As companies struggle to adapt to disruptions caused by the coronavirus pandemic and other unanticipated factors, they’re finding value in the deployment of a digital twin. The exercise allows them to gain visibility of suppliers, and acquire vital data needed to respond to any number of events throughout the supply chain. In this conversation with SupplyChainBrain Editor-in-Chief Bob Bowman, Karen Panetta, IEEE fellow and Dean of Graduate Engineering at Tufts University, explains the concept behind the digital twin.
SCB: What is a digital twin?
Panetta: In its purest form, a digital twin is essentially a model of a physical system or process that you emulate. It's like a simulated copy of a real device or product. When I was back at Digital Equipment Corporation, we just called them simulators.
SCB: How broad a representation of a supply chain is it? Does it incorporate multiple tiers of suppliers upstream, and customers downstream?
Panetta: Yes. If you think at the most atomic level within a company, they're going to have their products and services all simulated in as much detail as possible. But as you move outside of your domain to your customers, or things that feed into the supply chain, those might be modeled at a higher level — what we call abstraction levels. They're more behavioral and mathematical, maybe using more current information streams from trends and statistics. You can include anything from transportation, logistics, and other entities. Some people are even modeling things from what’s going on with Wall Street, or trending in the news.
SCB: Before you can create a digital twin, is it first necessary that you map out what your supply chain looks like?
Panetta: Yes. You cannot build a robust digital twin unless you completely understand your process. Otherwise those models are not going to be accurate. One of the first things you do as you build a digital twin is make sure it’s matching what you’re seeing. With most digital twins today, where they do fall apart, it’s because they don’t have an accurate representation of the influencers on the supply chain.
SCB: Are companies successful in incorporating multiple tiers of suppliers into their digital twins?
Panetta: They haven't got there yet. They’re using a high-level model. The biggest piece to developing these things is being able to get what I call ground truth data. You can use lots of statistical models, but you have to run them on some sort of data.
SCB: How are they getting this data?
Panetta: That's one of the biggest problems right now with a lot of supply chains. When you’re talking about different levels of suppliers, somebody has to be able to share that information. And a lot of people consider it to be proprietary, so companies are left guessing. In the future, when somebody engages with a prospect, they're probably going to build in data agreements. They’ll say, if you want to be part of our supply chain, we need to be able to model your process, and you need to give us the data to validate it. It’s like saying, if I'm going to partner with you, you’re going to have to let me look behind the curtain.
SCB: How can a digital twin help a company to address disruptions in its supply chain, such as we're now seeing with the coronavirus pandemic?
Panetta: Say we’re looking for masks. Where are they coming from? We can use that information to help us decide where to manufacture. One of the things that's happening right now is companies are becoming able to consider more catastrophic events. Digital twins used to look at small disruptions. I don't think anybody has ever modeled what happens if the entire country stops production, or the transportation system shuts down. Now we will. What happens if I don't get the batteries or a part I need in time? We might not be getting it for a year. That's a huge change in modeling.
SCB: How do you anticipate and identify the stress points of a particular supply chain? How do you reveal those with the use of a digital twin?
Panetta: Again, in your modeling you have to think about the input source. You have say, if I've got 15 different vendors, here's the past performance of those vendors. You're adding weight to your input, so it's not like one known and continuous stream. You’re breaking it down into sub-components. So you know that if this supplier falls apart, it’s not going to impact your chain as much. But if this major supplier does, that has a much bigger impact on the supply chain.
SCB: Given the complexity of today's global supply chains, is artificial intelligence an essential element in building digital twins?
Panetta: Absolutely. If you're not doing that, you might as well pack up now, because that's the way of the future. The companies that are going to lead the way and be the most successful aren’t just optimizing getting products out the door; they’re anticipating fluctuations in the supply chain, as well as identifying new avenues to mitigate these risks.
SCB: And AI plays a role in that?
Panetta: AI isn’t a magic box that you put stuff in and it tells you the answer. If you don't have robust, validated data on which to build your digital twin, with inputs from all these different entities, then your AI is going to be wrong.
Timely, incisive articles delivered directly to your inbox.