EAGE Digital 2022

OSDU and Total’s digital transformation. IFPen’s open source ELK-based geo data framework. INT’s OSDU front-end. Naftagas’ EPP (Esri not OSDU!). Schlumberger proposes ‘Shapeley additive explanations’ for AI black boxes. OMV – APC cornerstone of digitalization. Svalbox digital geology. EU Geological Data Infrastructure.

Mathieu Terrisse presented TotalEnergies’s digital transformation program that leverages the OSDU data infrastructure to underpin its in-house developed Sismage-CIG geosciences and reservoir platform. Terrisse’s presentation is a great sales pitch for OSDU, presented as ‘liberating and standardizing data’. While digital transformation is a long road, early results show promise. The OSDU wellbore domain data management services, DDMS, enabled scenario sharing between geoscientists and drillers, with bidirectional transfer of well information between the Sismage geosteering module and ‘DrillX’, another of Total’s in-house developments and a component of Adept, its ‘advanced drilling engineering platform’. At present OSDU’s live data exchange capability is limited as is version management but overall, OSDU has shown that different applications can collaborate on the same data ‘as per the theoretical model’.

Interestingly, Total presented Sismage-CIG before OSDU as ‘a platform with access to every piece of data’.

Tatiana Akimova presented another of Total’s OSDU trials, an ‘ambitious’ use case where an OSDU data platform running in the Microsoft Azure cloud was used in an eight month study of Total’s Suriname Golden Block 58. Here OSDU improved collaboration across drilling, geosciences and development accelerated exploration and appraisal drilling. The Suriname project saw Total extending the OSDU API with a ‘robust’ entitlement model to align with its security policies. Microsoft, Wipro, INT, Emerson and Thales were involved in the project which is described as ‘a major breakthrough for geosciences data management in Total’.

But not all things Total are OSDU. Antoine Bouziat (IFPen) presented a joint IFPen/TotalEnergies investigation into the use of opensource frameworks to manage geoscience knowledge. The ‘ELK’ stack* of open-source data management tools includes Elasticsearch (search), Logstash (data ingestion) and Kibana (visualization). ELK is used to manage and explore large collections of structured and unstructured data. The software is used in financial services and webstores but not so far in geosciences. The proposed architecture includes an Elasticsearch cluster and an FSCrawler application that scans and indexes files in the network. Other functions include language detection, de-dupe and OCR. The results are exposed in data-driven interactive visualization dashboards developed with Kibana. The authors are very positive as to the potential of ELK in the subsurface and encourage geoscientists to ‘consider open-source projects and hands-on programming experiments’. The work was carried out as a R&D collaboration between TotalEnergies and IFPen under the auspices of the Tellus consortium for geoscience digitalization .

* According to Wikipedia, ELK is now to be termed the Elastic Stack. While the components are open source, enterprise users may be interested in commercial bundles from Elasticsearch BV, Amazon and others.

Frederic Desloges (INT) gave an enthusiastic plug for OSDU that might have been taken as a commercial presentation if it was not for the fact that OSDU is ‘open source‘. Desloges has it that ‘moving data to the cloud is key to the future of the digital workspace. Companies ‘must be able to access their data in the cloud to generate insights quickly and make critical business decisions in real-time’. Enter the OSDU platform that makes it ‘easy to set up a secure data lake in just a few days’. For visualizing data in the lake Desloges unsurprisingly advocates INT’s IVAAP, a ‘modular platform’ (another one!) that supports ‘multiple sources of data such as OSDU, custom data catalogues, and data lakes’.

Kseniya Kultysheva reported on similar ‘fundamental’ data management difficulties which led Naftagas to kick off an ‘Exploration Platform Project’. This identified challenges that went beyond the ‘simple implementation or creation of a single software’. A novel approach to managing data across domains was proposed that leveraged, not OSDU, but a ‘geo-information web system’ that was already deployed in the legal department for land management. The product was rebaptized the ‘Exploration Platform’ and is under active development by Naftagas’ geoscientists.

OK so if it is not OSDU what is it? A little googling suggests that the ‘product’ used to manage Naftagas’ land is Esri ArcGIS Enterprise.

Schlumberger’s Amir Shamsa observed that while ‘ML solutions are now used everywhere’, they are frequently rejected by domains experts due to the ‘black box syndrome’. Shamsa proposes the ‘Shapley additive explanations’ (SHAP) visualization tool that computes the contribution of each feature to the model. The approach was used on an 800 wells dataset of the Duvernay shale gas field in Canada. The approach was tested on an Xgboost ML model of the field and used to identify the contribution of each modeling parameter to predicted production. The results were claimed to assist in selecting completion parameters and picking infill drilling locations. For more on the Shapley plot in oil and gas see Keith Holdaway’s LinkedIn paper Explainable Artificial Intelligence: The Shapley Values.

Georgia Kotsiopoulou related OMV’s deployment of advanced process control (APC) (a.k.a. MPC - model predictive control) as a ‘cornerstone of digitalization and operational excellence’. APC has been deployed at two of MV’s gas production facilities in an ‘almost completely remote manner’. The controllers are based on dynamic process models that autonomously optimize the system using real-time data. OMV carried out ‘thorough market research to find the supplier of the software’. Unfortunately the authors did not share what software was selected but the references point to earlier APC work by AspenTech and KBC/Yokogawa.

Kim Senger (University of Svalbard) presented ‘Svalbox’ a research tool for the study of digital geology outcrop models, samples and drill cores. Svalbox includes a subsurface project database built in Petrel along with virtual field trips to illustrate aspects of Svalbard geology. Apart from Petrel and ArcGIS, Svalbox uses ‘mostly open-source’ software available on GitHub. Svalbox is said to ‘bridge the gap between outcrops and subsurface data’ and exposes the Svalbard archipelago as a geoscientific playground.

Karen Lyng Anthonsen (Denmark & Greenland Geological Survey) presented the EGDI, the EU Geological Data Infrastructure that is to make European geological data accessible. EGDI was launched in 2016 and has since been extended with data and results from the GeoERA research projects. A ‘web GIS system*’ provides access to pan-European and national geological datasets connecting to platforms such as Destination Earth, the European Open Science Cloud (EOSC) , the European Plate Observing System (EPOS) and the EU Raw Material Information System (RMIS). EGDI also provides access to other geodata resources such as OneGeology-Europe (geological mapping), EuroGeoSource (energy and minerals) and others.

* A combination of OpenLayers, MapServer and MapScript.

More from the EAGE Digital home page. Proceedings (for those with a subscription) are on EarthDoc.

Click here to comment on this article

Click here to view this article in context on a desktop

© Oil IT Journal - all rights reserved.