3 ways to drive profitable plastic recovery with AI waste analytics

Tiziana Giordano

Tiziana Giordano

Jun 19, 2024

8 min read

AI waste analytics bridge a data gap that has long challenged plastic recycling.

Manual sampling provides detailed insight into waste objects, but typically covers less than 1% of a facility's material. Optical sorters, which use near-infrared (NIR) sensors to identify polymer types, struggle to identify sleeved packaging, black plastics, food-grade objects, and composite packaging. Much of that valuable material is lost to landfills and incinerators as a result.

AI waste analytics complements existing technology and manual processes by identifying waste like a human: visually. By automating that process, AI provides real-time composition data for the remaining 99% of material missed by manual samplers.

How AI waste analytics tracks plant profitability

The profitability of a plastics recovery facility (PRF) hinges on a delicate balance of throughput rates, product purity, and minimising the loss of valuable material to residue.

By gathering real-time data on material composition, AI waste analytics provide instant visibility into a facility’s key revenue-driving performance indicators:

  • Plant availability
  • Production throughput
  • Sorting yield
  • Loss rate
  • Product purity

Facility operators are leveraging these insights to take three key actions that help them drive profitability:

  1. Optimising throughput rates and sorting yield
  2. Maximising plant capacity and reduce losses
  3. Accurately monitoring final product quality


1. Optimise throughput rates and sorting yield

GreyParrot-facility-monitoring-IFAT-newsroom (1)-1

Increasing a PRF’s plant availability and throughput rate drives profit. Both contribute to a higher sorting yield (the amount of value a plant captures from the total material it processes) which is a measure of a PRF’s most essential function: resource recovery.

When AI monitoring units like Greyparrot Analyzer are deployed throughout a PRF, facility operators gain access to live insight into plant availability, throughput rates, and waste stream composition — from infeed to final product outfeed.

Detailed waste composition data on infeed and outfeed lines allow operators to track how much value they extract from the material they receive: by comparing the amount of valuable material on the infeed belt with the amount captured on product lines, giving operators a clear picture of yield.

With AI waste analytics, they don’t even have to make that calculation themselves: instead, systems like Greyparrot Analyzer do it automatically. Analyzer’s Facility Dashboard provides a unified view of a plant’s KPIs — on a single screen. By transforming raw data into actionable insights, it enables operators to adapt faster, reaching new levels of efficiency and profitability through lean management principles.

How a real-world PRF uses AI to increase yield

One PRF in the UK installed an Analyzer on their infeed belt to measure the amount of valuable material entering their facility, providing a baseline reading of the maximum possible yield. They also installed three Analyzers on key product lines (PET, HDPE, and aluminium) to gather accurate data on the amount of material they recovered at the end of the sorting process.

Using AI waste analytics, the facility measured the yield of these three materials, comparing the number of valuable items at the infeed stage with the number of items successfully separated. This constant monitoring allowed operators to track performance, and assess how changes to infeed blends and throughput rates affected yield.

They then used that information to adapt their operations to specific suppliers, modulate throughput rates, and tweak infeed blends to maximise the amount of value they recovered.

“Greyparrot allows us to gather a lot of valuable information from our sorting facilities and enhance our knowledge of packaging, sorting technologies and the sorting yield of the installation.”

Vincent Mooij, Director of SUEZ.circpack®, now part of Veolia Group


2. Maximise plant capacity and reduce losses

GreyParrot - Web Quality Still-12 (2)

Another way to increase sorting yield — and therefore profits — is by maximising plant capacity while reducing losses. PRF residue lines are a hidden source of lost value, and operators that monitor residue belts alongside infeed and product lines are effectively giving their facilities a “blood test” — the more valuable materials in residue, the more attention sorting processes need to increase yield.

How a real-world PRF uses AI to maximise capacity and reduce losses

Another of the UK’s leading PRFs was relying on visual checks to blend infeed material, resulting in inconsistent composition that overburdened sorting machinery, and resulted in large amounts of valuable material ending up on residue lines.

To diagnose the challenge, they installed three Analyzer units at key points: infeed, residue, and product outfeed lines. Using the Analyzer portal, they reviewed live data that revealed how changes in infeed materials directly affected output purity and residue. Over five years, this inefficiency was causing in an £8.2 million loss for the facility.

The data also revealed that a balanced infeed blend maximised the capacity of their existing sorting machines — and resulted in 18% less valuable material on their residue line. Their operators were able to control fluctuations and reduce future losses in response, and make significant savings.

With Analyzer units deployed across their facility, their team are now able to continuously monitor the entire sorting operation from the control room, making frequent adjustments to maintain stable throughput.

“There’s no obligation in the MRF code of practice [in the UK] to sample residue lines; which is some of the most important sampling when it comes to plant performance. A consistent flow of data across the entire facility would pay dividends when it comes to identifying areas for optimisation.”

Jonathan Caesar, Senior Technical Plant Engineer at Suez UK


3. Accurately monitor final product quality

GreyParrot - Web Quality Still-1_compressed

PRFs that can provide accurate data on product purity often command higher prices for their material, but that extra profit can be lost due to the high cost of manual quality control checks.

AI waste analytics is a scalable, cost-effective alternative that meets the accuracy requirements of recyclers. Systems like Greyparrot Analyzer can also differentiate between food-grade and non-food grade materials, helping operators remain compliant with stringent regulation, such as by the European Food Safety Authority (EFSA).

How a real-world PRF uses AI to track product purity

With Analyzers, operators can monitor purity in real-time using live alerts that flag when purity dips below pre-selected quality thresholds.

A large waste management company in Europe relies on stringent purity levels of 98% for their final plastic output streams. Their team use alerts to take immediate action when quality drops, and avoid ruining an entire bale.

When a recent alert indicated a drop in purity, they stopped the plant. By diagnosing the issue with performance data in the Analyzer portal and acting to fix it, they had the plant up and running again within 12 minutes. Without the alerts, they would have contaminated the entire day’s production — and only discovered the issue through manual quality inspections at a much later stage.

How AI fills the food-grade reporting gap left by NIR and digital watermarks


Another large waste management company in the UK successfully uses AI waste analytics to report food-grade purity data to EFSA, ensuring the compliance and high standards of their output materials.

In the UK and Europe, recycled food-grade material must meet EFSA’s stringent regulatory standards: no more than 5% of the plastic waste used for recycling can come from non-food grade objects. Differentiating between food-grade and non-food-grade materials is therefore crucial — but remains a challenge for existing tools and processes.

While NIR technology provides valuable polymer identification, it lacks the cognitive ability to differentiate between food-grade and non-food-grade items. AI waste analytics fills that critical gap by analysing visual information 62 times faster — and 250 times cheaper — than manual sorters.

“Fifteen years ago we adopted NIR technology, the best tech at the time. Requirements have now changed, particularly when it comes to distinguishing between food and non-food grade material.

AI bridges the gap that NIR can’t, and tells us what waste objects were used for. It completely expands what we can do in terms of sorting.”


Miguel Rosa, Technical Manager at Viridor

Digital watermarking systems like HolyGrail 2.0 offer detailed data on the lifecycle of specific products, helping trace their use in food- or non-food applications. Like manual sorters, though, they are expensive to implement on a large scale. While digital watermarking remains valuable for specific use-cases, AI achieves similar results at a lower cost and larger scale.

By leveraging AI waste analytics to track product purity, facilities can ensure regulatory compliance, maximise operational efficiency, and avoid jeopardising profitability with costly manual quality control solutions.


Why choose Greyparrot’s waste analytics?

GreyParrot - Web Quality Still-28_compressed

In 2023 alone, Greyparrot Analyzers tracked over 25 billion waste objects across 20 countries, with companies like Veolia, SUEZ, KSI, A2A, Biffa, Viridor, Berry Global and more relying on our data. That same year, our global data showed that over a quarter of the average residue line was recoverable plastic — a huge opportunity cost for the sector, and the environment.

Instead of creating robotic solutions, we’ve focused our efforts on developing the most accurate AI waste recognition possible. As a result, we’ve solved some of our customers’ most complex plastic monitoring challenges, with a growing recognition library including 43 kinds of hard plastics and 9 categories of flexibles. Greyparrot Analyzer can track and report on materials that present a challenge to technology like NIR (including black plastics and flexible films) while accurately identifying each object’s food-grade status. To help them turn that insight into action, we’ve developed the sector’s most intuitive waste data dashboards, enabling operators to track an entire facility’s operations and KPIs from a single screen.

PRF operators and recyclers around the world are using those features to improve sorting efficiency, product quality, and the broader economics of plastics recycling.

If you’re ready to start introducing AI waste analytics to your plastics recovery facility, you can learn how to choose the right provider in just 4 steps, or book a consultation with our team.


Keep up to date with AI waste analytics and circular economy events with our monthly newsletter.