Transforming Gas Flow Management

Situation

A leading Oil and Gas company struggled with unpredictable Acid Gas Removal Unit (AGRU) breakthroughs causing CO2 residuals, unplanned shutdowns, and costly manual interventions. Efficient operations required precise breakthrough prediction and automation to maintain seamless gas flow adjustments.

Challenge

The company faced:

• Scarcity of event data, making predictive modeling difficult.
• High rates of false alarms from existing systems, hampering trust and efficiency.
• A lack of integrated tools to deliver actionable insights for proactive decision-making.

Approach

1. Optimized Data and Hypotheses
   • Applied advanced resampling techniques to structure high-quality datasets tailored for AGRU sensor data.
   • Conducted rigorous hypothesis testing to pinpoint the ideal alert interval, optimizing response times to breakthroughs.
2. Predictive Modeling with Synthetic Data
   • Created synthetic event data to address data scarcity, improving model training and prediction accuracy.
   • Introduced a probability optimization algorithm to reduce false alarms by 86%, ensuring operational reliability.
3. Seamless System Integration
   • Developed a dedicated conversion pipeline to integrate predictive models into the native Advance Process Control System.
4. Empowered Engineers with Insights
   • Designed a Model Monitoring UI Tool for production engineers, offering real-time event insights and enabling proactive adjustments.

Impact

• $30M+ in deferred cost savings per year by preventing unplanned shutdowns.
• 10-minute notification lead time reduction for CO2 breakthrough events.
• 86% reduction in false alarms, fostering trust in automation and predictive analytics.