Smart Waste: how AI can increase the recycling rate of plastics

Nine percent – that is the proportion of all plastic waste that was recycled from 1950 to 2015. This must change urgently. Artificial intelligence can help. EREMA and the Software Competence Center Hagenberg (SCCH) are researching exactly how in a joint project.

In an interview with K-MAG, Dr Sophie Pachner talks about why so little is recycled and what potential AI holds in plastics recycling.

Dr Pachner, of 6,300 million tonnes of plastic waste generated from 1950 to 2015, only 570 million tonnes were recycled. Why is the recycling rate so low? And how can we change that?

Dr Sophie Pachner: The challenge with plastics recycling is that the incoming material stream is very heterogeneous in its composition – shape, degree of contamination, etc. – but in the end the recyclates have to be of a consistently high quality so that they can be reused at all. In order to be able to produce a high-quality recyclate, not only precise waste sorting is required, but also flexible adaptation of the recycling processes to changing material flows.

However, these recycling processes are very complex: the feedstock is sorted, crushed, washed, prepared, extruded, degassed, filtered and processed into regranulate. Artificial intelligence can help us optimise these processes.

To what extent can AI optimise plastics recycling?

Pachner: Various companies work together along the value chain: Recycling collection points, companies that buy the waste, do the sorting and then the companies that produce the recyclates. A particular challenge in data management is the traceability of material flows along the value chain. However, the problem with cross-company data analysis is often that the companies do not want to disclose the data.

Here, universities and competence centres are developing privacy-preserving methods for data collection in order to obtain a holistic view of value chains in the future without having to exchange data across company boundaries. With the help of AI, waste will be networked and thus become "smart waste".

Artificial intelligence recognises patterns in production data, can warn in the event of anomalies, develop forecasting models, thus providing decision support for the customer and ultimately ensuring consistent product quality.

Your project partner is the Software Competence Center Hagenberg (SCCH). What expertise do you each bring to the project?

Pachner: Our innovative recycling solutions for different requirements make us the world market leader in plastics recycling machines and system components. There are currently around 8,000 EREMA machines in use, which together produce more than 21 million tonnes of high-quality granulate every year. In this way, we are making a significant contribution to the Circular Economy.

SCCH contributes its expertise in the areas of data integration, knowledge extraction, process modelling and process optimisation. Machine learning methods are used to analyse the data. SCCH has developed a dashboard to visualise and analyse the process data. Here, it contributes its know-how of automated pattern recognition and analysis of complex correlations in the field of data science, as well as its many years of experience of machine learning methods for the analysis of process data.

AI in the plastics industry – what else could be possible in the future?

Pachner: There is definitely still a lot of potential in our industry to improve processes, products and services with the help of AI technologies. There are a multitude of possibilities along the value chain, from waste collection and sorting to extrusion and post-treatment of the re-granulate. Here, it is primarily a matter of making good use of AI and data analyses for more resource protection and recycling.

EREMA and SCCH are currently project partners in the lead project circPLAST-mr, which deals with the mechanical recycling of plastics, and in the CHASE project, which deals with the chemical process industry.