The chain is long, and every link in the chain is crucial to the whole: the supply chains for plastic packaging offer numerous opportunities to make improvements - and ultimately get closer to the ultimate goal of a circular economy.
Artificial intelligence finds an ideal playground here. Above all, its outstanding ability to handle big data, which humans are unable to handle in large quantities, makes it a beacon of hope in plastics recycling.
But the possibilities, says Pacoon Managing Director Peter Désilets, are far from exhausted: "AI is already being used for machine control in many production plants, but the opportunities for utilisation extend across the entire supply chain."
Twelve points where the AI lever can be applied.
1. raw materials: moving towards a circular economy with new materials
Even if the name suggests otherwise, Gnome is a giant. Last year, DeepMind's AI tool identified the structure of over two million new crystals, of which around 380,000 could be produced in the laboratory and are stable. According to DeepMind, one goal is to discover materials that enable more environmentally friendly technologies. Gnome has not discovered any plastics, but the message is clear: AI has the potential to research and simulate material properties before production and thus revolutionise material development.
Two years earlier, the US-based Berkeley Lab had caused quite a stir. The researchers had developed a plastic that could, at least in theory, be recycled an infinite number of times.
Parallel to the research on virgin material, artificial intelligence is also being applied to the recyclates themselves. In particular, the funding project with the innovation labs K3Icycling and KIOptiPack is highly active in this area. Around 40 partners from research and industry (including Pacoon) have joined forces to develop practice-ready, AI-supported tools that enable the product design and production of plastic packaging with a high proportion of recycled materials. As the use of recyclates is made more difficult by fluctuating properties or the lack of chemically high-purity batches, among other things, KIOptiPack starts at the product design stage. Building on this, the research looks at different processes that build on each other, from production to extrusion and moulding. Energy and sustainability aspects are also taken into account and consumer acceptance is analysed.
2. extrusion: AI controls the mixture
KIOptiPack is not alone in the field of extrusion either. The Digit Rubber research project of the German Institute of Rubber Technology, for example, uses artificial intelligence in the extrusion of rubber profiles. The AI analyses the rubber compound data in real time and intervenes in the event of deviations from the reference compound: The system then automatically adjusts the process parameters to keep the compound within the specification range. At the same time, the iProfilControl inspection system from Pixargus monitors the extrudate surfaces in real time with AI support.
With Detact, Symate has developed an AI that collects, analyses and processes production and quality data from ongoing processes. Detact is already being used in extrusion processes and makes them significantly more controllable.
The pacoon team has also looked into the possibilities of using AI. What is astonishing is that it can be used in machine control and the calculation of new variants at relatively short notice. This means that production can be optimised after just a few months. A pleasant side effect: New answers are also possible to the challenge of the shortage of skilled labour or training.
3rd functions: Simulating new packaging solutions
AI also has an influence on the functional details of plastic packaging. The software can be used to simulate the diffusion values of barriers, for example. The effects of new barrier combinations can also be determined virtually or the barrier values of the packaging can be adapted to different applications.
For example, the Fraunhofer IVV developed an app years ago that calculates how the necessary shelf life can be achieved for different foods based on the materials and thicknesses.
4. compounding: artificial intelligence finds new recipes
AI also demonstrates its ability to process enormous amounts of data in compounding. AI systems can help compounders to develop optimal formulations for plastic compounds by analysing millions of possible formulations and identifying the most promising candidates that have a certain probability of achieving the desired material properties. This means one thing above all: the number of experiments actually carried out is radically reduced. The models take into account the entire history of the material, including the ingredients, processing steps, process parameters and test results.
KIOptiPack and its sister project K3I-Cycling are making valuable contributions in this area: Several project members specialising in AI are jointly developing prediction models for the target compound properties and for film extrusion. Specifically, this includes the creation of digital material data sheets with property predictions.
Artificial intelligence is also working on topics that do not necessarily seem obvious. A self-learning system, for example, is able to assess the odour of a new plastic. And with success: the deviations between the machine and human sensory scores are in some cases smaller than those between different human odour testers.
5. production: automated increase in efficiency
The contribution of AI to the production of the films is not a specific one: production optimisation has been a core competence of artificial intelligence for years, contributing to the efficiency of the entire process by significantly reducing errors and through almost seamless quality control, among other things.
6. construction: optimising the packaging
The impact on the sustainability of a project is greatest at the beginning of a process. Around 80 per cent of the environmental impact of products can be attributed to decisions made in the design phase. AI-supported packaging design helps to optimise the design on several levels.
The self-learning systems create designs that are optimised in terms of wall thickness, structure, stability or material savings in a way that neither human experience nor the finite element method can achieve. AI can be used to virtually install stabilisers, test lamination, adhesives and various openings - and, of course, significantly reduce overpacking. Solutions such as OptimAI, which Pacoon uses together with partners for rigid packaging, also make retrospective design optimisation possible through AI assessments.
AI even opens up new possibilities in market observation and trend analysis as well as in the area of consumer insights: Virtual product tests without haptic packaging save time, material and energy.
7 Printing: AI optimises the job
In printing processes, AI ensures optimisation in both barrier and ink application. After all, colours are an important factor for recyclate quality. Together with an AI-based recyclability assessment, it is now possible to improve and compare recyclate quality based on the amount of ink used, print quantities, number of colours and finishing. AI can also be used to better coordinate print jobs and dramatically speed up processes in areas of lean management, such as those already practised by packaging specialist maag. This leads to fewer rejects and shorter delivery times.
8 Logistics: AI-controlled processes
The fact that logistics is both one of the strongest drivers and one of the most grateful consumers of artificial intelligence also has an impact in this specific supply chain. The entire arsenal also benefits plastics recycling.
The software ensures optimised warehousing and facilitates just-in-time delivery. It decides on the most optimised loading and unloading of vehicles and revolutionises traditional route planning. It also facilitates the integration of new business models such as lorry sharing platforms. Without exception, these are effects that have a direct impact on the ecological footprint.
9 Logistics data: Crucial knowledge for everyone
Another factor is the simplified exchange of logistics data between players in the supply chain. AI facilitates the automated and targeted exchange of detailed data on primary packaging, secondary packaging and load carriers between manufacturers, packers, logistics providers and recipients. Pack sizes and even pallet utilisation can be calculated in seconds in advance, enabling cost savings and CO2 calculations to be quantified.
10. processing: sustainable through the use of AI
The same applies to the processing of packaging materials as to their production. Properties and machine settings optimised by AI also make this step more efficient and therefore more sustainable. Changed material properties such as sealing temperatures, tear and puncture resistance, behaviour on the forming shoulders of films or papers or processing speeds will be 'supplied' in future, thus reducing errors or learning phases during processing. New sealing technologies such as those from watttron already allow high energy savings and material reductions. If these conditions are taken into account in advance in AI, materials can produce completely different results.
11. communication: automation of legal certainty
New and existing legal frameworks require a high level of communication - especially as the relevant laws and standards change frequently. Whether packaging fulfils the legal requirements must therefore be verified again and again. Artificial intelligence makes this communication much easier: the data is compiled and processed as required and can be automatically forwarded to the relevant authorities. Particularly in view of the Green Claim Directive, which must be implemented by March 2026, data can be derived from this that will be taken into account in communication or in the digital product passports.
12th specification: Sustainable up to the last step
And finally, the last step in the supply chain, which could also be the first: specification. AI not only makes it easier to archive existing solutions or find suitable new ones. The systems also take into account specific variations depending on the raw material supplier used and evaluate the material properties of the individual film layers.
The Pacoon vision
And what does this look like at Pacoon? "There are very simple and fast AI tools for integration into the entire supply chain process. We ourselves will soon be increasingly using AI to analyse documents. This will also reduce a lot of manpower and speed up the process. In production, there are tools that have already proven themselves countless times for machine control and production optimisation and that will achieve a positive ROI in just a few months."
But you have to proceed with caution, says Peter Désilets: "The fear of becoming redundant is undeniable. But I see the example of industrialisation. Instead of the feared job losses due to machine production, production costs and sales prices fell so sharply that significantly more goods had to be produced. The staff were still needed because they were used for machine control".
Désilets pleads for a realistic assessment of the opportunities: "In times of a shortage of skilled labour and low production costs in Asia and Turkey, streamlining and cost savings in the supply chain will provide the answer. Let's imagine that, thanks to data linking, possible solutions and their effects on the entire supply chain can be discussed directly at the customer meeting, instead of choosing one of the standard solutions 'after consultation with technology'. That's the future."