Transfer learning for O&G wells:unlocking the collective potential of prod.data from multiple fields

Offentlig prosjektsammendrag

Knowledge about oil, gas, and water flow rates is of crucial importance to the operation of a petroleum asset, and especially to the optimization of production and recovery. The economic upside of improving these operations on the Norwegian Continental Shelf is in the order of tens of billions NOK. In the past two decades the oil and gas industry has invested heavily in sensors and infrastructure for data capture, generating ever increasing amounts of data. This has prepared the industry for the recent advances in AI technology. At the same time, this has led to high, but yet unfulfilled expectations of impact, especially for high value applications like virtual flow metering. This project proposes a completely new approach to virtual flow metering based on machine learning algorithms that learn from production data. The underlying idea is to combine and leverage production data from different assets and oil companies to enable cross-learning, using a concept known as transfer learning. Well production data has never been collected and used across operators and regions in this way before, a process Solution Seeker has already started and is uniquely positioned to realize. The innovation will enable us to unlock the collective potential of production data collected around the world, to deliver flow rate estimates with higher accuracy and much less effort than traditional virtual flow metering systems. The primary objective of this project is to develop a prototype virtual flow meter based on transfer learning and test it in live operations together with participating partners. The most critical R&D challenges that the project will face are related to: dataset standardization across assets; creating new model architectures for cross-well learning; and, developing online calibration strategies to keep models up to date. Solving these challenges will lead to a highly automated virtual flow meter that learns from continuous streams of data from multiple oil fields.