Generative Well Pattern Design Applied to a Giant Mature Field Leads to the Identification of Major Drilling Expenditure Reduction Opportunity

Abstract

Abstract Well pattern design is difficult due to the very large number of possible solutions, the complexity of constraints deriving from drilling and completion, the nonlinear nature of fluid flow in porous media, the difficulty in ascertaining reservoir properties and, for mature reservoirs, development history. What matters the most is case dependent. This task is therefore traditionally conducted through manual processes with little or no help of computers, leading often to exceedingly conservative or simplistic designs whatever the reservoir heterogeneities. The increase in computational power and algorithmic advances are opening the door to a new "Generative Design" approach, already used by other industries (car, aerospace). It consists in exploring a larger number of computer generated design possibilities more quickly and efficiently than what human can do by combining: i- "technical and experience" rules to automatically build a large number of "designs candidate", ii- workflows that qualify the performance of the designs, iii- selection criteria to identify the best design(s). An innovative Generative Well Pattern Design Workflow named GWPD-WISH was benchmarked against traditional "manual" designs to leverage three reservoir development planning opportunities applicable to a giant mature middle eastern carbonate field already developed by hundreds of wells for which a reliable model was available: Locate 2×15 in-fill producers to be drilled from 2 freely chosen platform locations Locate 2×15 in-fill producers to be drilled from 2 pre-determined platform locations Select 16 multi-string water injector wells for re-drill to improve recovery through better control of local reservoir pressure balance The study was conducted using a large operational compositional model considering complex constraints in a limited time. The workflow proved able to identify substantially better patterns than the traditional approach for each of the three opportunities at the costs of only few hundreds of simulations. Pattern improvement was measured in term of reserves per incremental well and plateau duration extension. It corresponds to an opportunity for reducing drilling expenditures on a rolling basis by 30% or more.