Frank van de Winkel is Business Development Manager at TOMRA Recycling’s Metals sorting division. With a career starting in 2005 at CommoDaS, where he progressed from applications development engineer to test-centre manager, he transitioned into technical sales support after TOMRA acquired the company. He later took charge of sales in North America and Australia and now applies his extensive expertise in metals recycling to drive new markets, segments and applications globally.

In the e‑Magazine “Advanced Industrial Technologies in the Aluminium Industry (Part II)”, Frank van de Winkel highlights TOMRA’s pivotal role in enhancing metals sorting efficiency and sustainability, a key focus as the aluminium sector seeks greener, more digitised processing solutions.

AL Circle: How have recent advancements in sensor-based sorting technologies transformed the efficiency and quality of aluminium recycling, particularly in distinguishing between different aluminium alloys or coatings?

Frank van de Winkel: Recent advancements in sensor-based sorting technologies have dramatically transformed the efficiency and quality of aluminium recycling. A key innovation in this field is Dynamic LIBS (LaserInduced Breakdown Spectroscopy) technology, featured in TOMRA’s AUTOSORT™ PULSE. Unlike static LIBS systems, Dynamic LIBS leverages sophisticated 3D laser measurements and AI algorithms for optimal laser targeting. Its unique single-point focus mode allows the laser to repeatedly target the exact same spot on a moving object. This enhances material penetration and generates significantly more spectral data, resulting in the highest possible sorting accuracy, even for items with specific coatings, complex 3D shapes or surface irregularities where single-pass systems historically struggled. This precision directly translates to superior sorting outcomes. For example, with mixed 5xxx and 6xxx stamping scrap, the AUTOSORT™ PULSE has achieved impressive recovery and purity rates exceeding 95% in a single step. For more complex streams, such as Taint Tabor or mixed wrought fractions from Twitch, purity levels can surpass 97% for 6xxx, 5xxx or 3xxx series alloys, and even specific alloys like 6063, while maintaining recovery rates over 95%, reaching as high as 97-98% for 4xxx/cast alloys. This unparalleled accuracy in distinguishing alloys and handling challenging coatings vastly improves the quality and value of recycled aluminium.

AL Circle: What role is artificial intelligence (AI) playing in the next generation of aluminium sorting systems, and how is it improving material recovery rates and reducing contamination?

Frank van de Winkel: Artificial intelligence (AI), specifically deep learning, plays a crucial role in the next generation of aluminium sorting systems. Deep learning algorithms are integrated into systems like TOMRA’s AUTOSORT™ PULSE, to significantly enhance the system’s ability to identify the shape and position of items on the conveyor belt. The AI-based object singulation function, for instance, can accurately determine whether multiple overlapping pieces form a single object or multiple ones, improving single object detection beyond traditional processing technology. This enables higher belt occupancy and faster processing speeds while maintaining exceptional sorting purity. As a subcategory of AI, deep learning imitates the human brain’s information processing. It uses artificial neural networks trained on vast amounts of data to recognise and store complex patterns. This enables deep learning to solve highly intricate sorting tasks, leading to higher sorting granularity and a significant reduction in manual labour. For example, TOMRA’s GAINnext™ system, a deep learning-based solution, is already demonstrating its value in removing unwanted objects and materials from used beverage can (UBC) fractions, and new applications are continually being developed to address complex sorting needs in the metals sector that currently rely on manual intervention. By accurately identifying and separating materials, AI significantly improves material recovery rates and reduces contamination.

AL Circle: Can you shed light on how digitalisation and material traceability technologies are influencing closed loop recycling systems and supply chain transparency in the aluminium sector?

Frank van de Winkel: Digitalisation, particularly through the application of deep learning, is fundamentally influencing closed loop recycling systems and supply chain transparency. Deep learning, as discussed in Q2, relies on feeding vast amounts of digital visual data into neural networks to recognise and store patterns, thereby enhancing sorting capabilities to the purity levels needed for true closed-loop recycling. The success story of Gerhard Lang Recycling GmbH, detailed in Q3, demonstrates how these digital technologies enable access to previously untapped scrap sources and create new markets by sorting stamping scrap into distinct, high purity 5xxx and 6xxx products. By preventing downcycling and preserving material value, AUTOSORT™ PULSE directly supports full material circularity and a less carbon intensive aluminium supply chain.

To explore the full interview and gain deeper insights into the global aluminium recycling technologies and techniques, click here.