Designing digital twins

17 March 2020

Interview with 

Mark Girolami, University of Cambridge, the Alan Turing Institute


a view of a glass skyscraper shot from ground level


From the old to the new, what if instead of building new parts to a city, you could go into a computer and tweak bits of the city, and see what happens before laying a brick, a little like SimCity, it's called a digital twin and Mark Girolami, professor of civil engineering at Cambridge, joined Adam Murphy to chat about this new technology...

Mark - First of all, you need to take two perspectives here. The first is, what is digital and what is the twin? Now, engineering and science, for centuries, have always thought of the idea of a twin, an abstracted notion of some reality that you would want to understand and ultimately control. If you think of things such as the tide predicting machines from William Thompson going way back to the 1840s, that was a twin of the tidal systems that allowed individuals to predict when tides would be in or out, and they could then decide on when to set sail. So that's the whole notion of the twin, the digital component of course is that moving from some sort of physical and mechanical twin of a system that we want to study, to a digital, a computer based representation. And having those two things together, coupling the digital representation with the physical via data, provides us with this whole notion of a digital twin, which is much more different than a classical mathematical model where we would run simulations a bit like SimCity, but we'll never actually be going out and building our city from scratch and seeing if reality actually mirrors what the simulation suggested it would be.

Adam - And then whereabouts do we see this being used? What kind of applications does it have?

Mark - Digital twins are being used right across all of the sciences, right across all of the engineering disciplines. And of course what we're talking about today is urban infrastructure, city planning, regional planning. And we are seeing digital twins from rooms in our building. So what are the environmental conditions like? We heard about the underground farms, there are digital twins of those. How do you remove the heat, how do you control the heat, and how can you use those digital twins to predict the type of yield that the farm would produce? Right up to whole buildings where we have digital replicates of the geometry of the construction of the building, and then of course of the physics as well. And that then extends to whole urban central areas, it extends to cities, to transportation systems, to distribution of energy to the measurement and hopefully the improvement of air conditions.

Adam - How do you go about getting the data that you actually then put into these digital twins?

Mark - So data comes from a number of sources, and the obvious one is sensors. And what we are seeing certainly within city environments and urban environments is an almost "Renaissance" of being able to obtain data at unprecedented scales, that we've never had before. So whole pieces of infrastructure are getting sensors, networks of sensors, placed upon them and generating that data. And furthermore, citizens are also being able to contribute data. If you think of the iPhone I've got, if you look at that little sticker at the back, that measures nitrous oxide levels, and it communicates to the iPhone. And this measures the amount of nitrous oxide that I am experiencing in the location that I am in, at that particular time. And that then gets shared to help people that are, for example, in this case here, trying to improve upon the way in which air quality is measured and ultimately the actual improvement of the quality of air itself.

Adam - So taking that example where you have people with these sensors on the backs of their phone, what can you actually do in the digital twin to improve the city?

Mark - So as soon as you start getting measurements, you can take those measurements, you can take that data, and you can incorporate it into your digital representation. That can then be used to refine the predictions that come from the model. It can be used to validate the predictions that come from the model, and then ultimately guide policy makers and decision makers as to whether the predictions that are being made, and the levels of uncertainty associated with those, are something that should be acted upon. Because let's remember the aphorism that all models are wrong, but some are useful. And the key is that, you know, we need to understand that our digital twins are based on models that are wrong, but hopefully they are useful.

Adam - How far away for recently, like putting on a virtual reality headset and being able to walk around the digital twin of the city?

Mark - Well, we're actually there. You can put virtual reality headsets on and walk around whole buildings, you can take pieces apart and so on. It's the exact same. I just saw some work coming out of Berlin, where you're sitting as the driver in a train and actually going right through the city and experiencing the whole environment, the whole urban environment, and traveling through it. So that sort of technology is there. It emerged from the sixties, seventies, and eighties, in the technology that was used in developing flight simulators. And now, that's being used in being able to plan, design and develop and test cities.

Adam - And then if that's what we're doing now, what is the future of a digital twin?

Mark - Well, the fantastic thing that you would want from a digital twin is to have something that is going to be predictive. Something that is going to allow us to take into account those events that are going to have really high consequences. For example, the period of time that we're living through at the moment. Having digital twins that would enable us to forecast, and to look at multiple types of different scenarios, and act upon them. Now, that's at the global level, at the public level, but even at the individual level. If I go back to my little sticker on the back of my iPhone, a federated digital twin could then send a message to my phone to say, you should probably stop walking down Euston road because the pollution levels are too high. It's going to aggravate your asthma. I suggest you turn right here and go in to Costa coffee or whatever for five minutes.


Add a comment