Heaps of recycled urban metal sorted into cubes
CC0 via Wikimedia Commons

ISCN Global Mixer: Urban Mining with AI

Urban mining is a circular economy approach that aims for systematically reusing waste as a resource, instead of disposing it. In this ISCN Global Mixer, Tanya Tsui from the MIT Senseable Amsterdam Lab illustrates how AI can be used to close material cycles in cities and invites discussion which potential applications the approach offers.

Event details

Paragraphs

Key takeaways

  • From circular to linear: Urban Mining can shift cities’ role from sole consumers of the world’s extracted materials to providers of it as well.
  • From nominal to real values: Tanya Tsui’s research approach aspires to train algorithmic models in order to more closely estimate real values of the prevalence of materials in buildings over their lifetime, rather than nominal values emerging at demolition.
  • Who wants to compete with the Netherlands? While data availability in the Netherlands for building materials is already quite strong and in high spatial resolution, the presented research project would be interested to know and source from other regions as well and thus widen the scope.
Global Mixer Urban Mining

Global materials demand is estimated to more than double from 79 billion tons to 167 billion tons till 2060 (OECD, 2019) and global urbanization is a main driver. At the same time, the immense negative environmental and socioeconomic consequences of virgin mining and material extraction render a shift to approaches of circular economy more and more urgent. Cities can thereby assume not only the role of absorbers of material demand, but also as providers of it – some of the biggest mines in the world are arguably big metropolises!

Against this background, Tanya Tsui, Postdoc researcher at the MIT Senseable City Lab in Amsterdam presented her project on facilitating urban mining with Big Data and AI models.

She set the scene with referencing three examples of AI-supported urban data science

  • With “Treepedia”, streetview images are algorithmically analysed to estimate green coverage in streets and urban quarters. 
  • “Skyview” uses streetview images facing the sky from among buildings, to estimate radiation on areas and thus provide proxies for urban heat islands. 
  • In order to assess the prevalence of secondary building materials, the approach has been so far to work with “archetypes” of buildings to be found in the city. Their extrapolation on the whole building stock returns an estimate of volumes to be expected. 

However, so far these values are nominal and not yet "real". And it estimates largely waste from demolition of buildings but is not granular enough to catch material flows from renovations and the entire lifecycle of buildings. 
For this next step, Tanya tries to link building data and (disaggregated) waste data that exist in cities, always with the necessity of three core data points: material type, amount and location

In the Netherlands, building data are organized and provided in already relatively good spatial resolution through the Pdok data base and Register of Buildings and Addresses (BAG). Waste data on the other hand is more difficult to constitute and access. Its tracking usually begins only beyond certain thresholds and often doesn’t entail exact typological and locational breakdowns. In her quest for suitable sources and stakeholders she is thinking about demolition contractors, building material exchange platforms, building or environmental permits but is very open for other suggestions as well.

In the lively discussion after her keynote, Tanya reflected on several questions. One was why she prioritizes building material for urban mining rather than e.g. household wares (refrigerators, washing machines, etc.). To her, this focus is warranted by the sheer scale of the building sector and its climate impact, but also because data is much more accessible for it. Asked about her challenges on an operational level when contacting stakeholders, she stressed that getting relevant and accurate data remains hard to get by. But in order to motivate data provision she also underlined that insights could be properly translated into better allocation and matching of material needs. Especially if the real-time analysis reached down to component levels, quicker and precise market clearance would be a big step towards a circular economy. 

She thus subsumed her input with a call for action to the audience as she is particularly interested in additional data sources for urban building and waste data from Germany and beyond. 

If a cooperation might be of interest to you, you can reach out directly to her or via the International Smart Cities Network at iscn@giz.de 

Further links:

https://senseable.mit.edu/

Contacts