25.05.2021. Outokumpu has been working to digitalise its biggest plant in Tornio, Finland into the most digitalised and most cost-competitive stainless steel operation in the industry. In a partnership with Microsoft, the digitalisation has progressed as planned. Their overall targets are to improve reliability, quality and supply chain management, and to gain up freed capacity.
Outokumpu is leveraging the power of analytics and machine learning (ML) to monitor and improve production from end to end. First, they developed the Outokumpu Digital Platform in the cloud to combine data siloes from their Tornio operations. The stored data is used for example for quality optimisation, quality tracking, efficiency improvement, defect sources identification, and best practice comparisons.
At the same time, Outokumpu installed in parts of their production lines internet-of-things (IoT) sensors. They are capable of tracking variables such as temperature and location. These sensors feed the data to machine learning models that recommend actions to machinery operators on how to optimise timing, temperature, and other controls. They have also installed for example surface inspection cameras in certain lines, feeding quality data into the systems and detecting surface defects which enable line operators to take immediate action. This improves the quality, resulting better yield and eventually better customer satisfaction.
Empowering employees with machine learning
“Our operators are absolutely needed with all the experience they have,” says Niklas Wass, Head of Production in Europe at Outokumpu. “What we’re adding is the data scientist and the programmers and all the brain power of the cloud system. The combination of process knowledge, computer power and computing knowledge, and data science is bringing us into a new level of the game.”
The idea in machine learning is to empower employees. When the machine model identifies a potential action, it recommends that to an operator, rather than triggering an automatic action. For the operators, the key is that they can improve their own performance. This implies making better decisions that enable higher output and lower costs. It is especially helpful for those that are new to the industry.
Outokumpu is already getting benefits from machine learning. They have been able to free up significant amount of capacity, and for certain processes, output has increased by 10 to 15 % and quality defects are down as much as 40 %.
Much greener process
Machine learning is one step towards their goal of becoming carbon neutral by 2050: it helps lower electricity usage, leading to reduced CO2 emissions. It can also help to improve the quality of the production, resulting in better yield.
“We owe it to our families – especially to our children,” says CTO Stefan Erdmann. “We want our aspirations to have a legacy, to leave something behind better than what it was before. And I think with digitalisation, the biggest opportunity that we have in front of us is to increase the efficiency in such a way that we will leave a much greener process behind.”
During 2021, the digitalisation will move on to other Outokumpu sites internationally, but Tornio will remain as our “incubation site” for new innovations also in the future.