Protecting Product Quality
How Early Anomaly Detection Prevented Oil Degradation
The Challenge
A cooking oil production facility was facing a recurring quality issue that affected the value of its finished product.
Although the process appeared stable most of the time, occasional deviations during production caused the oil to degrade. When this happened, the product could not be sold at its intended premium price point and instead had to be downgraded to a lower value category.
After analyzing the process data, several key issues became clear.
Delayed Detection of Process Deviations
Operators relied on standard alarms and periodic checks to monitor production. However, these tools only detected issues once the process had already drifted significantly from its optimal conditions.
By the time the deviation was identified, the oil quality had already been affected.
Subtle Changes During Critical Process Stages
Small variations in temperature control, circulation, or residence time during critical phases of production were difficult to detect with traditional monitoring.
These small shifts accumulated over time, gradually degrading the oil.
Product Downgrades and Revenue Loss
When degradation occurred, the oil could no longer be sold in the premium category and had to be reclassified into a lower price segment.
Even when the volume loss seemed small, the financial impact accumulated over multiple batches.
The facility needed a way to detect these deviations earlier before product quality was compromised.
The Solution
The facility implemented MontBlancAI’s anomaly detection platform to monitor the production process in real time.
Using historical batch data, machine learning models were trained to understand the normal operating patterns of the process and detect subtle deviations as they emerged.
Real-Time Process Anomaly Detection
The ML models continuously monitored live process signals, identifying deviations from expected patterns long before traditional alarms were triggered.
Operators received early alerts when the process began drifting away from its normal behavior.
This allowed teams to intervene immediately adjusting process conditions before the oil quality was affected.
Early Warning of Degradation Risk
By analyzing historical production patterns, the system learned which signal behaviors were historically associated with oil degradation.
When these patterns began to appear in a new batch, the platform alerted operators early enough for corrective action.
Instead of discovering the problem after production finished, the team could stabilize the process in real time.
Faster Operational Response
With clear anomaly alerts and contextual data, operators no longer had to manually interpret dozens of signals.
The system highlighted exactly where the deviation was occurring, allowing faster troubleshooting and quicker corrective actions.
The Outcome
Within a short period of deployment, the facility began seeing clear improvements in product quality and operational control.
Reduced Product Downgrades
Early alerts allowed operators to intervene before degradation occurred, preventing oil from dropping into lower-value categories.
During the first deployment period alone, this early intervention helped prevent over 120 tons of cooking oil from being degraded, preserving its premium market value.
Faster Reaction to Process Issues
Instead of identifying problems after the batch was completed, operators could respond within minutes of the first deviation.
Higher Product Consistency
Maintaining tighter control over critical process stages improved overall batch stability and quality consistency.
Improved Operational Visibility
The team gained a clearer understanding of how small process changes affected product outcomes, allowing them to operate with greater confidence.
Key Takeaways
The deployment of MontBlancAI’s anomaly detection platform fundamentally changed how the facility managed product quality. Instead of reacting after problems occurred, the team could proactively protect product value.
Small Deviations Can Have Big Consequences
Subtle process shifts that previously went unnoticed were shown to have significant impacts on product quality.
Early Detection Protects Product Value
Catching anomalies early prevented degradation events and preserved premium product classification.
Real-Time Monitoring Enables Faster Decisions
With clear alerts and contextual insights, operators could act immediately instead of investigating issues after production.
Protect Product Quality with AI-Powered Anomaly Detection
MontBlancAI helped this facility prevent oil degradation, protect product value, and improve operational visibility.
Our platform can help manufacturers detect issues earlier, respond faster, and maintain consistent product quality.
Discover how AI-driven anomaly detection can transform your operations.
More cases
Explore Additional Success Stories
Identifying Defects Early
A huge range of issues has the potential to delay, disrupt, and derail manufacturing operations. If you can identify these issues at the earliest opportunity, you can maintain an efficient production process. Traditional monitoring methods struggle to pick up issues such as performance deviations, short-term disruptions, and the failure of critical components such as filtration membranes. Our approach to predictive solutions provides a proven answer to this problem.
Optimize Dairy Production
Germany is home to one of the world’s most competitive dairy industries. The small and moderately-sized dairies in the country have to take every opportunity to make their operations more efficient and productive. This is a cost-effective and reasonably feasible way to compete with the large corporations. The pursuit of efficiencies and increased capacity plays a crucial role in this process, but traditional assessment methods are costly, time-consuming, and relatively unreliable.One relatively small dairy approached us due to our reputation for powerful production process monitoring software. The brief was simple: To decide whether optimizing existing equipment or investing in new machinery was the most cost-effective way forward.
UF System Optimization
A major production facility approached us to address the operational issues behind a significant downtime rating. When the facility’s management first assessed the situation, the maximum continuous runtime was a paltry 13 hours before a Cleaning-in-Place cycle was needed. The relatively low uptime witnessed was having a seriously detrimental impact on both operational efficiency and capacity. With the help of MontBlancAI, the team implemented a series of targeted optimizations using, among other things,AI-powered operational intelligence tools. As a result, the facility significantly increased overall efficiency and capacity. Here’s how they did it with the help of our cutting-edge operational intelligence tools.
Enhanced OEE
A manufacturing facility was struggling with filling line inefficiencies, despite seemingly stable monthly performance reports. Frequent, untracked short stops and unclear performance fluctuations were silently eroding productivity. By leveraging MontBlancAI’s advanced production monitoring tools, the facility uncovered and addressed these issues, delivering a remarkable 12.4% increase in Overall Equipment Effectiveness (OEE) within just six weeks. This case study details the challenge, the innovative approach, and the eventual results.
Continuous CIP Validation
Monitoring and validating Clean-in-Place (CIP) processes can be a complex, time-consuming, and costly process, particularly for relatively modest operations. However, production intelligence software powered by AI can make it significantly easier. That’s why a German dairy, processing more than two million liters of milk, turned to MontBlancAI for help. Producing high-quality milk, cream, yogurt, cheese, and butter, the dairy relies on CIP to clean and sanitize equipment and pipelines, ensuring product hygiene, quality, and safety.Was there a way to streamline and enhance the existing CIP processes? Let’s take a look.
Drive measurable results. Without operational friction.
