Evergreen Forecast & Predictive LTRO Using Machine Learning – Case Study from PDO South

Abstract

Abstract Digital transformation (Dx) is increasingly becoming a key enabler in oil and gas industry to reduce costs, make faster and better decisions and increase productivity. The difference between leading the next innovation wave or being left behind may depend on how proficiently we embrace digital enablers, and how quickly we can test, prototype and scale these digital solutions to create value for the business. Digital technologies are not new to Petroleum Development Oman (PDO). In fact, the company has a track record of testing and adopting a wide range of new technology and integrated organisational capabilities to improve its business performance. Significant investments have been made into instrumenting its fields, including the IT infrastructure, Real-Time Operations, Exception Based Surveillance, Collaborative Work Environment (CWE), Smart Fields, NIBRAS, data management, analytics trials, to name a few. Yet consensus that Dx has significant further upside across PDO, led to the initiation of an asset-led pilot to digitally transform an existing PDO South Field – "S". The focus of the pilot was to identify new Dx opportunities while leveraging on existing PDO investments into digitalization, leading to quantified improvement in business performance of field –S. The project workscope was based on the outcome of an Opportunity Framing Event (OFE), in which a total of 27 opportunities were identified and ranked in terms of business value vs. feasibility or cost of implementation (Figure 1). Technical Subject Matter Experts (SMEs), asset field - surface, sub-surface, data management teams and other relevant support functions participated in the OFE so that business improvement synergies could be identified across the multiple disciplines in an integrated fashion. Following an agile approach, 5 Valuestreams (VS) were selected for Minimum Viable Product (MVP) implementation, in phase 1 of the pilot. Focus of this paper, however, is to elaborate further only one of the 5 VSs i.e. use of machine learning for "Evergreen Production Forecast for Field Development Plan (FDP) optimization and Locate the Remaining Oil (LTRO)".