Title: Machine Learning with Physics & Data Driven Modeling

Why Understanding Advanced Analytics and Machine Learning is Important for Engineering Professionals

Abstract:

Digital data collected by many industries support many, if not, most of its activities.  Data integrity, security, mining, analysis, and transfer are critical to its particular goal of providing insight.  Related topics such as the use and management of massive data sets (“big data”), data value and ownership, cybersecurity, cloud computing, machine learning, and virtual twin modeling and simulation represent only a small subset of the derivative uses of digital data being contemplated by all industries. Particular questions to be answered by data analysts, data scientists, and/or subject material experts are:

• Once data has been collected, what do we do with it?


• How do we extract knowledge and value from collected data to benefit operations, gain efficiencies, and improve safety and security?


• What are the business propositions that sensed data collection, storage, and analysis bring to the table?


• How effectively is data analyzed?


• What data analysis methods and tools should the industry be adopting that aren’t now being used?

This talk attempts to highlight the importance of advanced analytics and machine learning as a means to provide answers to these questions, thus providing valuable insights with business value. The future of Big Data and Data Analytics is not only for the Energy Industry (all forms of energy), but all industries in general.

Bio:

Ed Marotta has held faculty positions as an Assistant Professor at Clemson University (1997) and Associate Teaching & Research Professor at Texas A&M (2003), all within the Mechanical Engineering Department. He has held the position as Technical Manager for the Multi-Physics Simulation Group within the North America Technology Center, FMC Technologies Inc. In this capacity, he was responsible for developing a Center of Excellence for modeling and simulation of multi-physics phenomena for Surface and Subsea applications. Ed led the Systems Analytics, Modeling & Analysis group within GE HQ where data-driven and physics models were developed for System of Systems analysis. Most recently, Ed was the Chief Advisor for Landmark Graphic Inc. strategy for development of Digital Twins for reservoir, drilling, and production operations.

In addition, he has published over 100 Journal, Conference, and white-paper papers within the area of Thermo-Fluid Sciences and Digital Twins, and holds numerous patents. He currently holds the position of Adjunct Professor with the ME (Subsea Engineering Program) and MET departments at the University of Houston teaching courses in Data Science and Computational Methods.

Ed received a B.S. in Chemistry from the University of Albany (SUNY) and a M.S. and Ph.D. in Mechanical Engineering with specialization in Thermo-Fluid Sciences from Texas A&M University. He holds the grade of Fellow in the American Society of Mechanical Engineers (ASME) and Associate Fellow in the American Institute of Aeronautics and Astronautics (AIAA). Also, Ed formerly held the position of Associate Editor for a major ASME journal. He is actively involved in local ASME Chapters as well as the ASME & SPE OTC technical subcommittees. Ed has been tasked with leading the ASME Petroleum Division initiative on Big Data & Digital Transformation as the Chair for development and creation of guidelines for the Oil & Gas Industry.

Title: A Robust Approach to Post Well Drilling Data Analysis at Scale

Abstract:

In today’s fast paced drilling environment, emphasis is on execution. Many operators and service providers are using real time analytics solutions to optimize and improve drilling. However due to the lack of real time streaming of high resolution downhole data from each tool, the operators and service providers are unable to consistently capture, analyze, aggregate data and learnings after each run. These learnings are lost and cannot be transferred to the next well and monitored real time. Furthermore, it takes tremendous effort to clean, organize, manipulate, and structure the data well-after-well to be able to make comparisons and aggregate statistical expectations. This has led to invisible lost time (ILT) becoming a leading source of inefficiency for our industry.

The authors present a novel cloud based platform that addresses this issue. This platform is loaded with data that characterizes the drilling environment as well as drilling performance.  This data includes drilling operating parameters and performance data such as WOB, RPM, ROP, and many other characteristics from the Surface Electronic Data Recorder or EDR, information that characterizes the formations being drilled as well as the geometric profile of the well and finally, downhole dynamics measurements. Such comprehensive data builds a rich downhole view of how efficiently the energy being put into the system is being consumed to drill. The authors will discuss various technical issues they encountered building such a system and algorithms used to clean and organize the information. Case studies presented will show how analytics performed on the clean data create a robust roadmap to drill more efficiently. The authors will also show examples of how the use of this platform has led to improved drilling performance.

Bio:

Chaitanya Vempati is Digital Development Manager, Drill Bits product line at Baker Hughes. In his current role Vempati leads the digital development group that develops the in-bit measurement platform and digital tools that derive insights through data driven approaches. He is passionate about challenging conventional thinking and developing innovative products and solutions. Vempati holds a Master’s degree in Mechanical Engineering from University of Texas at Austin focusing on product design and innovation, artificial intelligence and neural networks, graph theory and numerical optimization. He has co-authored five publications and holds 15 US patents.