Abstract

The predictive ability of software continues to advance, but one of the weak points is still data quality coupled with the audit trail. While there may be an overwhelming amount of data, it’s only as good as the sensors/devices that collect it and the software along with the methodologies (“recipes”) that interprets it. These can be hampered by the level of the directness of the measurements, environmental and mechanical interference but are also vulnerable to security risks (external and internal). We will discuss some of these elements and how to structure predictive analytics to avoid reliance on erroneous or false information. In addition, we will also show some elements of application of data analytics techniques on the differential operators of the transport equations.

We will show specific examples in the areas of fluid properties and relative permeability data which are used in multi-disciplinary applications, from reservoir evaluation to reservoir performance management. Interpreted material is utilized, in a broader sense, to predict flow in porous media for various applications. We will look at various aspects of multi-faceted data quality and how to apply the physics-based content knowledge and data-centric interpretation to ensure modeling fidelity. For the final goal, which is the generation of the use case(s) in an integrated manner, following steps needs to be covered:

• Current Landscape and workflows:
  • Collection of relevant data
  • When and where to collect the data
  • Metadata
  • Data sharing, security, use and its effect on the end-product
  • Quality/Quantity (how do we measure these?)
  • Compatibility
  • Balancing act: Technology-Daily Work-Culture
• Deficiencies in physical and correlative models
• Identifying quantitative proxies. Do they exist?
• Product: Physics augmented/hybrid approaches by examples (will also include some of the deployed versions of our work via web-apps)

Bio

Dr. Birol Dindoruk is currently AADE Endowed Professor of Petroleum Engineering at University of Houston, previously he was the Chief Scientist of Reservoir Physics and the Principal Technical Expert of Reservoir Engineering in Shell.

His technical contributions have been acknowledged with many awards during his career, including SPE Lester C. Uren Award (2014), Cedric K. Ferguson Medal (1994), and Distinguished Membership. In 2017, he was elected as a member of the NAE for his significant theoretical and practical contributions to EOR & CO2 sequestration.

He was one of the Distinguished Lecturers of SPE for 2010-2011 term.

Dr. Dindoruk was Data Science and Engineering Analytics Technical Director of the SPE and a member of the Advisory Committee of the SPE Reservoir Dynamics and Description Technical Discipline. He has been active in various editorial positions under SPE and also Elsevier. Currently he is the Editor In Chief for all SPE Journals and as well as Editor In Chief of Journal of Natural Gas and Engineering of Elsevier.

Dr. Dindoruk is well-known for his extensive work on thermodynamics of phase behavior/EOS development and experimental work, interaction of phase behavior and flow in porous media, enhanced oil recovery and CO2 sequestration, and correlative methodologies.

Recently, Dr. Dindoruk has also been working in the area of data analytics, artificial intelligence, and machine learning and focusing on effective incorporation of data sciences into the oil and natural gas industry practices and energy systems. In recent years, he has authored/co-authored various articles for hydrogen, geothermal systems and adsorptive storage.

Dindoruk has 28 years of industrial experience, holds a BSc Degree from Technical University of Istanbul in Petroleum Engineering, MSc Degree from The University of Alabama in petroleum engineering and also a PhD from Stanford University in Petroleum Engineering and Mathematics, and an MBA from University of Houston.