Barcelona is trying to create its digital twin. After mundane matters are sorted out — how to collect real-time data from sensors around the city, how to make a computer understand them, how to analyse and predict traffic and energy usage with machine learning and artificial intelligence — what is the overall aim?
“To build an oracle,” says Jordi Cirera Gonzalez, director of the Knowledge Society at Barcelona City Council, and a man not short on ambition. “Like the ancient Greeks’: a place where you can ask anything you can imagine and it’s possible to find some answer.”
Barcelona’s digital twin project will harness the power of the city’s supercomputer. Its latest version, MareNostrum 5, unveiled in December, has the ability to perform 314 million billion calculations per second. It lives within the deconsecrated Torre Girona chapel, on the campus of the Barcelona Polytechnic. Where once one might have prayed to God for an answer, now one goes to a computer.
In recent years, digital city-building has become a legitimate part of urban planning. Barcelona, Cambridge and Helsinki are among a number of European cities exploring how copies of themselves could prove useful in making their built environments sharper, faster, cleaner and greener.
What exists in real life is being rendered a second time in the digital space: creating a library of the past, an eagle’s-eye view of the present and, potentially, a vision of the future.
One of the most striking projects has been happening in Ukraine, where technology company Skeiron has, since 2022, been mapping the country’s monuments, under threat from bombing.
The project #SaveUkrainianHeritage has recorded 60 buildings, from the St Sofia Cathedral in Kyiv and the Chernivtsi National University — both Unesco world heritage sites — to wooden churches across the country, something Skeiron’s co-founder Yurii Prepodobnyi mentions with pride. There are thousands of them. “Some are only 20 or 30 square metres,” he says. “But Ukrainian churches keep Ukrainian identity.”
With laser measurements, drone photography and photogrammetry — the art of stitching photographs together — Prepodobnyi and his team can produce highly detailed 3D models.
They have even managed to recreate the exterior of the Mariupol drama theatre, destroyed in the early days of the Ukraine war, after calling for photographs and drone footage.
Another project, in Pompeii, has been using similar digital techniques to capture the evolution of excavations into a 3D model. The Pompeii I. 14 Project, led by Tulane University and Indiana State University, takes the process of excavating buildings within one block of Pompeii, Insula 14, and turns it into a digital representation. Using laser measurements, iPad Pros, a consumer drone and handheld cameras, a space can be measured to within a couple of millimetres. What is relayed back along the stream is a visual record of how a room changes over thousands of years, as the debris, volcanic eruption and layers of life that went before are revealed.
“So we can literally rewind time by removing layers, to show what was being done in this location before the eruption at Pompeii,” says Steve Aldrich, who was professor of geography at Indiana State University when working on the project. This Pompeii team could, in theory, go all the way back to a time before human settlement began on the site, then show the process of the city being built.
The concept of the digital twin started in the predigital age. Nasa used the technique — then called mirroring — to save the Apollo 13 spaceship. Before the mission, astronauts were trained on the ground using simulators. Later, in space, when oxygen tanks exploded, crippling the spacecraft, those same simulators were adapted to mirror the damage and to trial rescue techniques.
Mirroring engineering problems digitally developed in the early 2000s and in 2010 Nasa’s John Vickers first crystallised the term “digital twin” in one of the space organisation’s road map reports.
They are already used to monitor systems in healthcare, production lines, aircraft engines and product design. But human ambition means that eventually the question has arisen: could this technology be applied to cities?
The city-building simulation game SimCity launched 35 years ago next month. It works on preprogrammed algorithms. If you put a police station on a block, crime drops and properties nearby become more valuable.
The challenge for urban planners is to discover the underlying factors that would cause a particular street to become congested or one district to become dysfunctional. What the rules or algorithms are for this, they don’t know. But with data and computing power, they can begin to work backwards.
“Thanks to AI, we can answer questions about what is going to happen without knowing exactly the law that drives the system,” says Cirera. “But you need good data. Without it, you cannot train an artificial intelligence system.”
In city terms, twinning means recreating its assets — buildings, roads, trees — or its processes in digital format and then keeping the continually updated connection between this digital version and the physical one, like the project in Pompeii. If the data is uploaded in near real time, that would make it a “true” digital twin.
Early starters on digital twins are Helsinki and Singapore, which have created 3D representations of their cities with existing data. Helsinki modelled its railway station in 3D back in the 1980s. Its current mesh includes block plans and versions to show roofs that would suit solar panels, building energy use and it even makes its cityscape available to video game developers. It is updated regularly using city databases and is open to the public to view.
The next stage in the development of a twin is looking at what could be done with dynamic live data — trying to capture the motion of the city, interpret it, then model what will happen next. This can be done with sensors, which can decipher visual data or energy usage, or through data transfers such as mobile phones linking to masts.
Cambridge is a city suffering from its own success. Despite having a top university, with life sciences and technology at its heart, it also has an increasing population, traffic congestion and a mission to reduce car usage by 20 per cent. To add to its challenges, in December UK housing secretary Michael Gove announced 150,000 new homes for the Cambridge area to turn the area into Europe’s Silicon Valley. Cambridge city’s current population is only about 150,000. Where could a digital twin help?
Traffic is the starting point for data in the city. There are already a series of low-resolution cameras across the city, distinguishing people from trucks. That is being fed into a transport data dashboard, now under development. “So we can see, as we put in physical interventions, how that changes the counts and the number of people walking, cycling, using scooters or driving,” says Daniel Clarke, head of technology and innovation for the Greater Cambridge Partnership. Those interventions include new cycle paths — such as the Chisholm Trail on the northern side of Cambridge — or when Mill Road bridge had a long-term closure: how would it affect traffic in the area? Not as badly as feared, it turned out.
Part of the problem with developing a twin for cities is that it involves local governments, democratic accountability and multi-layered targets, which means many are stuck at an early stage. The data is there but what is the question you want to answer? And is it the right question? For example, if you are fixated on reducing traffic, should you look at more than just modes of transport?
“If you understand the propensity of people in an area to work from home but, actually, they’re not working from home, is that an issue with connectivity?” says Clarke of a Cambridge scenario. “If we improve the broadband in that area, would that mean fewer people travelling?”
Entopy, based near Cambridge, is run by Toby Mills. He builds platforms that cut across silos of information and connect various pieces of data together. Ontology — the philosophical concept dating back to Aristotle that asks, “What is the nature of a thing, and what is its relationship to another?” — has come back into digital twin tech-community parlance.
“At its end stage, [a digital twin] has to be sophisticated, using modelling, artificial intelligence, machine learning right the way through to data visualisation, virtual and augmented reality,” says Mills. “That’s where the term ontology comes in. Imagine trying to get 20, 30 or 40 different systems that aren’t designed to work together to integrate in a way that you can get meaningful intelligence from. [It] is quite a tricky thing to do.”
Where digital twins have had more solid progress is when there is a specific question to answer. One of Entopy’s projects was for a big British port (which Mills declines to name) that, as a result of Brexit, was going to be under considerable increased stress. Could its traffic be anticipated?
Using weather data, seasonal changes, CCTV, live action and AI, the team created a true digital twin that allows the port to prepare for variations in arrivals. “They are modelling the amount of traffic that is predicted for a period of about 15 days,” says Mills. Weather forecasts are broadly reliable for about the same length of time.
Another example from Entopy was a food market of 25 concessions, a commercial project and one that sought to understand people’s behaviour. A combination of point-of-sales data, CCTV and WiFi beacons were used to discover who was a local customer, a repeat customer, the volume of sales and where people liked to “dwell” — hang around — in the market, and whether being too crowded affected sales. Its visualisation was a map with hotspots. Its outcomes: queue management, targeted advertising to either a local or tourist audience and advising retailers on ordering and preparation of food. “This is a piece of infrastructure in the market that starts to lend itself more towards cities,” Mills says.
What both examples demonstrate is that a twin can go well beyond just measuring what is going on. With sufficient data and understanding between different components of the system, and sufficient computing power, it can begin to provide information about how best to act in the future. If X happens, then we should do Y. If we are to build new housing in the northern quarter of the city, we should rework this road junction and a new school will be required here.
But there is another concern. “We are living in very complex societies and sometimes we really have super-polarised opinions on what should be done,” says Stefania Paolazzi, a policy adviser in Bologna, which is working on its own digital twin project. Issues of traffic restriction, for example, are also about freedom. Bologna recently introduced a citywide 30kmh zone, which was met with some local unrest. One way of countering the controversy might have been to show what would happen once the policy was enacted, in an accessible visual format. “It could be a tool for transparency and could be a tool that citizens could use to be more informed on some specific problem,” says Paolazzi.
In Barcelona’s case, high on the agenda are extreme temperatures. “How do we adapt to fight climate change in terms of public policies?” asks Michael Donaldson Carbón, digital innovation commissioner for Barcelona City Council. That could take the form of how to design a street, restrict traffic or introduce green space, based on existing knowledge.
Better information is going to be what we have in the end, says Cirera. “In the city, you don’t have a development environment; you only have one city. The laboratory is the place where the planners go to test. So test in a digital twin and then develop or implant in the city. That’s going to be the value.”
While the Ukraine project is capturing what is present, and Pompeii what is past, city-scale projects in Barcelona and Cambridge are trying to guess what will happen next. And what will happen if you make an intervention, or don’t. It is “a philosophical more than a technical question”, says Donaldson.
Even so, we need to ensure we are asking the right question first, and second, that we understand and accept what it tells us. After all, the oracle’s answer was often a riddle in itself.
Find out about our latest stories first — follow @FTProperty on X or @ft_houseandhome on Instagram
