Joint production with multiple determining products: the case of platinum-group metals

Abstract Purpose Platinum-group metals (PGM) constitute an extreme example of a joint production process since mines, mainly in Africa and Russia, co-produce up to eight valuable metals simultaneously. While this multifunctionality has been traditionally dealt with by allocation (partitioning), in this article we aim at developing a life cycle inventory (LCI) model following consequential modelling principles, covering supply of five of the six metals in the group (platinum, palladium, rhodium, ruthenium, iridium). Methods An analysis of data from 33 mines around the world suggests that PGM production can be considered a case of a ‘type 3’ situation, where there is not one, but several determining products, in this case the five PGMs. Following this finding, we develop a consequential LCI model with a cradle-to-gate scope, considering supply from mines in South Africa, Russia and Zimbabwe, based on existing data for PGM mining in the ecoinvent database, together with public statistics on global PGM production trends and prices in the period 2019–2023. Results and discussion The model is evaluated at the impact assessment level, focusing only on greenhouse-gas (GHG) emissions per kg metal. The ranking of metals, from higher to lower emissions is rhodium, iridium, platinum, palladium and ruthenium. Sensitivity analyses show that results are mainly influenced by the marginal supplying countries, but especially by metal prices. When the model is evaluated for the period 2014–2018, GHG emissions per kg metal are substantially affected. Conclusions A key finding of this research is that, from a consequential modelling standpoint, PGM mining constitutes a case of ‘type 3’ situation in joint production, where all PGM can be considered co-determining products in mines. The main limitation of the developed model arises from the need to address the intrinsic variability in prices associated to the global PGM market, as a result of imbalances in supply and demand. While our model has been based on historical 5-year time series on production and prices, an alternative for future improvement would be to use longer time series, or more advanced forecasting such as Autoregressive integrated moving average (ARIMA) techniques for the future producer mine supply as well as on prices.