Stochastic inversion of facies and reservoir properties based on multi-point geostatistics. (3rd August 2018)

• Record Type: Journal Article
• Authors: Liu, Xingye; Li, Jingye; Chen, Xiaohong; Guo, Kangkang; Li, Chao; Zhou, Lin; Cheng, Jiwei
• Is Part Of: Journal of geophysics and engineering. Volume 15:Number 6(2018:Dec.)
• Page Start: 2455
• Page End: 2468
• Publication Date: 2018-08-03
• Subjects: stochastic inversion -- multi-point geostatistics -- reservoir characterization -- probability perturbation method -- quantum annealing algorithm; Geophysics -- Periodicals; Prospecting -- Geophysical methods -- Periodicals; Engineering -- Periodicals; 622.1505
• DOI: 10.1088/1742-2140/aac694
• Languages: English
• ISSNs: 1742-2132

Abstract

Abstract: A stochastic inversion method that can provide reservoir models of facies and reservoir properties is proposed for seismic reservoir characterization. It is an iterative optimization process achieved by the probability perturbation method on the basis of conditional simulation with multi-point geostatistics, and it also incorporates a quantum annealing algorithm to improve the accuracy of inverted reservoir properties. The usual practice of stochastic inversion based on multi-point geostatistics is time-consuming and only generates the inverted facies. The probability perturbation method can help to reduce the number of simulations and speed up the inversion process. An inverted model of facies is generated, and a volume of pseudo reservoir properties met with a fixed error is also produced in this stage. However, the continuity of this volume is poor because random sampling is used when transforming facies models into reservoir property volumes. Moreover, the characteristics of reservoir properties in the same facies cannot be reflected well. We extract the volume of pseudo reservoir properties as an initial model for optimization to improve the inversion accuracy of reservoir properties. A quantum annealing algorithm is utilized for its fast convergence rate and ability to avoid falling into a local minimum. The presented method can produce fine-scaled reservoir models of facies, and produce reservoir properties in a reasonable amount of computation time. Tests on model data demonstrate the feasibility and reliability of this technique. Facies and porosity inversion results with relatively higher accuracy are acquired, which verifies the advantages and feasibility of the proposed method. Successful application to field data further validates the effectiveness of this method.