Smart recycling of packaging waste

Plastics have long been seen as a valuable raw material that can be used again following its initial use thanks to a sophisticated recycling process – but the potential of plastics recycling has yet to be fully exploited. "It’s a real shame that, according to the Federal Environmental Agency, over 60 percent of plastic waste still ends up in thermal waste treatment," says Alexander Ennen, Head of Department at the Development Center for X-Ray Technology of Fraunhofer IIS and coordinator of the K3I-Cycling project.

Broad network for the plastics recycling of the future

K3I-Cycling hopes to improve the recycling rate of plastic waste and deliver greater sustainability. "We’re seeking to optimize the recycling of plastic packaging using artificial intelligence," says Ennen. The project name, K3I-Cycling, is pronounced "Key-Cycling," with the number three referring to three key conceptual properties and technologies: artificial intelligence, recirculation, and plastic packaging. The aim is to increase the mechanical recycling rate being implemented on a pilot basis at a partner company of the consortium. As part of the funding measure "AI Application Hub on Plastic Packaging – Sustainable Circular Economy through Artificial Intelligence," the Federal Ministry of Education and Research (BMBF) is promoting these developments at two innovation labs with a budget of some 13.8 million euros each. The KIOptiPack lab deals with the design and production of plastic packaging, while K3I-Cycling focuses on closing the cycle through the collection, logistics, sorting, separation, and processing of plastics. Implementation work by the two labs is overseen by Project Management Jülich.

Ennen knows that it is essential to take an holistic view of the recycling process. After disposal, the waste arrives at light packaging plants, where the materials are automatically separated and sorted according to the type of plastic. That is why it is so important to completely remove the metal foil from the yogurt pot – otherwise, it is not possible to recycle. What emerges at the end of the recycling process is a reusable, recycled material.

K3I-Cycling brings together AI and digital twins

As part of this project, we at the Development Center for X-Ray Technology of Fraunhofer IIS are creating a platform for digitally modeling the various process steps. Here, the concepts of artificial intelligence and digital twins come together to produce an "Artificial Neural Twin" (ANT). This ANT creates an AI-based, self-regulating process that works like a control loop, allowing us to optimize processes based on the data obtained while simultaneously maintaining the system’s efficiency. Ennen gives an example: If the company decides or external guidelines stipulate that the recycling rate must be increased, the AI indicates which process steps are most affected by the decision. Overall, particular emphasis is placed on data security. "Everything that’s calculated or used goes on locally at the company," says Ennen. When it comes to communication with downstream or upstream process steps, only abstract values are used to model the processes with the ANT. This means that none of the actors or processes need to disclose internal data such as process parameters. Another focus topic is sensor-based sorting. Here, a sensor line is being established and tested as part of the project with a view to using the data obtained as a reference dataset for sensor-based sorting of light packaging.

The BMBF, Fraunhofer IIS, and a total of 20 partners are therefore well on the way to improving sustainability in the recycling of plastic packaging. "Development is at an advanced stage," says Ennen. The project runs until September 2025.