Generative Well-Pattern Design Presents Opportunities To Reduce Drilling Expenditures

Abstract

This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper SPE 203152, “Generative Well-Pattern Design Applied to a Giant Mature Field Leads to the Identification of Major Drilling Expenditure Reduction Opportunities,” by Maddalen Lepphaille, Total; Arthur Thenon, Modis; and Pierre Bergey, Total, et al., prepared for the 2020 Abu Dhabi International Petroleum Exhibition and Conference, Abu Dhabi, held virtually 9–12 November. The paper has not been peer reviewed. A generative well-pattern-design (GWPD) work flow was benchmarked against traditional manual designs to leverage three reservoir-development planning opportunities applicable to a giant mature Middle Eastern carbonate field. People remained central in ensuring efficient problem setup and exploration guidance, but the work flow proved able to identify substantially better patterns than the traditional approach for each of the opportunities at the cost of only a few hundred simulations. GWPD Overview The work flow used to tackle the different problems of this study, which the authors call the GWPD “well-improvement scheme” (WISH), consists of the following steps. Each step is detailed in the complete paper. 1. Definition of design space 2. Constraint of design space 3. Qualification of design space 4. Nondominated sorting (a specific ranking of all of the cells of the con-strained design space according to the value of their quality indicators) 5. Candidate design investigation 6. Investigation of preferred designs 7. Optimization of preferred design GWPD Application Context. The application study was con-ducted at the beginning of the industrialization of WISH, a proprietary software tool dedicated to GWPD. The authors call the work flow GWPD-WISH. In the studied oil field, more than 400 oil producers and water- or gas-injector strings have been drilled from approximately 100 platforms in a series of reservoirs. The study focused upon two specific reservoirs holding most of the field reserves. These reservoirs are developed with peripheral water injection and gas injection into the gas cap. According to the latest development plan, hundreds of wells will be drilled during the next 40 years in order to maintain a production plateau. The context was deemed favorable because a 3D gridded dynamic reservoir model was available, the geology and development history defined a large and complex design problem, and large liquid hydrocarbon reserves were thought to remain. While the software and method used enables considering an ensemble of realizations capturing reservoir uncertainties, only a single history-matched realization of the model was available. Consequently, the study did not deal with reservoir uncertainties.