Inside the Development of PLAATO’s Predictive Brewing Models
Inside the Development of PLAATO’s Predictive Brewing Models
By Erik Magnus Eriksen Olseng and Kristian Fjelde Pedersen
Inside the Development of PLAATO’s Predictive Brewing Models
Introduction
At PLAATO, our development process is grounded in the practical application and use of our products. We continuously test and refine through first-hand experience gained in our in-house breweries, in addition to our development partner breweries. During this process, we have brewed our PLAATO House Pilsner recipe over 10 times, gathering more than 7,000 density and temperature measurements using our PLAATO Pros. As data engineers, this amount of data has opened up new possibilities for innovative uses.
Batch Comparison
It all started with our batch comparison tool, where we identified clear trends between the unique batches. They were mostly quite similar, but one of the batches deviated from the norm - we had underestimated the pitch rate and cell counts, causing an extended lag phase. We wanted to see if we could automatically detect these kinds of deviances, and through a weekend “hackathon” (developer-speak for nerding out on something we enjoy, similar to a 90’s LAN party) our first deviation detection model was born. This allowed us to visualize if something was off, but only through manual comparisons between the PLAATO Pro sensor data and the “benchmark” for the recipe that the model created.
Fast forward a couple of months and this has evolved into an automated warning system, immediately alerting us to deviations and allowing us to act quickly. Building on this foundation, we have also developed a predictive model to accurately forecast when and at what FG a fermentation will end.
Empowering Brewers with Predictive Brewing
By harnessing these predictive capabilities brewers are equipped with practical tools to help streamline production, enhance quality, and reduce costs. Here’s how:
1. Early Deviance Detection
With the automated warning system, brewers can immediately detect deviations in fermentation behavior. This allows for early intervention, preventing potential issues from escalating to a potentially spoiled batch.
2. Improved Consistency
The benchmark model helps brewers establish consistent quality standards by comparing current batches against historical data in an easy-to-digest way. This capability is particularly valuable for scaling up production or in maintaining consistency across different production sites.
3. Precise Fermentation Control
The predictive model enables brewers to anticipate when fermentation will end, allowing them to plan their schedules more effectively. Whether managing production timelines or ensuring optimal conditioning time, brewers can schedule tasks with confidence, improving efficiency and beer quality.
4. Optimized Spunding Process
By predicting the final gravity (FG) accurately, brewers can know exactly when to spund to reach the desired carbonation naturally. This reduces CO2 usage and minimizes yeast stress, leading to better-tasting beer and significant savings.
Learn how PLAATO Pro can help you maintain consistency by comparing batches
A Glimpse into the Future
These predictive features have proven invaluable internally, and we’re excited to see how they will benefit your brewing process. We’re still a couple of months away from getting the systems fully tested and ready. In the meantime, each batch brewed will improve its accuracy and value once it goes live.
If you’re interested in leveraging this data-driven approach to improving your brewing process, reach out to us.
Simillar articlesview more → 28th May 2024 10th May 2024 3rd May 2024 9th April 2024 4th April 2024 4th April 2024Inside the Development of PLAATO’s Predictive Brewing Models
Accelerate Off-Grain Distillation: Plaato Pro’s Real-Time Monitoring
Lagers, Grapes and Portuguese Beer
Consistent Brewing: Never Spill the Difference
The Digital Twin of your Fermenter – Part 1
The Digital Twin of your Fermentor – Part 2
