Inversion of fracture weakness parameters based on 3C vertical seismic profile data — Part I: Horizontal transversely isotropic media composed of a single fracture set in the isotropic background media

Abstract

Natural fractures in oil and gas reservoirs are a crucial factor that cannot be ignored because they significantly influence the reservoir’s petrophysical properties and hydrocarbon development. A horizontal transversely isotropic (HTI) medium composed of a single fracture set in an isotropic background is a typical anisotropic medium. Meanwhile, the S-wave splitting is a sensitive response of such anisotropic media, resulting in the generation of fast and slow S waves. The normal and tangential fracture weaknesses are crucial parameters that characterize the anisotropy of fractured media. We develop an inversion method for fracture weakness based on 3C vertical seismic profiling (3CVSP) data. First, assuming weak anisotropy and an HTI medium containing a single fracture set, we derive a first-order linear approximation of the traveltimes of the converted fast and slow S waves (PS1 and PS2 waves) with respect to fracture weakness parameters in the phase velocity domain. By solving for the horizontal projection of the slowness vector, approximate equations of the traveltimes of the PS1 and PS2 waves are converted from the phase velocity domain to the group velocity domain. Furthermore, we devise an inversion workflow consisting of three primary steps: (1) preprocessing the VSP data to derive the traveltimes and azimuth of the HTI medium, (2) constructing a forward model with undetermined fracture weakness parameters, and (3) following the establishment of the objective function, conducting the inversion for the fracture weakness parameters. We demonstrate the reliability of the method through numerical examples and synthetic 3CVSP data. The inversion errors are primarily influenced by the azimuth angle, with minimal influence from the receiver depth. Furthermore, a collective set of inverted results derived from all geophones is more stable and accurate than individual geophones. The application to actual 3CVSP data further confirms the effectiveness of our approach.