Fast inversion method of two-dimensional nuclear magnetic resonance based on random singular value decomposition (RSVD)
Li Haitao,Deng Shaogui,Hu Xufei,Yuan Xiyong,Zhang Pan
DOI: https://doi.org/10.1190/segam2019-3211181.1
2019-01-01
SEG Technical Program Expanded Abstracts
Abstract:PreviousNext No AccessSEG Technical Program Expanded Abstracts 2019Fast inversion method of two-dimensional nuclear magnetic resonance based on random singular value decomposition (RSVD)Authors: Li HaitaoDeng ShaoguiHu XufeiYuan XiyongZhang PanLi HaitaoSchool of Geosciences, China University of Petroleum (East China), Qingdao 266580, ChinaSearch for more papers by this author, Deng ShaoguiSchool of Geosciences, China University of Petroleum (East China), Qingdao 266580, ChinaSearch for more papers by this author, Hu XufeiSchool of Geosciences, China University of Petroleum (East China), Qingdao 266580, ChinaSearch for more papers by this author, Yuan XiyongSchool of Geosciences, China University of Petroleum (East China), Qingdao 266580, ChinaSearch for more papers by this author, and Zhang PanSchool of Geosciences, China University of Petroleum (East China), Qingdao 266580, ChinaSearch for more papers by this authorhttps://doi.org/10.1190/segam2019-3211181.1 SectionsSupplemental MaterialAboutPDF/ePub ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InRedditEmail AbstractAims to accurately evaluate the formation fluid and make up for the limitations of one-dimensional nuclear magnetic resonance logging in practical applications, twodimensional nuclear magnetic logging has long been the focus of research. Increasing the sequence and inversion parameters of 2D NMR can effectively improve the antinoise and resolution of the inversion, but at the same time it will reduce the inversion speed and significantly increase the computer memory occupancy, which will put forward higher requirements for computer configuration. The computational cost poses a challenge to traditional 2D NMR inversion algorithms. Based on this, a new fast 2D NMR inversion method based on random singular value decomposition (RSVD) is proposed. The author first compares the computational efficiency and memory occupancy of RSVD and traditional SVD, and takes the oilwater and gas-water model as an example to analyze the twodimensional nuclear magnetic inversion effects of (T2, D), (T1, T2) in detail, and (T1, D) are also discussed with multiple sets of echo spacing (TE) and waiting time (TW). Numerical simulations show that compared with currently used LSQRSVD algorithm, RSVD can significantly reduce memory requirements, speed up the inversion, and ensure the accuracy of the inversion, effectively reducing the computation time and hardware cost. Due to the significant advantages of the algorithm in terms of computational speed and memory requirements, the application of twodimensional and even multi-dimensional NMR logging in practice can be further promoted.Presentation Date: Wednesday, September 18, 2019Session Start Time: 8:30 AMPresentation Time: 8:30 AMLocation: 217CPresentation Type: OralKeywords: 2D, inversion, nuclear magnetic resonance, optimization, high-resolutionPermalink: https://doi.org/10.1190/segam2019-3211181.1FiguresReferencesRelatedDetails SEG Technical Program Expanded Abstracts 2019ISSN (print):1052-3812 ISSN (online):1949-4645Copyright: 2019 Pages: 5407 publication data© 2019 Published in electronic format with permission by the Society of Exploration GeophysicistsPublisher:Society of Exploration Geophysicists HistoryPublished Online: 10 Aug 2019 CITATION INFORMATION Li Haitao, Deng Shaogui, Hu Xufei, Yuan Xiyong, and Zhang Pan, (2019), "Fast inversion method of two-dimensional nuclear magnetic resonance based on random singular value decomposition (RSVD)," SEG Technical Program Expanded Abstracts : 804-808. https://doi.org/10.1190/segam2019-3211181.1 Plain-Language Summary Keywords2Dinversionnuclear magnetic resonanceoptimizationhigh-resolutionPDF DownloadLoading ...