Life Cycle Assessment of Recycled Aggregate Concrete: The Moderating Role of Strength Class in Recycling–Impact Relationships

Recycled aggregate concrete (RAC) is often presented as a lower-impact alternative to conventional concrete, yet whether higher recycling rates yield consistent environmental gains across strength classes has rarely been tested. This study examines that question using 23 Environmental Product Declaration (EPD) datasets from eight Beton Eisack batching plants in Alto Adige, northern Italy. The dataset spans three strength classes (C12/15, C25/30, C30/37), with recycling rates ranging from approximately 7% to 38%, and ten environmental indicators reported within the EN 15804 cradle-to-gate A1–A3 framework. An interaction regression model was used to test whether strength class moderates the slope of the recycling-impact relationship. For C12/15 mixes, higher recycling rates are associated with lower values for global warming potential, primary energy, abiotic resource use and acidification. The slopes flatten in C25/30 and reverse sign for several indicators in C30/37, where higher recycled content corresponds to higher rather than lower impact. The pattern is consistent with a cement compensation effect: higher strength classes typically require a richer binder system, and the resulting cement-related burden offsets the savings from aggregate substitution. These results indicate that recycling rate on its own does not reliably reflect the environmental performance of RAC. Strength class, cement content and mix-design assumptions need to be assessed together, particularly when EPD-based results inform benchmarking schemes or procurement criteria. The regional scope, the eight plants and the sample size of N = 23 make the coefficients reported here highly regionalised, so they cannot be generalised to other production contexts. These limits do not change the main methodological implication: strength class acts as a moderating variable in the recycling-impact relationship and should be modelled explicitly, rather than averaged out in future RAC-LCA work.