A hot topic in academia is how AI can benefit different scientific disciplines – and process industry research is no exception.
To increase AI competence in academia, Linköping University, led by AI professor Fredrik Heintz, has tailored a course to the doctoral students in the business research school Resource Smart Processes.
– “Used correctly, AI can improve existing process research and also open up research that was previously unthinkable. In the course focusing on AI in the process industry, participants have gotten the basics and started using AI in their own projects, says Fredrik Heintz.
The course, which was held in the spring of 2025, focused on relevant application areas for the process industry. Extra focus was on data analysis to process large amounts of data and explainable AI, which means that the AI tool not only presents a result but also explains how it reasoned, something that is extra important in the complex environment of the process industry. Students also learned more about how language models can facilitate literature studies and scientific writing.
Using AI to analyze data from ForMAX
One of the participants now using AI in her research is Carolina Marion de Godoy, a PhD student at Chalmers. Her research can help to increase quality and resource efficiency in the production of paper pulp by exploring how things like tree species and wood chip size affect the smoothness of the pulp. She does this by studying how the wood fiber changes during the sulfate cooking process.
The data needed from different stages of the cooking process have been obtained in experiments at the ForMAX research station, in collaboration with research groups from the universities of Lund and Uppsala. ForMAX is part of the MAX IV laboratory, which has the world’s strongest X-ray light with the ability to study materials down to the nano level.
– Each individual measurement generated a 3D image, created from almost two thousand images, as well as the precise measurement data I needed to follow the change of the wood fiber during cooking. With such extensive data, AI was a good choice of method for processing and analysis,” says Carolina Marion de Godoy.
To help her, she used an open source image segmentation tool, a type of image analysis where AI divides an image into different segments based on color and texture, for example. With the tool, she could first select suitable images and mark what in the image represented wood fiber. The selection then formed the basis for using machine learning to create a classifier that distinguishes wood fiber and, once trained, can analyze large amounts of image data.
– The approach is interesting, it’s faster than doing the segmentation manually while I can contribute with my own input. I definitely see that AI has the potential to speed up the analysis of large amounts of data and overcome some limitations of more conventional methods,” says Carolina Marion de Godoy.
Calls for more skills and better computing resources
Fredrik Heintz points out that AI can contribute to high-quality research and new scientific discoveries in the process area, but that universities need to strengthen their skills and acquire better computing resources, preferably in the form of local computing capacity instead of purchasing expensive services with license agreements.
His two PhD students, Ibrahim Delibasoglu and Sanjay Chakraborty, also taught this spring’s course. Both have focused their research on AI in the process industry.
– Scientific disciplines differ greatly, and to make AI relevant to the process industry, this kind of expertise is needed, with one foot in AI research and one in the specific research area,” says Fredrik Heintz.
Now the ambition is to further develop the course and open it up to more people.
– “Right now, ‘everyone’ wants to work with AI, which means that AI is sometimes used where it is not necessary. With this course and similar initiatives, we make sure to strengthen the competence so that AI-based methods and tools make a real difference for science,” Fredrik Heintz concludes.
