Chan Chang, Shing Cheng
Data-driven reservoir management using deep learning techniques
Development of unconventional resources is faced with many challenges due to the lack of reliable numerical/physical models. However, the low cost of monitoring sensors enables the collection of large amounts of real time data that can be analysed or interrogated to extract valuable knowledge about subsurface reservoir within a data-driven framework.
The success of this data-driven model relies on utilising a set of signal/image processing tools to capture useful signal-level attributes, referred to as features. However, crafting good features is a particularly challenging process underpinning the success of the entire framework.
In the current project, recent advances in deep learning algorithms will be applied for automatic feature discovery for subsurface reservoir.
Dr Ahmed Elsheikh and Professor Mike Christie