![]()
Abstract The objective of this work was to provide high-resolution elastic property volume through the Modified Stochastic Inversion (MSI) workflow with careful consideration of the geologic context and availability of the data. Advances in drilling and completions have unlocked large amounts of hydrocarbons, but geology remains the strongest driver of long-term production. Carbonate debris flows are an example of the significance that geology plays. These debris flows may represent drilling hazards and stress barriers, and when they are not avoided, they may increase rig time and costs. From a previous deterministic inversion, a carbonate debris flow within the target was identified but needed more information about its thickness and extent.
Stochastic inversion is a modeling process that generates multiple property realizations at higher resolutions than what is available from deterministic inversion and enables geoscientists and engineers to estimate subsurface uncertainty. This work is based on a technology known as Modified Stochastic Inversion (MSI) workflow where the deterministic inversion result is spectrally merged with the desired number of simulations to generate several broadband realizations of the reservoir. High and low frequency bands are created from Sequential Gaussian Simulations (SGS) from the well data. MSI simplifies the stochastic inversion process through explicit use of the inversion, satisfying the seismic data constraint and significantly reducing the computational time.
We were able to create high-resolution property volumes models that better constrained the carbonate debris flow detecting layers as thin as 8 ft. These volumes not only captured a higher vertical resolution, but also a measure of uncertainty from geologic plausible scenarios. The realization with the best correlation to the blind wells can be selected as the high-resolution subsurface model and uses to represent the reservoir. To validate the results quantitatively, the correlation factor between the MSI results and the upscaled logs in the interval of interest was estimated, obtaining a 0.75 value versus 0.65 from the deterministic inversion. New insights about the debris flow were provided using average thickness and probability maps considering the realizations collectively.
A better understanding of the debris flow's position, thickness, and extent was gained through the high-resolution results and uncertainty. Geological features as thin as 8 feet could now be modeled with confidence, rather than the original 40 ft resolution from seismic data alone. Reservoir specific accommodations could also be made to calibrate hydraulic stimulation parameters to optimize rock failure in the presence of this stress barrier, thus improving the performance of the wells. This technology can also serve as a step toward the direct use of high-resolution geomechanical properties derived from seismic to build 3D wellbore stability models in developing fields, characterizing the lateral variation of mechanical properties and providing better prediction throughout the full field.
Site News