Multi-National oil & gas company: pump jack predictive maintenance
- Large oil producer is interested in understanding the ongoing effectiveness of their downhole pumps to lift fluid to the surface.
- Pumpjack failures proved difficult to detect using rule-based diagnostic methods and often led to expensive field failures and lost production
- Created a unified view for wells over time by combining dynamometer, operational, well maintenance, downtime, and performance failure data.
- Using Azure Machine Learning, Neal Analytics classified failure types to identify patterns of failure occurrences and allow for remote diagnostics of pre-failure conditions
- Reduced cost of nonproductive time and operations
- Increased Return on Assets by avoiding severe pump jack failures.
- Optimized maintenance schedules based on risk probabilities.