Background and activities
Autumn 2018, I started a PhD at the department of Engineering Cybernetics. The PhD is a part of a larger project, BRU21 (Better Resources Utilization in the 21 century) and will regard production optimisation within the oil and gas industry.
Title: "A hybrid data-driven and mechanistic model for production optimisation in the oil and gas industry".
For a petroleum asset to succeed economically, the operators must daily make crucial decisions regarding optimization of the asset. Knowledge regarding the multiphase flowrates in the asset is therefore of high importance. One way to obtain this knowledge is through virtual flow meters (VFM) which takes advantage of existing measurements to describe the input-output relationship of a system with a mathematical model.
There are several ways to model a VFM, and most common in today's oil and gas industry are mechanistic models based on first principles, where prior knowledge about the system is an important factor. However, in order for these models to be computationally feasible in real-time optimization, simplifications are a necessity and plant-model mismatch unavoidable. Another alternative is data-driven models which are generic mathematical models fitted to input-output data and where no prior knowledge about the system is needed. However, data-driven models have less interpretability than mechanistic models as there are no physical relations represented and they often have less extrapolation power.
Hybrid modeling aims aims to utilize the best of both worlds by combining mechanistic and data-driven methods. This PhD research will investigate different variants of hybrid models for production optimization in the oil and gas industry.
Lundin Norway AS is my main industry partner providing a real-case scenario for the research. Supporting companies are Solution Seeker and TechnipFMC.