In Stay Sharp Episode 93, Unpacking Digital Twins Series: Myths and Misconceptions, co-hosts Jonathan Scott and Juliann Grant are joined by the Director of Mission Integration at SAIC, David Ewing. Dave has worked in industry and at a software OEM, and he’s the architect of ReadyOne, SAIC’s award-winning digital engineering platform for the U.S. Department of Defense. Together, the trio tackle what a digital twin is and isn’t, as well as what misinterpretations exist.
A Model is Not a Digital Twin
Dave starts with his definition of a digital twin. “The key thing is to have the physical, the thing, the asset,” he says. “To have a digital twin, you’ve got to have the physical asset.” Models, on the other hand, are idealized virtual representations of a product, a piece of getting you to a twin. While some people think a model is a virtual twin, Jonathan notes a virtual twin is “a big collection of models that represent the whole behavior or the whole something of the physical asset. It’s just the physical one hasn’t been built yet.”
When a physical asset is built, the ideal becomes reality. At the point it rolls off the production line, it’s a perfect twin to your virtual twin, but then things change. Dave gives the example of aircraft, which require parts to be maintained or replaced, even sometimes with different equipment. He says, “Now that asset is subtly different, doesn’t match the CAD anymore. Doesn’t match the system model anymore. This is where now that twin is taking a life on its own. Now I need to rev the CAD into a unique instance.” That’s when a unique digital twin is born, as is the ability to track each specific instance of your physical assets. “That’s the point of the value of a twin over the value of the model,” Jonathan points out. “Because the value of the model is that idealized answer the question, but the twin is that specific answer the question.”
Demystifying the Digital Thread
To create your digital twin, and all the unique instances of it, you need to capture all the data about it from a variety of sources and relate it together—that’s the digital thread. In a diagram, a digital thread looks like boxes connected by little lines, but the context those lines represent is critical. “It’s a ton of stuff on that little line behind the scenes from a configuration perspective,” Dave explains. “Business rules, all kinds of things that happen when you make that connection.” Dave also cautions against thinking that because you have data and models in some sort of storage structure, and can export a file and import it elsewhere, you’ve got a thread. “No, you don’t,” he says, “The whole point was the linkage.”
Dave does admit it’s better to do something with what you have, rather than doing nothing—and you don’t have to start by pouring human and financial resources into creating the perfect giant system to manage your digital twins. The terms or systems may be new, Jonathan notes, but the concepts and practices aren’t. “It’s traceability, but with digital added in so we can do it more easily,” he says. “A document-based digital thread is better than no digital thread, but a data-centric digital thread is really what you want, but you might have to take steps to get there.” Start small, use what you’ve got, Dave says, just remember, “Context is a simple word, but that’s part of what changes the game.”
Using AI with Digital Twins
Jonathan asks the ubiquitous question: what about AI? Specifically, will it work and is it a good idea? Dave answers yes to both, saying, “I actually think AI is going to become our connectors of tomorrow.” His company is already using AI successfully to research and chew through large documents or datasets, though the baseline assumption is that you have to train the AI and verify your models to do so. Jonathan summarizes Dave’s beliefs: “AI’s going to help us get there, but it’s not going to do it for us and we shouldn’t let it do it for us. We’ve got to trust but verify.” The trio agrees, AI isn’t a silver bullet or magic, but it can be an accelerator.
Learn More About an Expert’s Take on Digital Twins
The full podcast contains more discussion and details on a variety of topics, including the exception where a model actually can be a twin in specific situations, when and why you need configuration management to support your digital twins, the evolution of digitize to digitalize to digital engineering or transformation, and more.
Be sure to check out Stay Sharp Episode 93: Unpacking Digital Twins Series: Myths and Misconceptions and join us each week for a new podcast.
Follow the Razorleaf Podcast, Stay Sharp in Digital Engineering on:
