How Physics and Machine Learning Combined Provide Unmatched Capabilities of Real-time Flow Simulations
Oil & gas operators are continuously aiming to achieve optimized production throughout the lifetime of their producing assets. Applying machine learning on real-life data and simulations from physics models can allow for improved agility, accuracy and robustness to flow simulation solutions in the market.
Agenda:
- Physics-based and data-driven modeling approaches
- Hybrid modeling philosophy – combining the best of the two worlds
- Hybrid modeling applied – use cases
- Maximize production and avoid operational upsets due to slugging
Approach – Physics to produce data - Improve production insights for mature fields with high water rates
Approach – Transfer learning to support physics - Monitor production to reduce well testing frequency
Approach – Apply Physics to improve ML model, then complement the physics model
- Maximize production and avoid operational upsets due to slugging
Speakers:
