The uncertainty arising from laboratory sampling (sub-sampling) can compromise the accuracy of analytical results in highly inherent heterogeneous materials, such as solid waste. Here, we aim at advancing our fundamental understanding on the possibility for relatively unbiased, yet affordable and practicable sub-sampling, benefiting from state of the art equipment, theoretical calculations by the theory of sampling (ToS) and implementation of best sub-sampling practices. Solid recovered fuel (SRF) was selected as a case of a solid waste sample with intermediate heterogeneity and chlorine (Cl) as an analyte with intermediate variability amongst waste properties. ToS nomographs were constructed for different sample preparation scenarios presenting the trend of uncertainty during sub-sampling. Nomographs showed that primary shredding (final d90 ≤ 0.4 cm) can reduce the uncertainty 11 times compared to an unshredded final sub-sample (d ≈ 3 cm), whereas cryogenic shredding in the final sub-sample can decrease the uncertainty more than three times compared to primary shredding (final d90 ≤ 0.015 cm). Practices that can introduce bias during sub-sampling, such as mass loss, moisture loss and insufficient Cl recovery were negligible. Experimental results indicated a substantial possibility to obtain a representative final sub-sample (uncertainty ≤ 15%) with the established sub-sampling plan (57-93%, with 95% confidence), although this possibility can be considerably improved by drawing two final sub-samples instead (91-98%, with 95% confidence). The applicability of ToS formula in waste-derived materials has to be investigated: theoretical ToS calculations assume a poorer performance of the sub-sampling plan than evidenced by the experimental results.
Keywords: Resource recovery; Shredding; Solid recovered fuel (SRF); Sub-sampling; Theory of sampling (ToS); Uncertainty.
Copyright © 2020. Published by Elsevier Ltd.