E N D
1. A Review of Geological Modeling A. Keith Turner
Colorado School of Mines
&
Carl W. Gable
Los Alamos National Laboratory
2. Why Is Modeling & Visualization Important? The World of the Geoscientist Is Multi-dimensional Current interpretation methods limit this view
Digital versions of traditional maps are not sufficient
Increased efficiency demands computer-based methods to:
Integrate and Manage the data
Interpret geological features
Visualize attributes spatially and temporally
Model dynamic Earth processes
3. Importance of This Topic … 3-D subsurface modeling first became feasible in late 1980’s with the introduction of high-performance Unix-based graphical workstations
Developing digital representations of the subsurface does not ensure high-quality and efficiently managed projects
Society is increasingly demanding:
multi-scale, multidisciplinary, integrative projects
a shift from passive data collection and archiving to dynamic information management and dissemination
4. Over The Past DecadeEnormously more powerful computers anddata storage have vastly reduced costs! Continuing, rapid advances in computer HARDWARE and SOFTWARE technologies
Modeling & visualization increasingly integrated
Increasingly realistic models possible
5. Problems in Subsurface Investigation are Unique... Subsurface information is often incomplete and conflicting;
The subsurface is naturally complex and heterogeneous;
Sampling is most often insufficient to resolve all uncertainties; and
Scale effects on rock , fluid, and other properties are usually unknown. TO BEGIN....
There are many problems in subsurface investigation which are unique only to the geosciences. These problems will dictate many of the requirements for an n-Dimensional Earth Modeler.
To begin, the subsurface information with which we have to work is often incomplete and conflicting. Even with 3D seismic, the results are often ambigious, and conflict with well information. This is both a matter of the scale of the volume of rock measured, and the resolution at which it is measured by the tool being used. At its best, the resolution of 3D seismic is no more than 45 feet (+/- 15 m), yet a very large volume of the subsurface can be measured. A well on the otherhand can be logged to a few inches or even mm’s of resolution (if for instance cores are taken), but the volume of the subsurface actually sampled is very small on a regional scale. Yet, in nD modeling, these two disparate data types will be combined into a final interpretation.
The natural complexity of the subsurface adds yet another problem. The complex and heterogeneous nature of the subsurface makes modeling of attribute distribution and visualization very difficult. We are often forced to make ad hoc simplifications in the model which reduce its validity and the usefulness of the model for the purpose for which it was constructed.
Our inability to sufficiently sample the subsurface is constrained by both cost and the availability of tools with which to measure the desired property or properties.
Scale effects on rock, fluid, and other geologic processes are usually unknown......more ??????
TO BEGIN....
There are many problems in subsurface investigation which are unique only to the geosciences. These problems will dictate many of the requirements for an n-Dimensional Earth Modeler.
To begin, the subsurface information with which we have to work is often incomplete and conflicting. Even with 3D seismic, the results are often ambigious, and conflict with well information. This is both a matter of the scale of the volume of rock measured, and the resolution at which it is measured by the tool being used. At its best, the resolution of 3D seismic is no more than 45 feet (+/- 15 m), yet a very large volume of the subsurface can be measured. A well on the otherhand can be logged to a few inches or even mm’s of resolution (if for instance cores are taken), but the volume of the subsurface actually sampled is very small on a regional scale. Yet, in nD modeling, these two disparate data types will be combined into a final interpretation.
The natural complexity of the subsurface adds yet another problem. The complex and heterogeneous nature of the subsurface makes modeling of attribute distribution and visualization very difficult. We are often forced to make ad hoc simplifications in the model which reduce its validity and the usefulness of the model for the purpose for which it was constructed.
Our inability to sufficiently sample the subsurface is constrained by both cost and the availability of tools with which to measure the desired property or properties.
Scale effects on rock, fluid, and other geologic processes are usually unknown......more ??????
6. Why We Need Special Modeling and Visualization Tools and Not Just CAD
7. A Typical Modeling Project
8. Geometry (Descriptive) Modeling Definition: Geometry (Descriptive) Modeling involves visually describing, through various means such as computer graphics and modeling:
The geologic framework
Distribution and propagation
of attributes
9. 3-D Model Involves Two Stages Framework Definition
Borehole and isolated sample data
Triangulated surfaces
2-D grids and meshes
Iso-volumetric models
– from triangulated surfaces
– from cross-sections
– from grids and meshes
– parametric (NURBS, etc)
– Boundary Representations Discretisation and
Property Distribution
3-D grids and meshes
– regular hexahedral
– octree variable
– geocellular
– tetrahedral unstructured
meshes
10. Modeling Often Begins with Borehole Data
11. Geometry Models can be Constructed Using Cross-Sections
12. 3-D Solid Models can bedeveloped from Multiple Surfaces
13. Layered Models may involve many surfacesComplex channels and “pinch-outs” are difficult to model
14. Regional (Volumetric) SubdivisionFeasible for Non-stratigraphic Cases
15. Framework Models require Grids or Meshes to assign Property Distributions
16. Geological Framework Defined First –then Grid Resolution
17. Discretisation may involve“structured” and “unstructured” meshes
18. “Geocellular” Volumetric Model
19. Yucca Mountain Represented by a Tetrahedral Mesh Model
20. Accurately Modeling Faults is a Challenge Near-horizontal thrust faults form additional surfaces
Steeply inclined Faults commonly shown as vertical
21. Advanced Fault Modeling
22. Models may be “Nested”from Regional to Local Scales
23. Purpose of Modeling is Prediction… Prediction has an extrapolative rather than interpolative character…
Involves risk
Leads to Decision-making
24. Predictive Modeling Definition: Predictive Modeling involves prediction and/or simulation of events, dynamic and other types of processes occurring in the
geological subsurface:
Solve equations, or other
numerical analyses
Forward and inverse
modeling techniques
25. Another example is given in this model. This is a again a relatively layer cake type geology, but now we want to represent, model and visualize the distribution of a particular contaminant in the subsurface…..particularly within the blue layer, here presumed to be an aquifer.Another example is given in this model. This is a again a relatively layer cake type geology, but now we want to represent, model and visualize the distribution of a particular contaminant in the subsurface…..particularly within the blue layer, here presumed to be an aquifer.
26. This image shows the overall volume and distribution of the contamination, in this case, lead….within the geologic strataThis image shows the overall volume and distribution of the contamination, in this case, lead….within the geologic strata
27. If we peel away the strata, we can see only the contaminant volume itself and we can begin to slice through the volume to see the concentration levels of lead. The sticks represent borehole data points. The coloring on the sticks represents the thickness and nature of the geological strata at each penetration. Hotter colors are used to represent the concentrations of lead. In most cases like this some multivariate statistical technique or more commonly, multiple conditional probabilities (I.e., stochastic techniques, geostatistics) are used to determine the concentrations between wells.If we peel away the strata, we can see only the contaminant volume itself and we can begin to slice through the volume to see the concentration levels of lead. The sticks represent borehole data points. The coloring on the sticks represents the thickness and nature of the geological strata at each penetration. Hotter colors are used to represent the concentrations of lead. In most cases like this some multivariate statistical technique or more commonly, multiple conditional probabilities (I.e., stochastic techniques, geostatistics) are used to determine the concentrations between wells.
28. The Bottom Line … Trends in 3-D Modeling Continuing, rapid advances in computer technologies
Increasingly realistic models possible
“Coarse-” vs. “Fine-scale” Models
Cannot easily incorporate geologic knowledge (i.e. geologic interpretations)
Difficulties representing uncertainties
Hampered by imprecise data and inability to adequately sample the prototype
29. The Modeling Challenge Build 3-D models from isolated bore data and “soft” information
Connect known conditions between the boreholes
Model real-world complexity
Verify the resulting models
Represent the uncertainties of the geologic framework
Reduce cost and time of model development
30. An Integrated Data Management Process
31. What Benefits will an Integrative Approach Provide?? To the Individual Scientist –
To the Scientist’s Organization –
To the “Client” Organizations and Their Staff –
To Society at Large –
32. Benefits to the Individual Scientist – More time for science, so….
More interactions with colleagues
More interesting projects/studies
More job satisfaction
More publications
More rapid promotions
33. Benefits to the Scientist’s Organization – Society
Maximize Organization’s
Value to “Customers”
Maximize Strategic Value – BETTER!
Maximize Financial Value – CHEAPER!
Maximize Operational Effectiveness – FASTER!
34. Benefits to the “Client” Organizations and Their Staff – Model Consistency
Data versioning, audit trail, documentation
QA compliance
Improved Predictive Capability & Model Maintenance
Efficiency: streamlining processes, updating databases
Auto-updating decision support tools
35. Benefits to Society at Large – Relevant science that supports decision-making
Clearly communicated results
Cost reductions
36. Questions???