Digital twins

As part of our R&D project on digital twins, we chatted with Mark Enzer, Chair of the Centre for Digital Built Britain’s Digital Framework Task Group (DFTG), about the purpose of digital twins, the national digital twin, and the spectrum of possible futures

Transcript of the podcast episode

Intro: Welcome to the Open Data Institute podcast, where we share inspiring stories about data use and its impacts. I’m Rachel Wilson. Today’s episode is all about digital twins. Which are, as one definition goes: “realistic digital representations of assets, processes or systems in the built or natural environment”.

There’s a whole range of systems we could call digital twins, from a simple 2D model of a network all the way up to real-time sensors in buildings using complex prediction models.

We at the ODI have been researching digital twins and how they can connect and interact.

The ODI sits on a body called the Digital Framework Task Group which was set up by the Centre for Digital Built Britain. The DFTG, as the task group is known, has been given the goal of developing a national digital twin by the year 2050.

Now, this national digital twin isn’t a big central brain of the country. Rather think of it as a decentralised ecosystem of digital twins that are connected and share data across different sectors. For instance, if you wanted to see how transport in an area affects local electricity usage, the national digital twin could help you.

The DFTG recently published a collection of 9 principles, known as the Gemini Principles, that are intended to guide the development of this national digital twin. So it includes principles like public good, openness, and federation.

To learn more, we sent our ODI colleague Fionntán out to talk to Mark Enzer, who is the Chair of the DFTG.

Interview: First question off the top is what are digital twins and why do we need them? I think a nice way into describing digital twins is to say that they’re a realistic digital representation of something physical. And that’s obviously a very basic definition but in the context of the built environment, I think what that means is it’s a realistic digital representation of assets and processes and systems within the built environment. So it’s not just meant to be just focused on individual assets but bigger things. And I think what makes a digital twin a twin rather than just any other model is this two-way connection between the physical world and the digital world. So you have data coming from the physical world into the digital twin. That can help make sense of that, help generate insights to enable better decisions and then those decisions drive better interventions and the interventions are on the physical world and make a difference. So that’s really the core of it. It’s a two-way connection between digital and physical. I don’t want you to pick favourites or anything like that but what are some good examples, or good illustrative examples, of digital twins out there today in the UK. So I guess the best examples that we see are not actually in infrastructure. This is where there are established examples of digital twins. And probably at the top of that pile are digital twins related to Formula 1 cars and people may well be familiar with that. Which basically take data from the car as it’s going around the track, helping to facilitate better decisions to drive winning races. So some of the best examples are not in infrastructure. However, within infrastructure there are emerging all sorts of very good examples and so you could take as one of them in a nuclear power plant working out how best to move around reactor rods or contaminated radioactive waste. Which you’d want to make sure that you can do it safely before putting people in. And so that kind of thing enables us to do what-if scenarios and planning in the digital world to enable these better decisions back in the physical. Doing our research, we found that there is a kind of core group of people who seem to understand or have a clear idea of what digital twins are and then beyond that there seems to be a lot of maybe noise or misconceptions and why do you think that is? I think it’s kind of inevitable at the beginning of something, where we haven’t all settled on a single definition and to a certain extent we have some people competing on the definitions and some saying my definition is better than yours. And where we’ve got to get to nationally is having a shared definition that we’re happy with. I think it’s kind of inevitable that we’ll go through a little bit of a period of turmoil before a definition we’re all happy with will emerge. But I think it’s really healthy to have the debate so that we collectively agree what we mean by digital twin. But you could also imagine having a digital twin at a network level and that could drive better decisions relating to resilience around that network and trade-offs that have to be made in it. Or alternatively you could imagine a digital twin at a whole system or sector level that might be used for forward planning and what-if scenarios. So really I think we can see the concept of digital twins applying at different levels, at different time scales but at the core of it the idea is the same. And that is you take data from the physical world, do something useful with it to drive decisions which mean you make better interventions back in the physical. We know from our research that there’s talk, at least there’s ideas, of trying to start to connect up, or at least think about connecting up, these digital twins and just wondering how exactly would that happen or what do you think that will look like? I think that really goes to the central tenet of the national digital twin, which we’re defining as this ecosystem of connected twins. And if we can see the value of individual digital twins, which we just talked about, helping make better operational maintenance, investment, planning decisions. That shows us how we can derive value from individual twins. But where we are moving beyond that and talking about connecting twins is really down to data sharing. Because the connection between twins will be about sharing data from one to the other. And where that becomes important is where there’s relevant data from one organisation or one sector in the decisions of another. So for example if you have a connection between the energy sector and the water sector that starts to point towards where it makes sense to share potentially live data across those kind of organisational boundaries. And where we think this is likely to be most important is where all of these sectors come together and that’s principally in the context of a city. So in a city is where all of these infrastructure sectors meet and it clearly has to make sense that there is data shared between those sector twins that can then facilitate cross-sector decisions for the benefit of the people in the city. Just on the data itself that’s being shared. So there’s layers of abstraction of information as you go up, so you have raw sensor data, you have maybe models that are built, you might have visualisations or whatever, do you see the majority of initial sharing of data to just be the low level sensor data or is there things further up the chain of abstraction that might be shared? Yes it could be all of those. What we think though is it should be driven by purpose. This is one of the key principles within the Gemini Principles. That rather than just building twins just because we can or because it’s fun that it should be because there’s a purpose driving it. And so I think that data that would be shared would really depend on the purpose to which its put. But we would definitely anticipate that different layers of data at different levels of abstraction, as per your question, would be relevant for certain purposes. And it wouldn’t just be asset data. So for example geospatial data is going to be very important to this because everything happens somewhere but also you can also imagine that weather data is going to be something that’s going to be important across many different twins. And so it’s not just the asset data and it’s not just the sensor data from those assets, or the performance data from those assets, you can imagine lots of different datasets coming to bear in a single digital twin. Do you think it’s possible to have a digital twin of humans? Is that a slightly scary idea or is it something you could imagine that would be part of the national digital twin? A digital twin of all the public? So that’s certainly not any part of our scope, what we’re looking at, digital twins relating to assets, processes and systems within the built environment. I think theoretically it is entirely possible to come up with a digital twin of a person but I think that then you end up moving in to some much more complicated ethical issues that are not part of the national digital twin and certainly what we’re not trying to do is model individual behaviour. Because a large part of what we’re trying to do with the national digital twin and it’s built right into the Gemini Principles is it’s got to be trustworthy and if people think there’s a digital version of them walking around with a predictive model of what their behaviour is that’s not going to be trusted. So I think we’ve got to be pretty clear we’re not going there, that’s not what this is about. You talked there of the national digital twin and I guess you’ve defined it somewhat and what do you hope to achieve with it, why was it felt that there was a need to define this national digital twin as opposed to just let things, let the build up of digital twins just happen by itself? Yeah it’s a great question. I think that we can see that there is real potential value in having this data sharing between twins. That facilitates different and better decisions at system and sector level. And what we could have done is just encourage people to develop their own individual twins. But then what we would be likely to have a number of years down the line is people saying if only we had thought earlier, we would have done something to enable these twins to connect. So I think what we are doing here really is starting with the end in mind. Because it would be much more difficult at a later stage to backtrack and build in the kind of consistency of data we would need to have connections between twins. So rather than just encourage people to develop their own digital twins in their own way with their own formats which potentially couldn’t talk to each other we feel that it’s good to start with that end in mind and facilitate the connection between twins early on which will then make our lives much easier later down the line when we want to connect them. You mentioned the long term vision for the national digital twin and I believe the timescale you’re thinking of is 30 years? 20-30 years? I guess we’re just wondering why such a long timescale? Why not 5, why 10? Is that you’re dealing with the built environment that things would move much slower than we could compare to the software world for example. Yes. So I think that when we’re looking at the overall timescale for this program, that’s just recognising that the built environment is a big thing and we haven’t finished building it yet. And so it might take many decades and a number of generations to build out the idea of a digital twin or a national digital twin across the whole of the built environment. However, what we would see is that there needs to be value released to show that we’re on the right journey well before that. I think that what we’re looking to drive here is an identifiable release of value within a couple of years which then will encourage us on the rest of the journey to build out the national digital twin across the rest of the built environment. You mentioned it earlier of the Gemini Principles as they’re called. And, just wondering could you tell us about them, why even make principles, what do they address and what are these Gemini Principles for. The Gemini Principles are values-based principles that will guide us on the journey. The idea of addressing those right up front before we’ve even really started on the technical issues about how you share data is that we feel that this whole journey needs to be values based. Because technology on its own is kind of amoral. You can use it for good or for ill. But what has kicked us off on this journey, is the National Infrastructure Commission’s report 'Data for the Public Good'. And so right from the beginning we’re intending that this will turn out for public good. In other words, the national digital twin has to be a good twin. It needs to be informed by values that we all ascribe to. So very early on in the journey we wanted to engage with the industry on what values we think should guide us on that journey and what we’ve come up with as a kind of discussion document are these Gemini Principles which effectively become the conscience of the national digital twin. And what that’s trying to do is give us something that we can carry with us through the journey to keep checking against to make sure that we’re still doing it in the way we started out to. Because we have a feeling technology will change rapidly but values shouldn’t change that rapidly. There’s something which is likely to be longer lasting and if we have a values-based national digital twin it’s much more likely to turn out well. I notice that one of the Gemini Principles is 'Federation'. Can you speak to the thinking behind that? What is maybe the worry with centralisation or whatever, and why emphasise the federation of this national digital twin? Yeah so there’s quite a strong perception that a centralised solution wouldn’t work. Because what I think what we’re looking for… Perception from whom? Perception from the industry and pretty much anyone we talked about, didn’t like the idea of one massive, monolithic model with some kind of heavy-handed, top-down control. That doesn’t suit our national character. But the idea of something that’s federated where there’s control at a lower level, at the level of where it makes sense but having some kind of coordination so that an ecosystem can be built, so there can be connections. That feels much more doable but also much more in line with the way that the industry actually works. So centralised, top-down, authoritarian control bears very little prospect of working whereas bottom-up, experience-driven, learning by doing and progressing by sharing a kind of a more organic way of growing a national digital twin connecting things together, that just has the feel of something that is more likely to succeed. But I guess bottom-up, are you also slightly worried about too bottom-up? Or it becoming this wild unfettered garden, or certain players becoming monopolies say? Yes. You’re exactly right. I think there is a danger at the other end of the spectrum. Another extreme of just having complete chaos where everyone does their own thing and there’s no kind of coordination. And I think that more Darwinian solution is likely to take longer to get to the right answer because it’s not necessarily the biggest or the loudest that are going to be the best. I think if we describe these two ends of a spectrum which we think don’t work. The top-down authoritarian approach, which is to have some experts come up with the perfect answer, impose it on the industry. It’s not going to work because the experts aren’t going to come up with the right answer in the first place, they’ll take too long, and industry won’t take it if they try to impose it. So that’s one end of the extreme. The other end of the extreme where it’s fully Darwinian and unguided evolution could potentially take too long and the answer that ends up dominating might not be the right answer. However in both of those extremes there is something which is good. And I think in the top-down extreme having experts who know what they’re talking about who can come together and help see the big picture of what might be best for the nation, there’s something good in there. Likewise there’s something very good in the bottom-up, Darwinian approach of learning by doing. This whole thing about a practitioner-based approach to growing standards from experience. So looking at the Gemini Principles there’s two in particular that maybe might be seen as there’s a certain tension possibly between them, between 'Public Good' and 'Value Creation'. And I wonder, do you have any thoughts on the balance between those principles particularly as it plays out between commercial organisations who have their own idea of value creation versus public bodies within the national digital twin who have their idea of public good? I really like that question because what I think you pointed to is something which we intentionally built in to the Gemini Principles and so we recognise within there, there’s potential tension. And you’ve pointed to an important one between value creation and public good. And there’s another one potentially between security and openness. So what we’ve tried to do in the Gemini Principles is indicate what’s important for us for the national digital twin. We’re not trying to say where the balance point should sit but where there is a tension, what we think is very healthy is to engage some kind of dialogue some kind of public debate about where that balance point should sit and where we want it to sit. And so I think when it comes to a value creation and public good we can say we want both. We definitely want public good, that’s what this is all about, that’s what the original National Infrastructure Commission report was all about but we also wanted to create value. And creating value for companies is a good thing, a certain level of profit is very healthy. So we’re not trying to say what the balance point is but when it comes to something like value creation you can kind of see that creating a whole new market where there can be new players working in creating digital twins helping to facilitate data sharing between twins. Putting in place all sorts of methodologies for not just sharing data but managing reference data etc. There’s a lot of work to be done in there which is value creating. And what we would hope is it can be an open market with low barrier to entry with many players. Probably what wouldn’t be value creating is having just a very small number of dominant players. And so I think it’s really the debate around the industry as to what balance point we want, is part of building that tension in to the principles. So say we were 30 years hence and in the future and the national digital twin has gone fantastically well. What would this look like? I mean what services, products, decisions would be enabled because of that? Or what benefits to society maybe on a higher level do you think would come from it? I think it really is all about that, it’s all about the benefits to society. The National Digital Twin programme being successful would be society benefiting from it and us enabling decisions that facilitate human flourishing and outcomes that are what we want in terms of economic, social and environmental outcomes. That’s the point of doing this. So it’s really not about just doing some geeky playing around with data. So I think a successful National Digital Twin programme would be enabling more human flourishing within the built environment. Ok so that’s the good future in 30 years time what would the bad future look like? What would that it’s not gone well? I think there’s probably a spectrum of bad futures. And there’s probably more ways of it going wrong than there are going right. That’s why we need to put the energy in to make it go right. And I guess at one end of it if we just haven’t grasped the value of data and information at all and all we’re doing is just playing around trying to make some kind of sense of physical infrastructure but not having the tools to do it, that would feel like a very bad future. I think that if we do grasp how valuable information is within the built environment and what kind of value we can release for the citizens of the UK, then we could get halfway to the national digital twin, which wouldn’t be good enough. And that halfway would be where everyone has gone off and developed their own digital twins and they’re getting some value out of the individual digital twins but we haven’t cracked the ability to share data between them. And I think that would be halfway, it might feel as if we’ve done some success because there would definitely be value from the view of the digital twins but somehow we would have let ourselves down by not being able to have this interoperability, this integration that we’re going for that is implicit in the national digital twin. Well I guess let’s hope for the good future then. Let’s go for the good future.