires are good. So I didn't have to go out and actually, you know, go ding, ding, ding, ding, ding on my tires. I could just look at the app on the phone, which is the IOT form, right? Which is digital twin, right? Which is connected content. I also, when driving the vehicle, If I'm in traffic and everyone stops like not wanting to hit a cat, of course, I mean, we're saving the cats. But my brakes will automatically lock up and a thing go across the front saying, hey, jump out of the car. Yeah, no, it doesn't say that, but you know, mean, it locks up keeps me from making contact. OK, so those are things where within the ecosystem of the vehicle, I've got things that say, oops, you know, jam on the brakes or.

Michael Finocchiaro

you

Brion Carroll

I can look at my phone and I can see things that keep me from, I think that anything that people trust, that's an airplane, that's a boat, that's a train, that's a car, any kind of thing that people get into, trusting they will live through the experience, should have continuous IOT because it's important to save humans, right? Cats, yes, we should be saving the cats, but the fact is, The humans in a car or a plane or a train or a boat should always be the of utmost importance. And therefore, regardless of what it takes, IOT should be constantly monitoring as Kenn brought up, is the black box being monitored? It should be. And Oleg, know, OpenBOM has all this graph database and I don't know enough about it because I'm not that deep into it. But the fact is that data can go jamming pretty huge. And at some point you recycle and say after 10 years or after a year, we don't care about what happened before. And you can begin to undo and, you know, deprecate that content. But there is ways to be able to take things that people trust, IOT them. And at the same time, know, assure product life cycle management.

Patrick Hillberg

Well, let me get to the last question, which does actually get to the trust part of, and it was my final question on the thing is in that dual V scenario where you've got the systems V working versus the revenue V. And there's this point where the engineers are, they can't quite get to completion. They can't quite fulfill the requirement. The technology really is just too hard. And yet revenue is going more and more and more more negative. And the financial people are saying, you got to release the product or the company is going to go out of business. You've got to. And that's what was the case at Boeing and that's the case at GM and that's the case in the auto in the autonomous vehicles. And just this morning, it was the case in humanoid robots. I read that article. So does the product developer have a moral responsibility to tell truth to power?

Brion Carroll

Yeah, yeah, Yes. Yes.

Kenn Hartman

Yes.

Patrick Hillberg

Okay, give me the thumbs up, thumb down and recognize it and it's telling truth to power. You're probably going to give up your career.

Jos Voskuil

Yes.

Kenn Hartman

Yes.

Michael Finocchiaro

Yep,

Brion Carroll

Hey, look at the justice department. There are people flying out the windows in that place because they're not doing things they're not supposed to, or they feel they're not supposed to. So if an engineer says we don't have the technology to protect humans, and yet we still got to get out the launch door, then they should be saying caveat, safe harbor statement. They should be doing something to augment. I couldn't fix this problem. All humans beware or something that caveats or fixes whatever it is that could be an issue.

Michael Finocchiaro

yep absolutely.

Oleg Shilovitsky

Yes.

Kenn Hartman

Thank you very much.

Martin Eigner

But Patrick and Michael, could be a wonderful discussion. I have started a link in. I do not agree to the standard V model because it comes out of the software. And software is design and testing. Software is not produced. I think the standard V model is not suited for megatronic interdisciplinary products. So we miss totally on the left side. We have in the early phase with Cameo, with this, we have

Patrick Hillberg

Yeah.

Martin Eigner

We have a small V where we can simulate in system architecture. Later we can simulate with simulation system. And then on the right wing we produce. So I think we have to think it's a V model, a good model to protect us against failure. think, yeah.

Patrick Hillberg

It was a cut. had a colleague at Siemens who said the right side of the V gets very heavy. It gets harder and harder to get to the left side of the V is an easy piece of cake, right?

Martin Eigner

Yeah.

Brion Carroll

Well, you gotta look at a multi V. You

Martin Eigner

Exactly. ⁓

Brion Carroll

gotta look at multi V. It's a multi V. The little V and then the bigger V and then the very, very big V. I mean, it's a multi V.

Martin Eigner

Meike, next discussion is the V model appropriated for the actual V2. ⁓

Michael Finocchiaro

you

Patrick Hillberg

Oh yes, I would love that discussion. Because I'm

Brion Carroll

Hahaha

Patrick Hillberg

in the thumbs down, but yes, I would love that discussion. we're at 57 minutes.

Michael Finocchiaro

Well, maybe I should

Martin Eigner

Me too. Me too.

Michael Finocchiaro

get the audience to take a vote because I also think that the problem of sovereignty ⁓ is really important too, right? ⁓ But we could also do a part two of this one on the V and then talk about sovereignty. Although we are running out of runway for 2025, either of those could be really good discussions. ⁓

Jos Voskuil

Yep.

Michael Finocchiaro

I thought it was great. Did you guys have a good time? I thought that we got through some of that stuff. That was cool. I was worried it could have been a little bit more volatile disagreement, but actually it turned out pretty well.

Rob Ferrone

I've got a question for the next time to Brian about the data flowing between companies. Would Brian pay $40 a month for the privilege of sending his vehicle data to Toyota so that they could analyze it and track it and store it? Most people wouldn't.

Brion Carroll

Well, I would say I wouldn't pay that,

Michael Finocchiaro

Hmm.

Brion Carroll

they should pay me for giving them the data so that they can develop and augment and roll forward how a Toyota Avalon operates in the United on the street. Let them pay me for giving them data. Right. That's what I would say.

Michael Finocchiaro

Hahaha.

Oleg Shilovitsky

That's a very... ⁓ That's a very interesting point, Brian, because for many companies today to show what you presented on your screen, they charge customer subscription.

Brion Carroll

Well, that's because I didn't have to go down and check my tires. I just did it on the phone. So I do pay something for this. But the fact is, if you want to look at it as something that allows people to manage in use content, the value of everyone that drives an Avalon, feeding back to let them see how the car maintains.

Oleg Shilovitsky

Exactly.

Brion Carroll

is something that's of value to them as long as they feed it back into PLM to correct the design of some flaw like tires or suspension or who knows what. So I think.

Oleg Shilovitsky

And this is called regulation, right? So you put regulation on the companies and on the people to run specific software, which is fine, but who will be doing it to connect all these pieces of data that we discussed? ⁓

Brion Carroll

But that should be the designer of Toyota. Toyota should do that. mean, you know, it's all PLM, product lifecycle management is the life cycle of the product from its ideation through its commercialization and the use of it. If it's a fan, you might not care the use of it. But if it's a plane, you definitely should care the use of it. This goes back to Kenn's point. Isn't the black box talking all the time to the back end system? The answer is no. Cause you got a service organization, you got the original design, you got the resellers, you got all the shit. me. Sorry. First, and all of that should be connected. So that the life of humans is protected as best can be based on all the content coming together to, you know, Michael's point of agents, ⁓ AI agents going and figuring things out while you're sleeping. All that stuff should be happening. We have the technology. We're just laying down going, oh yeah, help me get me out of bed. No, it should be, we're out of bed. We're getting this stuff done and all the stuff should be connected. That's product life cycle management. Not this, it's this.

Michael Finocchiaro

⁓ I think that one of the comments, and I should have been a little more vigilant in throwing in some of the comments, but we didn't have a lot of time to even talk about these very big questions. ⁓ One of the comments was also about ⁓ recycling, so the afterlife of the product, which is I know a big subject for Rob and Yas in particular, but the whole, I don't think that, I mean, I think Brian is correct that probably like so management goes beyond design and manufacturing. But I think it goes beyond service to recycling as well, particularly for products that have dangerous materials or just plain huge like cars and airplanes. But I think the recycling angle is super important. And Europe has at least put that into law with the DPP and all that kind of stuff. So I just wanted to put that. ⁓ yeah, Alex got his hand up.

Brion Carroll

Yes.

Patrick Hillberg

Oleg, what are you thinking?

Oleg Shilovitsky

About what? About recycling? ⁓ I already, I'm sorry, I already just forgot to switch it off. Yeah.

Michael Finocchiaro

You have your hand up.

Jos Voskuil

you

Patrick Hillberg

Well, you just raised your hand.

Jos Voskuil

Time is

Brion Carroll

Well,

Jos Voskuil

flying.

Brion Carroll

I gotta jump. It's been a fantastic hang with you guys. Looking forward to the next time.

Michael Finocchiaro

Okay everybody, thank you and we'll see you guys in a couple of weeks.

Patrick Hillberg

Sounds good, yes, we are just beyond the top of the hour, so.

Kenn Hartman

Thanks everyone for inviting me.

Oleg Shilovitsky

Thank you very much. Thank you.

Martin Eigner

Thank you.

Jos Voskuil

Yeah. Bye bye.

Kenn Hartman

Bye.

Patrick Hillberg

Bye, guys.

Oleg Shilovitsky

Bye bye.

Michael Finocchiaro

Anybody?