Real-Time Brewery optimization with the fermSYS Brew Buddy
Updated: 4 days ago
How AI & VOC Sensing Transform Fermentation and Yeast Management
The thousand-year-old art of brewing is undergoing a historic transformation. Driven by the transition from Industry 4.0 to Industry 5.0, cutting-edge technologies like Artificial Intelligence (AI) and Volatile Organic Compound (VOC) sensing are moving into the cellar [1, 2].
With the Brew Buddy (combining precision hardware like the fermSENSE - PRO with intelligent control software fermHUB), breweries can now measure the "breath of beer" in real time, digitize fermentation charts, and accurately predict apparent extract dynamics [1].
Here you will find a detailed insight into how Brew Buddy combines traditional brewing craftsmanship with modern AI data science, based on real-world validation data obtained as part of our collaboration with the Rolinck Brewery (Krombacher Group) and the Department of Bioprocess Engineering at the Westphalian University of Applied Sciences in Recklinghausen [1].
1. Measuring the "Breath of Beer": Advanced VOC Off-Gas Sensing
Traditional fermentation monitoring relies heavily on manual sampling and hydrometer (spindle) measurements. While effective, manual readings introduce operational lag, human error, and labor costs.
fermSYS Brew Buddy solves this by monitoring the volatile organic compounds (VOCs) released during yeast metabolism [2]:
Gas-Phase Ethanol & Alcohol Tracking: Using heated metal oxide semiconductor sensors, fermSYS measures volatile off-gas components—such as ethanol, methanol, and isopropyl alcohol—directly in the vessel off-gas stream.
Real-Time Growth & Peak Detection: During yeast propagation, the continuous off-gas signal precisely detects peak cell counts and optical density (OD600). This enables brewers to pinpoint the exact optimal harvest time for yeast propagation.
Hygenic & CIP-Capable Hardware: To ensure industrial applicability, sensor prototypes are integrated into standard, CIP-cleanable Varivent® steel connections, preventing contamination risk during routine cleaning.
"Direct measurement of the ethanol content by these sensors provides us with real-time data that can be relevant to the optimisation of yeast propagation... By using sensors, manual sampling and analysis can be reduced, which saves cost and effort."— Sven Detering, Master Brewer, Rolinck Brewery [2]
2. AI-Powered Fermentation Forecasting (LSTM Models)
Hardware sensors provide real-time data, but AI turns that raw data into actionable process intelligence. Integrated into the Brew Buddy, our Long Short-Term Memory (LSTM) neural network architecture continuously calculates and predicts the apparent extract content throughout the fermentation process. [1]
Why LSTM?
Standard static neural networks struggle with sequential time-series data. LSTM networks utilize specialized memory cells with input, forget, and output gates. This structure allows the AI to "remember" historical fermentation trends (temperature curves, gravity changes, pitch rates) and accurately forecast future fermentation behavior. [1]
Key Performance Metrics:
Historical Training: Trained on over 200 historical fermentation batches including variable parameters like temperature, initial extract, and yeast generations (harvested vs. propagated yeast) [1].
High Predictive Accuracy: Reached an R² value of 99.58% in validation testing, predicting apparent extract with an average deviation of just 0.23° P (4.6%) [1].
Virtual "Soft Sensor" Data: In continuous off-gas monitoring mode, the soft sensor updates extract estimations approximately once every minute [2].
3. Seamless Integration & Digitalization (brewTwin / fermHUB)
A key barrier to brewery digitalization is user complexity. The Brew Buddy addresses this with an intuitive interface and easy system integration [1]:
Quality-Assured (QM) Digitalization: Replaces pen-and-paper fermentation logs with automated, QM-compliant data storage.
Anomaly Detection & Real-Time Alerts: Operators immediately see if the current fermentation batch deviates from the golden batch curve, allowing rapid interventions (e.g., dynamic PID temperature adjustments or blending).
Industrial SCADA Compatibility: Designed to interface seamlessly with modern brewing control systems (such as ProLeiT brewmaxx) and inline hardware (e.g., Endress+Hauser QWX43 Fermentation Monitors).
"The brewTwin [fermSYS platform and part of the Brew Buddy] is a practical way for us to digitize fermentation documentation. The entered data is immediately available for AI refinement, while our staff can instantly see how current fermentation deviates from the optimum. It has catapulted our work in the cellar into the digital age."— Sven Detering, Production Manager, Rolinck Brewery [1]
4. Business Impact: Quality, Cost Savings, & Predictive Maintenance
Deploying fermSYS - Brew Buddy across the brewhouse and fermentation cellar delivers clear operational benefits [1, 2]:
Feature | Traditional Approach | fermSYS - Brew Buddy Approach |
Fermentation Monitoring | Manual spindling (e.g. daily) [2] | 24/7 continuous real-time AI prediction & VOC tracking [1,2]
|
Yeast Propagation | Fixed time schedules or manual lab counts [2] | Real-time peak cell count detection via off-gas peak [1]
|
Data Management | Paper logs, siloed spreadsheets | Automated cloud or local data storage [2]
|
Maintenance | Reactive / Fixed schedules | Predictive maintenance via off-gas flow & pressure anomaly detection [2]
|
Furthermore, academic research led by Prof. Frank Eiden (Westphalian University of Applied Sciences) demonstrates that continuous sensor monitoring unlocks deep insights into process dynamics, stabilizing product flavor profiles while lowering energy and labor overhead [1,2].
What’s Next? The Future of Smart Brewing
As Marcel ter Steege (Plant Manager, Rolinck Brewery) notes: "We are just at the beginning of what is possible with this technology."[2]
Future roadmap updates for fermSYS - Brew Buddy include [1]:
Aroma Profile Prediction: Machine learning models trained to anticipate ester and higher alcohol profiles during active fermentation.
Automated Recipe Adaptation: Deep learning algorithms adjusting mashing and hopping schedules based on annual raw material crop variations.
Reinforcement Learning Control: Closed-loop, self-optimizing fermentation control via dynamic PID pump strategies.
Upgrade Your Brewery Operations Today
Experience how the Brew Buddy transforms complex fermentation data into maximum yield, reduced manual labor, and rock-solid beer consistency [1,2].
Contact the fermSYS Team today for a demo or pilot trial
[1] Dymek, F. (WFH); Eiden F. (WFH): Künstliche Intelligenz im Gärkeller: Vorhersage von Prozessdaten. Brauwelt | No. 18 (2024). Available at: DE: https://brauwelt.com/de/themen/gaerung-lagerung/647283-kuenstliche-intelligenz-im-gaerkeller-vorhersage-von-prozessdaten
[2] Janetzky, R. (WFH); Eiden, F. (WFH): Measuring the breath of
beer | No. 6 (2024). Available at:
Eiden, F. (WFH): Prozessoptimierung im Brauereiwesen. GIT (21 February 2024). Available at: https://analyticalscience.wiley.com/content/article-do/prozessoptimierung-im-brauereiwesen
Detering, S. (Rolinck): Der Atem des Bieres: Wie moderne Technologie das Brauen revolutioniert. LinkedIn (2025). Available at LinkedIn: https://www.linkedin.com/posts/krombacher_atem-des-bieres-moderne-technologie-trifft-activity-7287378470706245632-OCYK
Comments