Methodology paper - Composition data collection and consolidation
Abstract
The FutuRaM project aims to develop the Urban Mine Platform, a comprehensive knowledge base on the availability and recoverability of secondary raw materials across the European Union, Iceland, Norway, Switzerland, and the United Kingdom. The project focuses on six main waste streams: waste batteries, end-of-life vehicles, waste electrical and electronic equipment, construction and demolition waste from buildings and wind turbines, slags and ashes, and mining waste. These waste streams represent significant sources of critical and strategic raw materials. To estimate the availability of secondary raw materials from 2010 to 2050 under different scenarios, the project develops stock, flow, and recovery models based on detailed composition data. This data describes the structure of products, their components and material fractions, and elemental content, providing essential insights into the quantity, distribution, and potential recovery of critical raw materials in waste streams. The aim of milestone 23 of the FutuRaM project is to provide consolidated composition datasets for all waste streams that can be used for subsequent stock and flow and recovery modelling as well as for recoverability and secondary raw material assessment. The FutuRaM composition data structure distinguishes between hierarchical layers within a waste stock or flow. Each layer in the hierarchy is composed of the entities of the previous one. This hierarchical structure enables a systematic and scalable description of composition across diverse waste streams. To better reflect the varying characteristics of the waste streams, the composition data model employs two distinct approaches, a product-centric approach and a deposit-centric approach. Based on this structure, composition data was collected using specifically developed templates enabling the collection of data according to the composition data model from various sources. For the consolidation of the collected raw data, a harmonized method was developed in order to transform the scattered raw data into consistent and generic datasets with the average composition of relevant products (or product groups) and waste stocks and flows. Data points with different data structures and levels of detail can be combined into generic compositions for products and waste stocks and flows by transforming them into the desired data structure and defining appropriate levels of detail. This methodology paper provides an overview of the composition data collection for each waste stream. The consolidation method is described and specifications for each waste stream illustrated. Challenges during the process are highlighted and recommendations for further improvement of the processes are provided. Furthermore, in this methodology paper, the considerations regarding future composition and composition data validation are presented.
Full text
Methodology paper Composition data collection and consolidation
Methodology paper – Consolidated composition datasets | www.futuram.eu | 2 Project: Future Availability of Secondary Raw Materials Acronym: FutuRaM Grant Agreement: 1010585 Funding Scheme: Horizon Europe Webpage: www.futuram.eu Work Package: WP3 Work Package Leader: Vera Susanne Rotter, TUB Document Title: Methodology paper Composition data collection and consolidation Report Leader: Katharina Kippert, TUB Version: V2 Status: Final version Author(S): Katharina Kippert, Han-Jung Hu, Nathalie Korf, Ronald Arvidsson, Catrin Böcher, Manuele Capelli, Vanessa Forti, Giulia Iattoni, Franziska Maisel, Daniel Monfort Climent, Audrey Philippe, Matthias Rösslein, Lorena Sanz Prada, Maximilian Schubert, Max Tippner, Liesbet van den Abeele, Joren van Stee, Vera Susanne Rotter E-Mail: [email protected] Reviewed By: Cornelis Peter Baldé, Biborka Boga, Audrey Philippe, Kirsten Remmen, Patrick Wäger, Tales Yamamoto Due Date: n/a Date Of Submission: 01.10.2025 Dissemination Level: public
Methodology paper – Consolidated composition datasets | www.futuram.eu | 3 Version History VER. NO. DATE REASONS FOR RELEASE RESPONSIBLE 1 20.05.2025 Revision by consortium Katharina Kippert, Vera Susanne Rotter 2 01.10.2025 Final version Katharina Kippert, Vera Susanne Rotter Notice The contents of this document are the copyright of the FutuRaM consortium and shall not be copied in whole, in part, or otherwise reproduced (whether by photographic, reprographic or any other method), and the contents thereof shall not be divulged to any other person or organisation without prior written permission. Such consent is hereby automatically given to all members who have entered into the FutuRaM Consortium Agreement, dated 1st June 2022, and to the European Commission to use and disseminate this information. This information and content of this report is the sole responsibility of the FutuRaM consortium members and does not necessarily represent the views expressed by the European Commission or its services. Whilst the information contained in the documents and webpages of the project is believed to be accurate, the author(s) or any other participant in the FutuRaM consortium makes no warranty of any of any kind with regard to this material.
Methodology paper – Consolidated composition datasets | www.futuram.eu | 4 Content EXECUTIVE SUMMARY ......................................................................................................................................... 8 1 INTRODUCTION ............................................................................................................................................ 10 2 PURPOSE OF THE METHODOLOGY PAPER ........................................................................................ 11 3 COMPOSITION DATA COLLECTION ....................................................................................................... 12 3.1 COMPOSITION DATA MODEL ................................................................................................................................ 12 3.2 COMPOSITION DATA COLLECTION TEMPLATES.................................................................................................16 3.2.1 Data and metadata types .......................................................................................................................................17 3.2.2 Harmonized code lists ............................................................................................................................................... 18 3.2.3 Data quality assessment ........................................................................................................................................20 4 WASTE STREAM-SPECIFIC COMPOSITION DATA COLLECTION ................................................ 21 4.1 WASTE BATTERIES .................................................................................................................................................. 21 4.2 END-OF-LIFE VEHICLES ........................................................................................................................................24 4.3 WASTE ELECTRICAL AND ELECTRONIC EQUIPMENT ........................................................................................ 27 4.4 CONSTRUCTION AND DEMOLITION WASTE ..................................................................................................... 30 4.4.1 Buildings ......................................................................................................................................................................... 30 4.4.2 Wind turbines................................................................................................................................................................ 32 4.5 SLAGS AND ASHES ................................................................................................................................................ 35 4.6 MINING WASTE ....................................................................................................................................................... 37 4.7 DATASET SCREENING AND HARMONIZATION .................................................................................................. 40 5 COMPOSITION DATA CONSOLIDATION.............................................................................................. 41 5.1 GENERAL APPROACH ............................................................................................................................................ 41 5.1.1 Specifications for the product-centric approach ......................................................................................... 41 5.1.2 Specifications for the deposit-centric approach ......................................................................................... 45 6 WASTE STREAM-SPECIFIC COMPOSITION DATA CONSOLIDATION ...................................... 48 6.1 WASTE BATTERIES ................................................................................................................................................ 48 6.2 END-OF-LIFE VEHICLES ......................................................................................................................................... 51 6.3 WASTE ELECTRICAL AND ELECTRONIC EQUIPMENT ........................................................................................ 52 6.4 CONSTRUCTION AND DEMOLITION WASTE ...................................................................................................... 57 6.4.1 Buildings .......................................................................................................................................................................... 57 6.4.2 Wind turbines................................................................................................................................................................ 58 6.5 OVERVIEW OF RESULTS OF THE PRODUCT-CENTRIC APPROACH ................................................................ 62 6.6 SLAGS AND ASHES ................................................................................................................................................ 62 6.7 MINING WASTE ...................................................................................................................................................... 65 7 FUTURE COMPOSITION AND PRODUCTS ......................................................................................... 66 7.1 WASTE BATTERIES ................................................................................................................................................ 66
Methodology paper – Consolidated composition datasets | www.futuram.eu | 5 7.2 END-OF-LIFE VEHICLES ....................................................................................................................................... 66 7.3 WASTE ELECTRICAL AND ELECTRONIC EQUIPMENT ....................................................................................... 66 7.4 CONSTRUCTION AND DEMOLITION WASTE ...................................................................................................... 67 7.5 SLAGS AND ASHES ................................................................................................................................................ 67 7.6 MINING WASTE ...................................................................................................................................................... 68 8 (EXTERNAL) COMPOSITION DATA VALIDATION ............................................................................. 69 8.1 WASTE BATTERIES ................................................................................................................................................ 69 8.2 END-OF-LIFE VEHICLES ....................................................................................................................................... 69 8.3 WASTE ELECTRICAL AND ELECTRONIC EQUIPMENT ....................................................................................... 69 8.4 CONSTRUCTION AND DEMOLITION WASTE ..................................................................................................... 70 8.5 SLAGS AND ASHES ............................................................................................................................................... 70 8.6 MINING WASTE ...................................................................................................................................................... 70 8.7 VARIABILITY AND UNCERTAINTY ........................................................................................................................ 70 9 RECOMMENDATIONS FOR COMPOSITION DATA COLLECTION AND CONSOLIDATION 72 10 CONCLUSIONS .............................................................................................................................................. 74 11 LITERATURE ................................................................................................................................................... 75 ANNEX........................................................................................................................................................................77 ANNEX 1 - GLOSSARY ......................................................................................................................................... 78
Methodology paper – Consolidated composition datasets | www.futuram.eu | 6 Tables Table 1: Overview of parameter codes and required and optional information .................................... 14 Table 2: FAIR principles and implementation in FutuRaM ........................................................................... 17 Table 3: Data quality dimensions ......................................................................................................................... 20 Table 4: Data quality evaluation battery composition ................................................................................. 22 Table 5: Data quality evaluation vehicle composition .................................................................................. 25 Table 6: Data quality evaluation WEEE composition .................................................................................... 29 Table 7: Data quality evaluation building composition ................................................................................. 31 Table 8: Data quality evaluation wind turbine composition ....................................................................... 33 Table 9: LoW keys included in the slags and ashes scope (additional codes in the FutuRaM project in italic) .......................................................................................................................................................................... 35 Table 10: Data quality evaluation slags and ashes composition ............................................................... 36 Table 11: Level of detail consolidated results BATT ........................................................................................ 49 Table 12: Level of detail consolidated results end-of-life vehicles ............................................................ 51 Table 13: Level of detail consolidated results WEEE ..................................................................................... 55 Table 14: Level of detail for the consolidated results of buildings ............................................................ 58 Table 15: Product key level for consolidation of different components and materials in wind turbines ........................................................................................................................................................................ 60 Table 16: Level of detail consolidated results for wind turbines ................................................................. 61 Table 17: Overview of results of the product-centric approach ................................................................. 62 Table 18: Level of detail consolidated results slags and ashes ................................................................... 63
Methodology paper – Consolidated composition datasets | www.futuram.eu | 7 Figures Figure 1: Processes, procedures, and usage of composition datasets in FutuRaM ................................ 11 Figure 2: Structure of composition data under (a) product-centric approach (applied to waste batteries, end-of-life vehicles, WEEE, and CDW from buildings and wind turbines), and (b) deposit-centric approach (applied to slags and ashes and mining waste) .............................................. 13 Figure 3: Entity-relationship-diagram showing all entities and attributes of the composition data model in the product-centric approach .............................................................................................................. 15 Figure 4: Entity-relationship-diagram showing all entities and attributes of the composition data model in the deposit-centric approach ............................................................................................................... 16 Figure 5: Data (light green) and metadata (dark blue) types of the composition data collection templates....................................................................................................................................................................... 18 Figure 6: Overview of the levels in product, component, and material keys .......................................... 19 Figure 7: Concept of “Subdivision" vs. "subcomponent”............................................................................... 19 Figure 8: Transferring ProSUM composition data structure to FutuRaM composition data structure ........................................................................................................................................................................ 28 Figure 9: Consolidation concept for components in wind turbines .......................................................... 34 Figure 10: Data structure of MINERALS4EU database.................................................................................. 38 Figure 11: Overview mining waste data in MINERALS4EU database ........................................................ 39 Figure 12: Overview of consolidation method for the product-centric approach ................................ 42 Figure 13: Required sum-up before median/mean determination............................................................44 Figure 14: Overview of consolidation method for deposit-centric approach ........................................ 46
Methodology paper – Consolidated composition datasets | www.futuram.eu | 8 Executive Summary The aim of the FutuRaM project is to develop a Secondary Raw Materials Knowledge Base – the Urban Mine Platform – on the availability and recoverability of secondary raw materials within the European Union (EU27+4), with a special focus on critical raw materials. The project focuses on six different waste streams, namely waste batteries, end-of-life vehicles, waste electrical and electronic equipment, construction and demolition waste from buildings and wind turbines, slags and ashes, and mining waste. All these waste streams represent an important source of critical and strategic raw materials. Stock and flow, as well as recovery models, are developed within the project to estimate the availability of secondary raw materials from 2010-2050 for three different scenarios. For these models, composition information of products placed on the market and waste stocks and flows is required as input data. The composition data provides information on the structure of products and on mass fractions and content of elements, materials and components in products and waste stocks and flows. This data is crucial to understanding how many, and which strategic and critical raw materials are present in the waste streams, where they are located, e.g. in which components or materials, and how they can be recovered. Therefore, composition data is the basis for recoverability and secondary raw material assessments. The aim of milestone 23 of the FutuRaM project is to provide consolidated composition datasets for all waste streams that can be used for subsequent stock and flow and recovery modelling as well as for recoverability and secondary raw material assessment. Therefore, harmonized and complete composition datasets are needed. This means that (a) the composition datasets need to include all relevant products (or product groups) and waste stocks and flows that will be modelled and analyzed in the waste streams, (b) the composition datasets are provided in a fixed harmonized structure, and (c) the composition datasets have the necessary level of detail required for recoverability assessment. The FutuRaM composition data structure distinguishes between hierarchical layers within a waste stock or flow. Each layer in the hierarchy is composed of the entities of the previous one. This hierarchical structure enables a systematic and scalable description of composition across diverse waste streams. To better reflect the varying characteristics of the waste streams, the composition data model employs two distinct approaches. The product-centric approach, where the waste stream can be described as stocks or flows of products, which are composed of components, and subsequent materials and elements, was applied to waste batteries, waste electrical and electronic equipment, end-of-life vehicles, and construction and demolition waste from buildings and wind turbines. The deposit-centric approach, where the waste stream can be described as different stocks and flows which are composed of minerals or compounds and subsequent elements, was applied to slags and ashes and mining waste. Based on this structure, composition data was collected using specifically developed templates enabling the collection of data according to the composition data model from various sources including scientific publications, reports, databases, manufacturer information sheets such as environmental product declarations (EPD), digital product passports (DPP) or bills of materials (BOM), and results of dedicated sampling campaigns. For the consolidation of the collected raw data, a harmonized method was developed in order to transform the scattered raw data into consistent and generic datasets with the average composition
Methodology paper – Consolidated composition datasets | www.futuram.eu | 9 of relevant products (or product groups) and waste stocks and flows. Data points with different data structures and levels of detail can be combined into generic compositions for products and waste stocks and flows by transforming them into the desired data structure and defining the level of detail that can be reached with all data points. The composition datasets generated for waste streams such as waste batteries, end-of-life vehicles, WEEE and CDW from buildings and wind turbines, allow the mapping of respective components, materials, and elements within the product. During the consolidation process, multiple data gaps were encountered that needed to be filled with assumptions or required consolidation on a higher level of aggregation. These data gaps included missing temporal and spatial granularity as well as missing links for a complete composition description resulting in skipped layers or layers without sufficient information. For mining waste and slags and ashes, waste composition was determined using available sitespecific data from geological surveys complemented by information from relevant literature sources. Composition data is frequently reported on the element or mineral/compound layer. Due to limited and non-harmonized or -standardized reporting, presentation, and generation of composition data, it was not possible to include the mineral/compound layer in the consolidated dataset and therefore the dataset only includes the elemental content of different slags and ashes or mining wastes without considering the mineralogical state or associations in detail. This methodology paper provides an overview of the composition data collection for each waste stream. The consolidation method is described and specifications for each waste stream illustrated. Challenges during the process are highlighted and recommendations for further improvement of the processes are provided. Furthermore, in this methodology paper, the considerations regarding future composition and composition data validation are presented.
Methodology paper – Consolidated composition datasets | www.futuram.eu | 16 Figure 4: Entity-relationship-diagram showing all entities and attributes of the composition data model in the deposit-centric approach 3.2 Composition data collection templates For both waste types described in chapter 3.1, a composition data collection template was developed based on the entity-relationship-diagrams shown in Figure 3 and Figure 4 (see annex 2). The entityrelationship-diagrams are the result of a revision of the composition data collection template from the ProSUM project. Whereas waste batteries, end-of-life vehicles and WEEE can be clearly allocated to the productcentric approach and slags and ashes as well as mining waste to the deposit-centric approach, the CDW stream can be allocated to both types of waste streams. A building can be considered as product. However, CDW is generated during construction, renovation, and demolition processes and is not in the form of discarded products. Due to the modelling approach where CDW is modelled for buildings and wind turbines as output of lifetime models, building and wind turbine composition was required and composition data in the form of material intensities was collected with the template for the product-centric approach.
Methodology paper – Consolidated composition datasets | www.futuram.eu | 17 The templates are designed in the long table format, i.e. one row per data point, which facilitates automated data collection and further processing with Python, R, etc.Fehler! Verweisquelle konnte nicht gefunden werden.. The composition data collection templates enabled the collection of data from various sources including scientific publications, reports, databases, producer information sheets such as environmental product declarations (EPD), digital product passports (DPP) or bills of materials (BOM), and results of dedicated sampling campaigns. All references used were collected in a bibliography library (Mendeley). A list of all references can be found in annex 5. The data collection essentially follows FAIR principles to ensure FAIR datasets. The principles and their implementation for the composition data collection and the resulting consolidated composition datasets are described in Table 2. Some data was provided under non-disclosure agreements (NDA) and therefore does not comply with all FAIR principles. Table 2: FAIR principles and implementation in FutuRaM Principle Implementation in FutuRaM Findable • Descriptive metadata • Either a globally unique, citable and persistent identifier (DOI) or a web address (URL) is associated with the data, e.g. in open access repositories (zenodo) and UMP Accessible • Accessible data(sets) and metadata • Accessible and comprehensive protocols/templates Interoperable • The data is available in an open, standardized, structured, and machine-readable format • Harmonized code lists (existing, official, and newly developed code lists/classification systems) are used Reusable • Data licenses and usage rights (e.g. open access, CC BY 4.0) 3.2.1 Data and metadata types An overview of the included data and metadata categories in the composition data collection templates is provided in Figure 5. Detailed information of the data and metadata types can be found in the templates in annex 2. The core information is the value and the parameter code which are included in the category “parameter”. Here, the value itself and its relation to the structure of composition data indicated. How the value was derived is described with the metadata in “value generation”.
Methodology paper – Consolidated composition datasets | www.futuram.eu | 18 Figure 5: Data (light green) and metadata (dark blue) types of the composition data collection templates 3.2.2 Harmonized code lists To have comparable and interoperable results for all waste streams and enable processing of the composition data in the waste stream models, harmonized code lists were created to fill the composition data collection templates, see code list repository in annex 3. A set of code lists called product key, component key and material key were developed to describe the different products, components and materials in the waste streams. Based on the ProSUM approach, product and component keys were created for all waste streams applying the productcentric approach. The product and component keys are hierarchical classifications to describe the products in a waste stream and the components in these products. The classification starts at a general level (upper level) gaining more detail in each following (sub) level (lower level). Each code in a sub level is linked to a code in the upper level. This enables the collection of data with different levels of detail. Accordingly, a material key was created overarching for the product-centric approach. An overview of the levels in the product, component, and material key is shown in Figure 6.
Methodology paper – Consolidated composition datasets | www.futuram.eu | 19 Figure 6: Overview of the levels in product, component, and material keys Besides the product keys, all component keys are waste stream-specific, i.e. even if the same component exists in more than one waste stream, they do not have the same code, as their composition can vary heavily between Waste streams (e.g. a door of a car is different than a door of a house or of a fridge). Even for components which have a composition that is not exactly waste stream-dependent (e.g. fan, permanent magnet, etc.), it was decided to keep the differentiation to be consistent. The codes in material key on the other hand are waste stream-unspecific, i.e. all Waste streams use the same codes. However, especially on the more general (upper) levels in the material key, the elemental material composition can be waste stream-specific and even component-specific (e.g. for ferrous metals) resulting in the need to provide all links in the composition data structure, i.e. for an e-m (element in material) composition such as iron (Fe) in ferrous metals, the link to the component it is found in needs to be provided. The material key follows the principle of “subdivision”, i.e. the lower levels are more detailed categories of the upper level (e.g. metals are detailed into different types like aluminum, iron, copper, etc. and then into different alloy types and so on). For the components however, in most waste streams the “subcomponent” principle was used, i.e. there are main components which are then divided into subcomponents and so on, compare Figure 7. The components often represent a component group instead of a single component, e.g. all PCBs in a product. For the deposit-centric approach, product, component and material keys are not needed. Figure 7: Concept of “Subdivision" vs. "subcomponent”
Methodology paper – Consolidated composition datasets | www.futuram.eu | 20 3.2.3 Data quality assessment The determination of data quality included six dimensions: accuracy, validity, consistency, timeliness, completeness, and integrity. Accuracy and validity were evaluated for each data point, timeliness, consistency and completeness for the whole dataset or a section like e.g. a product group within the dataset. Integrity was calculated as the mean of accuracy, completeness, and consistency to give these dimensions more weight. Table 3 includes an overview of the data quality dimensions. Each dimension was ranked with a data quality score (DQS) between 1 (very reliable) and 4 (dubious: no information or no reliable information concerning the value of property). Afterwards the mean of all dimensions was calculated and subsequently the overall DQS determined as follows: • 2.3 ≤ mean value → DQS 3 • 1.3 ≤ mean value < 2.3 → DQS 2 • mean value < 1.3 → DQS 1 Chapters 4.1 to 4.6 include further explanations on the data quality assessment and scoring concept in each waste stream. Table 3: Data quality dimensions Dimension Description Validity Degree data is within defined requirements Accuracy Degree data represents the reality Timeliness Degree data is available at the time needed Consistency Degree of data being equal within and between datasets Completeness Degree necessary data is available for use Integrity Overall accuracy, completeness, and consistency of data
Methodology paper – Consolidated composition datasets | www.futuram.eu | 21 4 Waste stream-specific composition data collection 4.1 Waste batteries Scope The scope of the battery waste stream includes all battery types and chemistries that are currently available on the market or are expected to be available until 2050. Battery types and chemistries that are only in early stages of development are not included. The waste stream is divided into batteries in general for different applications, such as light means of transport or portable devices, and electric vehicle batteries as they differ significantly in size and capacity. Product and component key The battery key was developed based on the different electrochemical systems of batteries. The highest level includes the different battery types (lithium ion primary and rechargeable batteries, natrium ion rechargeable batteries, zinc batteries, lead batteries, NiMH, and NiCd) which are then further divided according to their chemistries (suband sub-subtypes). The component key includes the main components of the battery cell namely casing, anode, cathode, separator, binder, and electrolyte. Some of the main components are then further divided into subcomponents. The casing is not a direct part of the battery cell, but it is often mentioned within the different sources and therefore part of the component key. The component key for electric vehicle (EV) batteries also includes the battery pack housing. Tree diagrams of the battery product and component key can be found in annex 4. Data sources For the batteries in general, the ProSUM raw data was utilized as the starting point for data collection and was reviewed in terms of consistency. The already existing dataset was transferred to the composition data collection template and extended by conducting extensive literature research. For this purpose, the following sources were screened: • Published data from the industry including data from recyclers • Books about batteries • Research papers and scientific publications • Battery material safety data sheets published by battery manufacturers In total, 156 data sources with composition information were found. Within these sources are 6 books, 73 research papers, 64 battery material safety data sheets, 10 other sources like reports, internet information, product information sheets or theses and 3 software models. From these sources, composition data was systematically extracted in three steps. Firstly, the addressed battery chemistry was derived from the source and the respective product key code assigned. Secondly, the composition information was allocated within the data composition model. Thirdly, the value and the unit of measurement were extracted as well as how the data was obtained. The composition data is given in wt% of the battery cell in most sources. In case of need, the value was converted to wt%.
Methodology paper – Consolidated composition datasets | www.futuram.eu | 22 For EV batteries, a different approach was used. The composition is solely based on the BatPaC model from the “Argonne National Laboratory” (Knehr et al., 2022) to guarantee the desired level of detail. In the first step, the model results were analyzed using standard settings for seven different capacities (30kWh, 60kWh, 90kWh, 120kWh, 150kWh, 180kWh, 210kWh) to depict a broad range of potential battery packs. For each battery chemistry (product sub-sub-key), the elements, materials and components were then extracted from the model. Element values in the electrodes remain constant with changing battery pack capacity and therefore an average value was generated. For the other elements, materials, and components, a single value was extracted for each battery pack capacity and product sub-sub-key, which allowed to conduct a polynomial regression to create a new model enabling the extraction of composition data for the desired capacities. In alignment with the end-oflife vehicles waste stream, the composition data for seven different capacities was extracted from the newly created model (25 kWh, 45 kWh, 60 kWh, 80 kWh, and 100 kWh for BEV, 1 kWh for HEV, and 20 kWh for PHEV). The composition data is given in kg/kWh. Data quality assessment Data quality of the references in the two datasets was evaluated as indicated in Table 4. Table 4: Data quality evaluation battery composition Dimension DQS Description Validity 1 2 3 4 Bill of materials (data directly comes from manufacturers and needs to be validated in a proper way) Values measured by own investigations and measurements; Values generated by expert opinions or manufacturer information. More than one source used for data generation (average of more than one source); Own investigations, but not further explained. No information on how the value is measured or where the data comes from (no primary source). Accuracy 1 2 3 4 Values are given in a range of wt% per battery cell or kg/kWh for EV battery packs Value with one decimal place; value with more than one decimal place but given uncertainty; range of values. Value with more than one decimal place (too accurate) without uncertainty. Only single value without decimal places and without range or uncertainties. Timeliness (for the whole dataset) As battery chemistry and composition do not change over time, the timeliness of the dataset was ranked with DQS 1.
Methodology paper – Consolidated composition datasets | www.futuram.eu | 23 Dimension DQS Description Consistency (per product sub-sub-key) 1 2 3 4 Mean of the relative deviations between minimum and maximum (max-min/max) per product sub-sub-key < 25% 26% ≤ mean of the relative deviations between minimum and maximum per product sub-sub-key < 50% 51% ≤ mean of the relative deviations between minimum and maximum per product sub-sub-key < 75% Mean of the relative deviations between minimum and maximum per product sub-sub-key ≥ 75% Completeness (for the whole dataset) For the general batteries dataset, there is sufficient data for 29 out of 46 product sub-sub-keys. Data for 8 additional sub-sub-keys could be derived from similar battery chemistries. From the left 8 sub-sub-keys, 3 will not have a significant market share until 2050, 1 is not needed, and for 3 there is no data available. Therefore, the dataset was ranked with DQS 2. The electric vehicle batteries dataset was also ranked with DQS 2. The used BatPaC5.1 model provides detailed composition data for a range of battery chemistries. However, newer battery chemistries are missing and had to be derived from the general batteries dataset and adjusted to the electric vehicle batteries dataset. Raw dataset Out of the 156 identified sources containing information on battery composition data, 117 were included in the dataset. The remaining sources were excluded either because they cited already included references or provided incorrect units of measurement. The number of sources per battery chemistry and product sub-sub-key varies significantly, with the lowest number of references being 1 and the highest 38. The DQS of all references per product sub-sub-keys is either 1 or 2. Only a few references with DQS 3 are included. This DQS is usually a result of a low data quality score in the dimensions accuracy or validity. Certain data points from the raw dataset were excluded from consolidation as they would negatively affect the consolidation results: • Data points for combined components or elements • Data points where it is known that they reflect contamination • Data points with no information on product key levels 2 and 3 • Data points with no information on component key level 2 (cathode and anode without differentiation in active material and current collector) The electric vehicle batteries dataset is, as mentioned before, only based on a single reference (BatPaC5.1) which has an overall DQS of 2. This reference does not include all considered battery chemistries in EV (lithium ion and sodium ion rechargeable batteries). Therefore, the compositions of the cathode active material, anode active material and electrolyte from the missing ones (sodium ion batteries and Lithium-Iron-Manganese-Phosphate (LFMP) batteries) were taken from the general
Methodology paper – Consolidated composition datasets | www.futuram.eu | 24 batteries dataset and adjusted to the electric vehicle batteries dataset using their respective energy densities of 230 Wh/kg (LFMP) and 145 Wh/kg (SIBs). Many of the data points collected in both datasets are not in the required composition data structure format (e-m, m-c and c-p). In many cases, the references report elements next to materials to describe the composition (e.g. cathode as elements, but electrolyte as material) making it difficult to distinguish between a material and an element layer. It was therefore decided to avoid the material layer as much as possible in the battery composition and aim for an e-c and c-p composition. Many materials are directly reflected as components in the component layer (e.g. cathode material) and the allocation of elements to specific materials is not possible in most cases. On the other hand, it is possible to transform many of the materials into their elemental composition when the specific chemical formulas are known. To enable data transformation in the consolidation process, for parameter codes m-p and e-p, materials and elements were allocated in components with knowledge on battery chemistry. 4.2 End-of-life vehicles Scope 1 The end-of-life vehicles scope includes all types of passenger cars (up to 3.5 tons) except types which are very uncommon. This includes for example sportsand hypercars, custom-made luxury cars, and fuel-cell hydrogen powered cars. Product and component key For the vehicle key, cars were firstly grouped according to their fuel-type/drive train (petrol, diesel, battery electric, hybrid electric, plug-in hybrid electric and other vehicles) and then according to the market segments (ABCDEF) of models which correlate with length/weight of the vehicle. In addition, there is a separation between Sport Utility Vehicles (SUVs) 2 and non-SUVs for passenger vehicles as they are getting more common especially for battery electric vehicles and between cars and vans. The highest level of the component key differentiates into component groups which are recycled together, the following level includes overarching component groups, and the lowest level detailed components. Tree diagrams of the vehicle product and component key can be found in annex 4. Data sources The data collection builds on existing data from previous projects with the addition of systematic literature review and desktop research. The composition for vehicles from the time period between 1980 and 2010 is based on ProSUM data and their continuation/ update for a JRC repository (European Commission, 2017).1 1 The composition data collected for WP3 was not applied in the stock and flow and recovery modelling (WP4) of end-oflife vehicles. Here, the composition data from the ProSUM project and its continuation/update for a JRC repository (European Commission, 2017) was used. 2 An SUV (Sport Utility Vehicle) is a spacious, versatile car with higher ground clearance, often featuring all-wheel or fourwheel drive, making it ideal for both urban and off-road travel.
Methodology paper – Consolidated composition datasets | www.futuram.eu | 25 For the years after 2010, the data collection approach is based on two steps: a) deriving representative and complete sets of products, components, and materials used in the vehicle industry from industrywide reports and b) collect composition data for all e-m-c-p relations to describe the complete composition data for the whole set of products, components, and materials. The data collected is based on numerous industrywide reports, which were compiled by consulting companies that monitor the past, current and future development of the vehicle sector with all their specific aspects. The product key was derived from and is linked to the data being provided by these reports of different consulting companies that selected the representative cars of each segment and performed data consolidation. The industrywide reports are focusing besides the product-specific overall composition on component groups and component-specific information. The granularity of the component groups and components are mainly based on interesting material compositions. Data on components was obtained from the previously mentioned reports, product-specific information, which was found on different websites of the vehicle industry and their suppliers, and from different own studies. Furthermore, information on component groups and components is used to verify the product-specific overall composition and to detail the material composition. The material composition provided in the mentioned sources is on a general material group level which focuses on production methods (e.g. sheet, extruded aluminum) or on specific material properties, which are essential for the physical properties of the specific components (e.g. ultra-high strength steel). They do not go into further details concerning the elemental material composition. For this, international standards describing the elemental material composition and numerous articles, industry blogs etc. specifying commonly used standard materials in the automotive sector were gathered. Based on the elemental material composition from international standards and the vehicle sector specific selection of certain alloys, an overall elemental material composition was calculated as a (weighted) mean value with an R-script. The derived material list does not include newly developed material compositions, as they are not published by the industry and are rarely investigated by independent analytical laboratories. Therefore, the most recent developments are not included. Additionally, only materials containing CRMs are included individually in this material list and only for them elemental composition was collected in detail. Materials such as fabrics, wood or plastics are aggregated in a “rest” category. Data quality assessment Data quality of the references in the dataset was evaluated as indicated in Table 5. Table 5: Data quality evaluation vehicle composition Dimension DQS Evaluation Validity 1 2 3 4 Numbers directly from a report of company, which assesses the composition and weight of a standard vehicle. They looked at around 50 different cars. Therefore, these data are already consolidated. Derived from a standard car, which include some assumption. Estimates based on different information on the internet without direct information Does not make sense to define for this dimension
Methodology paper – Consolidated composition datasets | www.futuram.eu | 32 The material intensities in the database are often given for construction time spans. These time spans needed to be separated into data points for single years. This was done by replicating the entries for all years included in the time spans. A few raw data points were excluded from the consolidation as they are not relevant for the consolidation results and specific to certain references (MI for asphalt, linoleum, heraklith, carpet, clay, and straw) or would cause double counting (aggregates and cement; instead concrete was used). The elemental material composition (parameter code e-m) is only included for metals in the dataset. It was calculated independently and did not follow the consolidation procedure due to the way of its creation. It is based on a report from the German Environment Protection Agency (Raatz et al., 2022) which provides an overview of the different metal alloys or alloy groups used and their shares within different industry sectors with one of them being the construction sector. Partly, the report also provides elemental composition of the alloys or alloy groups. Missing information was added with data from producer information and websites. Based on the shares of different alloys or alloy groups in the construction sector and their elemental composition, average metal compositions (ferrous, aluminum and copper) were calculated for buildings. 4.4.2 Wind turbines Scope The scope covers the main wind turbine types for onshore and offshore applications which include two main technical designs – direct drive and gearbox. These two designs can be further divided into subtypes and have significantly different constructions, differing in generator design and drivetrain system (Carrara et al., 2020). Product and component key The wind turbine product key was developed based on an approach outlined by (Carrara et al., 2020) and the SURFER project reports (Laurent et al., 2019). Both are using nine different wind turbine types. The nine types are reduced to eight in the FutuRaM project by putting together SCIG fullconverter and SCIG without full-converter under the name SCIG. However, HTS wind turbine type is not considered in FutuRaM due to its low relevance. These eight, respectively seven, keys resemble the lowest product key level together with an additional unspecified category for parks and turbines with gearboxes. Subsequently, they are aggregated to more general groups in the upper levels. There are two different strings, one for single wind turbines and one for wind parks. The wind turbine component key is based on Vestas LCA’s components (Vestas Wind Systems A/S, n.d.), considering only the most common ones that can be also found in other LCAs. As wind turbines have over 10.000 components, which vary between model types and manufacturers, the components used in the FutuRaM project are generic components representing functional units of wind turbines. Tree diagrams of the wind turbines product and component key can be found in annex 4.
Methodology paper – Consolidated composition datasets | www.futuram.eu | 33 Data sources The collected data includes all references collected for the SURFER project (Laurent et al., 2019) and from Li et al 2022 (Li et al., 2022). This article only focuses on metals, so the data was completed with further sources to also include other materials, like plastics or glass fiber reinforced plastics. Additionally, different LCAs from Vestas were incorporated in the dataset. Data quality assessment Data quality of the references in the dataset was evaluated as indicated in Table 8. Table 8: Data quality evaluation wind turbine composition Dimension DQS Evaluation Accuracy 1 2 3 4 All compulsory fields + other fields are filled, and the product key is not wind turbine/park unspecified. All compulsory fields are filled, and the product key is not wind turbine/park unspecified. All compulsory fields are filled, and the product key is wind turbine/park unspecified. Not all compulsory fields are filled (not used) Validity 1 2 3 4 Measurements from a constructor. Values were calculated or estimated by an expert. Other value generation. Unknown value generation. Consistency (for the whole dataset) As all the value generation methods differed from each other, the dataset was ranked with DQS 3, except the Vestas LCA references which were ranked with DQS 2 as the same method was applied in all cases. Timeliness (for the whole dataset) Most of the wind turbines/parks that will be decommissioned before 2050 are already existing. Therefore, the composition data is representative also for the future and was ranked with DQS 2. Completeness (for the whole dataset) There is data for almost all product sub-sub-keys. Therefore, the dataset was ranked with DQS 2. Raw dataset The wind turbine dataset is based on 72 references. The DQS for the references ranges from 2 to 3 which is a result of the difficulty and non-harmonized methods to generate material intensities. The material intensities from all references were harmonized to the unit of measurement kg/MW as far as possible.
Methodology paper – Consolidated composition datasets | www.futuram.eu | 34 Only a limited number of references include information on components. Therefore, most elements and materials could not be allocated in components and it was decided to only distinguish three components (permanent magnet in the generator, foundations and external cables) and combine all the other components with, respectively in, the shadow component, resembling the turbine itself (see Figure 9). The elemental material composition (parameter code e-m) is only included for the permanent magnet and metals in the dataset. If sufficient information was included in the dataset, the e-m composition of the respective material was determined during the consolidated process. Otherwise, the approach described for buildings in chapter 4.4.1 was applied. Certain raw data points were excluded from the consolidation as they would negatively affect the consolidation results: • Data points with the unit of measurement wt% as they cannot be consolidated together with data points with the unit of measurement kg/MW • Certain components like switch gear, transmission insulation, conductor, transformer and internal wiring as only a few references mentioned these components • Certain materials which are not relevant (like gas, oil, lubricants, etc.) or would lead to double counting (like cement and aggregates if also concrete is considered) Figure 9: Consolidation concept for components in wind turbines
Methodology paper – Consolidated composition datasets | www.futuram.eu | 35 4.5 Slags and ashes Scope The scope includes slags and sludges from metal industries as well as ashes from energy generation and waste incineration processes. The waste stocks and flows are described by their List of Waste (LoW) key (European Commission, 2014). Some of the LoW keys are further differentiated according to the input material in energy generation or the furnace type in steel production. An overview of the included LoW keys is given in Table 9. Table 9: LoW keys included in the slags and ashes scope (additional codes in the FutuRaM project in italic) LoW key Description 01 03 Wastes from physical and chemical processing of metalliferous minerals 01 03 09 Red mud from alumina production other than the wastes mentioned in 01 03 10 10 01 Wastes from power stations and other combustion plants (except 19) 10 01 01_B Bottom ash from bioenergy (pure wood biomass combustion) 10 01 01_C Coal bottom ash 10 01 02 Coal fly ash 10 01 03 Fly ash from peat and untreated wood 10 01 15_A Bottom ash from incineration of agricultural biomass 10 01 15_B Bottom ash from incineration of other biowaste 10 01 15_P Bottom ash from incineration of paper sludge 10 01 16* Fly ash from co-incineration containing hazardous substances 10 01 17_A Fly ash from incineration of agricultural biomass 10 01 17_B Fly ash from incineration of other biowaste 10 01 17_P Fly ash from incineration of paper sludge 10 02 Wastes from the iron and steel industry 10 02 02 Unprocessed slag 10 02 02_AOD Unprocessed slag, from argon oxygen decarburization 10 02 02_BF Unprocessed slag, from blast furnace, including Gilchrist-Thomas process 10 02 02_BOF Unprocessed slag, from basic oxygene furnace, included LD converter 10 02 02_EAF Unprocessed slag, from electric arc furnace 10 02 02_Ladle Unprocessed slag, from ladle 10 04 Wastes from lead thermal metallurgy 10 04 01* Slags from primary and secondary production 10 05 Wastes from zinc thermal metallurgy 10 05 01 Slags from primary and secondary production 10 06 Wastes from copper thermal metallurgy 10 06 01 Slags from primary and secondary production 10 08 Wastes from other non-ferrous thermal metallurgy 10 08 08* Salt slag from primary and secondary production 11 02 Wastes from non-ferrous hydrometallurgical processes 11 02 02* Sludges from zinc hydrometallurgy (including jarosite, goethite) 19 01 Wastes from incineration or pyrolysis of waste 19 01 11* Bottom ash and slag containing hazardous substances 19 01 12 Bottom ash and slag other than those mentioned in 19 01 11 19 01 13* Fly ash containing hazardous substances 19 01 14 Fly ash other than those mentioned in 19 01 13
Methodology paper – Consolidated composition datasets | www.futuram.eu | 36 Data sources A search of the English language peer-reviewed literature was conducted initially using protocoldriven search strategies with the electronic databases Scopus, Web of Science, and Google Scholar, also using a “snowballing” approach (Wohlin, 2014). A series of keywords were used as search strings, and these keywords were also entered as search strings into the Google search engine, to seek nonacademic data about the wastes of interest, e.g., from government and industrial research reports. Data quality assessment Data quality of the references in the dataset was evaluated as indicated in Table 10. Table 10: Data quality evaluation slags and ashes composition Dimension DQS Evaluation Accuracy 1 2 3 4 Based on value generation variable: Value was measured. Value was estimated based on experts or calculated. Other value generation approaches. Dubious data which have no information about value generation. Validity 1 2 3 4 Based on three factors (CRM/SRM, EU, mandatory cells): More data than required for the three factors Data only for the three factors Data for mandatory cells and CRM/SRM, but data is not from EU Data only for mandatory cells and EU, but not about CRM/SRM elements Consistency (per LoW key) 1 2 3 4 Same kind of sample, same method, and accuracy value =1 Different sample, same method, and accuracy value = 1 or 2 Different sample, different method, and accuracy value = 1 or 2 Different sample, different method, and accuracy value > 2 Timeliness (per LoW key) 1 2 3 4 Estimation, present and past values. Present and past values. More than 6 different years. Present and past values. More than 3 different years including some of the 5 last years. Present and past values. Less than 3 different years including some of the 5 last years. Completeness (per LoW key) 1 2 3 4 More than 10 different references. Less than 10 different references or more than 50 different samples. Between 10 and 5 different references or more than 25 different samples. Less than 5 different references or less than 25 different samples.
Methodology paper – Consolidated composition datasets | www.futuram.eu | 37 Raw dataset The slags and ashes dataset includes 185 data sources for composition data. Most of these data sources were evaluated with DQS 2, a few of them with DQS 3 and only one with DQS 1. Especially the dimensions validity, timeliness and completeness are responsible for low DQSs. Composition data is usually reported as elements in a stock or flow. If compounds are given, this only includes the oxide form of these elements. Therefore, it was decided to skip the compound layer and generate an e-f/s composition. All oxide data points with the parameter code com-f/s were transformed e-f/s compositions, i.e., oxide compositions were converted into elemental compositions. Multiple units of measurements are used in the references and need to be harmonized. In the dataset, basically the whole elemental system can be found. However, the mass fractions of certain elements are neglectable low in some cases. Certain entries for municipal solid waste as well as for certain slags were exclude from consolidation as they were not within the boundaries of the FutuRaM scope. 4.6 Mining waste For mining waste, the composition data collection template was not used to collect composition data. The composition data is instead taken directly from the MINERALS4EU database. The MINERALS4EU database (EGDI, 2022) builds up on the data structure shown in Figure 10 and was developed in previous Horizon 2020 projects. It is a reference database for mineral occurrences and deposits in Europe including mining wastes. In the ProSUM project, the data structure was further developed for mining waste. The MINERALS4EU database contains information at site level and covers at the same time composition data and volumes of waste. During the FutuRaM project, this dataset was extended by inclusion of further sites.
Methodology paper – Consolidated composition datasets | www.futuram.eu | 38 Figure 10: Data structure of MINERALS4EU database The CRM Act identifies two main activities for mining waste: obligations for operators (current mining activities) and obligations for member states concerning historical or legacy sites. In the FutuRaM project, the main effort in data acquisition is done for historical sites. All member states in the EU have an obligation to create inventories of historical mining waste (Extractive Waste Directive (2006/21/EC)). As the main purpose of this directive was to identify the sites with the highest environmental risks, the data in these mining wastes inventories, however, is not always sufficient for the assessment of CRM recovery plans. Figure 11 shows the status of member states’ data on historical mining wastes in the MINERALS4EU database. Within the FutuRaM project and GSEU projects, several workshops were performed to acquire additional data from the member states including creation of a tool to enter the data into the now named EGDI (earlier MINERALS4EU) database. At the beginning of the FutuRaM project, only 8 EU member states had data in the MINERALS4EU database with only Portugal including composition data. Currently, 11 member states have added data on composition or data upload is in progress. For a few member states, data currently consists of only information on waste sites, sometimes including tonnage, but without composition data. For the remaining 8 member states, no mining waste data has been provided. For the member states without composition data, acquisition is ongoing. For the +4 countries outside of the EU, Norway has delivered site locations, and the generation of composition data is in progress. Composition data might be available at the end of the project. The UK has so far not delivered data and Switzerland and Iceland have no mining waste data of consequence.
Methodology paper – Consolidated composition datasets | www.futuram.eu | 39 Figure 11: Overview mining waste data in MINERALS4EU database 3 In blue color: countries with data in the MINERALS4EU database already at the beginning of FutuRaM project (8 member states) In yellow color: countries with data added within the FutuRaM project or data upload in progress (10 member states) In white color: countries with data acquisition still in progress (9 member states) With blue diamond: data includes composition data With x: no mining waste 3 The designations employed and the presentation of the material in this publication do not imply the expression of any opinion whatsoever on the part of FutuRaM consortium members concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries.
Methodology paper – Consolidated composition datasets | www.futuram.eu | 40 4.7 Dataset screening and harmonization Data screening and harmonization is a crucial step to have consistent datasets enabling further processing. It included the following steps: • Check for the existence of mandatory information • Check for empty cells/blanks • Check for required information related to parameter code • Check for consistency with codes in code lists • Check for consistency within product, component, and material key • Check for consistency in similarity rating • Removal of unnecessary space blanks • Formatting of numbers • Check for the possibility to allocate elements and/or materials to a specific material and/or component for e-p, e-c and m-p parameter codes • Check for the requirement for additional codes to better describe undefined and other materials • Harmonization of unit of measurement (UoM)
Methodology paper – Consolidated composition datasets | www.futuram.eu | 41 5 Composition data consolidation 5.1 General approach For the consolidation of the collected composition data, a harmonized method was developed for all waste streams. As the consolidation for the deposit-centric approach includes fewer steps than the one for the product-centric approach, they are described in different chapters. For all waste streams, the consolidation was executed multiple times due to gradual corrections and completions in the raw datasets. For the battery waste stream, the construction and demolition waste stream, and slags and ashes, the consolidation steps were carried out in Excel. For the WEEE stream and end-of-life vehicles, an R-script was used. 5.1.1 Specifications for the product-centric approach The consolidation for the product-centric approach includes five steps: 1. Data transformation 2. Inventories (c and m) 3. Median or mean determination 4. Data reconciliation and filling data gaps 5. Data compilation in portrayals as input file for waste stream models An overview of the method is shown in Figure 12. Each step is described in detail below.
Methodology paper – Consolidated composition datasets | www.futuram.eu | 48 6 Waste stream-specific composition data consolidation 6.1 Waste batteries The two different datasets, general batteries and electric vehicle batteries, were consolidated individually due to different component sets and units of measurement (wt% and kg/kWh). 1) Data transformation It was decided to aim for an e-c and c-p composition for the battery waste stream. The reasons are described in chapter 4.1. However, a few materials which cannot be transformed into elements are kept besides the elements. Therefore, the consolidated composition is an e/m-c and c-p composition depending on the transformation possibilities. For the components cathode, anode and electrolyte, for most battery chemistries an e-c composition was generated. Binder, separator and casing are always m-c compositions. The data transformation includes the following steps: • As much as possible allocation of materials and elements to specific components for data points with parameter code m-p and e-p. Material and elements that could not be allocated, were allocated in the shadow component. The issues related to this concept are described in chapter 5.1.1. • Transformation of m-p and e-p data points to m-c, e-c and c-p values by summing up all material and elements allocated in one component (c-p) and dividing the specific share in the whole product by the share of the component (value m-p or e-p / value c-p = value m-c or e-c). For the EV battery composition, which is in kg/kWh, the m-p or e-p composition can be directly used as m-c or e-c composition when the element or material is allocated in a component (m-p ≙ m-c and e-p ≙ e-c). The c-p values are also derived by summing up. • Transformation of materials to elements when the elemental composition was known for data points with the parameter code m-c. The transformed dataset includes parameter codes e-c, m-c (rest that could not be transformed to elements) and c-p. 2) Inventories (c and m) As data points without information on component key level 2 were excluded from the consolidation, component key level 2 could be reached for all components. It was decided to neglect the component “cell terminals” in the general batteries dataset as it was not present or mentioned separately in many battery types and therefore would require significant data gap filling. For the materials, detailed information is available for the cathode, anode, and oftentimes electrolyte material. The composition of casing, binder, and separator, however, is often on a less specific level.
Methodology paper – Consolidated composition datasets | www.futuram.eu | 49 3) Median determination For the median determination, product key level 3 was chosen, which is in line with the level used in the stock and flow modelling. The median determination was done for components in products (c-p) and elements or materials in components (e/m-c). The used unit of measurement is wt% for batteries in general and kg/kWh for EV batteries. No differentiation in time and location was considered. Battery chemistry does not change over time and with location, there are only new chemistry types entering the market. Therefore, the composition will also stay the same in the future. Data points for POM and waste generated were consolidated together. 4) Data reconciliation and filling data gaps Due to the various transformation steps and the variety of composition reporting in the references, extensive data reconciliation and validation needed to be carried out for the general batteries dataset. This was done with expert knowledge on battery chemistry and mainly included the removal of double/side-by-side compositions, e.g. of materials and elements in components due to the fact that not all materials could be transformed to elements (e.g. 100% Cu and 100% CuAndCuAlloys for the anode current collector) or the removal of the shadow component if compositions for all components were available or could be derived from similar chemistries (which was the case for all battery types). A few c-p and e/m-c data gaps needed to be filled. For the c-p data gaps, the average value for the respective c-p composition over all comparable battery types was used. For the e/m-c data gaps, the composition was indicated with 100% undefined materials or elements unless the specific material or element was known. The composition on a level was normalized to 100%. All results were validated by comparing them to materials and elements expected in the different components of each battery type and chemistry. The electric vehicle batteries dataset did not require data reconciliation as it is based on a single reference (compare chapter 4.1. However, the c-p values needed to be calculated from the e/m-c values as they are not given in the reference. It is expected that in the upcoming years the anode active material in certain NMC batteries will be partly replaced with silicon. Therefore, additional product codes for these batteries with a 10% and 20% share of silicon were added to the consolidated results in the general batteries dataset by dividing the e-c composition of the anode active material to carbon and silicon. For the EV batteries dataset, their composition could be derived from the used reference. 5) Data compilation in portrayals as input file for waste stream models The consolidated results reach the level of detail indicated in Table 11. There are two different versions for the battery waste stream, one with e-/m-c and c-p and one with all e-m-c-p relations by adding a “shadow material” where necessary. Table 11: Level of detail consolidated results BATT Product key level 3 Components Materials or elements battLiCFx_subsub battLiCO_subsub battLiFP_subsub General: anodeActiveMaterial anodeAndCathodeActiveMaterial additives AlAndAlAlloys CFx
Methodology paper – Consolidated composition datasets | www.futuram.eu | 50 Product key level 3 Components Materials or elements battLiMFP_subsub battLiMnO2_subsub battLiMO_subsub battLiNCA_subsub battLiNCA5 battLiNCA15 battLiNMC111 battLiNMC532 battLiNMC622 battLiNMC811 battLiNMC955 battLiSO2_subsub battLiSOCl2_subsub battNaMMO_subsub battNaMVP_subsub battNaNiCl2_subsub battNaNMC_subsub battNaNMMT_subsub battNaPBA_subsub battNaVPF_subsub battNiCd_subsub battNiMH_subsub battPb_subsub battZn_ZnAgO2_subsub battZn_ZnC_subsub battZn_ZnHgO_subsub battZn_ZnMnO2_subsub battZn_ZnO2_subsub battLiNMC622_Si10 battLiNMC811_Si10 battLiNMC955_Si10 battLiNMC622_Si20 battLiNMC811_Si20 battLiNMC955_Si20 batteryCellBinder batteryCellCasing batteryCellElectrolyte batteryCellSeparator cathodeActiveMaterial cathodeAdditive currentCollectorAnode currentCollectorAnodeAndCathode currentCollectorCathode EV: anodeActiveMaterial batteryCellCasing batteryCellElectrolyte batteryCellSeperator batteryCellTerminals batteryPackCables batteryPackModuleEnclosuersAndCo olantManifolds batteryPackSupportFrame batteryPackTerminals batteryPackThermalInsulation cathodeActiveMaterial currentCollectorAnode currentCollectorCathode ferrousMetals glass graphite mixedMaterials organicSolvents paperCarton plastics undefinedMaterials β“-alumina Ag Al As B Br C Cd Ce Cl Co Cr Cu F Fe Ga H Hg K La Li Mg Mn N Na Nd Ni O otherElements P Pb Pr S Sb Si Sm Sn Ti undefinedElements V Zn
Methodology paper – Consolidated composition datasets | www.futuram.eu | 51 6.2 End-of-life vehicles As the data collection approach for end-of-life vehicles differed from the approach in the other waste streams (refer to chapter 4.2), data collection and consolidation were executed simultaneously in an iterative way with the developed R-script. Data transformation was not needed as data was only/directly collected in the required format. Inventories resemble the product key level 2, component key level 1 and the pre-set material list as described in chapter 4.2. The composition is calculated separately for each production year from 1980 to 2010. Data reconciliation and validation included mainly the comparison of the different data sources and balancing differences within product, component, or material composition. The consolidated results reach the level of detail indicated in Table 12. Table 12: Level of detail consolidated results end-of-life vehicles Product key level 2 Components Materials Elements Combination of “fuel type”, “type” (car/van), “length”, “segment”, and “car body” (sedan/SUV) --> Set of 84 combinations of different vehicle keys (a few of them are excluded due to marginal market shares) elvCatalyticConverters elvElectricMotor elvEmbeddedElectronicsActuators elvEmbeddedElectronicsCables elvEmbeddedElectronicsControllers elvEmbeddedElectronicsHeadlights elvEVbattery/elvEVspecificOther elvGeneralComponents elvPowerElectronics elvRest 2xxxAlAlloys 5xxxAlAlloys 6xxxAlAlloys AHSS batteryMaterials castAlAlloys castIron catalysts HSS magnetAlloysNdFeB MgAndMgAlloys mildSteel PCBelectronics powerElectronics pureCu UHSS undefinedMaterials Ag Al Au B C Co Cr Cr_Mo Cu Dy Fe Mg Mn N Nb Nd Ni otherOrUndefined Elements P Pd Pr Pt Rh S Si Ti Ti_Nb V Zn
Methodology paper – Consolidated composition datasets | www.futuram.eu | 52 6.3 Waste electrical and electronic equipment 1) Data transformation The WEEE stream was subject to significant and wide-reaching data transformation activities. This is mainly since different datasets with different data structures and standards for data assessment were used. The data transformation was conducted with an R-script to prevent manual mistakes and allow adaptations of the script in the future. In detail, the data used can be classified into three data structure types, FutuRaM-adherent (complete), inconsistent (partially complete but missing entries), and divergent (structurally different). The FutuRaM-adherent datasets were collected for FutuRaM specifically and collected with the FutuRaM composition data collection template. Therefore, these datasets, mostly stemming from literature research, were consistent with the FutuRaM composition data structure and did not require transformation. Inconsistent datasets describe the data from other and previous projects, i.e., the Desire4Electronics project with Fraunhofer Institute for Advanced Manufacturing, where data is assessed in a way like in the FutuRaM project but was systematically missing certain entries, such as the overall product weights or the component as an individual entry. These entries could be generated through the Rscript from the respective existing product entries. Divergent datasets describe mainly the data from Ecosystem (Ecosystem, 2023) and ProSUM, which used different composition data structures and thus required significant adaptation. Additionally, these constitute by far the largest share of data points, which make these highly relevant to the process. In these specific instances, the data transformation posed three issues: (1) variety of materials with missing component link, (2) variety of elements in components with missing material link, (3) no mass balance. Each problem was addressed specifically and systematically in the R-script. 1. No component link: Both, the ecosystem dataset as well as the ProSUM dataset, link certain materials directly to products without a component link (m-p relations). This leads on the one hand to the problem that information on components is missing, as materials are present without a corresponding component, therefore leading to an incomplete component inventory of these products. On the other hand, this makes it difficult to determine from which components these materials came originally. To resolve this problem, the “shadow component” concept was applied (refer to chapter 5.1.1), which exacts the mass of all materials in a product without a corresponding component (c-p). Subsequently, these materials were linked to the “shadow component” and changed the m-p relations to m-c. 2. No material link: Besides the m-p relations, the ProSUM dataset uses e-c relations. Therefore, a material plug was needed to enable the connection of the elements to the components. This was particularly relevant as the element information from ProSUM builds the foundation to the element-content calculations within FutuRaM. To resolve this problem, the “shadow material” as an equivalent to the “shadow component” was introduced. However, unlike “shadow component”, “shadow material” does not equate the mass of all elements as the elements from ProSUM are normalized and represent already a percentage-value. Thus, the “shadow material” has the mass of 100% or 1 kg/kg for the m-c relations.
Methodology paper – Consolidated composition datasets | www.futuram.eu | 53 3. No mass balance: Across these datasets, issues with mass balance on every level of the composition occurred. In ProSUM, unknown or missing masses were not reflected at all, which leads to the element level of components or materials adding up to less than 100%. In these cases, a “unknown or missing element” was introduced, making up for the difference of the aggregated sum of elements within a component (in this case transformed into “shadow material”, see above) or within a material. Similarly, this was done for components if the overall sum of components and materials added up to less than the product weight in ProSUM. For the Ecosystem dataset, a different approach was chosen. As the product weights partially diverged significantly (>10%), the existing product weights were overridden with the sum of all components and materials in the product, if the divergence was larger than 5%. Beyond these structural transformation steps, generic transformation steps were conducted. Except for the ProSUM dataset, all weights received were in g/unit. These were converted to proportional (g/kg) values by dividing by the overall weight of the respective upper layer relation (normalization). This approach was conducted bottom-up (materials, components) and includes the shadow materials and components. Element information from ProSUM was isolated and introduced at this stage of the process. The elements in materials in ProSUM are to a vast extent stable across time and products. Therefore, the element information was matched in a simple 1:1 matching process with the new datasets. Manual corrections were made to ensure that all known material compositions were matched successfully. Elements in components presented a larger challenge as they, albeit stable over time, are strongly differing for the same components within different products. Therefore, a simple 1:1 matching process did not yield desired results, as similar components within comparable but not similar products would not be matched and thus crucial and reasonably accurate information would be lost. Therefore, a new database for the average element composition per collection category was created and matched up for all components which would not have a direct match from the 1:1 matching process. This critically enhanced information detail. Additionally, some elements would be already logged with complete material information and would get their element information from their corresponding materials. In these cases, the materialelement matching was prioritized, and element composition information was not leveraged from components. Nonetheless, gaps remain as not all materials have available element information yet. Moreover, targeted improvements were performed through the reassigning of similar compositions to materials and components without adequate data matching. This was done with expert consultations. In essence, similar materials and components received element composition data that was highly similar but would not be assigned through the automatic algorithm in the script. This further helped to close data gaps and omitted issues which stemmed from a too generic algorithm which would assign broad averages to information-lacking components and materials. 2) Inventories (c and m) Due to the nature of the data and the variety of datasets, many products within the same UNU key would have a varying number and kind of components and materials, and even when applying the same material or component key, would reflect a different level of detail within that same key. Therefore, all datasets were merged to create component and material inventory based on the lowest common key level within a material or component type. The ProSUM dataset was included at this stage, as well, to enable a complete merger in subsequent steps. Most components in the component
Methodology paper – Consolidated composition datasets | www.futuram.eu | 54 inventory could reach level 2. In the material inventory, most materials remain on level 1, whereas in the raw data a plurality of levels up to material key level 4 is available for many but not all data points. 3) Mean determination For the mean determination, product key level 1 (UNU key) was chosen, which is in line with the level used in the stock and flow modelling. The UNU sub-keys and sub-sub-keys were disregarded at this step, as they would disrupt the dataset coherence too strongly. Nonetheless, they are present in the raw data. The consolidated composition at UNU-key level (level 1) was then further aggregated to WEEE collection category level (level 0). The composition is calculated for 5-year intervals of the production years from 1981 to 2050. To generate a truly representative value for each product, all data points across the sample years within one product key were added together and then divided by their respective upper layer (e.g., component-weight divided by product-weight; material-weight divided by component-weight). This approach was deemed necessary and the only feasible one to generate a coherent and useful dataset. Other previous approaches, such as generating time segment-dependent values, or the median led to inconsistencies or skewed values in the dataset. There were elaborate issues, however, the most notable was the inconsistency of the data points. In particular, the inclusion of zero values for components, materials, or elements, which were found in a sample of a given product key but not found in another sample of the same product key led to an overestimation or underestimation of information. UNU-keys are an accumulation of similar, yet distinctive products. The data used for the analysis does not represent adequate market distribution and thus waste stream distribution but are large point samples. Using the median while including the zero values leads to the potential exclusion of data points, since the most middle value can be 0%. This leads to several problems, most prominently: (1) the disregard of valid data points such as materials that can be found in some of the UNU key products and (2) an upset mass balance (due to disregarded information) and thus an overrepresentation of the included information after the normalization. The advantages of the chosen approach are the consistent results, the smoothing of outliers through the large number of computed data points, and an accurate representation of the sample quantities in the final values. The drawbacks are that no change over time is immediately visible in the time segments due to the disregard of the time segment portion within the computation step (i.e., all results for time segments are the same, starting from the first new data point (e.g., segment 2010 onward, or segment 2015 onward). These changes can be modelled and/or added through an extra step in the computation and rely on expert knowledge. The mean was generated for each layer of the data model (e-m, m-c, c-p) and product weights. For sake of consistency, the same time horizon as used in ProSUM was chosen with time increments starting in 1981 (ProSUM: 1980) to 2050, in 5-year intervals. However, as the new data had a strongly varying number of data points per category, adding ProSUM as single data points per producttimespan combination, would either lead to barely any impact at all, or to being the only data point. Therefore, a fixed weighting to ProSUM data points was assigned, which can be changed at any point but currently stands at 50%. The mean considers ProSUM data points therefore always in a 1:1 ratio.
Methodology paper – Consolidated composition datasets | www.futuram.eu | 55 4) Data reconciliation and filling data gaps In case there was no data available for a certain product-timespan combination, similar to ProSUM, the existing preceding or succeeding data was extended back to 1980 or forward to 2050, respectively. 5) Data compilation in portrayals as input file for waste stream models The consolidated results reach the level of detail indicated in Table 13. Table 13: Level of detail consolidated results WEEE Product key level 1 Components Materials Elements 0001 0002 0101 0102 0103 0104 0105 0106 0108 0109 0111 0112 0113 0113_large 0114 0201 0202 0203 0204 0205 0301 0302 0303 0304 0305 0306 0307 0308 0309 0401 0402 0403 0404 0405 0406 0501 0502 0503 0504 0505 brush burnerNiobiumTube cables cameraIntegrated capacitor capUnspecified casingPlasticBased casingSteelBased casingUnspecified compressor condenserEEE CRTwhole displayCRT displayLCD displayOther displayPlasma displayTFT driveCDDExt driveCDDInt driveFDD driveHDD driveUnspecified electricMotorUnspecified evaporatorEEE fan filterFoamBased fluorescentPowder foil fuse gasDischargeLamp heatingResistor heatSinkEEE HgVapourInLamps inductor insulation keyboard knob lenseAssembly lighting magnetPermanent additives aggregateBinderMixedMaterials AlAndAlAlloys batteryMaterials bitumen carbon composites constructionMinerals CuAndCuAlloys ferrousMetals fluids glassAndCeramics mixedMaterials nonCuAndAlAlloys otherMaterials otherMinerals otherOrganics paperCarton permanentMagneticMaterials plastics shadowMaterial SiSemiconductor wood Mo Ag Al As Au B Ba Be Bi Br C Ca Cd Ce Cl Co Cr Cu Dy Er Eu Fe Ga Gd Ge Hf Hg In K La Mg Mn Na Nb Nd Ni O P Pb Pd
Methodology paper – Consolidated composition datasets | www.futuram.eu | 56 Product key level 1 Components Materials Elements 0506 0507 0601 0602 0701 0702 0703 0801 0802 0901 0902 1001 1002 magnetron magnetSensor nobleGases nonElectricCoil otherElectricCompUnspecified otherNonElectricCompUnspecified passiveJunctionBox PCBCat1 PCBCat2 PCBCat3 PCBCat4 PCBCat5 PCBCat6 plug powerSupply pump PVBacksheet PVcellCdTe PVcellCIGS PVcellSi PVEncapsulationEVA PVFrameAl PVSolarGlass refrigerant remoteControl rotor rotorGear screw sensorUnspecified shadowComponentEEE solenoidValve speakerEEE switch thermostat tonerCartridge touchSensor transformerEEE transmission transmitter wheel Pr Pt Rh S Sb Sc Se Si Sm Sn Sr Ta Tb Te Th Ti unknown OrMissing Element V W Y Zn Zr The consolidated composition data was then further aggregated on collection category level considering the market shares of the different UNU keys within the categories.
Methodology paper – Consolidated composition datasets | www.futuram.eu | 57 6.4 Construction and demolition waste 6.4.1 Buildings 1) Data transformation All data points have the parameter code m-p. An allocation of the material in components was not possible at this point. Therefore, no data transformation is conducted. 2) Inventories (c and m) Currently, components are not included in the dataset for buildings. Therefore, no component inventory was created. The material inventory includes materials on different levels of the material key. 3) Median determination Data gaps only allowed consolidation on product key level 1. The median determination was done for materials in products (m-p). The used unit of measurement is kg/m². Initially, it was planned to have spatially explicit composition. However, the amount of data gaps was huge and therefore a generic composition for the whole EU27+4 was created. The differentiation in time includes three different construction time periods (before 1945 (before World War II), 1945-1995 (post-war architecture), after 1995 (modern architecture)). 4) Data reconciliation and filling data gaps Due to the high aggregation level, no data reconciliation needed to be carried out. The missing elemental material composition (e-m) was attached with the approach described in chapter 4.4.1. 5) Data compilation in portrayals as input file for waste stream models The consolidated results reach the level of detail indicated in Table 14.
Methodology paper – Consolidated composition datasets | www.futuram.eu | 64 Flow (LoW key) elements 19 01 12 19 01 13* 19 01 14 Gd Ge Hf Hg Ho In K La Li Lu Mg Mn Mo Na Nb Nd Ni otherOrUndefinedElements P Pb Pr Rb REE Rh S Sb Sc Se Si Sm Sn Sr Ta Tb Te Th Ti Tl Tm U V W Y Yb Zn Zr
Methodology paper – Consolidated composition datasets | www.futuram.eu | 65 6.7 Mining waste No consolidation steps were carried out for mining waste (see different approach explained in chapter 4.6). Composition data for mining waste is strongly site-specific. Therefore, no average values per LoW key like for the slags and ashes were calculated.
Methodology paper – Consolidated composition datasets | www.futuram.eu | 66 7 Future composition and products The future composition is based on the evaluation of new products and new production technologies. In the following chapters, the approaches of the waste streams regarding the inclusion of future developments and new products in the composition dataset are described. 7.1 Waste batteries Battery chemistries do not change over time. Therefore, the product composition is not timedependent on the battery sub-sub-key which was chosen for composition data consolidation and describes a specific battery chemistry. However, new chemistries are in development and already have a small market share or are expected to enter the market (e.g. SIB – sodium ion batteries, LIB with silicon replacing carbon in the anode active material). These are already included in the product key and composition data. While battery chemistries, respectively product compositions, do not change, market shares of the different chemistries are expected to change until 2050. Therefore, for each individual application, different market shares of battery sub-sub-keys are defined. The analysis of market shares and their future development is based on expert interviews and extensive market research using scientific papers, online sources and reports. 7.2 End-of-life vehicles The trends in composition changes for passenger vehicles, which are seen currently, are expected to continue: An example of a current observation is the trend towards a more complex material composition which includes less steel and more composite and plastics due to weight saving measures and efficiency goals. Additionally, increasingly new digital features and monitoring of car functions including a move towards automatic driving is increasing the amount of electronics built in a passenger car. Following these observations, the end-of-life vehicles composition model extrapolates current trends up to a certain point in the future (until 2035), after which we estimate that there is a levelling off. Future composition of vehicles is then locked from 2035 onwards, as we see high uncertainties involved regarding making assumption how the composition will change between 2035-2050. New product keys are not expected for vehicles in the future within our scope. 7.3 Waste electrical and electronic equipment WEEE is a complex waste stream, and composition varies substantially per category. Various stakeholders were consulted during the implementation of the project to obtain robust data to be included in the composition model. However, it was not possible to obtain quantitative inputs but rather qualitative considerations that are summarized in the next paragraphs. Quantitative changes of the composition in the future are not included. Large household appliances, cooling and freezing equipment are not expected to fundamentally change over time as their technology has reached a certain level of maturity. On the other end, small household appliances, IT equipment, screens and monitors may experience changes in composition over time, especially in the case of ultra-thin laptops, tablets, mobile phones or in virtual reality equipment such as drones. The trend that can be observed is that a smaller quantity of CRMs is used for producing the same components by maintaining or improving their performances. On the other
Methodology paper – Consolidated composition datasets | www.futuram.eu | 67 hand, more products in the future will include sensors, PCBs, or integrated cameras which will contribute to increase the materials demand for such products. Lights and luminaries are already observing a change in the composition mainly referring to LEDs using more plastics instead of metal or glass. PV panels underwent major changes in terms of technology and composition in recent years. In the future, most PVs are expected to utilize monocrystalline silicon technology. 7.4 Construction and demolition waste Both buildings and wind turbines have relatively long lifespans compared to the timeline of the model. As a result, the composition of future outflows will primarily depend on the composition of past and current stocks. Consequently, changes in the composition of future inflows are unlikely to significantly affect the composition of future outflows. The composition of future outflows will change as different shares of each building type cohort or type of wind turbine reach their end-of-life, according to the lifetime curve. In summary, over the short to medium term, the composition of the input for waste generation is expected to remain largely unaffected by variations in the composition of future products. The overall waste stream composition changes only according to the shares of cohorts and wind turbine types in the outflows. 7.5 Slags and ashes Compositions of the following slags and ashes are not expected to change in the future, as the composition of the materials will not change. • Ashes from biomass incineration • Ashes from coal incineration • Ashes from sewage sludge incineration Compositions of the following slags and ashes are expected to change in the next 50 years. • Slags and sludges o Optimization in the different industries can lead to small reductions in metal mass fractions in the slags and sludges. The use of other types of ores can also influence the composition. However, in the short term, no major changes are expected. o In general, the slags from secondary raw materials contain higher mass fractions of CRMs than the ones from primary raw materials. When both streams are processed in the same plant, we can expect that the share of the secondary raw material input streams will increase in the future, which will result in higher mass fractions of CRMs in the slags. o The expected changes in the composition of slags from primary raw materials and secondary raw materials will be incorporated in the scenarios. For example, for steel it is the consequence of the share between BOF vs. EAF steel plants, which are associated to different slag compositions. ▪ For the BAU scenario: no changes will be incorporated ▪ For the REC and CIR scenario: more production in EAF steel plants
Methodology paper – Consolidated composition datasets | www.futuram.eu | 68 o Small changes linked to optimizations in the metallurgical industry are not incorporated as this industry is out of the scope of the project. - Ashes from waste incineration o Under the BAU, no changes are expected. o Under the REC and CIR, when products are better recycled, or the life span will increase, it is expected that the concentration of CRMs in the incinerated municipal solid waste will decrease. However, small mass fractions will always be present due to sorting losses and dissipation. o There is insufficient data to accurately characterize the current waste composition. Changes in future composition are difficult to predict. Hence, no changes are modelled due to lack of data. 7.6 Mining waste Composition of mining waste currently and in the future depends upon different factors. The profit a mining company can make with a raw material steers its content in the mining waste. There is a general trend that raw materials that were not economically interesting and considered impurities initially become available due to new technologies of separation and recovery in the different steps like flotation, hydrometallurgy and smelting. Therefore, the composition of mining waste is directly linked to available technologies for raw material extraction and the value of a raw material so the mining company can generate profit with its extraction. A side effect of this is that some mining waste landfills become economically interesting due to the increased value of the raw materials they contain, and which can be recovered with profit with todays or future technologies. Especially tailings from mines that have been used for decades and will be abandoned in the next years or decades might be of interest due to increased value of raw materials they contain and advances in technology in comparison with their age.
Methodology paper – Consolidated composition datasets | www.futuram.eu | 69 8 (External) composition data validation Certain data validation is already done during the consolidation process (compare step 4 in chapter 5.1.1). Additionally, further validation of the consolidated composition data as described in the following chapters was or will be considered. 8.1 Waste batteries The consolidated composition data was reviewed by an internal expert and resulted in the following modifications: • All lithium rechargeable batteries should have the same c-p and m-p values for casing, so the average c-p value among all and the average m-c value among all is used for all battery chemistries (normalize other c-p values after setting the c-p value for casing: (c-p value) / (all c-p values) * (100 - c-p value casing)) • The electrolyte in lithium rechargeable batteries should have salt and organic solvent in all battery subtypes. The composition of the electrolyte in the LiCO battery type was applied to all other types excluding additives. Stoichiometric distribution for LiPF6 (Li: 5%, P: 20%, F: 75%) was carried out after normalization with organic solvents. • The cathode additive was added to the NMC955 battery type (average value from the other NMC subtypes) • The electrolyte in sodium rechargeable should have organic solvent and additives in all battery subtypes. The relative value of organic solvent and additives in the electrolyte of the NaVPF battery was applied to the other sodium rechargeable batteries and the share of the elements from the salt were reduced accordingly. 8.2 End-of-life vehicles Further validation of the composition data via external stakeholder involvement in a Delphi like workshop is planned outside of the scope of the presented work. 8.3 Waste electrical and electronic equipment A small Delphi-like workshop with internal experts/stakeholders/partners was conducted to validate certain aspects of the consolidated composition data. Composition data aspects were discussed in the expert round until agreement on how to further proceed was reached. Different aspects which need a more in-depth analysis were identified and will be tackled within other tasks of the project. All elemental material composition is based on ProSUM data. However, the elemental PCB composition included questionable values and was therefore revised and replaced with a newly created composition which is based on the evaluation of the ProSUM raw data sources and additional literature research. Additionally, it will be investigated if further Delphi studies could be used to validate other composition data aspects and/or determine future composition.
Methodology paper – Consolidated composition datasets | www.futuram.eu | 70 8.4 Construction and demolition waste Further validation of the composition data is currently not planned. The focus lays on expanding the existing dataset, especially regarding components in buildings. 8.5 Slags and ashes For each of the waste flows an internal workshop was organized to review the data on composition, and stocks & flows (Antwerp – 22 March 2024, Lille – 10 September 2024, Antwerp – 25 October 2024). Composition data, represented for all member states, is limited to all slags and ashes waste flows. Thus, simplifications were unavoidable. Ashes from coal, biomass and sewage sludges are not verified as their composition should not vary with time and geographic location. For municipal solid waste incineration bottom ashes, data from various perspectives (scientific literature and sector federations) were compared. Composition was then chosen such that it aligns with various sources. No external verification was provided. If industrial data (from sector federations) is available, a similar alignment will be performed for the slags. 8.6 Mining waste The validation of the data is done by following standard protocols for sampling and analysis. Uncertainties will be determined based on different aspects of the data: • Density of sampling • Geochemical analysis • Tonnage 8.7 Variability and uncertainty A concept to account for data variability and uncertainty was developed. This approach differentiates based on the number of data points used to define the median or mean, as well as minimum and maximum value during consolidation. The proposed implementation strategy could be the following: • If the data count is ≥5, the minimum and maximum value were chosen as range. • If the data count is <5, a range reflecting the uncertainty was estimated based on the data quality. The lower the data quality the higher the range. Each waste stream has the flexibility to define its own criteria when deciding on which method should be applied, depending on the collected data. Moreover, decisions regarding the treatment of outliers should be made based on the availability of data and specific knowledge of the waste stream. The ranges reflect on one hand naturally occurring variability in the results, e.g. due to products from different brands, different construction techniques for buildings, or different MSW compositions resulting in different ash compositions after incineration. On the other hand, they also account for uncertainty related to low data quality.
Methodology paper – Consolidated composition datasets | www.futuram.eu | 71 The methodology ensures that each waste stream independently determines its ranges, incorporating both variability and uncertainty. Consequently, each dataset yields a minimum, maximum, and median/mean value, which serve as inputs for subsequent modules. The next modules, including stock-and-flow modelling, are executed three times using the minimum, maximum, and median/mean values. This approach enables the assessment of the influence of compositional data on the model outcomes. Where feasible, a comprehensive Monte Carlo analysis is recommended to enhance robustness.
Methodology paper – Consolidated composition datasets | www.futuram.eu | 72 9 Recommendations for composition data collection and consolidation During data collection and consolidation, different non-foreseeable challenges and issues were faced. The following list of “lessons learned” should be considered for any update of the composition data or similar procedures in the FutuRaM project: • Importance of a reasonable, comprehensive, and hierarchical product key: o Representation of groups of products with a similar function and composition. o Sensible questioning/challenging of established product key code lists. • Importance of a reasonable, comprehensive, and hierarchical component key including decision on “subdivision” or “subcomponent” concept (see Figure 7, page 19). • Importance of a reasonable, comprehensive, and hierarchical material key • Codes with the same name at different levels (only distinguished by the respective level number) should have the same composition. • Importance of distinguishing if the reference is referring to an element or material e.g. in case of copper, aluminum, or iron. • Importance of clear instructions on data quality assessment. • Importance of clarification of the composition data structure and required information for each parameter code. • Hesitancy of rating data quality and similarity with the best score. • For the consolidation, production/generation year and location cannot be displayed as time span or a list of countries. If the value is connected to different years and locations, a separate row for each year and location needs to be inserted. • Varying or wrong formatting of values hinders further data processing (decimal point or comma, blanks, usage of < and >). • Blanks before the first and after the last word in a cell hinder further data processing. • If a reference includes < or > information, this can be displayed using upper and lower limits. • Advantages when units of measurement (UoM) are directly harmonized during data collection; harmonization afterwards is possible if UoM are of the same type, but problems occur with UoM of different types (e.g. wt% and kg/MW). • Component and material key should be always filled up to the lowest possible level to simplify the creation of component and material inventories for consolidation. • Data with parameter code e-c, e-p or m-p: as much as possible allocation to a component and/or material, otherwise apply “shadow or placeholder material and component concept” (see chapter 5.1.1). • Consistency within the dataset plays a key role in the generation of consolidation results. • To make the consolidation approach completely automatable, a complete raw dataset is required. • The type of central value is dependent on data availability and on further research on variability. The current consolidation method aims to consolidate the information from as many data points as possible regardless of their data structure and their level of detail. For this, the level of detail that can be reached with all included data points is chosen for the consolidation. However, with this method, more granular information that is only given in a few data points is not included with this level of
Methodology paper – Consolidated composition datasets | www.futuram.eu | 73 detail in the consolidated composition results. Therefore, the development of a method which enables additional inclusion of more granular information should be considered. Furthermore, various metadata of the raw data was not yet used in the consolidation process. It is planned to investigate possibilities to further include them and conduct further in-depth analyses of the raw data. The data collection and consolidation process revealed that the currently used classification systems and available code lists are not always appropriate to collect and display composition data in the desired way. Therefore, it is recommended to check and improve the classification systems used and code lists for further data collection. The classification systems should be reasonable, comprehensive, and hierarchical. One code should resemble a unit which is similar in application, weight, and composition. More detailed codes are only required if summarizing has an impact on the composition. During the consolidation process, multiple data gaps were encountered that needed to be filled with assumptions or required consolidation on a higher level of aggregation. For each waste stream, it should be evaluated if additional data is available and can be collected and/or if the data gaps can be closed by additional own test/analyses (e.g. sampling campaigns or batch tests). This includes temporal and spatial granularity as well as complete product composition (reduce missing links). For remaining data gaps, recommendation for the EU on how to close them by e.g. reporting requirements will be compiled. It is further aimed at harmonizing the elemental material composition among components, products and even waste streams. For this, a high material granularity is required (e.g. on alloy level). It will be investigated how material granularity can be enhanced in each waste stream.
Methodology paper – Consolidated composition datasets | www.futuram.eu | 80 Term Definition Reference DPP (Digital Product Passport) A Digital Product Passport (DPP) is a structured collection of product related data with predefined scope and agreed data management and access rights conveyed through a unique identifier and that is accessible via electronic means through a data carrier. The intended scope of the DPP is information related to sustainability, circularity, value retention for reuse, remanufacturing, and recycling. From: https://cirpassproject.eu/dpp-in-a-nutshell/ Digital Building Logbook (DBL) A Digital Building Logbook (DBL) is a common repository for all relevant building data. It facilitates transparency, trust, informed decision making and information sharing within the construction sector, among building owners and occupants, financial institutions and public authorities. Dourlens-Quaranta et al. (2020) Products Usually refers to anything that is made to be sold including components and materials. Here it only refers to assemblies that are made out of components which are made out of materials (e.g. batteries, vehicles, EEE, buildings). Based on Cambridge Dictionary: https://dictionary.cambridge.org/dictionary/ english/product Components or component groups Uniquely identifiable parts or subunits of products. Components are usually mechanically removable in one piece and are considered indivisible for a particular function or use. A component can consist of other subcomponents e.g. a printed circuit board may contain a capacitor which is also a component. Some products may contain other products as components, for instance, a car has a battery. Based on ProSUM Harmonisation paper for external feedback and consultation Annex 1 ( Huisman et al., 2016 ) Materials Refers to ‘engineered materials’ that are composed, manufactured and processed to achieve intended properties. Based on ProSUM Harmonisation paper for external feedback and consultation Annex 1 ( Huisman et al., 2016 ) Minerals A valuable or useful chemical substance that is formed naturally in the ground Based on Cambridge Dictionary: https://dictionary.cambridge.org/dictionary/ english/mineral Compounds A chemical substance that combines two or more elements. Based on Cambridge Dictionary: https://dictionary.cambridge.org/dictionary/ english/compound Mass The mass of an entire component or product (subscript c or p respectively), kg or also kg/unit, kg/piece. Based on ProSUM Harmonisation paper for external feedback and consultation Annex 1 ( Huisman et al., 2016 ) Volume The volume of an entire component or product (subscript c or p respectively), m³ or also m³/unit, m³/piece. Based on ProSUM Harmonisation paper for external feedback and consultation Annex 1 ( Huisman et al., 2016 ) Area The area of an entire component or product (subscript c or p respectively), m² or also m²/unit, m²/piece. Based on ProSUM Harmonisation paper for external feedback and consultation Annex 1 ( Huisman et al., 2016 )
Methodology paper – Consolidated composition datasets | www.futuram.eu | 81 Term Definition Reference Mass fraction The mass fraction of an element, material, component or product in a material, component, product or stock/flow (subscript e.g. e-m, e-c, e-p, m-c, m-p, or c-p), kg/kg, mg/kg (ppm), wt%. Based on ProSUM Harmonisation paper for external feedback and consultation Annex 1 ( Huisman et al., 2016 ) Concentration The concentration of an element, material, component or product in a material, component, product or stock/flow (subscript e.g. e-m, e-c, e-p, m-c, m-p, or c-p), mg/l, mg/m³, vol%. Based on ProSUM Harmonisation paper for external feedback and consultation Annex 1 ( Huisman et al., 2016 ) “Density”, Intensity” The "intensity" or "density" of an element, material, component or product in a material, component, product or stock/flow (subscript e.g. e-m, e-c, e-p, m-c, m-p, or c-p), kg/m², kg/MW, kg/MWh. Based on ProSUM Harmonisation paper for external feedback and consultation Annex 1 ( Huisman et al., 2016 ) Number content The number content of an element, material, or component in a product (subscript e-m, e-c, m-c, e-p, m-p or c-p), units/unit, pieces/piece. Based on ProSUM Harmonisation paper for external feedback and consultation Annex 1 ( Huisman et al., 2016 ) Mass content The mass content of an element, material, or component in a product (subscript e-m, e-c, m-c, e-p, m-p, or c-p), kg/unit, kg/piece. Based on ProSUM Harmonisation paper for external feedback and consultation Annex 1 ( Huisman et al., 2016 )