About
Building on the breakthroughs made in the field of computer science over the last decade, operating companies and service providers have shown a renewed interest in computational statistics, automation and self-learning algorithms. However, unlike a number of other industries, significant segments of the E&P still require substantial transformations before IOT-ready and self-updating digital oil fields become reality – a transition for which the subsurface realm is no exception.

While a successful implementation of IR 4.0 requires reservoir and production specificities to be acknowledged, various solutions are available to overcome current showstoppers such as restricted data access, spatio-temporal autocorrelation and small size sample. What is special about spatial data? What are the alternatives when limited labelled data is available? And why cannot we simply let the data decide? are among the questions which will be discussed and supported by a number of references and case studies.

Concluding on the need for humans-in-the-loop for sense-checking and accountability, the talk will touch upon how hybrid physics-based machine-learning approaches can speed up projects’ learning curve and increase generalizability of statistical predictions. A selection of open-source resources, books and code libraries will also be provided.
Presenter
1597241255-1b3f7dfdc00842c0
Nicolas Leseur
Petroleum Engineering Manager, Baker Hughes
Nicolas Leseur is the regional Petroleum Engineering Manager for Baker Hughes, looking after the execution of surface and subsurface reservoir consulting services in MENATI. He holds a master’s degree in Geological Engineering from the E.N.S.G., France, and started his career as a Geomodeller in Canada after that, he later worked with various operators in Bahrain, India, Saudi Arabia, Nigeria and UAE.

Throughout his career, Nicolas held various positions such as Sr. Reservoir Petrophysicist and Development Geologist with a particular emphasis on probabilistic multi-phase flow petrophysics, pore-scale-to-field-scale modelling and multi-variate 3D geostatistical modelling of static and dynamic reservoir properties. Besides its collaboration to various field development plans, prospect generation and the economic evaluation of new ventures, Nicolas’ current interest lies in exploring how the recent advances in machine-learning can jointly be used with physics-based approaches to achieve more precise and accurate geo-predictions.
Register To Watch Recording
Full name*
Email Address*
Country*
Company*
Job Title*
We use BigMarker as our webinar platform. By clicking Register, you acknowledge that the information you provide will be transferred to BigMarker processing in accordance with their Terms of Service and Privacy Policy.