Facilitating the Use of the Theory of Representative Sampling for Granular Waste: Number of Particles, Sample Size and Uncertainty Arising From Sampling
Résumé
Environmental forensics assessments are often based on few analytical measurements. Biased data can, therefore, led the involved actors to be improperly charged or acquitted. Data reliability, considered in terms of uncertainty and variability, is determined by the analytical methods and the performed sampling procedures. The influence of this latter increases with highly heterogeneous materials characterized by the presence of trace, “rare” elements (both considered resources or contaminants), as occurring in the field of waste management. Nevertheless, the theoretical principles behind representative sampling are still not well understood in the waste community, which often underestimates its influence on the reliability of the resulting data and consequent decisions. Applying the binomial probability distribution, the theory of representative sampling is reviewed to facilitate its use for granular solid wastes. The concept of the number of particles n that shall compose a representative sample is introduced. Within n, the same fraction p shall be included of “rare” particles present in the lot to be characterized. Also, n depends on the minimum achievable variability of test results, i.e., coefficient of variation CV computable on a set of equivalent samples. Data are presented quantifying occurring intra and inter-laboratory variability of the most used chemical-physical and ecotoxicological analyses of homogeneous material. Accordingly, a specific formula to calculate n is proposed. Three requirements are then laid down to transform n into mass of sample. Representative samples must include a fraction p of particles considered “rare” in terms of i) presence/concentration of substances and ii) particles size distribution, since large waste particles are seldom characterized by the same composition of fine particles. Further, iii) the size of the sampling equipment shall prevent segregation of larger particles. The mass of a representative sample is calculated by reconciling these three conditions. According to the dedicated technical standards, the layout of a replication experiment is developed to measure the uncertainty arising from a sampling plan designed for waste classified with European Waste Code 191212 “other wastes including mixtures of materials from mechanical treatment of wastes”.Practical and design challenges that can affect the correct performance of the related experimental activities and the reliability of the outcomes are discussed.