First EAGE/PESGB Workshop on Machine Learning

Date
29 - 30 November
Location
London, United Kingdom
Registration
Open
Call for papers
Closed

Keynote Speakers

Thursday, 29 November 2018

13:00-17:00 hrs

The first day of the workshop will consist of keynote presentations by innovators in the field. Scroll down for an overview of the titles, biographies and abstracts of each talk.

 


Artificial Intelligence in E&P - Perspectives on an upcoming exciting journey! 
Yves Le Stunff 
Digital Officer Subsurface/Digital Coordinator E&P, TOTAL

Biography: Yves Le Stunff has been working for Total Exploration Production for more than 20 years. As a geoscientist he has held various positions both in operations and in research & development. He has been involved in Exploration and Field Development activities in various countries.

From 2012 to 2015 he has been in charge of Prospective and R&D Strategy for Total E&P. In September 2015 he has been appointed Digital Officer and is in charge of developing Artificial Intelligence and Machine Learning for E&P activities.

He is a graduate in Physics from the Ecole Normale Supérieure of Paris (France) and holds a PhD in Geophysics from the University of California at Berkeley (USA).


Achieving Success in AI Adoption - What Can We Learn From Other Industries?
Mark Roberts
Analytics Consultant and AI Lead, Tessela

Abstract: In some industries, such as the life-sciences, AI and machine learning have a long and successful history having been used in R&D and operations for many decades. The recent explosion in interest in AI means that applications are emerging across all industries, often in areas that have varying readiness and maturity levels to take on this transformative technology. Given this disparity, it is prudent to try to learn lessons by examining how other more mature industries are using AI. In this talk we will examine the typical maturity curve of AI adoption, discuss the level at which different industries are currently operating, and look at what lessons can be learned by examination of the field as a whole. Based on these observations, we can draw some general conclusions about what is required to be successful in the adoption of AI, many of which involve cutting through the current hype, recognising the decades of best-practice that already exist, and employing rigorous engineering principles to the design, deployment and support of AI and machine learning solutions.


How Do We Realize The True Value From Artificial Intelligence?
Steve Freeman
Director of Artificial Intelligence and ML, Schlumberger 

Biography: After gaining a PhD in Geology from the University of Leeds in the UK, Steve worked within an O&G and minerals exploration and development consultancy. He became technical director in 2007, specializing in software solutions, before becoming CEO in 2012. In 2014, the group was acquired by Schlumberger, where he became a chief scientific advisor for SIS. Becoming VP of Digital Solutions in 2016, he was tasked with development and delivery of Schlumberger’s future software cloud solutions in the form of DELFI in 2017. Steve’s current focus is in developing and deploying Artificial Intelligence solutions.

Abstract: Our industry continuously faces the challenge of delivering greater productivity and efficiency gains, while at the same time demonstrating ever greater safety and environmental sustainability. Artificial intelligence (AI) is one of the main enablers that has been identified as key to unlocking this future state.

As an industry, there is a mindset and technology shift that needs to happen in order for us to reap the full value of AI. We need to fully understand how we can develop, deliver, and embrace this technology to create better outcomes. In this presentation we will explore many aspects around these challenges and steps that we can take as an E&P community to ensure that we all benefit and take part of this technology evolution.

There are several components that need to come together to yield the value that our industry is expecting. These include i) ensuring the technology is fit-for-purpose for our industry, and if so ii) how to scale the solutions across the industry, then, iii) how can we gain broad adoption.

It is only when all these pieces come together that we will see a real step-change in performance and insights. These components are complex and involve not only technology but changing deep routed processes, workflows, and—probably most critically—cultural perceptions and responses.

Our experience to date has been that AI can drive substantial and meaningful insights across all domains within our businesses, this includes Petrophysics, Geophysics, Geology, Drilling and Production. We, as an industry, are on the journey to explore how those capabilities can best be integrated and scaled to realize the true value.


Geocomputing Skills for Digital Geoscientists
Matt Hall
(Founder & CEO, Agile) & Jo Bagguley (Principal Regional Geoscientist, Oil & Gas Authority)

Biography Jo Bagguley: Jo Bagguley is a geologist who has been working in the oil and gas industry for 20 years. Throughout her career Jo has worked for a number of operating companies primarily in exploration but has also spent time working on both operated and non-operated producing assets. Jo joined the OGA as their Principal Regional Geologist in March 2016 and now leads their regional exploration team in delivering the OGA's regional exploration initiatives across the UKCS.

Abstract: Significant volumes of industry data collected as part of the lifecycle of hydrocarbon exploration and production are held in large, often unstructured, data stores where the data held is variable in vintage, quality and format. For the subsurface practitioner, this makes it difficult to extract and analyse the data and thus significant potential value remains inaccessible to exploration and production evaluations at both the regional, prospect/discovery and well levels. It has been demonstrated by some UKCS operators that the application of various machine learning techniques can help unlock the potential value from unstructured data but this approach has not yet been widely adopted by the diverse cross-section of companies working in the UK.


Moore's Law in Geo Machine Learning?
Eirik Larsen
Founder & CEO, Earth Science Analytics

Biography: Dr. Eirik Larsen is cofounder and CEO of Earth Science Analytics, a company focusing on the commercial application of AI in petroleum geoscience (earthanalytics.ai). He has 19 years’ experience from the E&P industry. He has held various technical and managerial roles in oil companies and consultancy firms including Statoil, Rocksource, and Geokonsulentene. He has experience from research, exploration, field development, and production on the Norwegian Continental Shelf as well as internationally. He holds a PhD in Petroleum Geology from the University of Bergen and is now laser focused on implementation of AI in petroleum geoscience.

Abstract: Machine Learning (ML) has been capable for three decades, to infer lithology, sedimentary facies, porosity, and fluid saturation as functions of wireline logs. Now, ML is moving from R&D projects and into the tool box of the generalist, transforming the subsurface workflow. In addition to being fueled by algorithmic development, data, and high-performance compute; this transformation is enabled by the emergence of data analytics platforms, that facilitate; i) practical use of ML methods by the generalist geoscientist, ii) integration of data analytics with structured data in databases, iii) semi-automated data management, quality-control and -improvement, and iv) tracking of data provenance, enabling reproducible scientific workflows.

At well scale, the generalist geoscientist can now make reliable predictions of rock and fluid properties derived directly from data, quickly and repeatably without needing to rely on approximate models or inefficient and error prone manual interpretation. Geoscientists now have tools that they can seed with their expertise and creativity, and use to interrogate data and answer questions. We can now start to appreciate the impact of this development in the daily business of exploration and production teams.

At seismic scale, we see the beginnings of ML being used for both prediction of geological structure and of rock and fluid properties. Initially this is happening though ML based approaches to inversion and seismic interpretation. As researchers apply ML to known problems, new ways to constrain and quantify uncertainty are also emerging.

Beyond these initial steps, and as development accelerates, there is a huge and real opportunity to redefine subsurface work. By embracing this technology, which could and probably should change how we interact with our data, we will learn new ways to incorporate our expert knowledge and change our relationship with subsurface uncertainty. Leveraging such composite data analytics methods and workflows, we can start making truly data-driven decisions.