Machine learning and model building

The second step is all about uncovering the real dynamics, correlations and uncertainties in your production system - to truly understand your system's behavior.

The Compact Database can be post-processed through multiple machine learning systems working in parallel and together, by combining hierarchical neural networks with first principle physics to learn the real dynamics, correlations and uncertainties in the production system and for auto-calibration of data-driven models.

This opens up for a broad spectrum of production enhancing applications of varying complexity.

One example is real-time analysis of well deduction tests, automating live analysis of such tests; measuring the uncertainty and giving constant updates on test progress . This facilitates reduced test times, which is directly linked to reduction of production losses (since the well is shut-in during the entire test). 

ProductionCompass can also assess and monitor the quality of production settings in real-time, leading to improved situational awareness. To exemplify, underperforming wells can be flagged.