Integration of Machine Learning with Distributed Temperature and Acoustic Sensing to Build Data-Driven Dynamic Reservoir Model

This project will develop practical workflows, algorithms and programming codes for inferring unknown reservoir properties from distributed temperature and acoustic sensing data. In-situ pressure and flow conditions can be interpreted from downhole fiber signals gathered in real time, which are used to estimate unknown heterogeneous reservoir parameters continuously. Machine learning methods will be incorporated to facilitate the handling of large amount of measured data and computations more efficiently. The project outcomes will help to advance the deployment of fiber-based instrumentation and optimize operations of inflow/outflow control devices for downhole monitoring and production diagnoses of oil and gas wells. One PhD student will be trained.

Hossein Izadi
Faculty Supervisor: 
Juliana Leung
Partner University: