Setting Course, an ABS Podcast

Demystifying Digital Twins for Maritime

• American Bureau of Shipping • Season 1 • Episode 11

Share episode

Digital twins have captured the maritime industry’s attention amid the global journey toward digitalization and decarbonization. However, confusion remains about what exactly a digital twin is, what it isn’t and how it can be used.

On this episode of Setting Course, an ABS Podcast, Eric VanDerHorn, ABS Technology Manager, joins host John Snyder, Managing Editor of Riviera Maritime Media, to discuss digital twins and the potential benefits for the maritime industry.

If you like this episode, share it with your friends, leave a review, or send feedback to podcast@eagle.org. More episodes are available at www.eagle.org/podcast and all major podcast platforms.

Key Points

  • Digital twins are a hot topic in the maritime industry, with over 20,000 publications mentioning them in 2023 alone.
  • The concept of a digital twin can be traced back to the use of physical twins throughout history to better understand the real world.
  • Building a digital twin involves integrating various data sources and models, but there is no grand unified model yet.
  • Digital twins offer potential advantages in areas such as diagnostics, predictive maintenance, performance optimization and training.
  • Digital twins can also be used to support decision making in the decarbonization journey of shipowners and operators.

Guest

Eric VanDerHorn is a Technology Manager for ABS. He supports digital transformation in the maritime industry by developing and qualifying cutting-edge technologies such as digital twins, smart functions, structural health monitoring, asset integrity management, predictive analytics, and condition-based Class programs. Eric holds a bachelor’s and master’s in mechanical engineering from Washington University in St. Louis and a doctorate in civil engineering with a focus on uncertainty quantification from Vanderbilt University.

John Snyder (0:00.00)

You're listening to Setting Course, an ABS Podcast. Join us as we navigate the latest trends, developments, and challenges facing the rapidly evolving maritime and offshore industries. Catch every episode at www.eagle.org and podcast platforms everywhere. Hello and welcome to the show. I'm John Snyder, managing editor of Riviera Maritime Media, and I'll be your host.

Today we're going to talk about digital twins, what they are and how they're being used in the maritime industry. Joining us for today's podcast is Eric VanDerHorn. Eric is the digital product line manager at ABS. Eric, welcome to the show.

Eric VanDerHorn (0:50.855)
Thanks, John. Excited to be here today. 

John Snyder (0:52.988)
So, Eric, tell me about digital twins. What are they? What are they all about?

Eric VanDerHorn (0:58.105)
Starting, starting with the really hard questions, I think.

Digital twins are really a hot topic these days and I think really kind of kicked off when Gartner put them on the hype cycle in 2018. Just kind of as a measure of it, I went and looked at Google Scholar, which you can search for how many academic works are mentioning digital twin. And back in 2018, that number was about 3000 for the year. And I checked this morning for 2023 alone, year to date, over 20,000 publications talking about digital twins. So, a lot of people talking about it. And I think in some ways, because there's been such an explosion in interest, that's leading to lots of different kind of variations and definitions and a lot of confusion about what it is and what it isn't.

I was even listening to an interview with Dr. Michael Grieves, who is generally known as the originator of the digital twin concept, and a colleague of his, John Vickers, who's a technologist at NASA. So, Dr. Grieves came up with the concept in 2002 as part of a PLM concept within manufacturing and then had a bunch of various names. So, it wasn't originally digital twin. He used terms like mirrored spaces model and information mirroring model, things that really didn't kind of catch on.

But it was John Vickers who, in 2010, used and coined the term digital twin as part of a NASA technology roadmap. But in an interview just earlier this year, even they were kind of debating some of the nuances of the definition, which I think just highlights how hard it is to come to a really clear consensus on what it is and what it isn't.

But the interesting thing is, I think at the core concept, there's something that underlies all of these definitions that people are using. And if you really look back, it starts with why people use twins in the first place, going back to the physical twins that we use to better understand the real world. So I think, you know, you could probably go back to caves in prehistoric times and, you know, the early humans were drawing pictures of cliffs and talking about how they're going to force the buffalo over them. Crude and rudimentary, but a form of a physical twin to use to plan and understand what could happen in the future. And then you think maybe more modern, in the early 1800s is when military started adopting sandboards, so having kind of a geographic or topographical layout. And as they got information in from the field as dispatches, move the indicators of the troops around and use that to come up with tactical plans, right? So, this kind of concept of, how can I bring in information about the real thing to some sort of twin that I can then interrogate, manipulate, play around with to help make better decisions. And then that evolves into, well, now I can start bringing in data. 

So, as I was kind of preparing for this podcast, when you look for what's the history of a digital twin, there are hundreds of hits that talk about the Apollo program. And the reason why is in the development of the Apollo program, NASA had about 15 simulators that they used for both training purposes and to support operations. This all really came to a head with the Apollo 13 crisis. After the explosion on the command module, NASA started to use telemetry data coming from the actual asset and implementing those in the simulators. So instead of having the astronauts in the field do a bunch of trial and error with limited power and limited resources, they could then run those in the simulators on the ground. So, I would argue that's still a physical twin, but there was starting to have these elements of digital components to it, where you were modifying the simulation to reflect what was actually happening in reality.

And as we move forward, as more things become digital, as models and simulations start to advance, we can see how that continues to evolve. But I think it's really important to recognize that some things are going to be more physical twins and some things are going to be more digital twins. It's not a bright line separator. As I was thinking about that, I think the closest comparison is something like autonomous vehicles, where autonomous vehicles have been discussed and talked about since the 1920s when they started putting remote radio controls on cars in New York. And obviously here we are 100 years later, still talking about autonomous cars. We never got to that ideal end state for autonomous vehicles yet. It's been gradual step changes. And I think you can use that parallel quite nicely with digital twins. You know, we started in the past with physical twins and as we get better technology, we can incorporate more data, we move towards that kind of end state, ideal state of a digital twin, where I can have a full replica in a digital space of the physical asset of interest that I can manipulate, interrogate, as if I was doing that to the actual asset. But recognizing that, just like autonomous vehicles, it may be a while before we get to that end state, but there's still incremental value that can be seen in this kind of transitional gradient.

John Snyder (06:16.665)
How are digital twins built of physical assets? How does that process work?

Eric VanDerHorn (06:24.786)
Right. And I think this is why I think it's so important to kind of highlight that there is no grand unified model today, even though that may have been the vision, right? So, the original vision when Dr. Grieves laid it out was ultimately, how can I use information to replace wasting resources. And we already do a lot of that today with models and simulations. And I think there's almost too much of a focus these days on building out the perfect model or building out the perfect simulation. Really, the focus of digital twins is more so around, how can I start integrating all of these pieces that already exist? So, you have a vendor who has been collecting the data, a different vendor who's developed a finite element model, a different vendor who's developed a CAD model. How do you start to connect all of these things? Because that's where you're really going to start to see that value. I don't think that necessarily exists today for a number of reasons. There's challenges in getting the data, there's challenges in the lack of standards for interoperability between all of these different applications and functions.

So, we end up kind of with these intermediate steps of, how can I start to combine certain pieces that exist? How do I start to break down some silos? My recommendation to anyone who's thinking about a digital twin is, your goal shouldn't be to build a digital twin. Your goal should be to have a use case where you have expected outcomes, and the digital twin is a tool to help you start achieving those outcomes.

How to determine whether a digital twin is the right tool for that really comes down to some of the core tenants of a digital twin. I think it boils down to three things. One is, you need to be talking about something in the physical world. There has to be a physical asset that you're trying to represent. Otherwise, there's no point in having a twin. Two, there needs to be some sort of information exchange between that twin and its virtual representation. Then finally, again, the goal here should be to, how can I use this information to reduce resources, whether that's time, level of effort, building something, etc.

As part of that same interview with Grieves and Vickers, John Vickers is talking about the current Artemis program to build the next rocket to go to the moon. And obviously compared to the old Apollo program, now everything's being done in a very digital and model-based way. One example of this is one of the fuel tanks. They built out the digital model of this fuel tank, had run all of the tests to make sure that it could sustain all of the operational loads and conditions. They wanted to understand the failure modes.

But, given the current paradigm, they still went out and spent hundreds of millions of dollars to build a physical version of it and test it. And when they tested it, they basically found that the failure limit was within 3% of what had been predicted by the model. So, you go back and say, well, aren't we doing this a little bit backward? We're testing to check to see if our model is right, when really we should be modeling to determine when we need to test. And I think that's the paradigm shift that we're trying to move forward, towards with digital twins. And recognizing that ultimately, the goal here should be, how can I be using this to replace something that may be time or resource intensive today? And the actual mechanism of building it out is going to require having all of that infrastructure in place to do that. And I think that that's where all industries that are looking at digital twins are struggling. And so, you end up with kind of more targeted and focused solutions.

The goal is not to have that digital twin for the entire asset that encompasses the structural system, the control system, the machinery system, but rather more focused on one specific subset of a system and one specific action of that system so that you can achieve a very specific goal. That's where I think you're seeing the most success today.

John Snyder (10:20.482)
What do you see the advantages of using digital twins in the marine industry? Where is it being applied?

Eric VanDerHorn (10:29.321)
I think you're seeing emergence in a number of areas. Obviously, again, you want to kind of tie to those key tenants. It needs to be asset specific. And so, you want it, and you're going to be making an investment in building a digital twin. You want it to be something that's being used over time, right?

Part of the benefits of having a digital twin is that it lives with that physical asset. And so, you want to be focused on something that is changing over time, is asset specific. And I would say it has two kind of key focuses. The first would be very much a current or backwards looking focus. So, I got data, but there's aspects about the physical asset that I can't directly observe. And how can I use the digital twin to help me understand those things that I can't observe. A really good example of this would be fatigue damage. It's really hard to directly measure fatigue damage, especially measure fatigue damage across the entire vessel. But we can measure things like the loading or the vessel response, and then use the digital twin to assess what the corresponding fatigue damage is.

So, that ability to have this diagnosis of what's happening, of things that I didn't directly measure, couldn't directly see, that would be focus number one. Focus number two would be the more forward-looking viewpoint of, okay, I have an understanding of the system. If I were to change something, how would that affect the vessel? It could be something around modifications or repairs. It could be more operational nature of, hey, if I operated things in this way, what would I see change?

Another way to look at it is, hey, based on this progression, so, consider a degradation mechanism like corrosion. As we see corrosion over time, I have the data to understand how it's progressed to date, what's that look like going into the future as well. So that kind of backwards looking and forwards looking approach, that's where I would say the biggest advantage is that digital twins exist.

John Snyder (12:19.201)
So, you can do some sort of predictive maintenance or performance optimization then, using a digital twin?

Eric VanDerHorn (12:27.194)
Certainly, certainly. When you have that asset specific nature, as well as that kind of change over time, that's really the ideal use case where digital twin fits in.

John Snyder (12:37.980)
How about using a digital twin for training purposes? Is that a possibility as well?

Eric VanDerHorn (12:44.795)
I think so. I think you're going to end up with kind of a balance. It kind of goes back to the NASA example of a combination of a physical twin and a digital one. If you think about it on that gradient. Because obviously, there are certainly advantages to having some of those physical interfaces so that you have that kind of more realistic understanding as the user.

But, I could certainly see in the future going forward with the advancements in AR and VR type technologies where that interface becomes solely a digital or virtual one. And that may align quite nicely with advancements in digital twins to the point where they are fully encompassed as kind of that grand unified model.

John Snyder (13:28.219)
What do you see as some of the key benefits then for using a digital twin?

Eric VanDerHorn (13:33.676)
In terms of the value proposition, I think ultimately the value proposition is defined by the use case or the outcome. I think it's really important to recognize that, again, the digital twin isn't your outcome. The goal isn't to have a digital twin. The goal is to achieve something of value, and a digital twin is a tool or mechanism to achieve that. Whether that could be in the space of operational effectiveness, predictive maintenance, life extension, there's certainly lots of applications that are either being implemented today or being looked at to be implemented in the future that have those value propositions. And people are recognizing that digital twins are an ideal tool to support that. 

John Snyder (14:14.692)
You mentioned as an ideal tool and of course, much of the discussion today for shipowners is their decarbonization journey. Do you see digital twins fitting into that as a decision-making tool? 

Eric VanDerHorn (14:29.090)
Absolutely. Absolutely, in a number of different ways. You could consider digital twins as part of that operational effectiveness. So, using a digital twin to track the operational behavior, either A, to understand it so that you can go and make decisions. You can run simulations to understand how that will impact outcomes in the future.

Certainly opportunities for digital twins in terms of alternative fuel sources. So, you want to see what the impact of a new engine or a fuel saving device is going to be on that specific asset. Perfect opportunity for digital twins. That's really, I think, one of the key use cases is again, that idea of how can I manipulate and interrogate this model before I touch the actual physical one to understand what the ramifications are. And so, with all of these kinds of what-ifs and trade-offs, really an ideal situation for digital twin.

John Snyder (15:25.503)
Interesting. I was going to say, Eric, now with only a few minutes left in our podcast, I was wondering if you had any kind of key takeaways for our listeners.

Eric VanDerHorn (15:35.674)
Certainly recognizing that digital twins can be a bit confusing, even though, you know, you'll hear lots of different definitions. I think remembering that the core tenants are having some sort of physical asset that you want to be able to interrogate, manipulate, etc. And you want to do that in a digital space. And then two, you want to be able to use that information in a way instead of wasting resources in the physical world. I think if you hold to those key tenants, the actual definition itself is less important.

If you've got that physical asset, if you've got that information exchange, and if you're holding to those core tenants, you're on that gradient of a digital twin and everything else really just becomes marketing. Next thing you really want to look at is just make sure that you have that use case in mind. You're using the right tool to obtain that use case. Then I think you're set. Those are really the key aspects, I think, to see success now. Then as we continue to work towards that ideal end state, we're only going to get better at all of the interconnections and interoperability that's going to be required to reach the end state. But, I don't think trying to achieve that today is realistic, but I still think we can find value today with the technologies at hand.

John Snyder (16:42.719)
Well, Eric, thank you for an enlightening discussion and taking the mystery out of digital twins.

Eric VanDerHorn (16:49.893)
Yep, thanks for having me, John.

John Snyder (16:59.474)
Thank you for joining us today on Setting Course, an ABS Podcast. If you're interested in learning more about today's topic or listening to more episodes, visit www.eagle.org.