WOODDEE
View original picture , opens in new tab/window

Efficient wood drying with AI

Published: 17 June 2020

Wood drying is the most energy demanding process in wood processing. By using automation and digitalisation – machine learning and AI – researchers at Luleå University of Technology will try to increase the energy and resource efficiency in the process.

All wood must dry before it can be processed and different varieties of kiln drying are used for this purpose. Wood drying is an energy-intensive process in the wood processing chain, and the drying process accounts for up to 90 percent of the total production energy. Modern sawmills can use and monitor up to 50 kilns with more than 100 schedules simultaneously. Tools are required for regulating the drying process, each individual kiln and the kilns interaction.

It is in the area of optimising control tools that the researchers, within the framework of the project WOODDEE, hope to propose new methods that further reduce energy consumption and increase the efficiency of wood kilns. Alent Dynamic, project partner to Luleå University of Technology, has in close cooperation with Stenvalls trä developed a new pumping technique that has revolutionized the wood drying industry. With this technology, new opportunities are created to further optimise the process.

– There are many reasons for energy losses in the drying process, for example, hidden faults on the dryers, inefficient schedules, inaccurate drying and uneven heat dissipation, says Wolfgang Birk, professor of Automatic Control at Luleå University of Technology.

– The purpose of our project is to develop new methods and tools that identify the flexibility of energy use that exists in the drying process, but also in an entire drying plant. We believe that there are many solutions that can demonstrate the potential that we believe is locked into the process.

Digitization provides new tools

In recent years, digitisation has been an important part of the sawmill industry and has created conditions for compiling and storing data. Most sawmills collect data from the various stages of the drying process. By using machine learning and AI together with domain knowledge the project will result in self-correcting tools that utilize the data stream from the process in real time. The tool base that will be used for realisation is a digital twin for monitoring, forecasting and control, which in turn allows for overall optimisation.

– We believe that many of the sawmill's problems are reflected in this process data, both in terms of drying but also subsequent processes. A digital twin with a high degree of autonomy is a tool that reflects the behavior of the process and thus detects problems and improves energy and resource efficiency. In addition, the idea is that the digital twins will also be able to interact and learn from each other, says Wolfgang Birk.

Eric Björkman at Alent Dynamic, shares Wolfgang Birk's expectations.

– We expect to achieve even higher performance and precision in the drying process and also be able to optimise the drying against quality requirements for different end products for the wood, says Eric Björkman.

– The project will make our control system self-adjusting and our pump drying can be more easily adapted to other types of wood and thus have a huge impact within the framework of Agenda 2030. Alent's and the researchers' joint work can contribute to several of the goals, such as sustainable industry, innovations and infrastructure, sustainable cities and communities, and a global partnership.

Contact

Wolfgang Birk

Wolfgang Birk, Professor

Phone: +46 (0)920 491965
Organisation: Control Engineering, Signals and Systems, Department of Computer Science, Electrical and Space Engineering