Christian Hinz presented Math 2 Market’s Digital Rock Physics. DRP is described as the ‘digitalization of special core analysis’ (SCAL). Numerical SCAL generates ‘statistical digital twins’ of the reservoir that save time and support automated workflows, controllable with a Python interface. Users include Shell, Petrobras and OMV.
Gwenole Tallec presented ThermoFisher’s Athena software for rock
imaging data management, illustrated with a rapid core analysis use
case on data supplied by KAUST.
Computer tomographic (CT) core data and metadata is captured
automatically for remote access and visualization along with
annotations. The PerGeos core profile assembly wizard creates a single
core image from partial scans for facies determination. Digital SCAL on
micro CT scanned plugs also ran. More on Athena from ThermoFisher.
Daniel Austin presented Earth Science Analytics EarthNET, a.k.a. ‘the digital underground’. The cloud based suite of software tools supports 3D earth model creation from a range of data sources and scales, from micro CT up. Data management is achieved through EarthBANK, a cloud-native data platform with API links to OSDU and Petrel. EarthAI adds automated and semi-automated AI workflows. ESA is project leader for the Norwegian NOROG released well project to analyze drill cuttings samples from some 1,900 wells drilled on the Norwegian Continental Shelf.
A panel discussion led by RISIG Founder Ross Davidson heard Daniel Austin argue for open standard data formats (such as those used in seismics) for core data. Austin’s ambition is for OSDU to make core data exchange easier. Davidson compared OSDU to POSC’s standardization effort of the last century which failed because of the inefficiency of an open model. Austin agreed, this is a big challenge, but modern data formats are much better. ESA’s bricking data format for seismic machine learning comes from outside of the incumbent geoscience domain, leveraging novel approaches from the likes of VTK, Software Underground, PyVista and GemPy. Davidson observed that data management is a huge topic. Regarding management of rock imaging data, does it have to be done in the cloud? Tallec said yes. Training ML models on the desktop is hard. The cloud is the future. ESA has been cloud-native for six years. But what of security? Clients seem nervous. Is this paranoia or just being careful? It does not have to be a public cloud. Cloud-agnostic software can be deployed anywhere. For Hinz, more users are now in a cloud environment, either on prem or public.
More from the Rock Imaging SIG.
© Oil IT Journal - all rights reserved.