Limitations of applying life cycle assessment to complex co-product systems: The case of an integrated precious metals smelter-refinery

Integrated smelter-refineries play an important role in the recovery of multiple metals from complex primary and secondary materials, and hence in closing metals cycles. Processes in these facilities are strongly interconnected, dynamic, and multifunctional, which challenges a typical representation in life cycle assessment (LCA). This is especially true when LCA is applied to calculate the environmental profile of single metals products.This study examines methodological requirements for assessing complex co-product systems using attributional LCA through a static, gate-to-gate inventory model that quantifies the environmental impacts of each of the metal products of an integrated precious metals smelter-refinery. The model is based on a large number of subprocesses and is formulated using detailed industry data, which allows quantification of the sensitivity of the results with respect to allocation rationales and the data collection period.The results within one impact category vary strongly among metals (up to four orders of magnitude for copper compared to rhodium). Moving from mass- to value-based allocation changes the result for a given metal by up to two orders of magnitude. If value-based allocation is used, the selected reference year for metals prices influences the results by up to a factor of two.Allocation rationales are critically analyzed, and it is shown that none reflect the business model or other system drivers. While the model is focused on quantifying environmental impacts of metal outputs, the actual process is economically driven to efficiently treat a continuously changing feed mix. The complexity of a smelter-refinery cannot be captured by static, attributional inventory models, which is why the choice of allocation rationale remains arbitrary. Instead, marginal, parameterized models are needed; however, such models are substantially more time and data intensive and require disclosure of more detailed, process specific data.