Title: A Digital Oil Field Solution for Unconventional Well Performance Analysis Combining a Hybrid Data and Physics Approach
In unconventional reservoirs, understanding well performance is essential to drive value but also very challenging.
Traditional forecasting methods (such as decline curve analysis and its variants) are often not fully representative and impacted by surface operations (e.g. constrained flow, choke changes etc.) or subsurface events (e.g. well interference, frac hits, depletion below saturation pressure, etc.). Analytical and numerical modeling methods address these issues by applying first principles and simplified physics, but they are also quite interpretive through not scalable manual analyses, requiring additional reservoir inputs that are often not collected or known for all wells. Consequently, these methods are not suitable for large scale, repetitive forecasting.
We desire a reliable, consistent and scalable well performance analysis method that can be usable from a practical standpoint, working with routinely measured data (i.e. surface pressures, flow rates and fluid data).
A data-driven procedure for estimating PVT properties for all wells is applied to convert surface pressures to bottomhole pressures. Pressure depletion and well deliverability are estimated using a novel continuous drainage volume estimation method based on the concept of diffusive time of flight in conjunction with material balance. A predicted rate profile can then be obtained from the productivity and depletion trends. Finally, a new stochastic method is used for a reliable uncertainty quantification in production forecasting.
The proposed workflow has been successfully applied to more than a thousand wells in a major unconventional field through an automated digital oil field platform, after applying necessary data processing and cleansing procedures. We present value-driven case studies demonstrating tangible ways in which these solutions can be successfully implemented.
Diego is a petroleum engineer with nine years of international oil and gas industry experience, and currently works with Anadarko's Advanced Analytics and Emerging Technologies (AAET) team, helping the company adopt data analytics and develop digital technologies to enable corporate goals. Diego is the product owner for IPSO Onshore, an in-house digital oilfield solution designed for well performance analysis and production optimization in unconventional fields, which has been successfully deployed in multiple business units in North America.
Diego has completed a MS Analytics degree from Texas A&M University, a MS in Petroleum Engineering in Italy and a BS Petroleum Engineering degree in Venezuela.
Title: Using AI to Extract and Structure Data from Documents
The critical strategic and tactical decisions that E&P companies take are based upon written interpretations of heterogeneous quantitative data, from proprietary internal and published external reports and slide presentations. Similarly, much non-technical content upon which E&P companies depend, such as contracts, lease agreements and regulatory filings, is naturally unstructured. In both cases, these documents have immediate value upon initial use; however, they quickly fade into the clutter of a company’s document management practices, and their information, knowledge, and learnings—the implicit value that the documents embody—are lost.
While the industry has used platforms that manage, search and analyze structured data for many years, platforms that manage text documents are still maturing, and largely provide capabilities centered upon indexing, versioning, searching, annotating (i.e., with metadata), and enforcing document retention and disposal policies. A key missing feature in most content management systems is the ability to enrich text by “reverse engineering” latent information from it.
We describe an AI Platform, an artificial intelligence application that uses natural language analysis, computer vision, and domain knowledge to structure and classify text, recognize entities, and extract their properties. The platform, which implements the SPE’s Research Portal, adopts a knowledge-centric approach to automated text enrichment, augmented by input from subject matter experts and from statistical machine learning. In recent work, we have extended the platform to detect document structure and to recognize, interpret and extract data points from text and tables. Our long-term goal is to be able to answer questions by referencing the knowledge that we can find and extrapolate from a large corpus of documents.
Eric Schoen is the Chief Technical Officer of i2k Connect Inc. Before joining i2k Connect, Eric spent over thirty years at Schlumberger, in both research and engineering functions, most recently as its Chief Software Architect. At Schlumberger, he contributed to a broad range of software, from the company’s early pioneering efforts to leverage knowledge-based systems, its GeoFrame and Ocean platforms for reservoir characterization, its software quality processes, and its strategies for enterprise-scale architecture for data acquisition, transmission, processing, and delivery. Eric holds a Ph.D. in Computer Science (Artificial Intelligence) from Stanford University.