Microseismic inversion for anisotropic velocity model in unconventional\nreservoirs

The main objective of this thesis is to develop a practical, geology- and rock\nphysics-oriented approach to constructing anisotropic velocity model for\nunconventional reservoirs using downhole microseismic datasets. The working\nprocedure of the approach starts by addressing the geological sources of\nanisotropy. A priori knowledge of anisotropy is obtained by integrating\ngeological information and rock physics studies. The prior knowledge serves as\nconstraint on the microseismic inversion. The anisotropic velocity model\nobtained by the approach can reflect the heterogeneity of anisotropic\nparameters and cover the anisotropic symmetries of most importance in seismic\nexploration and reservoir characterization. The optimal anisotropic velocity\nmodel not only minimizes the data misfit, but also is reasonable from the\nperspectives of geology and rock physics. The results derived from downhole\nmicroseismic dataset are comparable with laboratory experiments. This\ndemonstrates that the downhole microseismic monitoring, as a quasi in-situ\nexperiment, has a potential to contribute to a better understanding of\nsubsurface anisotropy beyond the laboratory. The approach developed in this\nthesis uses a layered velocity model. This approximation is adequate due to\nthe limited spatial range of microseismic monitoring and the relatively flat\nsedimentary background of unconventional reservoirs. The transverse isotropy\ncaused by the bedding-parallel fabric is defined by Thomsen parameters in each\nlayer. The lateral heterogeneities within each layer are dismissed, while the\nvertical gradients of transverse isotropic parameters are kept. The fracture-\ninduced anisotropy is only defined in a specific layer of high brittleness and\nis characterized by normal and tangential fracture compliance. The approach\nuses the arrival-time of seismic waves recorded by sensor arrays. An\nanisotropic ray-tracing algorithm is modified to calculate the synthesized\ntravel-time. Parallel computing is employed to accelerate the ray-tracing\nprogram. The inherent singularity problems in the ray-tracing method are fixed\nby applying numerical strategies. Two nonlinear inversion methods are involved\nin this approach to determine different components of anisotropy velocity\nmodel. The multi-layer TI model is inverted by an iterative gradient-based\noptimization (the Gauss-Newton method). The fracture-induced anisotropy\nrepresented only by two parameters is obtained by a global search method.\nBesides, as a possible source of uncertainties in the velocity model inversion\nand event locations, the issues of computing triggering time (T0) are analyzed\ntheoretically and illustrated with examples. The approach developed in this\nstudy is partially applied to a completed project of downhole microseismic\nmonitoring in a coalbed methane reservoir to verify the capability of\niterative gradient-based inversion for anisotropic velocity model and\nillustrate the T0 issue in the configuration of limited aperture. Then, the\napproach is fully applied to a downhole microseismic dataset from Horn River\nBasin in Canada to investigate the fabric anisotropy and fracture-induced\nanisotropy of shales. The fabric anisotropy of shale is caused by the\nalignment and lamination of the low aspect-ratio, compliant particles, such as\nclay minerals and organic matter. The existence of quartz minerals can prevent\nand interrupt such alignment and lamination and consequently weaken the fabric\nanisotropy of shale. Laboratory measurements show a strong positive\ncorrelation between the degree of fabric anisotropy and the volume contents of\nclay minerals and kerogen. Thomsen parameters ε and γ of shale samples are\nwell correlated with each other, but not with δ. By integrating the geological\ninformation and experimental studies, the fabric anisotropy of Horn River\nshales is initially estimated. The quartz-rich shale gas reservoir is expected\nto show much weaker transverse isotropy than the overlying clay-rich shale. An\niterative optimization using the gradient-based method is then implemented on\nthis initial model. The results derived from the downhole microseismic dataset\nare consistent with the laboratory measurements. The optimized VTI model\nreduces the time misfit by about 65% compared to the originally provided VTI\nmodel. The event locations are also significantly improved. The preferred-\noriented fracture set is another important source of shale anisotropy.\nMechanical analyses show that the fractures in Horn River shales mainly occur\nin the quartz-rich formation showing much higher brittleness. According to the\ncore analyses and fracture mechanism, the fracture planes are commonly\nperpendicular to the bedding plane and the dominant fracture set strikes to\nNE-SW direction which is parallel to the current maximum horizontal stress.\nThe elastic behaviors of the fracture are effectively described by the normal\nand tangential fracture compliance (i.e., ZN, ZT) regardless of any physical\ndetails of fracture. Theoretical modeling and experimental measurements show,\nthe magnitudes of ZN and ZT increase with the fracture dimension scale, and\nthe ZN/ZT ratio is sensitive to fluid fills and has the value less than or\nslightly larger than 1. These facts are used as physical constraints in the\ngrid search for the optimal fracture compliance. The magnitudes of ZN and ZT\ndefine the searching range and the ZN/ZT ratio is used as a quality control.\nThe optimal ZN and ZT have the same order of magnitude as other measurements\nin the crosshole and microseismic scale. The ZN/ZT ratio corresponds to the\nextreme cases of dry or gas saturated fractures.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.