Answers are grounded in this paper and cite their sources.
Modern sorting plants for lightweight packaging waste (mainly plastics, metals and compounds) can operate with up to 50 sensor-based sorters (SBS), generating large volumes of material flow data. This study presents the first systematic evaluation of SBS data for real-time, inline monitoring of throughput (0.1–17.5 t/h) and input composition (eject shares 5–50%). Two fractions were examined: larger polyethylene “chips” sorted by color via visible light (VIS) cameras, and smaller “flakes” of various polymers sorted by near-infrared (NIR) technology. Formulas converting pixel counts to mass-based metrics were developed, while artificial intelligence was deliberately avoided to highlight the inherent potential of pixel data. Monitoring accuracy depended strongly on particle overlap, measured by the superposition factor (fsp). For fsp<1.05, median throughput deviations were +0.3% (chips) and −11.6% (flakes); composition deviations were +3.9% and +2.4%, respectively. If the outlined challenges are considered, the technology can be used in realistic conditions of plant operation (fsp<1.25).
No takes yet. Share an insight, caveat, or question.
Schlögl et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: