Zoan’s Virtual Capital of the World, showing Helsinki Cathedral in Senate Square

How to decarbonise cities using digital twins

Image credit: Zoan; City Of Helsinki

Helsinki, a world leader in using digital models as tools to achieve its ambitious and rapid decarbonisation goal by 2030, is looking to harness the potential of geothermal energy with the help of local homeowners.

Two-thirds of the world’s people are expected to be living in cities by 2050, so urban planning for climate change is vital, both to mitigate extreme climate events and to realise a low-carbon future. At the forefront worldwide of deploying and using a digital twin model of their city is a small team in the Finnish capital Helsinki, led by architect and city planner Jarmo Suomisto.

The Helsinki digital twin is in fact two models, a Reality model, and a Semantic City GML – short for Geography Markup Language, an international standard in mapping software. Simply put, the reality model is the pretty pictures – made up of two billion polygons, and the semantic model is the underlying brains, with data layers that can be labelled, interrogated, added to, and changed. At the beginning the models were used to evaluate proposals in architectural competitions. The 1985 3D model took 12 hours to render a simple black and white street view image.

To give you an idea of the current complexity of the data, two years ago the 11 terabytes of oblique image data took five months to render the entire 400km² reality city model using a Parisian data farm. As Suomisto says: “The real power in the smart digital city is hidden beneath the fancy visuals. Much like an iceberg you only see 10 per cent, the rest is invisible, but is where the real power resides.”

Both models are accurate to better than 20cm. By comparison, Google Maps contains errors of up to 5m and therefore cannot be used for design purposes. One model feeds into the other, and artificial intelligence (AI) is beginning to make the reality model smarter by replacing the laborious and expensive job of manual labelling of, for example, windows, roofs, road signs etc.

Polygon mesh helsinki

Image credit: Jarmo Suomisto Helsinki 3D+

The reality model was created by flying a light aircraft at 1,200m above the city. The plane was equipped with five cameras that shot the same scenes from differing angles. Software combines these images by matching similarities to create a point cloud of the 3D city. All of Helsinki’s 80,000 buildings and over one million surfaces (roofs, windows etc) have been mapped in this way.

The beauty of the underlying semantic model is that it can be enriched endlessly, for example to reflect new roof construction materials and insulation, or renovations. It can be taught to understand the city better and better by using AI to label new elements, checked by ground-truthing against expensive manual labelling, to add further elements such as accessibility data like dropped kerbs. Day-to-day updates and maintenance of the semantic model are carried out by the City Hall surveyor.

In the future, there’s a desire to combine the aerially generated imagery with street-level scanning and images. “We have the technology to do this, but not the computing power to process the data for the whole of the city – it might take one or two years to do so,” says Suomisto, “but we can see what is possible by looking at small area projects.”

Suomisto is most proud of the Energy Atlas, based on the two models to help the city plan for carbon neutrality by 2030. It will not surprise you, given the climate in Finland, that district heating of buildings takes up 56 per cent of the carbon budget of the city. By comparison, transport is less than half that.

In the northern latitudes, the behaviour of solar radiation by season in the built environment is an important variable for the design of buildings and their heat demand, as the angle of incidence varies so much between the long summer days and short winter ones. It is also important for the quality of life and the comfort of their human inhabitants. The team has analysed the hours of solar radiation across the year, and also the shadows cast by built or to-be-built structures as they fall on parks, pavements, and other buildings. Both of these variables also impact on the efficacy of solar panels for energy production. The open-source platform allows any landlord or resident to consult the model and look up the solar potential of any given roof space on their mobile phone.

The city has conducted solar radiation and shadow analysis on one million surfaces in the built and to-be–built environment to allow developers and existing owners to calculate the actual benefits of solar panel installation, and the most efficient placements for those.

“We can examine each of the one million surfaces of the city to see which has the best solar energy potential now. It’s a great way to demonstrate to house owners how they can contribute to and support the decarbonisation goal plus save on energy inputs,” Suomisto explains. “We have also mapped the entire city for geothermal potential. Landlords can see at a glance if their property can access geothermal energy via a ground-source heat pump drilled to 150m below the surface in some areas of the city, or 300m deep in others. Obviously, depth of the shaft equates to the cost of tapping into geothermal resources.”

Helsinki Solar

Image credit: Jarmo Suomisto Helsinki 3D+

The data within the digital twin is vital to be able to demonstrate the benefit of tapping into these energy resources and to ultimately replace the current mix of coal-fired, nuclear, and hydroelectric sources for district heating. Equally, each building has an energy rating that enables the calculation of the effects of renovation or improved insulation against current fossil fuel-based energy consumption.

Juho-Pekka Virtanen, a geomatics expert at city-owned innovation company Forum Virium, spoke to E&T about areas where the Helsinki Digital Twin can be enhanced to become even more useful to urban planners and residents alike.

“One area in which the current models are deficient is the mapping of the green component of the city. Helsinki has enlisted the help of its residents in some pilot projects, to start collecting this mapping information via crowdsourcing, using phone cameras. We are also testing the use of embedded lidar sensors in some smartphones,” he says.

Projects have also allowed city dwellers to participate in green infrastructure planning using AR applications on their phones to visualise planting schemes devised by professional landscape architects, and to use these AR apps to help people envisage what final plantings will look like once complete. Both professionals and residents found the experience positive.

Green infrastructure – a city’s green spaces, parks, plants, and trees – play a key role in the lives of city residents, and their pets, but they are also critical in mitigating flash flooding by acting as buffering drains, which absorb sudden storm rainfall and release it slowly, quite the opposite of the way that storm runoff behaves in the built environment, which is dominated by hard surfaces like tarmac and brick. Combined with hydrological and geological data this modelling also allows the city to prepare for and predict flash flooding runoff and other extreme weather events, which are expected to become more frequent in the coming years. “Live sensors on new local weather stations could help us build simulations to understand urban flooding patterns better and help us to pinpoint areas where we need to do the most, earliest,” Virtanen adds.

Understanding the heat island effects in the city and how heavy storm winds will affect Helsinki are both future research priorities.

Another under-mapped part of the city has been dubbed ‘urban delivery pain points’ – a driver has reached a building address but does not know which of multiple entrances act as office reception, goods delivery doors, or loading bays. A pilot scheme run by local company Tietorahti Oy asks delivery drivers to input navigation tips to facilitate future deliveries. Data like this will be usefully deployed when robotic last-mile delivery services become more widespread, and can also be helpful for disabled people who may need to locate step-free entrances.

Forum Virium is also looking into terrestrial surveying technology, using either handheld mobile devices or phones, or vehicle-mounted lidars, which can yield much greater detail in a 3D model than the existing one, which was created from aerial images. If the contractual issues can be negotiated, the point clouds generated by passing autonomous vehicles could also be a new source of mapping data.

Another future ambition is to include the interior structures of buildings in the digital twin: again an aim that is technically achievable now, but in terms of data processing not practically feasible.

“We would also like to incorporate more and more live sensor data into the city models. Energy use by building is clearly available, but we need to get permissions to access it, and use it in a useful way – say to predict future energy demands ahead of a predicted changing climate and to run simulations based on different climate outcomes,” says Virtanen.

Another recent addition to the digital twin environment is advanced wind simulation software originally developed by German company CADFEM to refine the streamlining of automotive designs in wind tunnels. It works perfectly to simulate the behaviour of wind in the built environment as it travels over and around buildings. This allows the team to examine building heat loss seasonally, and to calculate the predicted energy loss as the wind cools surfaces.

Suomisto sees digital twin technology as an enabler of social change. “Our models show the city leaders and the citizens very clearly what they as residents or landlords can do to achieve our city’s goal of carbon neutrality in less than eight years from now. As a result, we are confident that we will get there by 2030,” he says. “We’ve come a long way from the first wooden 3D model of Helsinki produced by the city architect Eliel Saarinen in 1915.”

Digital twin

All fun and games

In the real world, the Helsinki city models have already been used to forecast the best places to see a fireworks display, or to plan crowd control and parking around a big music event. The model can be used to find the best site for the stage, simulating the noise nuisance and effects on local residents; or to predict traffic hotspots, and plan for traffic flows in advance of the actual concert. Marathon runners in the Helsinki Marathon can use the model to ‘run’ their route digitally in advance, planning toilet stops and being more aware of the city topography, arranging meeting points with supporters, and being clear as to feeding and water stations along their route.

Virtual Helsinki, the creation of Helsinki company Zoan, is a virtual space accessible online or via VR headsets for a more immersive experience. Virtual Helsinki opened the city to virtual tourism, especially welcome when we were locked down by Covid-19 restrictions. Virtual Helsinki was also the venue for popular lockdown concerts. The first was a May Day concert in the city’s Senate Square in 2020, by rapper duo JVG, attended by one and a half million fans internationally and deemed a huge success. “Virtual Helsinki demonstrates the opportunities for this kind of technology to change how we experience cities and destinations in the future,” said Jan Vapaavuori of Zoan at the time.  

The city digital twin has also been opened to ‘Minecraft’ for free and is very popular in Helsinki schools, where the kids can create and modify their own neighbourhoods. It is also possible to define an area of the digital map, download that to a USB stick, take it to your local library and have them print out the locality on their 3D printers.

And fully engaging in the virtual reality of the digital twin, Helsinki 3D+ leader Jarmo Suomisto will lead an architectural tour of Helsinki in UNREAL engine this September.


Sign up to the E&T News e-mail to get great stories like this delivered to your inbox every day.

Recent articles