Probabilistic model for compressive strength of recycled aggregate concrete accounting for uncertainty of recycled aggregates from different sources

Recycled aggregate concrete (RAC) structures have a promising future in the sustainable development of construction industry. However, recycled aggregates (RAs) may come from different sources and have fairly different properties. Such differences would significantly increase the variability in the mechanical properties of the resulting RAC. Available analysis on the variability of RAC properties only focused on RAC with RAs from a single or very limited sources, the variability was quantified by conducting standard compressive strength tests for concrete on more samples from the same concrete batch (i.e. on 20–30 samples instead of 3 samples). This method cannot account for the uncertainties of RA sources and the uncertainties of cement properties, natural aggregate properties, and water contents, etc., leading to the underestimation of the uncertainty of RAC properties and unsafe design. To date, no statistical analysis has been conducted with the account of the uncertainty induced by the difference in RA source. To address this gap, a new method is proposed in this paper to quantify the variability in the compressive strength of RAC accounting for uncertainty of recycled coarse aggregates (RCAs) from different sources. This method considers the uncertainty of RCAs from different sources by the probabilistic model for compressive strength ratio of RAC to natural aggregate concrete (NAC), which is determined using available test data with RCAs from 193 different sources. This probabilistic model is then introduced into the existing probabilistic model for compressive strength of NAC through uncertainty propagation to further account the uncertainties in the content and properties of cement, natural aggregates, water, etc. based on the statistical data of 680000 NAC samples. With such method, the probabilistic model for compressive strength of RAC is established, and the characteristic cubic and axial compressive strengths of RAC are determined. Analysis results indicate that: 1) the coefficient of variation (COV) of RAC compressive strength increases with the increasing RCA replacement ratio, the COV of the RAC cubic compressive strength with a full use of RCA is 6.3 %–22.8 % higher than that of NAC; 2) the COV of RAC compressive strength decreases with the increasing strength class, as the strength class increases from RC20 to RC50, the COV of the RAC cubic compressive strength decreases from 21.9 % to 18.3 % for a full use of RCA; 3) at the same concrete strength class, the mean value of the RAC cubic compressive strength with a full use of RCA need to be 2.8 %–8.0 % higher than that of the NAC due to the high scatter of the properties of RCAs from different sources; 4) the characteristic axial compressive strength is 21.0 %–21.6 % lower than that of the companion NAC.