A hybrid data-driven and mechanistic model for production optimization in the oil and gas industry
A hybrid data-driven and mechanistic model for production optimization in the oil and gas industry
PhD Candidate Mathilde Hotvedt
Main Supervisor Lars Struen Imsland
Sponsor: Lundin Energy Norway AS
Hybrid modelling, also called grey-box modelling, combines the principles of two modelling approaches; physics-based, first-principle, mechanistic modelling and data-driven modelling. The idea is for the hybrid model to preserve the favorable characteristics of mechanistic models, such as physical interpretability and good extrapolation abilities, while exploiting the ability of data-driven models to capture unknown/unmodelled phenomena. In that manner, the hybrid model should have the potential to achieve high levels of accuracy and still be computationally feasible for utilization in real-time optimization.
Project result: Predicting choke performance of wells in the Edvard Grieg Field
A study comparing the accuracy of mechanistic, data-driven and hybrid choke models. There is value in data!
BRU21 Conference 2022
