Yokogawa and Kyoto-based brewer test AI-guided fermentation optimization process

Craft Bank Co., Ltd., a beer brewer in Kyoto Prefecture, and Yokogawa Digital Corp. reported Sept. 26 that they’ve proof tested a manual temperature-setting schedule created by Yokogawa’s Factorial Kernel Dynamic Policy Programming (FKDPP) reinforcement learning-based AI algorithm.

The algorithm was applied to the fermentation process for the brewer’s award-winning Bank IPA 1 craft beer. By manually implementing this temperature-setting schedule developed by FKDPP’s AI algorithm, Craft Bank’s brewers shortened their fermentation time by 28% from 336 hours to 240 hours (Figure 1).

Previously, fermentation temperature was kept  constant, and the brewmaster manually measured sugar content each day, conducted sensory evaluations of aroma and taste, and checked for off-flavors. Consequently, Craft Bank and Yokogawa’s proof of concept (PoC) focused on temperature’s effect on fermentation speed. They confirmed that temperature-setting schedule AI algorithm would be valid and effective.

To implement the PoC, a simulation was developed that replicated Craft Bank’s beer production process. Using data from the brewmaster, including stress conditions on the yeast, the AI algorithm developed the temperature-setting schedule for the fermentation tank within the simulation. The brewmaster reviewed the appropriateness of this plan, and followed it to manually set temperatures. Using sensory evaluation during fermentation, it was confirmed that all quality criteria were met.

“As a brewer who’s responsible for everything onsite from preparation to fermentation every day, this initiative was very meaningful. Previously, fermentation tests were physically performed each time, so it took about a month,” says Daichi Haboshi, president, CEO and brewing manager at Craft Bank. “This time, by repeating simulations using data automatically collected by a sensor, we virtually verified fermentation many times, and conducted hypothesis-based tests on actual equipment. The expertise of our brewmasters in functions such as sugar content measurement, temperature control, and sensory evaluation of aroma can be complemented by AI. And, with them working as partners, we can see an unprecedented balance between quality and efficiency. I feel like I now have had a glimpse of one of the optimal solutions that will come through the relationship between people and AI in the AI era. We’ll explore new ways of producing craft beer that combines the characteristics of Kyoto with cutting-edge technology.”