How can AI improve industrial wastewater management? At Sappi's Carmignano mill, an AI-powered forecasting model predicts wastewater quality changes hours before they occur, enabling faster intervention, improved environmental compliance and greater operational efficiency. The initiative highlights how digital transformation and machine learning are helping manufacturers enhance sustainability, optimise processes and drive continuous improvement across their operations.
Sappi’s mills measure the quality of water that they discharge back into the environment very carefully. This is because papermaking relies on a constantly changing mix of chemicals including coatings, polymers and carbonates, depending on which paper grades are running through the machines on any given day.
Any unintentional chemical imbalance can affect the composition of the resulting effluent – making careful wastewater management a key part of Sappi's environmental efforts. Dedicated wastewater treatment systems follow strict operational controls to maintain environmental requirements.
Sappi’s Carmignano mill decided to take a more proactive approach to wastewater management by forecasting potential deviations in water quality parameters hours in advance, so enabling the team to act sooner and optimise process performance before intervention became necessary.
The aim was not only to continue to improve the mill’s operational excellence and environmental stewardship, but also to benefit customers and partners directly. By taking a technologically advanced approach to improving the chemical balance of the water used by the mill, it would be possible to improve production reliability and save costs – as well as embed new approaches that could lead to further results-focused mill improvements.
The project serves as a prime example of how Sappi is applying AI and advanced analytics to support smarter, more proactive decision making across the company’s manufacturing operations.
A tool to mirror the mill, not just the data
Like other mills, Carmignano measures the levels of total organic carbon (TOC) in its treated effluent water as part of its wastewater management and environmental compliance processes. TOC is a key water quality parameter used to assess treatment plant performance and ensure that discharge limits are consistently met.
When trends point to higher levels of TOC, the mill can take precautionary steps to optimise wastewater treatment performance and maintain compliance with discharge requirements. In some cases, this may require operational adjustments, which can affect production efficiency.
That’s why, back in 2024, the Carmignano leadership team launched a project to develop a forecasting tool. Built internally, the solution would be designed to accurately reflect the complexity of the mill's wastewater system, mapping its pathways and accounting for the interactions and dependencies that influence TOC levels.
The new tool was developed by the digital transformation team along with the Process Quality Management (PQM) wastewater plant foreman and shift supervisors. Together, they shared their operational knowledge of the plant's pathways and helped shape how the model was structured.
The development team then drew on a variety of AI and machine learning technologies to put together a tool that worked specifically for Carmignano. The best part of a year was spent developing, building and implementing the project leading up to the system being monitored and then integrated into the mill’s production process.
The successful realisation of the project reflects how Sappi combines advanced data analytics with deep operational expertise – how digital transformation can be used to solve complex manufacturing challenges.
A four-hour head start
Today, the model can flag a potential TOC deviation up to four hours before it occurs – giving the plant foreman time to act while also reducing the need for manual monitoring.
Twice a day, the foreman receives a report setting out current and predicted TOC levels alongside the key features driving them, information he can then weigh against his own experience to adjust chemical dosing, tubing flow rates and the plant's wider operating parameters.
Importantly, the same data is also helping to unpick why quality issues arise in the first place, allowing the foreman to cross-reference a water quality reading against earlier process changes, such as a pump replacement or a tube adjustment.
The AI’s additional advantages
Although predicting TOC levels was the project’s initial goal, the new model has wider benefits. It can also forecast turbidity (how murky the water is due to suspended particles) and is able to give early indications of phosphorus and nitrogen content – both also relevant to the broader question of water quality.
No surprise that other Sappi mills, including Condino in northern Italy, are already exploring the possibility of implementing the model for their own wastewater forecasting.
And the next step?
The team suggests that the same forecasting logic can be applied to improving the chemical dosage itself. Such a change would mean a more efficient use of chemicals and even cleaner water effluent – and could lead to further significant operational cost savings.
In this way, the project is a clear example of how AI can be used to unlock practical value across manufacturing processes. And of how, at Sappi, digital transformation is helping to improve environmental performance, operational efficiency and opportunities for innovation throughout the business.
Related articles
Taking action on SDG6: Water resilience
Just as water is essential for a thriving world, it is vital for our business too. Water plays a critical role in our manufacturing processes, and it’s never been more important to safeguard this invaluable resource to ensure its long-term sustainability.
How Sappi Europe uses AI to benefit both mills and customers
A new machine learning operations system is delivering smarter production and improved sustainability metrics – alongside a range of customer benefits
How AI is reinventing productivity for printers under pressure
As margins tighten across Europe’s print industry, automation, digitalisation and AI are becoming essential. Intergraf’s 2025 report and real-world case studies show how printers are streamlining workflows, reducing manual tasks and creating new opportunities for productivity and growth.