Introducing ODYM-RECC – A community model for circular economy and material efficiency assessments

The UNEP International Resource Panel Report (IRP) report on “RESOURCE EFFICIENCY AND CLIMATE CHANGE – Material Efficiency Strategies for a Low-Carbon Future” [1] released in December 2019 is the result of a major collaborative effort among industrial ecology researchers [2]. It features an assessment of the economy-wide impacts of material efficiency in passenger vehicles and residential buildings in the G7 countries, the EU, India, and China. For the first time, the quantitative link between service provision (mobility and thermal comfort), product stocks, product flows, material flows, energy demand, and GHG emissions is established at the full (country) scale. The model framework behind this report is both a large-scale prospective LCA and a dynamic MFA with links to service provision, energy and emissions. The framework represents a milestone in itself, which is why we present it in a separate journal paper manuscript [3]. At the core of the framework is the Open Dynamic Material Systems Model for the Resource Efficiency-Climate Change Nexus (ODYM-RECC), which we are now releasing as open source community tool [4,5].

In this piece, I write about the challenges we had to overcome when gathering the data and building the model framework and about the plans for its future development.

Data: For such assessment, a large number and variety of data from different disciplines are needed. Information such as the age-cohort breakdown of current in-use stocks, their energy consumption, energy carrier mix, and material composition needs to be known and calibrated against aggregate energy statistics to make sure our starting point is right. We then need scenarios: For future population, service levels, product and technology mix, product material composition and energy consumption, and the impacts of future energy and raw material supply. Then, data on current and future material cycle processes: the yield factors, energy consumption, and direct GHG emissions of material production, manufacturing, and the waste management industries. Finally, information about the improvement potential as a result of material efficiency all across the system. We were able to collect a large share of these and many other data from the literature and fill remaining gaps by reasonable assumptions. Some data are readily available for download and use, such as the different parameter of the shared socioeconomic pathways (SSP, [6]) or the EU Hotmaps project on the building stock [7], which saved us a lot of time. Still, many person-months went into the compilation of the project database. The details about the data gathering process are described in the supplementary documents of ref. [1]. The ODYM-RECC database now contains 56 parameters, and we applied the following principles to our data collection:

  • All data are structured into a common data model and format. With the variety of data gathered and the wish to easily improve existing data and expand the database, a common data model and data format is very helpful. This way, the information contained in each dataset and its link to the project’s system definition can be clearly defined and the common data format helps when several people work with the data. For the RECC assessment, database and model framework were co-designed to build a powerful dynamic MFA tool. Actually, one of the largest synergies of this project and the IE research community is that I managed to generalize the data model originally developed for this project into a comprehensive data model for socioeconomic metabolism [8] and that the data template used in this project is now also the data template used for the industrial ecology data commons prototype [9]. For me, this is a strong hint that we have done some really useful and scalable work here!
  • The project’s database is made available under a permissive license. We need to speed up knowledge accumulation if we want to continue the success story of the industrial ecology community. By releasing our database to the community [5], we contribute to knowledge accumulation and are confident that the improvement and additions made by others will eventually feed back into the project. Also, knowing that your data will be open source forces you to work properly with the data and provide sufficient documentation. The database contains data from proprietary databases, including ecoinvent and the IEA Energy Technology Perspectives, but we use only individual numbers taken from the original formatting so that these databases are not reproduced in our work.
  • A project-wide classification is used. In a master file, the classification items (which labels are used/available for a given data aspect, like countries, materials, and sectors) are defined. All data files must use these items, and if the raw data come in other classifications or use no classification at all, the data supplier must document how the original data were matched/aggregated/disaggregated to the project-wide classifications. This setup allows us to quickly add new data and to reduce the bilateral coordination needs between data suppliers. It also helps external colleagues to supply data – they already know which classifications and items therein are to be used.
  • All data handling is traceable. We need to make sure that the data gathering and processing is documented in a way that each individual number can be traced back to where it came from, either an external source or an internal calculation using other external data. To that end, each data file has a log sheet where all processing is documented and a ref sheet where all references are listed. Moreover, an internal database, that is not publicly available, was created to store the different raw data files obtained from the various web and other sources, as many of them are confidential. This archive also contains internal Excel workbooks and scripts used for data extraction, processing, and formatting. We also made an effort to structure all raw data files by sector, country, and datatype. While this process helped us to keep the overview during the project, it surely can be improved, as it is still difficult and requires inside expert knowledge to identify exactly which data where changed for which version of the dataset and how. Here, we see the need for more exchange of experiences within the community and the willingness to make an effort to develop and spread better ways of making data handling traceable.

Model development: We need to move material flow analysis tools to the next level. I got frustrated when trying to extract results from some of my old work and realized that I did not save all potentially interesting results. Neither did I archive all the raw data (because the UN Comtrade files were too large back then), nor can I actually rerun the model (because I don’t have a Matlab license anymore). This mess has to stop, and I swore myself to do it better this time. Over the years, I developed a model logging and self-archiving framework, where each model run is assigned a UUID and a separate result folder, to which the model script, the config file, and some core results are stored. The raw data are archived separately, and a particular model run can be recreated at any time if more results that what were archived are needed. A scenario overview file contains a list of all model runs and the parameters that change each time, and the model scripts themselves are version managed and available via GitHub. The core model, ODYM-RECC, was built using these insights and structure. It is a dynamic material cycle model built on the ODYM framework for dynamic MFA [10]. It is modular, the core dynamic stock routines are tested, and it combines a description of the stocks at the product, material, and chemical element layer. It is available under an open source license [4].

Vision and next steps: The framework is designed so that new sectors, countries, products, materials, and chemical elements can be added just by supplying the relevant data and changing the model config file. For some additions, like expanding the regional scope to countries not yet covered, the model code doesn’t even need to be changed. For adding new sectors, the relevant product in-use stock dynamics must be coded, but the material cycle model remains unchanged, it just gets a larger or different product output and scrap input. We want to use this flexibility of the model to add more sectors, including electricity generation technologies and their material implications, appliances, and nonresidential buildings. For the latter, new product archetypes that are then scaled up need to be modelled. The big gaps are infrastructure and industrial assets, for which not much literature exists and where substantial additional funding may be needed. As we expand the sectoral and regional coverage, more and more of the total global material production will be covered and the we approach an estimation of the total potential of material efficiency in the material cycles. At some point, when ca. 70% of the demand of a certain metal are captured, we can simply scale up to include the remaining 30%, even though the corresponding end-use sectors are not described in detail. This way, we can estimate the future total global recycling and circular economy potential for certain materials, which is another missing building block in the assessment of sustainable development strategies.

We also hope that the capabilities of this model framework will spur the interest of the community and beyond. Cumulative research should be a new norm, and this framework can help to compile, integrate, and link the different data on current socioeconomic metabolism and in-use stocks, in particular. It can also be used to calculate the material consequences of product and service scenarios from other model frameworks like energy system and integrated assessment models. Via contributions from the community, environmental impacts other than GHG, other material substitution options, and other material production technologies can be added and their system-wide consequences can be explored. It is also possible to narrow down the scope to individual countries, including those that have not been covered yet and those that are currently part of more aggregate regions.

Finally, we see a strong need to link our scenarios to macroeconomic assessments of future lifestyles and industrial production, as provided, for example, by CGE and macro-econometric models. A simple cost assessment of the different material efficiency strategies could be a first step, in particular, if that cost assessment can be link to questions of environmental and social justice.

If you are interested in using the data or the model or want to contribute to expanding the database and the capabilities of the model framework, let us know!



[1] IRP Report link:

[2] Author list of the modelling chapters of the IRP report and acknowledged colleagues:

Lead authors: Edgar Hertwich, Reid Lifset, Stefan Pauliuk, and Niko Heeren.

Contributing authors: Saleem Ali, Qingshi Tu, Fulvio Ardente, Peter Berrill, Tomer Fishman, Koichi Kanaoka, Joanna Kulczycka, Tamar Makov, Eric Masanet, Paul Wolfram.

Research assistance, feedback, data: Elvis Acheampong, Elisabeth Beardsley, Tzruya Calvão Chebach, Kimberly Cochran, Luca Ciacci, Martin Clifford, Matthew Eckelman, Seiji Hashimoto, Stephanie Hsiung, Beijia Huang, Aishwarya Iyer, Finnegan Kallmyer, Joanna Kul, Nauman Khursid, Stefanie Klose, Douglas Mainhart, Kamila Michalowska, Rupert Myers, Farnaz Nojavan Asghari, Elsa Olivetti, Sarah Pamenter, Jason Pearson Adam Stocker, Laurent Vandepaer, Shubhra Verma , Paula Vollmer, Eric Williams, Jeff Zabel, Sola Zheng and Bing Zhu.

[3] RECC framework paper: Linking Service Provision to Material Cycles – A New Scenario and Model Framework for Studying the Resource Efficiency-Climate Change Nexus. By Stefan Pauliuk, Tomer Fishman, Niko Heeren, Peter Berrill, Qingshi Tu, Paul Wolfram, Edgar G. Hertwich. In preparation.


[5] IRP Report input database:, DOI  10.5281/zenodo.3566865    IRP Report results: , DOI 10.5281/zenodo.3566859



[8] A General Data Model for Socioeconomic Metabolism and its Implementation in an Industrial Ecology Data Commons Prototype. Stefan Pauliuk, Niko Heeren, Mohammad Mahadi Hasan, and Daniel B Müller. Journal of Industrial Ecology, published online, DOI 10.1111/jiec.12890.


[10] ODYM – An Open Software Framework for Studying Dynamic Material Systems – Principles, Implementation, and Data Structures, by Stefan Pauliuk and Niko Heeren. Journal of Industrial Ecology, published online, DOI 10.1111/jiec.12952

3 thoughts on “Introducing ODYM-RECC – A community model for circular economy and material efficiency assessments

Leave a Reply

Your email address will not be published. Required fields are marked *