At geoFluxus, we make tremendous efforts to get our hands on the most detailed and precise data available, even if it involves tiring bureaucratic processes and tedious data crunching. From there, we incorporate the actual material needs of the country or region, including local production, imports and exports. In short, we consider the entire geographic metabolism, not just the waste.

Only the detailed information about the content of resource flows allows us to understand how to reduce the demand for resources in such a way that consumption and regeneration are at least as fast. Knowing which materials or products are discarded as waste in a region is not enough until we know who discarded them and why and where they came from. If we know that 78 000 tons of high-grade wood have been discarded by construction companies inAmsterdam, we can target specific actors and understand the reasons for their waste.

At the same time, we need to know the context of our resource flows well to ensure that positive changes do not come at the expense of everyone's right to constantly strive to improve their well-being. We make sure that all of our data analysis covers geographic location and time as precisely as possible. By knowing where and when positive changes occur, we can compare those changes with indicators of well-being in the same time and space. For example, the dip in waste generation in 2020 unfortunately correlates better with unemployment benefits and mortality rates than with positive environmental recovery.

We determine the areas of greatest impact by first gathering as much information as possible about the area in question. Then we estimate how much we know, and how much we don't know. We communicate the uncertainty between known and unknown to our clients, because identifying data gaps is a good way to determine practical next steps. We place all figures in the larger context by comparing them with each other and selecting the largest ones based on size or environmental impact.

Reducing household waste is high on the agenda of many governments. Although households and individuals are also responsible for reducing the impact of their consumption, their relative share of total waste in the Amsterdam Metropolitan Area is barely 11% compared to the 89% of commercial waste. The unique combination of data allowed us to accurately track this ratio for the first time for the Amsterdam municipality. This creates perspective on the relative contribution of household waste to the total amount of waste in the area. This leads to increasingly informed decision-making.

Our experience in data science, sustainable resource management and urban planning enables us to bridge the gap between policy makers and data providers. We formulate questions, and translate them into queries that the data can answer. The precision of our work is communicated to our clients and is inherent in our tool.

For example, if we are asked about a city's recycling rates, we must first establish our boundaries and terminology. Are we talking about the downtown area or the entire urban area? Are we looking at the waste collected in the city, or all the waste processed in the city? Do the recycling rates refer to municipal waste, industrial waste or both? This could also lead to the question of what is considered recycling in the first place; everything that is actually recycled, or does collected unusable waste also count?

Even if a number were calculated, it could be misinterpreted. A larger number or ratio here would often be interpreted as a positive change, even if it increases due to increasing total inputs - and thus does absolutely nothing to reduce resource extraction.

This example shows how a simple question about recycling rates quickly requires extra attention to be worded correctly in a search query.

We provide not only numbers, but also knowledge. We put our numerical findings in the right (human) context. This means we communicate what we think the numbers mean, what they don't mean, and where to start digging if we want to know more.

To give an example: through our reclassification algorithm, we found that 19.6% of all commercial waste in the Amsterdam Metropolitan Area was directly reusable. This was the first time such a summary statistic was calculated. Yet it also turned out that only a few economic sectors provided sufficient additional data to make the calculation possible, leading to the identification of sector-specific data gaps.

However, this did not mean that these materials were immediately reused. In some cases, they were simply stored, without knowing if and when they would be reused or disposed of. In some cases, we saw such streams going to incineration. It seems plausible that they could have been processed by a method that retains a higher value. The reasons for not doing so can be varied, from legal to economic. The difference between streams that were labeled as directly reusable and their final processing option allowed us to identify specific material streams that consistently lost value, such as A-rated wood from construction that was burned or composted.

These important lessons provided the municipality with new insights, statistics, and starting points for policy development.

We help identify and translate messy data sources into structured information. During our projects we have seen the difference between what private actors in the value chain know about their resources and what is captured in (centralized) government databases.

Often the available information is not centrally accessible. Partly because of the perceived (administrative) burden, partly because it is simply not mandatory. Waste registration systems are designed to record waste, not resources. Similarly, the function of invoices is to record financial rather than material flows. But if every material transaction has a corresponding (digital) financial flow, all the necessary data exist.

Such indirect data sources can be transformed using sophisticated data analysis methods and algorithms. We use machine learning, natural language processing and geospatial probability algorithms to transform and enrich relevant datasets.

While reviewing government reports on waste, we noticed that there was more information available than we thought. In addition to the mandatory European list of wastes, waste reports contain an optional free text field in which further information can be provided. In our experience, about 20% of all waste reports record useful information in that field. Using natural language processing algorithms, we analyze the free text fields, and use them to reclassify waste streams for their potential reuse in a circular economy.

Het is duidelijk dat niet alle afgedankte producten en materialen onder dezelfde doelstellingen en hetzelfde beleid kunnen vallen. Organisch afval kan uiteraard niet worden "opgeknapt" of "gerepareerd", schroot uit het productieproces kan niet worden "hergebruikt" en hernieuwbare energiebronnen mogen niet worden "geweigerd" als zij kritieke grondstoffen vervangen. Dit vraagt om het onderscheid van doelstellingen - en dus metrieken - ten minste per materiaalgroep en economische sector.

Hoewel dit een nogal voor de hand liggend voorbeeld lijkt, zien we dat de onverwachte effecten en mismatches in elke geanalyseerde dimensie voorkomen. Daarom benaderen wij het probleem met vijf lenzen en hun combinaties voor het verstrekken van de metriek. De lenzen stellen ons in staat elk gegevenspunt in zijn specifieke context te bekijken, generalisatie te minimaliseren en de blik op het grote geheel gericht te houden.

1. De geospatiale lens maakt het mogelijk de metriek te bekijken volgens de verschillende administratieve eenheden en te vergelijken met andere administratief beschikbare gegevens, met name sociale en milieukwaliteit. Aangezien afvalproductie en -verwerking vaak niet binnen dezelfde administratieve eenheden plaatsvinden, worden ook de stromen en relaties daartussen belangrijk.

2. De Temporele Lense maakt het mogelijk veranderingen in de tijd te volgen, seizoensgebonden patronen waar te nemen, veranderingen in verband te brengen met andere grootschalige gebeurtenissen, de oorzaak-gevolgrelaties tussen beleid en de effectieve veranderingen te bepalen.


3. De Economische Sector Lense maakt het mogelijk gericht beleid op te stellen, mogelijkheden voor industriële symbiose te identificeren en de milieueffecten van economische activiteiten zelf te onderscheiden van de effecten van hun materialen. Economische sectoren kunnen worden verkend aan de hand van internationaal erkende NACE-codes of speciale groeperingen, bv. op basis van de positie in de waardeketen.


4. De materiaallens onderscheidt stromen op basis van hun inhoud: oorspronkelijke grondstof, product, samenstelling, afvalclassificatie-taxonomie zijn allemaal verschillende parameters die via de lens kunnen worden verkend.


5. De proceslens richt zich vooral op de afgedankte producten die een bepaalde behandeling ondergaan. Deze lense maakt het mogelijk potentiële waardebehoud te zien en de effecten van de materiaalverwijdering te beoordelen.

The clearest example we can give to show that circularity can easily become a vanity metric is the proposal of building two houses, both using exactly the same raw materials, but with one being traditional and the other completely modular; built with standardized units or sizes for flexible reuse. The first house will stand for 100 years as originally built and then be demolished. The second house is torn down and rebuilt by each of the 5 new owners. After 100 years, all the modular components are so worn out that they are discarded, just like the first house. If we were to look at this example from the perspective of 100 years - both houses would be linear. But at any point in those 100 years, the modular house would have been called "a perfectly circular house" and included in all the statistics of "resources returned to the economy."

It is not to say that modular houses should not exist, but to emphasize that even the most circular solutions can be linear in the long run.

This is why we give more nuance to the monitoring of resource flows and stocks and use labels such as biotic / abiotic, organic / inorganic, pure / composite, modular / fixed, critical / non-critical, etc. The labels make it possible to determine which strategy can be applied to each of the potential risks of resource loss, how long the resources are expected to last, the urgency of replacing their use with an alternative, and the strategies needed to dispose of them.

We integrate the principle of identifying truly circular resource flows by engaging in networking. We literally map resource flows to understand which resources, in which context, can be considered circular. We convert all the data, using geospatial algorithms and specially developed correspondence tables, into a format that can tell us more about the content of the flows and spatial movements.

First, we break down the content of the flows into materials and conditions for high-value applications to reverse the process; from end-of-life to cradle. The material content helps us understand whether the material can make a full circle and be reused in the same product or process. Sometimes such a cycle is possible only at the substance level (e.g., phosphorus can be recovered from wastewater and reused for food production), sometimes at the material level (e.g., using recycled glass to produce bottles over and over again), rarely at the product level. By assigning labels to material streams such as pure/mixed, compound/raw, product/material, etc., we can see if the given stream can find its way back to its source.

The spatial context is important to follow the stream's journey back to its origin, to truly understand whether or not its use can be made geographically circular. Resources leaving a country's borders do not necessarily indicate resource loss or less sustainable use. At the same time, imported resources are not necessarily less sustainable than locally produced ones. Only by tracking global reach is it possible to make reliable assessments.

geoFluxus is a spin-off company of Delft University of Technology and Amsterdam Institute for Advanced Metropolitan Solutions. We started our careers as academic researchers and took academic values with us. Therefore, we believe that anything paid for with public money should be accessible to the public.

Clearly, governments have information that is too sensitive to disclose publicly. Yet this data and information can be used by governments themselves, under their strict conditions, for the public good. We see that often data from one department is not available to another simply because the purpose of the data collection was different. We take on the task of finding and connecting the missing pieces of information and initiating partnerships across departments.

Some of the data used by our platform in the Netherlands comes from a national database that records corporate waste. While its company-specific information is currently used primarily to drive regulatory agencies, we have extended its value for transition to private parties and nongovernmental organizations through user-specific levels of anonymization. Access to these data creates opportunities for cross-integration of information, such as enrichment with other data sources for monitoring material flows and identifying opportunities for industrial symbiosis.

Finally, although the data are not open, our methods for processing them are open source and accessible to the public. So are the conclusions and aggregate figures we extract from the raw data.

When it comes to resource flows, we gather as much information as possible to see the big picture. We combine knowledge and methods from different disciplines into one platform. At the same time, we ensure that our findings can be reused by other organizations and combined in the broader knowledge network.

Currently, in our platform we combine traditional Material Flow Analysis (which answers the question, "How many resources are there in a system?") with Geographic Information Systems ("Where do these flows go?), Input Output Analysis ("Which economic activities consume which resources?"), Life Cycle Assessment ("What is the environmental impact of these flows?") and data visualization ("How can we understand the patterns?").

The road network on the left represents the amount of carbon emissions produced in 2019 on each road segment from waste transport alone. Producing such a picture from the underlying data requires integrating all the analysis methods mentioned earlier.

The road network, metadata and all related data can be downloaded for each visualization, and the calculation methods are freely accessible in our code repository. Therefore, our approach is scalable, open-source and multidisciplinary.