Improving paper production quality with machine learning

The future of the paper production process is data-driven. Together with our client Kemira, we developed a machine learning-driven solution to improve production quality by predicting and preventing issues real-time, giving the human operators at-a-glance view to current situation and hours ahead. Kemira is a global chemicals company serving customers in the pulp and paper industry.

First steps for bringing AI to improve paper production quality

As a result of our collaboration, Kemira is now using an explainable AI solution to predict and prevent issues with paper and board quality and paper machine runnability. Our AI experts built a machine learning-driven solution that provides real-time predictions to a status dashboard, which is today in production and used as part of Kemira’s KemConnect™ Predictive Wet End digital service.

Chemical expertise turned into a digital service

Chemistry plays a key role in the stability of the papermaking process. With the help of the machine learning engine, Kemira is able to turn our chemistry expertise into a new digital service.

The service allows Kemira's customers to see all the likely or unlikely events in the process health that impact the quality of the product, hours into the future. It also helps them to understand where the disturbances are originating from, making it possible to prevent e.g. off-spec production.

Results

• Tested various machine learning models in the pilot phase
• Selected the best performing machine learning model that was put to production as part of Kemira's KemConnect dashboard
• Enables predicting and preventing issues real-time, giving the operators visibility into the current situation and the production hours ahead.

Read the press release for more information.