Science Inventory

Tracking end-of-life stage of chemicals: A scalable data-centric and chemical-centric approach

Citation:

Hernandez-Betancur, J., Gerardo J. Ruiz-Mercado, AND M. Martín. Tracking end-of-life stage of chemicals: A scalable data-centric and chemical-centric approach. Resources, Conservation and Recycling. Elsevier Science BV, Amsterdam, Netherlands, 196:107031, (2023). https://doi.org/10.1016/j.resconrec.2023.107031

Impact/Purpose:

Commercial, consumer, and industrial activities, including disposal, treatment, energy recovery, and recycling operations; in short, End-of-Life (EoL) operations involve chemical substances that may endanger human health and the environment. As a result, risk assessment for chemicals across their life cycles is a key step in determining unreasonable risks and assessing regulatory mechanisms to manage risks. Risk evaluation demands collecting comprehensive information, e.g., to estimate release quantities and identify potential exposure scenarios. Also, practitioners must obtain a chemical flow inventory or Life Cycle Inventory (LCI) as part of the risk evaluation process. Therefore, estimating LCIs of chemical releases is critical for environmental, health, and safety evaluations. This Manuscript shows a data mining framework to extract and transform data from publicly accessible, siloed, and multi-country database systems, for generating LCI of chemical transfers to off-site facilities for EoL management. The data mining approach provides information about chemicals, generator industry sectors, and EoL activities involved in the EoL transfers, as well as reporting years, annual transfer amounts, countries, and reliability scores for reporting transfer flows. These data entries describe potential EoL exposure scenarios due to off-site transfers and connect the enhanced framework with methodologies for quality assessment for LCI data. The information collected by the framework could be used for EoL chemical risk and exposure evaluation, supporting rapid LCI modeling, and using EoL sustainability indicators. Also, the framework could be connected to other data sources for gaining context about aspects that may affect the behavior of the EoL management chain, like the environmental stringency and molecular descriptors for cheminformatics. The framework moves forward to provide standardized LCI data that extends government actions and allows global-scale sustainability analysis at the EoL stage, considering progress over time, across countries, and with as much data granularity as possible.

Description:

Chemical flow analysis (CFA) can be used for collecting life-cycle inventory (LCI), estimating environmental releases, and identifying potential exposure scenarios for chemicals of concern at the end-of-life (EoL) stage. Nonetheless, the demand for comprehensive data and the epistemic uncertainties about the pathway taken by the chemical flows make CFA, LCI, and exposure assessment time-consuming and challenging tasks. Due to the continuous growth of computer power and the appearance of more robust algorithms, data-driven modelling represents an attractive tool for streamlining these tasks. However, a data ingestion pipeline is required for the deployment of serving data-driven models in the real world. Hence, this work moves forward by contributing a chemical-centric and data-centric approach to extract, transform, and load comprehensive data for CFA at the EoL, integrating cross-year and country data and its provenance as part of the data lifecycle. The framework is scalable and adaptable to production-level machine learning operations. The framework can supply data at an annual rate, making it possible to deal with changes in the statistical distributions of model predictors like transferred amount and target variables (e.g., EoL activity identification) to avoid potential data-driven model performance decay over time. For instance, it can detect that recycling transfers of 643 chemicals over the reporting years (1988 to 2020) are 29.87%, 17.79%, and 20.56% for Canada, Australia, and the U.S. Finally, the developed approach enables research advancements on data-driven modelling to easily connect with other data sources for economic information on industry sectors, the economic value of chemicals, and the environmental regulatory implications that may affect the occurrence of an EoL transfer class or activity like recycling of a chemical over years and countries. Finally, stakeholders gain more context about environmental regulation stringency and economic affairs that could affect environmental decision-making and EoL chemical exposure predictions.

Record Details:

Record Type:DOCUMENT( JOURNAL/ PEER REVIEWED JOURNAL)
Product Published Date:09/01/2023
Record Last Revised:09/01/2023
OMB Category:Other
Record ID: 357923