Rapid Anomaly Detection
In the ever-evolving landscape of industrial operations, we have implemented a groundbreaking approach to anomaly detection that is transforming the way our client manages their assets and optimize production. By leveraging the AVEVA PI System and implementing a sophisticated Rapid Anomaly Detection (RAD) system, we have achieved remarkable improvements in efficiency, cost savings, and operational excellence.
The Journey to RAD
Our journey with the PI System began in 1997, and over the years, it has grown into a robust infrastructure with over 600,000 PI tags, 50 interfaces, and a million streaming calculations and analytics1. This extensive data ecosystem laid the foundation for the development of RAD.
The genesis of RAD came from a simple yet crucial need: identifying leaking tanks as quickly as possible to reduce spill volumes. What started as an algorithm to monitor tank levels and detect slow rates of decline outside normal pump rates or truck hauls evolved into a comprehensive system for detecting various process anomalies1.
How RAD Works
RAD is built on a hybrid technology approach, utilizing multiple systems and “layers of analytics.” The process involves:
- Data Collection: Leveraging the AVEVA PI System for real-time data acquisition from various assets.
- Algorithm Development: Creating sophisticated algorithms using PI AF Analytics to detect anomalies.
- Event Frame Generation: Using PI AF to create event frames when possible anomalies occur.
- Notification System: Employing PI Notifications to alert the Integrated Operations Center (IOC) about potential issues.
- Data Processing: Utilizing Azure Data Factory for ETL processes and Databricks for advanced data engineering.
- Visualization: Presenting results through Power BI dashboards for easy interpretation and action.
Types of Anomalies Detected
RAD system monitors some key anomalies include:
- Separator health issues
- Compressor and engine vibration anomalies
- Gas lift problems
- Well loading issues
- Plunger lift anomalies
- Tank level anomalies
- Equipment failures and potential breakdowns1
The Impact of RAD
The implementation of RAD has yielded significant benefits:
- Cost Savings: By the end of September 2023, RAD had contributed to millions of dollars in savings across production increases, avoided losses, gas lift optimization, and reduced equipment failures1.
- Improved Response Time: Issues that previously took 6 hours to 3 days to notice are now identified within minutes, allowing for rapid intervention1.
- Operational Efficiency: The system has enabled Ovintiv to focus on the right opportunities as they emerge, reducing overall negative operational impact1.
- Proactive Maintenance: By detecting early signs of equipment degradation, RAD helps prevent costly breakdowns and extends asset life.
- Environmental Impact: Faster leak detection and equipment optimization contribute to reduced environmental risks and improved sustainability.
Lessons Learned and Best Practices
The experience with RAD has yielded valuable insights:
- Continuous Refinement: Tuning and refining anomalies is crucial for success and reducing false positives.
- Accountability: Assigning ownership of specific anomalies to individuals ensures thorough tuning and follow-through.
- User-Centric Approach: Defining acceptable metrics for daily anomaly counts and implementing feedback loops improves user adoption.
- Cross-Functional Collaboration: Strong teamwork across departments is essential for effective implementation.
- Gradual Implementation: Starting small with a few anomalies helps establish processes before scaling up1.
Future Plans
We are not resting on its laurels. The company has ambitious plans to further enhance RAD:
- Continuous addition of new anomalies and refinement of existing ones
- Integration of more data sources to create more sophisticated anomalies
- Phasing out anomalies where operational improvements have eliminated the original issue
- Expanding RAD to different user groups, including production engineering
- Implementing anomaly suppression techniques
- Further integration of machine learning algorithms for even more advanced detection capabilities1
Conclusion
The Rapid Anomaly Detection system here represents a significant leap forward in industrial operations management. By harnessing the power of real-time data, advanced analytics, and cross-functional collaboration, RAD has enabled Ovintiv to transition from a reactive to a proactive operational stance. This not only drives substantial cost savings but also enhances safety, efficiency, and environmental performance.
As the industrial landscape continues to evolve, systems like RAD will undoubtedly play an increasingly crucial role in maintaining competitive edge and operational excellence. Ovintiv’s success story serves as a compelling example of how embracing digital transformation can yield tangible, bottom-line results in the complex world of energy production and asset management.
