Modelling of the Water Cycle in a Paper Industry using Artificial Intelligence

Implementation period

September 2020 – March 2021

Scope of work

Digital technology


H2ï Analytics





This project has been funded by the national programme for clusters support “Ayudas a Agrupaciones Empresariales Innovadoras” (AEIs).


AIPIA has validated the modelling of water management processes with Artificial Intelligence. Throughout the project, treatments of both the water intake plant and the discharge treatment have been modelled and, predictions of the behaviour of the processes have been obtained up to 5 days in advance, which allows to adjust the processes efficiently and to work with higher stability than with the current systems based on real-time alarms. AIPIA has laid the foundations to control processes thanks to prediction in a relatively simple way as long as there is some digitisation. This control allows the industry to manage its water better and reduce consumption, opening the way to Water Sustainability.

The main challenge of the project was implementing Artificial Intelligence (AI) technology for water management in the water-intensive industry. Specifically, to model a paper industry’s water cycle (quality and quantity) using Artificial Intelligence (AI) to optimise the treatment time and e problems that may arise.


Once the treatments have been modelled, the advantages provided by neural networks will make it possible to recognise trends, predictive capacity and decision-making power. With all this information, the operational objectives of the project are to lay the foundations for optimising the management of the water cycle according to:

  • Addition of chemicals.
  • Increasing the useful life of membrane processes: Ultrafiltration and reverse osmosis.
  • Understanding the possibility of recirculating industrial wastewater to treat water entering the factory.


The advantages that Artificial Intelligence brings to management as it is done today are:

  • Real-time results, enabling trend analysis capability.
  • Pattern recognition and risk management.
  • Predictive and anticipatory capabilities.
  • Management and strategic decision support.
  • Optimising input water and wastewater.

The modelling of incoming water quality using public data from the river basin makes it possible to predict the treatment requirements five days in advance, avoiding problems in the treatment of incoming water at the plant. On the other hand, modelling the inlet water chlorination system (WTP) together with the discharge water model enables the reuse of water at the WTP, as it predicts the behaviour of the final inlet water and can therefore adjust flow rates and the optimal addition of chemicals.

Gaining confidence in processes is essential for the industry to embark on the path to reuse, the basis for water sustainability. AIPIA has demonstrated that process control using AI models is feasible and has great potential for application.

After verifying the efficiency of the models developed in a real environment, it can be seen that good digitisation is the basis of artificial intelligence. Those processes simulated with adequate quality and quantity of data have been able to predict the behaviour of the process up to 5 days in advance. Models based on scarce data or little variability are unreliable in their prediction.

Artificial intelligence provides information in advance of process changes occurring, and this advantage over a real-time alarm improves process stability, optimises operation and predicts the evolution of critical equipment. In short, it generates process reliability and opens a pathway to water sustainability for the industry.