Modelling releases from tailings in life cycle assessments of the mining sector: From generic models to reactive transport modelling

Extracting mineral ores to produce mineral concentrates and finally metals that enter the value chains of different sectors is a waste intensive activity. In particular, mineral processing activities generate yearly and globally several billion tons of wastes called tailings. Tailings storage is associated with significant environmental risks due to different sources such as releases to groundwater. Yet, these environmental impacts are poorly taken into account in the environmental life cycle assessment of a concentrate or metal production, in particular due to a lack of representative data in the inventory.This paper aims at coupling geochemical modelling geochemical modelling with life cycle assessment to quantify releases from tailings and the associated environmental impacts. In the present work, reactive transport modelling was implemented to quantify more accurately short-term (100-year timeframe) and long-term (10,000 year-timeframe) tailings releases. Reactive transport modelling allows modelling and estimating the distributions in space and in time of chemical reactions occurring in a specific environment. The resulting tailings releases are compared with those calculated through the approach, based on a first order law of kinetics, used in the ecoinvent database. The results from the ecoinvent approach are calculated both with generic and site specific parameters. These different approaches are applied on the case-study of a polymetallic sulphidic ore from a mine in Northern Europe.The tailings releases obtained through the different approaches differ. Overall, the reactive transport model leads to quantities released at both short-term and long-term timeframes being smaller than those calculated with the ecoinvent approach. As a result, the impacts on ecotoxicological and toxicological categories show differences between two to eight orders of magnitude. Each approach used to model the tailings releases differ in the nature and amount of data needed as well as in the sophistication of the model used. While the approach based on reactive transport modelling is exhaustive and presumably provides a more accurate prediction for the releases from tailings, it also requires massive computational resources and much larger volumes of data, some of which being difficult to obtain on-site. Consequently, a consensus has to be found between the complexity of the model used, the data availability and the expected reliability of the results.