Share with a co-worker
All Resources
Use case
The Challenge
In the Vacuum Distillation Unit, one critical temperature had been operating below its expected range for nearly six months. The deviation was subtle but persistent, and its impact was clear: product recovery was steadily declining.
The issue was not an acute failure but a gradual deterioration. With multiple interacting process variables and no obvious single cause, troubleshooting required more than intuition. The business impact was significant — a sustained drop in recovery translating into substantial financial losses.
- Persistent temperature deviation in the VDU
- Approximately 10% decrease in product recovery
- No clear single root cause
- Estimated annual loss of $300,000+
The Approach
The team structured the investigation to move from broad correlation screening to focused modeling.
- Full variable screening: All related tags including the key temperature were loaded and analyzed
- Correlation engine analysis: Correlations were run first against potentially related variables and then across the entire asset to uncover hidden relationships
- Layered comparison: High-recovery and low-recovery periods were separated and overlaid to compare statistics and trends of influencing parameters
- Root cause refinement: Two process variables were identified as the main influencing factors
- Interaction insight: It was discovered that the two variables did not act simultaneously but independently, explaining why individual correlations were moderate while their combined effect explained the temperature shift
- Predictive modeling: A linear regression model based on the two most influential variables was developed to predict temperature and recovery behavior
Key Insight
The performance loss was not driven by a single dominant variable, but by the combined effect of two independent factors. Once modeled together, the temperature deviation and recovery loss became explainable and predictable.
Results
The Takeaway
By moving from symptom observation to multivariable modeling, the team transformed a six-month performance decline into a quantifiable and explainable issue, enabling targeted corrective actions and providing a clear economic justification for restoring optimal recovery conditions.
The Challenge
In the Vacuum Distillation Unit, one critical temperature had been operating below its expected range for nearly six months. The deviation was subtle but persistent, and its impact was clear: product recovery was steadily declining.
The issue was not an acute failure but a gradual deterioration. With multiple interacting process variables and no obvious single cause, troubleshooting required more than intuition. The business impact was significant — a sustained drop in recovery translating into substantial financial losses.
- Persistent temperature deviation in the VDU
- Approximately 10% decrease in product recovery
- No clear single root cause
- Estimated annual loss of $300,000+
The Approach
The team structured the investigation to move from broad correlation screening to focused modeling.
- Full variable screening: All related tags including the key temperature were loaded and analyzed
- Correlation engine analysis: Correlations were run first against potentially related variables and then across the entire asset to uncover hidden relationships
- Layered comparison: High-recovery and low-recovery periods were separated and overlaid to compare statistics and trends of influencing parameters
- Root cause refinement: Two process variables were identified as the main influencing factors
- Interaction insight: It was discovered that the two variables did not act simultaneously but independently, explaining why individual correlations were moderate while their combined effect explained the temperature shift
- Predictive modeling: A linear regression model based on the two most influential variables was developed to predict temperature and recovery behavior
Key Insight
The performance loss was not driven by a single dominant variable, but by the combined effect of two independent factors. Once modeled together, the temperature deviation and recovery loss became explainable and predictable.
Results
The Takeaway
By moving from symptom observation to multivariable modeling, the team transformed a six-month performance decline into a quantifiable and explainable issue, enabling targeted corrective actions and providing a clear economic justification for restoring optimal recovery conditions.
Access now
Share with a co-worker
The Challenge
In the Vacuum Distillation Unit, one critical temperature had been operating below its expected range for nearly six months. The deviation was subtle but persistent, and its impact was clear: product recovery was steadily declining.
The issue was not an acute failure but a gradual deterioration. With multiple interacting process variables and no obvious single cause, troubleshooting required more than intuition. The business impact was significant — a sustained drop in recovery translating into substantial financial losses.
- Persistent temperature deviation in the VDU
- Approximately 10% decrease in product recovery
- No clear single root cause
- Estimated annual loss of $300,000+
The Approach
The team structured the investigation to move from broad correlation screening to focused modeling.
- Full variable screening: All related tags including the key temperature were loaded and analyzed
- Correlation engine analysis: Correlations were run first against potentially related variables and then across the entire asset to uncover hidden relationships
- Layered comparison: High-recovery and low-recovery periods were separated and overlaid to compare statistics and trends of influencing parameters
- Root cause refinement: Two process variables were identified as the main influencing factors
- Interaction insight: It was discovered that the two variables did not act simultaneously but independently, explaining why individual correlations were moderate while their combined effect explained the temperature shift
- Predictive modeling: A linear regression model based on the two most influential variables was developed to predict temperature and recovery behavior
Key Insight
The performance loss was not driven by a single dominant variable, but by the combined effect of two independent factors. Once modeled together, the temperature deviation and recovery loss became explainable and predictable.
Results
The Takeaway
By moving from symptom observation to multivariable modeling, the team transformed a six-month performance decline into a quantifiable and explainable issue, enabling targeted corrective actions and providing a clear economic justification for restoring optimal recovery conditions.
Access now
Subscribe to our newsletter
Stay up to date with our latest news and updates.
Other Webinars on Demand
Press Play on Operational Improvement
.jpg)
Other Resources
Explore Our Newest Content to Maximize Your Operational Efficiency
