Reducing Geologic Uncertainty in Seismic Interpretation* Jim Bock 1 Search and Discovery Article #41947 (2016)** Posted November 28, 2016 *Adapted from oral presentation given at 2016 AAPG Pacific Section and Rocky Mountain Section Joint Meeting, Las Vegas, Nevada, October 2-5, 2016 **Datapages 2016 Serial rights given by author. For all other rights contact author directly. 1 IHS, Centennial, CO, USA (jim.bock@ihsmarkit.com) Abstract When working with seismic and petrophysical data, both data types may pose a number of unique challenges to the interpreter. Although 3D seismic data provides wide data coverage, the information often lacks granularity and is not a direct measurement of the reservoir properties we are typically interested in obtaining. Petrophysical log data is quite nearly the opposite; in that, we may discern minor changes in reservoir properties along a well bore, but these measurements do not extend more than a few meters away from the logging tool. Combining the best of both data types, geologic models capable of filling in the gaps between seismic and petrophysical data sets have become exceeding valuable. This presentation will examine a number of uncertainty reducing workflows associated with both forward and inverse modeling techniques. Geophysical forward modeling techniques calculate a specific geophysical response given a well-defined physical property model. In the case of 2D seismic modeling, the physical property model can be taken directly from petrophysical log data, sonic and density logs that have been adequately tied to an existing seismic survey. Using both available log data combined with geologically reasonable model constraints, geomodelers may construct a number of modeled seismic responses that can be used to help validate or invalidate various working geologic models. In contrast, geophysical inverse modeling techniques attempt to construct a physical property model based off a geophysical response. In the case of seismic inversion, impedance values are calculated from an existing seismic data set. The largest challenge associated with inverse modeling is that there are multiple solutions available given an individual seismic data set. By using a simulated annealing (SA) inversion algorithm, geoscientists are able to greatly reduce the total number of possible solutions that are available by leveraging both a background model combined with efficient wavelet estimation for optimal tuning parameters.
Reducing Geologic Uncertainty in Seismic Interpretation AAPG Pacific/Rocky Mountain Section Jim Bock, Senior Technical Advisor jim.bock@ihsmarkit.com
Slide Shown with Image Agenda Example Subtitle goes here Data and Uncertainty Level 1 with a bullet point > Level 2 is indented with an arrow Geologic Models and Non- Level 3 is indented with an em dash Uniqueness Lorem Ipsum quasinto a telipto. Vitae tempor quam eu libeoro tortor quam Quam egestas am eu libobio. Model Constraints/Boundaries > Tortor quam, feugiat vitae, ultricies eget, Ipsum quasinto a telipt tempo onec eulibero amet quam. Modeling Examples Quam egestas semper tortor quam Ipsum a telipt uam egestas am eu libobio semper tortor quam. Summary
Data and Uncertainty Well Data High Level of Detail Limited Data Coverage Seismic Data Low Level of Detail Extensive Data Coverage
The need for Geologic Models Data Gaps Geophysical Ambiguity Verification of Interpretation Predicting Rock Behavior Optimize Survey Geometry Layouts Mitigate Risk and Reduce Uncertainty
Types of Geologic Modeling Forward Earth Model Geophysical Response Lithology Model Seismic Amplitude
Types of Geologic Modeling Inverse Geophysical Response Earth Model Seismic Amplitude Absolute Impedance
Depth (km) Forward Modeling Input: Earth Model Distance (km) 0 0 10 20 30 40 50 60 70 80 90 100 2.75 g/cm 3 10 20 30 2.80 g/cm 3 40
Gravity Response (mgals) Forward Modeling Output: Geophysical Response 0-1 -2-3 -4-5 -6 2.75 2.4 g/cm g/cm 3 3 2.76 g/cm 3 2.8 g/cm 3 0 10 20 30 40 50 60 70 80 90 100 Distance (km)
Gravity Response (mgals) Inverse Modeling Input: Geophysical Response 0-1 -2-3 -4-5 -6 0 10 20 30 40 50 60 70 80 90 100 Distance (km)
Depth (km) Inverse Modeling Output: Earth Model Distance (km) 0 0 10 20 30 40 50 60 70 80 90 100 2.75 g/cm 3 10 2.76 g/cm 3 20 30 2.80 g/cm 3 2.40 g/cm 3 40
Forward Model Earth Models are Non-Unique! Inverse Model
Dealing with Non-Uniqueness You need to set Boundaries Background Model Velocity Volume Log and Well Data Seismic Wavelet Additional Geologic Information Formation properties
Forward Modeling of Lower Mannville Channels in Southern Alberta Alberta Calgary
Depositional Environment Lower Mannville: Lower Cretaceous : 145 99.6 MYA Non-Marine Clastics Sourced from Up-Thrust Sedimentary Rocks from West Deposited along South to North Trending Drainage Pattern Drainage controlled by Paleotopography of Pre-Cretaceous Unconformity Map Credit: Ron Blakey, Colorado Plateau Geosystems, Arizona, USA.
2D Forward Modeling Workflow Tie Seismic to Log Data at Key Well Locations
2D Forward Modeling Workflow Construct and Modify 2D Model Channel
2D Modeling Results Model Validation/Invalidation Sand Silt Filled Filled Channel 2900 4100 m/sec
Inverse Modeling of the Rundle Group in Southern Alberta 16000 Alberta Calgary Seismic Amplitude: Top of Rundle 2000
Depositional Environment Rundle Group: Middle to Late Mississippian: 345.3 318.1 MYA Tropical /Shallow Marine Environment Carbonate Platform and Ramp Lithofacies Bounded at the top by an Unconformity Map Credit: Ron Blakey, Colorado Plateau Geosystems, Arizona, USA.
SA Inversion Workflow Sonic & (Density) Logs (T-D Chart) Seismic for Inversion LogSeisMatch Wavelet Estimation External Wavelet Seismic to Well Synthetic Scaling Well MacroModel Macromodel (Velocity Volume) parameter tuning Absolute Acoustic Impedance Volume Relative Acoustic Impedance Volume
SA Inversion Workflow LogSeis Match
SA Inversion Workflow Wavelet Estimation
SA Inversion Workflow Seismic to Well Synthetic Scaling
SA Inversion Workflow Parameter Tuning
SA Inversion Results Increased Resolution over Amplitude Data 16000 15000 Absolute Seismic Amplitude: Impedance: Top Top of of Rundle Rundle 11000 2000
SA Inversion Results Increased Resolution over Amplitude Data
Taking your Interpretation Further. Rock Layers not Rock Boundaries Better tie with Well Logs Reservoir Properties (Porosity, Permeability, etc.) Average Effective Porosity vs. Absolute Inversion Calculation of Reservoir Properties using Regression Analysis
Regression Calculated Reservoir Properties Inversion Porosity 1.203 -.008416*(abs inversion) 30 Structure Contour Interval = 25 meters 2
Geologic Modeling Summary Addressing Non- Uniqueness Background Model Additional Geologic Information
Geologic Modeling Summary Mitigate Risk and Reduce Uncertainty Sand Silt Filled Filled Channel Channel 2900 4100 m/sec Channel Silt Filled Channel 4100 m/sec
Geologic Modeling Summary Additional Insight with Proper Calibration
Final Thoughts All models are wrong, but some are useful George Box The truth..is much too complicated to allow anything but approximations John Von Neuman
Acknowledgements
Thank You. Questions? Jim Bock, Senior Technical Advisor jim.bock@ihsmarkit.com