Technical assessment of processing plants as exemplified by the sorting of beverage cartons from lightweight packaging wastes

- Methods for technical assessing of LP sorting plants are discussed.
- Established methods to determine the sampling effort were found impractical.
- However, the 3%-criterion did yield practical solutions.
- Performance and results of three plant assessments for the recycling of BC are given.
- Irrespective of amount and size of the samples, the CI intervals will remain large.
- The experiences can be used for a better reliable and reproducible sample taking.

The recovery of beverage cartons (BC) in three lightweight packaging waste processing plants (LP) was analyzed with different input materials and input masses in the area of 21-50 Mg. The data was generated by gravimetric determination of the sorting products, sampling and sorting analysis. Since the particle size of beverage cartons is larger than 120 mm, a modified sampling plan was implemented and targeted multiple sampling (3-11 individual samplings) and a total sample size of respectively 1200 l (ca. 60 kg) for the BC-products and of about 2400 l (ca. 120 kg) for material-heterogeneous mixed plastics (MP) and sorting residue products. The results infer that the quantification of the beverage carton yield in the process, i.e., by including all product-containing material streams, can be specified only with considerable fluctuation ranges. Consequently, the total assessment, regarding all product streams, is rather qualitative than quantitative. Irregular operation conditions as well as unfavorable sampling conditions and capacity overloads are likely causes for high confidence intervals. From the results of the current study, recommendations can basically be derived for a better sampling in LP-processing plants. Despite of the suboptimal statistical results, the results indicate very clear that the plants show definite optimisation potentials with regard to the yield of beverage cartons as well as the required product purity. Due to the test character of the sorting trials the plant parameterization was not ideal for this sorting task and consequently the results should be interpreted with care.