The digital twin is becoming a pervasive technology but what exactly can it do for manufacturing companies and how can they create value from it? Mark Venables spoke to Michael Campbell, executive vice president, augmented reality products at PTC to discover more.
Many companies talk about digital twins when they are in fact interactive 3D models. What are the properties that turn a 3D model into a digital twin?
Those companies that claim a 3D model is a digital twin are making the best of what they have, of what they can help customers with. What we would say is a digital twin is a comprehensive digital equivalent of a specific physical asset out in the real world. Nobody to our knowledge has ever created a completely comprehensive digital twin. Generally, people are creating a digital twin that represents enough information for them to solve specific business problems and address use cases.
How would you then define a digital twin?
We believe that a digital twin is only possible by intersecting the Project Object Model (POM) configuration and CAD data, with the experience that the actual physical device is having in the field. It requires two things. It requires context. What is this object? Say it is a tractor. I need to know what that tractor looked like when it was manufactured and delivered and then I need to know what is happening to it now.
I need to be able to understand sensor data, environmental data, so that I can have that rich digital definition that I can use to gain insight and perform analytics and drive action. We would say that you have got to have connectivity, so you know how it is behaving. You also would be well-served by having the complete digital definition.
Given that definition, how do you create value from digital twins?
It goes back to what is the value of industrial IoT; what problem are you trying to solve? If you are trying to maximise asset uptime, then you can use the digital twin to understand how the product is performing right now and predict what might happen to it in the future. Or at least understand what is happening to it right now and figure out why. A digital twin, just like a digital or a virtual prototype, on its own probably does not provide a lot of value. If it helps you address that business problem better, then that can be valuable, I would say.
Is it just a natural evolution of the digital thread companies are working with or is it a conscious decision they need to make to create digital twins?
It is a conscious decision that they that they need to make because it requires some effort. It requires some infrastructure, some instrumentation, some planning. I do not think that it happens by accident. Again, the idea of having this rich, complete digital definition without a plan to use it, without a problem to solve, can be a bit wasteful as well. We are really encouraging companies to think about what use is the case, what is the problem that they are trying to solve? Go start there and then scale out. Do not set off to create that comprehensive digital twin of the macro system.
Does a digital twin have to be a visual representation, or can it be a purely data model?
There are different lenses, there are problems I want to solve where I need a rich 3D digital definition. While I know who bought it and where it is in the world, what its serial number is, I’m beginning to piece together a lightweight digital twin. It’s not the complete definition but I think that data, pure data, as well as 3D definition, both contribute here.
Talk me through the basics to create a digital twin?
It is the definition intersected with the experience. I would say that is a digital photograph. That is a digital snapshot of the CAD data. If I do not know anything current about it, if I do not know what is happening to it now, where it is, whether it is on or off, those kinds of things, then it is not really a twin that is being kept up to date. It is more of a snapshot.
What were the challenges that you faced creating digital twins?
I think figuring out where and how you are going to capture this data is an important part of the question to answer. Intersecting that with the PLM definition because there are many different configurations of products. You want to get down to a serial number or a specific representation. Those are certainly some of the challenges that the technology PTC are developing is intended at to help address.
How far do you think we can go with these digital twins?
You can certainly imagine a rather comprehensive digital twin, but go back to what end? Why would you do that? What value are you getting out of doing that? That is the question you need to answer. Because it requires effort. It requires instrumentation and connectivity and if there is value in doing that then do it. I do not think you need to start by boiling the ocean.
Can you give me a couple of examples of work you have done?
A lot of people are starting around high valued assets where maximising uptime is critical. Think about heavy equipment tractors, manufacturing lines, and places where people want to ensure that things are up and running. They want to make sure that before something like a pump goes down, they know about it because that costs them lots of money.
Do you feel that the use of digital twins is going to continue to grow?
I think, absolutely. It is early days where we are just now beginning to appreciate the value of all this ubiquitous connectivity and all these sensors that people are putting into machines to help inform the digital twins. I think that some companies are starting out much more aggressively, others are starting out in a stepwise fashion. In general, I think that this is a trend that we will see continue in the future.