Austria must recycle more packaging materials. Especially for plastic packaging waste, significant increases are necessary to reach the EU recycling targets for 2025 and 2030. In addition to improving separate collection and introducing a deposit system for specific fractions, the share of plastic packaging in mixed municipal solid waste (MSW) could be utilized. In Austria, about 1.8milliontonnes of mixed MSW are generated. This includes about 110,000 t/a of plastic packaging waste. Most of the mixed MSW (94 %) is sent directly or via residues from pre-treatment, such as mechanical-biological treatment or waste sorting, to waste incineration. While materials such as glass and metals can also be recovered from the bottom ash, combustible materials such as plastics must be recovered before incineration. This work aims to evaluate the recovery potential of plastic packaging waste in mixed MSW with automated waste sorting. For this purpose, two of the largest Austrian waste sorting plants, with a total annual throughput of about 280,000 t/a, were investigated. The investigation included regular sampling of selected output streams and sorting analysis. The results show that the theoretical recovery potential of plastic packaging from these two plants is 6,500 t/a on average. An extrapolation to Austria results in a potential of about 83,000 t/a. If losses due to further treatment, such as sorting and recycling, are considered, about 30,000 t/a of recyclate could be returned to plastic production. This would correspond to an increase in plastic packaging recycling rate from 25 % to 35 %.

The low recycling rate of post-consumer plastic packaging waste (PPW), which is partly due to insufficient separate collection, heterogeneous composition and high levels of contamination, poses a challenge in Austria, where the recycling rate must double in order to meet the target of 55 %. This study analyzes key packaging characteristics of non-beverage plastic bottles influencing recyclability, using Vienna as a case study. Additionally, a net quantity indicator and separate collection rates were calculated. 738 bottles from mixed MSW and 1,159 bottles from separate PPW collection were analyzed. The main polymer's proportion described by the net quantity indicator was higher for bottles from separate collection (69-72 %) than from mixed MSW (58 %), showing that a large share of the foreign materials are residues and dirt, with significantly higher contents in mixed MSW (20 %) than in separate collection (11 %). With a separate collection rate of 19.2 %, the great potential for recycling currently lies in mixed MSW at 4,112 t/yr. Thereof, 46 % is uncolored, 54 % is colored/white and, in terms of material grade, 30 % is food grade. The most common filling volume for PET, PP and HDPE was 0.5 < x ≤ 1.5 L (23-59 %) and the most common decoration technology was label (60-85 %). PET and PP had the highest shares of food-grade bottles (37-46 %), while PP had the highest share of colored bottles (22-31 %). The mechanical recycling potential of bottles depends largely on packaging characteristics, influencing separate collection and also automatic sorting. Harmonized design specifications are therefore crucial for this heterogeneous PPW fraction.

Separating copper from iron scrap is a critical operation in metal recycling and achieving this with low cost sensoric equipment like RGB cameras instead of XRF/XRT is becoming increasingly attractive. In this article, the groundwork for creating an image classification model to separate copper from iron scrap has been performed. Twenty of the most common and most easily available CNN architectures were trained on 2200 metal scrap specimens and evaluated inline on a sensor-based sorting rig for their prediction accuracy and their inference latency to mimic real circumstances in an industrial setting. Out of these evaluated architectures, DenseNet-201 with 98% accuracy in inline tests is recommended if potent hardware is available. Otherwise AlexNet with 92% accuracy or MobileNet-V2 with 90% accuracy are recommended for further investigation and model creation if hardware restrictions apply. Based on the presented results in this article, the initial cumbersome surveyance of the most suitable network architecture can be substantially reduced and the creation of a sorting model can be streamlined. This article thus provides the basis for creating an inline applicable sorting method for scrap metal that uses low cost sensorics equipment and can provide reasonably high accuracy in its prediction.

Zusammenfassung: Die Textil- und Bekleidungsindustrie wird in Zukunft im Sinne einer Kreislaufwirtschaft zu gestalten sein. Für Textilien stehen bereits unterschiedliche Recyclingverfahren zur Verfügung, wenngleich die Kapazitäten in Europa derzeit noch gering sind. Für alle Recyclingverfahren sind jedoch textile Inputströme erforderlich, die bestimmte Spezifikationen (z. B. Zusammensetzung, Störstoffgehalt, Farbe) erfüllen müssen. Für die Sortierung von Alttextilien (Post Industrial, Pre Consumer, Post Consumer) ergeben sich daraus eine Reihe von Herausforderungen (z. B. Textilprodukte sind extrem vielseitig und individuell gestaltet, verfügen über mehrere Lagen, bestehen aus einem Gemisch verschiedenster Fasertypen und/oder enthalten eine Vielzahl von Störstoffen). Manuelle Sortierung allein kann dieser Aufgabe nicht gerecht werden, da die Recyclingindustrie Informationen benötigt, die über die Angaben auf Pflegeetiketten hinausgehen und diese Sortierung personal- und kostenintensiv ist. Zum jetzigen Zeitpunkt spielt die manuelle Sortierung eine tragende Rolle und wird – gemeinsam mit der Sammlung – in den nächsten Jahren eine wichtige Unterstützungsfunktion der automatisierten Sortierung übernehmen (müssen). Sensorgestützte – und damit automatisierte – Sortierung kann Merkmale und Eigenschaften erfassen, die sortierendem Personal aus unterschiedlichsten Gründen nicht zugänglich sind. Einige wenige automatisierte Sortieranlagen auf Basis von NIR wurden in Europa bereits umgesetzt. Technologien wie RFID, NFC, Fluoreszenzmarker oder Digitale Wasserzeichen erfordern stets auch die flächendeckende Einbindung der textilherstellenden Unternehmen und können – wenn überhaupt – erst in Zukunft eine Rolle spielen. Auch die Objekterkennung mithilfe von Künstlicher Intelligenz getriebenen Algorithmen und Modellen stellt noch eine Zukunftstechnologie der automatisierten Sortierung dar. Die Verwendung einer einzigen Technologie zur Sortierung von Textilien wird und kann in Zukunft nicht ausreichend sein, um den vielfältigen Sortieraufgabenstellungen in diesem Bereich zu begegnen.

Improving the sortability of plastic packaging film waste (PPFW) is crucial for increasing the recycling rate in Austria as they account for 150,000 t of the annually produced 300,000 t of plastic packaging waste. Currently PPFW is thermally recovered, as it is impossible to separate the mechanically recyclable monomaterial films from the non mechanically-recyclable multimaterial films. In this study, machine learning models capable of classifying inline into monolayer and multilayer films of PPFW according to their spectral fingerprint taken in transflection were created. Feature selection methods, like PCA and MRMR F-Tests, identified the most relevant spectral ranges for classification, that show the least redundancy and highest relevance. This effective subset of features decreases the required complexity of the model while reducing prediction time without compromising accuracy. The resulting models achieved a prediction accuracy of 85 % on unseen specimens with minimal prediction latency, effectively showing the inline applicability of these models in sorting aggregates.

Proper waste sorting is crucial for biodegradable plastics (BDPs) recycling, whose global production is increasing dynamically. BDPs can be sorted using near-infrared (NIR) sorting, but little research is available about the effect of surface contamination on their NIR spectrum, which affects their sortability. As BDPs are often heavily contaminated with food waste, understanding the effect of surface contamination is necessary. This paper reports on a study on the influence of artificially induced surface contamination using food waste and contamination from packaging waste, biowaste, and residual waste on the BDP spectra. In artificially contaminated samples, the absorption bands (ADs) changed due to the presence of moisture (1352–1424 nm) and fatty acids (1223 nm). In real-world contaminated samples, biowaste samples were most affected by contamination followed by residual waste, both having altered ADs at 1352–1424 nm (moisture). The packaging waste-contaminated sample spectra closely followed those of clean and washed samples, with a change in the intensity of ADs. Accordingly, two approaches could be followed in sorting: (i) affected wavelength ranges could be omitted, or (ii) contaminated samples could be used for optimizing the NIR database. Thus, surface contamination affected the spectra, and knowing the wavelength ranges containing this effect could be used to optimize the NIR database and improve BDP sorting.