top of page


Demonstrating the potential of AI assisted exploration and production studies

Commentary from Eirik Larsen, Chief Solutions Officer, Earth Science Analytics
17 January 2022

Demonstrating the potential of AI-assisted exploration and production studies. Eirik Larsen, Chief Solutions Officer, Earth Science Analytics, comments on ‘potential hydrocarbon pay’ and 3D property predictions.

Earlier in the summer, we were pleased to announce the extensive work we had been undertaking in collaboration with the Norwegian Petroleum Directorate (NPD) over the previous months. The work consisted of a comprehensive study utilising the power of EarthNET’s AI systems to generate3D property predictions for 545 Norwegian North Sea wells which had been identified as having potential ‘missed pay’ opportunities. At Earth Science Analytics, we prefer to label these reserves as ’potential hydrocarbon pay’, as it is not always the case that they have been missed, rather it is often that operators lack the information needed to demonstrate that reserves can be profitable. This is where EarthNET comes in.

The software was used on public well logs and CGG owned seismic, along with previous database learnings, to provide focused well studies on an enormous scale. The result: predictions of potential hydrocarbon pay within close proximity of existing wellbores and infrastructure. In real terms this meant a redefinition of the potential profitability of reserves, and a demonstration of the potential for operators to extract reduced cost and reduced risk production from existing projects.

More widely, the study offered an insight into the huge potential of AI to assist in oil and gas exploration, production, and field management by identifying new insights, streamlining workflows, and utilising the capabilities of non-biased machine led predictions.

So how did this come about and what does it mean for the industry?

In 2020, we took part in a project with the UK’s Oil and Gas Technology Centre (OGTC) to begin looking at potential hydrocarbon pay opportunities across the wider North Sea region. The project utilised EarthNET software to reveal areas of opportunity in c. 5000 wells. This was the ideal project to highlight the significant value that can be extracted when geoscience, data science and computer science are combined.

With the best data, the best software, and expert inputs, EarthNET generated property predictions for vast areas, with no compromise on the detail provided. The success of this project led the NPD to approach us to complete the picture of potential hydrocarbon pay on a further 545 Norwegian North Sea wells, and importantly, to predict hydrocarbon reserves away from the wellbore.

What came next was many months of hard work and collaboration between experts at Earth Science Analytics and the NPD. Among those working on the project was our Principal Data/Geoscientist Daniel Stoddart who was responsible for coordinating the workflows, QC’ing data, and presenting the outputs from the vast well and seismic datasets that were provided by the NPD.

These ML-driven workflows can easily be embraced by explorationists and domain specialists alike, allowing multi-skilled teams across the sector to reimagine their company's exploration strategy. The result is an optimisation of existing resources, which could pave the way for modernised decision-making processes in oil and gas exploration, development, and production.

As a cloud-native technology, EarthNET was able to seamlessly process the huge datasets provided by the NPD, which when combined with the platform’s insight and expert analysis, meant that the models could generate balanced and accurate predictions. This was confirmed by the comparison of EarthNET’s predicted vs actual findings on the 2009 Grosbeak discovery, which had been classified as a ‘missed pay’ opportunity.

With these two studies, we have developed a dynamic database of valuable well-data across the North Sea. This extensive feed of data has facilitated a continual enhancement of the capabilities and quality of EarthNET’s outputs and predictions. These capabilities indicate a huge potential for EarthNET to provide operators with the most accurate predictions of subsurface properties in potential hydrocarbon pay contexts using machine learning.

It has been a fantastic opportunity to work with the NPD to further explore the application of our industry-leading EarthNET software to predict hydrocarbon pay on such scale. The results demonstrate a roadmap to unlocking opportunities, which could revolutionise the way operators view hydrocarbon pay potential in the field

We look forward to further testing EarthNET’s abilities to push the boundaries of data-driven geoscience in the future.

bottom of page