|Theme||Visible||Selectable||Appearance||Zoom Range (now: 0)|
Skelt, Christopher (UNOCAL Corporation) | Glenn, David (UNOCAL Corporation) | Smith, Steven (UNOCAL Corporation) | Vu, Cung (UNOCAL Corporation) | Walden, Skip (UNOCAL Corporation) | Anderson, Paul (Fugro‐Jason) | Bacon, Brad (Fugro‐Jason)
Our industry increasingly relies on seismic data to appraise and develop new discoveries, particularly in deep water where drilling and development are costly. The improvement of reservoir characterization techniques based on seismic inversion offers immediate and significant impact on reserves estimation and quantification of uncertainty.
Experience shows that inversion results are strongly dependent on the low frequency background model derived from the wells in the study area, so an industry trend towards fewer wells and more reliance on seismic data prompts the question of how to address the associated degradation and uncertainty of inversion results. We will show how to use petrophysical analysis and geological insights to reduce the number of inversion parameters by creating a grid of virtual elastic property profiles or “pseudo wells” to supplement the real wells in the background model.
The algorithms used to generate elastic properties at the pseudo wells were derived from core analysis and petrophysical modeling. We observed usable correlations between the sedimentologist's core-based sand fraction log and both core-based reservoir properties and the gamma ray log. This allowed us to use the gamma ray alone to estimate all the key reservoir properties — porosity, permeability and water saturation. Density and compressional and shear sonic logs were reproduced from the gamma ray log and the associated reservoir properties using familiar equations, linking elastic and reservoir properties.
Virtual wells were created by drawing a reservoir sand quality profile that was then transformed to density and compressional and shear velocity for modeling synthetic seismic gathers. The profiles were compared with the 3D seismic gathers at the same location, and optimized by improving the match while honoring the geologist's views on allowable reservoir property profiles and stratigraphy. The resulting network of real and virtual wells was used as a constraint in the seismic inversion and produced a result consistent with petrophysical data and geological opinion.
Geostatistical inversion is applied on a Gulf-of-Mexico, 3D post-stack seismic data set to improve the vertical resolution of reservoir parameters yielded by trace-based inversion. Results are derived in the form of a set of equally probable realizations of acoustic impedance that honor both the well logs and the seismic data. Petrophysical analysis reveals two prospective gas-bearing sands. We focus our analysis to the determination of lateral and vertical resolution properties of the thicker of the two sands. Geostatistical inversion yields stochastic realizations of acoustic impedance and bulk density with vertical resolutions of 4 and 1 ms. Inversion of bulk density follows from the enforcement of a statistical relationship with acoustic impedance validated with wellbore data. In turn, bulk density lends itself to a relatively simple procedure to obtain high-resolution simulations of total porosity andabsolute permeability.