In Switzerland, the first cities are testing out how infrastructure plans can be evaluated better using anonymised mobile communications data. The Future Cities Laboratory at the Singapore ETH Centre is going one step further. Using big data and the simulation of the entire city population, it is optimising the planning of entire infrastructure networks and districts.
What is the best way to use the limited funds? All large Swiss cities are faced with the same challenges. The agglomerations are growing. Cities are being condensed, new districts are emerging, and more people moving in is transforming the existing districts. Not only do the future transport infrastructures have to meet the new mobility requirements in order for the current growth to remain sustainable, the quality of life for residents also needs to increase in centres that are becoming more and more compact. For Kay Axhausen, head of the Institute for Transport Planning and Systems at ETH Zurich, one thing is clear: «Infrastructures that will only be used to their full capacity in forty years, as was the case in Switzerland when the motorway network was generously planned in the 1950s, are no longer realistic.»
Data basis for investments of billions
Specifically, projects with a scope of dozens of billions of francs are currently under discussion in Zurich, Basel, Geneva and Lausanne. As well as the expansion of the pedestrian and bicycle networks, this includes new tram and urban train lines, the expansion of bypass motorways, tunnels under the city centres or passing over or underneath the lakes. In order to decide which of these expensive plans make sense, those responsible in the administrative departments – and, as a final instance, also the voting public – depend on reliable data.
Mapping the dynamics of city life
The city of Geneva already recognised the potential of big data for this type of city planning in 2011. The pilot project «Ville Vivante» impressively shows how realistic movement data from mobile phones can map the dynamics of city life. In the meantime, technology has moved a step forward. The first Swiss cities have started using the mobile communications network for fact-based transport planning. The greatest advantages of the mobile movement profiles: they map mobility practically in its entirety, the data is comparatively cheap to collect and they are available in close to real time.
Until now, the cities had to rely on surveys like the federal state’s microcensus on mobility and traffic. For reasons of cost, such surveys can only be conducted every few years and involve fundamental uncertainties. Both the catalogue of questions and the sample are limited. This means less common patterns of behaviour remain undiscovered and, over the course of the analyses, new questions that arise remain unanswered. In addition, people have a natural tendency to provide the most positive possible picture of themselves with their answers.
The uncertainties in today’s traffic models are correspondingly large. The anonymised and aggregated mobile phone data from the mobile communications network, on the other hand, provide practically a complete and current picture of reality. «A comparison of the previous models with our data shows significant differences,» as David Watrin, responsible for security and intelligence products at Swisscom, records.
Simulate every single resident
Work at ETH’s Future Cities Laboratory in Singapore shows how great the potential for such data analyses is, not only for traffic but for all areas of town planning. Here, traffic engineers, architects, sociologists, energy experts, designers, interior designers and modelling specialists work together on solutions for the city of the future. For this, they use sophisticated models like MATSim Singapore (Multi-Agent Transport Simulation), in which the specialists on Axhausen’s team record the entire mobility of the Asian city state.
In order to decide which of these expensive plans make sense, those responsible in the administrative departments – and, as a final instance, also the voting public – depend on reliable data.
Each of the more than five million residents is mapped in this simulation as a software agent with individual properties. These are sourced from the available statistical data relating to places of residence, workplaces and population structures, from the results of surveys and from the user profiles of the wireless smart cards with which the use of public transport in Singapore is automatically invoiced. «The numerous interactions between the various elements quickly become confusing when considered individually. With the help of models, we can understand them,» as Axhausen explains.
By not only simulating the entire mobility system in the MATSim model but also, for the first time, the behaviour of the residents, the ETH researchers receive additional key information. Among other things, they can estimate how passengers will react to a change in the offer. In Singapore, the researchers have, for example, run through on the computer how a very long bus route that is always out of sync with the timetable would have to be split up. With the separation into two partial routes, it was not only possible to stabilise the timetable. Thanks to the better service, the route also regained customers that it had previously lost – as predicted in the model.
Living space that works
However, the public and private transport that is as friction free as possible is not the only decisive factor for an attractive city. At least as important are attractive pedestrian zones in which neighbourhood life can develop. And here, too, simulations can provide decisive support so that the plans lead to living spaces that actually work in practice. How can subways and bridges over and under roads best be incorporated into the pedestrian network? Which type of road acts as a barrier to pedestrian traffic? Where are shops or street cafés needed to entice passers-by? How does a park need to be laid out in the surroundings so that people actually use it as an island of calm? The reactions of the virtual residents show which variants promise success in an interplay with the entire ecosystem of the city and which are hardly likely to work.
Control in the short term...
Currently, Axhausen sees the main benefit of big data in short-term control: «You recognise changes immediately and can thus react promptly.» Specifically, the movement data from mobile phones allows us to recognise traffic jams or problematic gatherings of people quickly and even to predict them within a certain period of time. In the Swiss cities, for example, the control of the traffic lights could be improved, as Watrin explains: «Unlike today’s firmly anchored counting sensors, our data not only records the major thoroughfares but all roads. This provides a much more complete picture of the traffic situation.»
...and continuously optimise
The fact that big data makes changes visible immediately also provides town planners with a further powerful lever: in the same way as Google and other large website operators currently improve the usability of new functions step by step by means of real-time analyses, in the future, town planners will also be able to optimise infrastructures and districts continuously. In the mobile communications data, it is possible to see promptly which measures achieve the desired effect and which do not, and changes in the use of city spaces also become visible much faster.