Towards an automated process for assessing reservoir rock quality from seismic imaging

4532424c-92f6-4592-baf1-73642eda6654

Closed

NERC
NERC
NERC
NERC
NERC

£164,405

Jan. 2, 2018

Nov. 30, 2018

This project will lay the basis of a training model for automated evaluation of georesources using the vast knowledge that can be acquired studying the global catalogue of subsurface datasets held by geo-energy companies.

The current global energy climate is characterised by readily available primary fuels at a relatively low cost. This has impacted the geo-energy business with a number of companies currently downsizing with reduction of their workforce. Yet, advances in computational power in the last 20 years has allowed acquisition and releasing of extensive subsurface datasets which provide a deeper understating on where georesources might be found. This means that automated processes will be increasingly important to efficiently analyse and tap the full potential of extensive subsurface datasets now available.

This project will focus on the analysis of reservoir units. These deposits represent a very valuable subsurface asset as they have elevated porous space, hence have the potential to contain hydrocarbons, water as well as to absorb carbon dioxide in 'Carbon Capture and Storage' operations. The amount of porous space in reservoir units, hence their efficiency to hold fluids and gas, is a function of a number of physical properties of such deposits. These same physical properties are also thought to control the overall shape and the size of the deposits.

With this project, we will analyse the morphology and physical properties of a number of reservoirs so to understand how they are interrelated. In doing so, we would be able to predict the reservoir quality by looking at seismic data, which is a method used in the industry to provide an image of the subsurface. The project will deliver a series of case studies that document how specific reservoir morphologies would indicate good or bad reservoir quality. This would provide the basics to construct, in the future, a training model for automated processes that will be able to efficiently scan large seismic dataset in search of rock with the right shape that indicates a high-quality reservoir.