Can fracture orientation and intensity be detected from seismic data?
The February issue of The Leading Edge includes a special section on “Unconventional case studies”. Devon Energy and SeismicCity contributed a case study article to this section addressing one of the key current interests of the industry - the ability to detect fracture orientation and density from seismic data. This key question can be finally answered by use of full elastic orthorhombic anisotropic simulation technology.
The article: “Can fracture orientation and intensity be detected from seismic data? Woodford formation, Anadarko basin, Oklahoma investigation.” is authored by Marianne Rauch-Davies and David Langton from Devon Energy, Michael Bradshaw, Allon Bartana, Dan Kosloff, Jeff Codd and David Kessler, from SeismicCity Inc., Jamie Rich from Cimarex and Gary Margrave from the University of Calgary.
The full text can be accessed here: The Leading Edge Article.
With readily available wide-azimuth, on-shore, 3D seismic data, the search for attributes utilizing the azimuthal information is ongoing. Theoretically, in the presence of ordered fracturing, the seismic wavefront shape changes from spherical to non-spherical with the propagation velocity being faster parallel to the fracturing and slower perpendicular to the fracture direction. This concept has been adopted and is used to map fracture direction and density within unconventional reservoirs. More specifically, azimuthal variations in normal moveout velocity or migration velocity, are often used to infer natural fracture orientation.
Significant evidence however suggests that these azimuthal variations are not always due to azimuthal anisotropy and are highly dependent on reservoir layer thickness, burial depth and strength of anisotropy. Yet there is a strong persistence of this interpretation. In order to address this issue, we built a model with varying levels of VTI and orthorhombic anisotropy, following the Woodford formation in the Anadarko Basin in Oklahoma. Having this model we applied full elastic orthorhombic anisotropic simulation creating a dataset that can be migrated using various time and depth migration algorithms. As the model and anisotropic characteristics are known, we can use the migrated data to accurately evaluate if fracture orientation and density extracted from the seismic data is actually matching the true geology.
The analysis of the depth migrated model data indicated that for the typical layer thicknesses of the Woodford shale layer in the Anadarko basin, observed and modeled percentage of anisotropy and target depth, the effect of azimuthal anisotropy is too small to be detected in real seismic data.
The study also demonstrated that if the geological layer that includes the azimuthal anisotropy is thicker then azimuth dependent Vnmo or migration velocity can be detected from the seismic data. However, the variation is small and in presence of noise and migrating the data with an approximated model, directly associating azimuthal variations in seismic moveout is not stable enough to be directly associated with geological fracture orientation and density.
With today’s availability of commercial full elastic solutions, the ability to link fracture orientation and density directly from seismic data can be analyzed for specific cases (i.e. layer thickness and depth of burial). Based on the analysis results we can decide if the real data acquired over a specific layer should be used for azimuthal velocity analysis and then linked to prediction of fracture orientation and density.
Earth Science Advisor
6 年Great paper. Another thing to keep in mind is the overall complexity of the system potentially having an impact on anisotropy: the presence of multiple fracture sets of different orientation (which is typically the case) as well as the maximum horizontal stress (making it rather challenging to differentiate between these variables).
Using geoscience to develop shale assets, conventional oil/gas, & subsurface storage
6 年NIce paper, confirms anecdotal experiences of lots of us who have tried to integrate these various 'anisotropic" attributes with other, real data.? Rarely have I seen it pay off.? Unfortunately, another example of the sales pitches outrunning the technology...? The best thing we have seen work for direct fracture detection is running Max Curvature or Dip Max Similarity on the highest power? frequency slice from spec decomp.? ?We were able to tie that to observed fractures from drilling events, and the Xmac fracture mapping tool.? ? Hopefully one day this can be a paper, also!
Sr. AI/ML Solutions Architect at Amazon Web Services (AWS)
6 年From past processing experience, most azimuthal anisotropy seems to be associated with surface stress and loading, rather than reservoir fracture. This article echoed with lots of processors’ experiences
Senior Advising Geophysicist at TGS
6 年Those interested in this might like to read an EAGE paper from last year... “extraction of azimuthal anisotropy parameters from a field scale ocean bottom seismic elastic finite difference study” by Tillotson et al.
DIrector at TNG Geophysics Limited
6 年Reflections from the Tiguentourine Seismic Fracture Detection Project. The modelling is a key part of the process. You have to know that you are in a should see environment. The next stage is to talk to people. Often they know why their attempt to do something failed. (Thanks to Mike Helton, Gordon Holmes and several others) The acquisition is key, dense full azimuth data needs to be acquired and BP and Axxis Geo Solutions OBN shows it can be even in marine environments. Then processing - surface fitting rather than sectoring? For our project (yes we tried both ways) surface fitting provided better result. What attribute to use - again we tried several and velocity was more stable. You need good well data to calibrate the results. Finally you need a thick skin to put up with all the people who tell you it can't be done, years after you did it.