Aalto University: Neste and FCAI collaboration: Optimizing chemical reactors with AI

When companies invest in close research collaboration with academia, it helps them stay up-to-date and benefit from the latest innovations. Neste realized that the use of artificial intelligence could bring significant benefits. “Artificial intelligence is useful for us because a flood of process data is generated at each chemical production facility every second. We needed AI to optimize the production and operation of chemical plants,” says Product Manager Samuli Bergman from Neste NAPCON.

Before initiating research collaboration with FCAI, Neste had made a framework agreement with Aalto University on strategic research collaboration. “It was very fortunate for this project that we had this agreement in place beforehand. Furthermore, FCAI’s Industry and Society Program gave us a good kick-start. It was easy to start with the practical project work and research collaboration,” says Bergman.
   
Neste NAPCON collaborated with FCAI researchers from Professor Alexander Ilin’s group at Aalto’s Department of Computer Science and Professor Francesco Corona from Aalto’s Department of Chemical and Metallurgical Engineering. Professor Simo Särkkä and his group from FCAI and Aalto’s Department of Electrical Engineering and Automation also contributed to the project.

The joint project demonstrated how prior knowledge about process dynamics can be combined with neural networks to monitor variables, such as catalyst activity, which can be costly or impossible to measure directly in real-world industrial reactors.

Academy-industry collaboration brings companies brand visibility and access to top talent in academia.

FCAI offers a wide range of collaboration opportunities. Typically, research collaboration starts with a joint project that involves master’s or doctoral theses.

For the collaboration with Neste, Professor Alexander Ilin recruited a doctoral student, Katsiaryna Haitsiukevich, to his research group. “Industrial projects can benefit students by giving access to real-world data, while companies get access to the state-of-the-art research in the field. In our project we had several research groups with diverse expertise. We had regular monthly meetings to exchange ideas, to share results and to agree on next steps. We also regularly received feedback from business and domain experts on how the model can be improved,“ says Haitsiukevich.

Academy-industry collaboration brings companies brand visibility and access to top talent in academia. In a field where experts are hard to come by, the recruitment possibilities that follow are invaluable. “During the project we developed a model of a tubular reactor which can be used to estimate unobserved reactor states. The obtained results were accepted and presented at an international conference,“ says Haitsiukevich.

Dozens of corporate-funded projects
During its first three years of operation, FCAI has launched dozens of corporate-funded projects. “As there is a strong demand for machine learning researchers, these projects are continually looking for new talent. Accordingly, doctoral students or postdoctoral researchers are offered multi-year contracts as an incentive to join FCAI,“ says Corporate Relations Manager Terhi Kajaste from FCAI.

According to Kajaste, companies finance these kinds of projects in many different ways. “Some have Business Finland funding that allows them to subcontract research, others conduct joint or commissioned research as a service purchase, and others make a donation for a targeted purpose. Donations are used surprisingly little, even though they are an excellent and simple tax-deductible way to advance research that is relevant to your business,” says Kajaste.

Potential collaboration starts with a discussion. “If there is no previous relationship with an FCAI researcher, then a few meetings are often needed to reach a sufficient consensus. The FCAI Industry and Society Program Team can assist in facilitating these meetings. Next, a joint research plan will be drawn up, the availability of data needed for the machine learning methodology research will be clarified, funding will be agreed and a project agreement will be drawn up,” explains Kajaste. She welcomes companies and the public sector to get involved and work with FCAI.

Comments are closed.