HolyGrail 2.0 completes industrial validation in advanced sortation of PET waste

The Digital Watermarks Initiative HolyGrail 2.0 has reached a new milestone with the successful validation of the digital watermark technology’s advanced sortation capability in an industrial environment, delivering positive results particularly in relation to the separation of food and non-food waste streams.

The industrial trials took place at the Wellman/Indorama recycling plant in Verdun, France, using a prototype detection module developed by machine vendor Pellenc ST and digital watermarks technology provider Digimarc, as part of the HolyGrail 2.0 Initiative.

This assessment focused on a PET waste stream, with two primary objectives - to remove watermarked non-food PET bottles from the stream to meet EFSA guidelines i.e to reduce non-food PET content in a food PET stream to less than 5%, and to create a non-food PET output stream.

The industrial validation of the HolyGrail 2.0 technology focuses on PET waste stream.

This industrial-level assessment, carried out in January and February 2023 under the supervision of the HolyGrail 2.0 technical team, built on the success of the previous Initiative’s semi-industrial trials performed in 2021 and 2022.

5.6 tons of watermarked non-food PET bottles were produced, and then mixed with post-consumer waste at material recovery facility SUEZ in Epinal, France, to mimic typical waste workflows. Multiple fractions of mixed waste containing around 200,000 watermarked bottles were created, baled, and shipped to the Wellman/Indorama recycling facility for sorting tests.

The trials were carried out on a fully operational line - equipped with the digital watermark detection module - working in standard operating conditions with nominal throughput of 3 tons/hour and belt speed of 3 meters/second. An NIR unit was also used in combination to assist with blowout.

Detection efficiency was 92.1%, while sorting efficiency was 88.3%, on average. For two-pass sorting, the detection and sorting efficiencies were 95.9% and 95.1% respectively.

Detection efficiency is 92.1% and sorting efficiency is 88.3%, on average.

These results yet again validate the efficacy of digital watermarks in separating with high granularity – in this case food vs. non-food separation. The high efficiencies indicate that use of digital watermarks can reduce impurities in food-grade PET output streams in recycling plants.

Additionally, it was found that even a single sorting pass can provide meaningful benefits: EFSA guidelines were always achieved in only one sorting pass, and that a non-food PET stream can also be created for circular use. Two-pass sorting, which is standard process in such recycling plants, offers further capabilities both for the food-grade stream and the non-food stream.

The results were very consistent and repeatable on each stream (standard deviations were around 1%). Overall, the high consistency of sorting behavior, together with reliability and continuity of the detection module in industrial conditions for almost two months, without interruption, were achieved.

The module will now be shipped to MRF Hündgen in Germany, where further industrial trials on different post-consumer waste material streams will be conducted in the second half of 2023.