Offshore staff
OSLO, Norway – The Norwegian Petroleum Directorate (NPD) has supported two studies investigating whether machine learning and artificial intelligence can be applied to logs from existing wells to determine whether hydrocarbons may have been overlooked.
The intervals of potential pay are in wells not registered as hydrocarbon-bearing when they were drilled; however, subsequent analyses suggest hydrocarbons could be present.
The NPD contributed to a study commissioned by UK authorities and completed last year which revealed that numerous wells contain the potential for missed pay.
Earth Science Analytics (ESA), on assignment from the NPD, has now reviewed 545 wells in the North Sea. To date around 1,250 exploration wells have been drilled in the North Sea, the NPD added, of which more than half were registered as dry.
“By releasing the analyses and underlying data from the studies, we want to challenge the industry to use this to discover new resources and gain new geological insight,” said Torgeir Stordal, director technology and coexistence at the NPD.
Well analyses can be time-consuming, but using new digital tools that employ AI and machine-learning algorithms, the industry could perform analyses for large volumes of data in a relatively short time span.
ESA also integrated seismic with well data in the Fram area in the northern Norwegian North Sea, creating various 3D datasets with the following reservoir properties: lithology, porosity, and water saturation.
The initial intention is to test a new technology that allows for quicker and less costly integration of well and seismic data that could provide new information about the subsurface, benefiting exploration and development of new and existing oil and gas fields.
ESA recorded potential missed pay intervals when criteria had been satisfied for good reservoir properties and possible hydrocarbon storage, identifying around 350 such intervals based on thickness.
The NPD plans to release the data from this program. However, the Fram area datasets will initially only be used for internal analyses.
07/07/2021
