SPE Distinguished Lecturer Program Primary funding is provided by The SPE Foundation through member donations and a contribution from Offshore Europe The Society is grateful to those companies that allow their professionals to serve as lecturers Additional support provided by AIME Society of Petroleum Engineers Distinguished Lecturer Program www.spe.org/dl 1
Let s Model It! 3D Geoscience Modeling Implications for Reserves Estimation and Field Development Planning Doug Peacock Gaffney, Cline & Associates Society of Petroleum Engineers Society of Petroleum Engineers Distinguished Lecturer Program www.spe.org/dl
Presentation Outline Development of 3D modeling techniques Current Problems and Issues Geoscience to Simulation Solutions and Best Practices Future Developments Summary & Conclusions 3
Why did 3D modeling become such a commonly used technique? It s 3D real world is 3D, not 2D Consistency of horizons, faults, picks etc into a single framework no more overlapping horizons, strange faults, unrealistic reservoir compartments t Common view of reservoir for all disciplines: Shared Earth Model concept Can be used as a basis for field activity appraisal, FDP, development, and updated 4
Why did 3D modeling become such a commonly used technique? Allows use of geostatistics, facies algorithms Evaluate heterogeneity in inter-well areas Analyze full range of uncertainty More meaningful volumetrics Dovetails the static / dynamic elements Allows iterative improvements It s addictive Biggest changes in 3D modeling have been Increased speed, detail Increased integration i across disciplinesi 5
Classical Modeling Workflow Well Correlation Mapping Structural Model Simulation Model Petrophysical Model Facies Model 6
Presentation Outline Development of 3D modeling techniques Current Problems and Issues Geoscience to Simulation Solutions and Best Practices Future Developments Summary & Conclusions 7
Gross Rock Volume GRV is typically the largest single factor in STOIIP uncertainty Often modeled only in relation to uncertainty in Top Reservoir What about: interpretation, isopachs, depth conversion, fault presence/position/ throw etc It requires more effort to model these uncertainties so it is easy to neglect them Especially important in the early / appraisal stage of field life when facilities design (capacity, lifespan) are being considered 8
Structural Issues Oil Water Contact 9
Fault Position Lost Volume Oil Water Contact 10
Fault Angle Oil Water Contact 11
Number of Faults Oil Water Contact t 12
Contacts 13
Problems and Issues with Modeling Techniques Predictions are extrapolative rather than interpretive Stochastic models alone do not utilize the skill and experience of the geologist Statistics (GSS) vs. Geology (Object Model) Assumptions usually have a large effect Results depend on experiences and preferences of modeler Although experienced geomodelers understand the effect of these assumptions 14
Modeling Assumptions Data available is never enough to provide full understanding of the subsurface Which Algorithm? Directional Variograms? Seismic Attributes? Stationarity? Vertical Proportions? Trends? 15
Algorithm Assumptions Which Algorithm? Moving Average Gaussian Sequential Simulation 16
Variogram Assumptions Data sampling is rarely sufficient to well define a variogram need to rely on experience, analogy, seismic i data, trial & error? Directional Variograms? Weak N-S Strong NE-SW 17
From the same data set. Which Algorithm? Variogram length and direction Trends / Seismic Data All models will match the input data; differences come from the decisions that are made about how to build the model 18
Presentation Outline Development of 3D modeling techniques Current Problems and Issues Geoscience to Simulation Solutions and Best Practices Future Developments Summary & Conclusions 19
Geoscience to Simulation Scale issues are still problematic Better History Matches may be achieved by correctly identifying contributing rock and honoring scale Feedback still required but earlier is better 20
Scale Issues Better History Matches may be achieved by correctly identifying contributing rock and honoring scale Scale issues are still problematic Relationships developed at a core or log scale are applied at a grid cell scale Log, core, geo cell, sim cell 21
Scales of Measurement Core Well Log Production Seismic Missing Scale Geo Model Sim Model 10-5 10-4 10-3 10-2 10-1 10 0 10 1 10 2 10 3 10 4 Measurement Volume m 3 22
log φ vs. log K at different scales Core scale Grid-cell scale log (Permeability, md) at core scale 10000 10000 log (Permeability, md) at grid-cell scale 1000 1000 100 100 10 10 1 1 0.1 0.1 010 0.10 013 0.13 016 0.16 020 0.20 025 0.25 032 0.32 040 0.40 010 0.10 013 0.13 016 0.16 020 0.20 025 0.25 032 0.32 040 0.40 log (Porosity) at grid-cell scale log (Porosity) at core scale Modified from: Worthington 2004 23
Presentation Outline Development of 3D modeling techniques Current Problems and Issues Geoscience to Simulation Solutions and Best Practices Future Developments Summary & Conclusions 24
Model Problems Too big, Too complex Too long to build Delivered late Don t meet business needs Difficult to update Difficult to History Match Homogenously Heterogeneous Don t necessarily give good predictions 25
Definitions Scenario Realization Different structural or One of a number of geological concept e.g. outputs from stochastic Fault Configuration modeling e.g. Gaussian Depositional Setting Sequential Simulation 26
Scenario Method Low Best High Scenario A Scenario B Scenario C 1 2 3 1 2 3 1 2 3 Realizations Scenarios may be variously defined Well suited to early field life Later field life may require fewer historymatched models 27
Scenario Method Cum Probability 100 90 80 70 60 50 40 30 20 10 0 Probabilistic Deterministic 0 50 100 150 200 STOIIP Uncertainty range often greater between scenarios than within them Risk of under-estimating range of uncertainty Modeling hundreds of realizations doesn t mean that all the uncertainty has been captured! 28
What makes a good model? Geologically Reasonable Represents geological l understanding di Honors available data Allows fast and accurate history match Assisted by accurate net pay, honoring scale (K, Sw vs h) Gives good predictions Of geology (in new wells, one by one removal) Of reservoir performance Fit for Purpose 29
Fit for Purpose Meets Business Needs e.g. time, budget, resources, technical Range of Uncertainties e.g. for Development Planning, Reserves Best Technical Case e.g. for well planning If simulation is involved, discuss all issues with reservoir engineer: Areal limits, it Orientation, ti Cell Size, Layering, Upscaling, Feedback, Key Issues, etc 30
Possible Solutions and Best Practices Top Down or First Pass modeling Capture key uncertainties ti with small number of simpler models detail added later Supports scenario modeling allowing different concepts & methods not just uncertainty First Pass / Top Down models allow Data Validation, Identify Data Gaps Early Results and Early Feedback Quantify Main Uncertainties and Risks Provide focus for more detailed modeling Business requirements define model purpose 31
Simulation Results can be Ambiguous A B Oil Water Contact Simulation Results indicate that Well A should have less Pore Volume and Well B should have more pore volume 32
Different Structure A Different interpretation and/or depth conversion B Simulation Results indicate that Well A should have less Pore Volume and Well B should have more pore volume 33
Thicker Sand A Different net pay cut-off results in inclusion of lower quality sands B Simulation Results indicate that Well A should have less Pore Volume and Well B should have more pore volume 34
Better Properties A Different property modeling assumptions e.g. channel width, depositional environment etc B Simulation Results indicate that Well A should have less Pore Volume and Well B should have more pore volume 35
A Different Contacts B Deeper contact e.g. different contacts in different fault blocks, contact not observed, uncertainty on pressure depth plots etc Simulation Results indicate that Well A should have less Pore Volume and Well B should have more pore volume 36
Move Fault A Different interpretation t ti and/or depth conversion B Simulation Results indicate that Well A should have less Pore Volume and Well B should have more pore volume 37
Staffing Issues Software tools are becoming increasingly complex Software tools are becoming easier to use Risk of becoming Nintendo GeoEngineers Specialists required to build a good model? Encourage generalists or specialists? Many large companies do have specialist geomodelers, with appropriate skills i.e. Software, Geology, Geostatistics, Experience. Dedicated group and outsourced to assets Or spread throughout assets Companies with limited staff, resources, time? 38
Presentation Outline Development of 3D modeling techniques Current Problems and Issues Geoscience to Simulation Solutions and Best Practices Future Developments Summary & Conclusions 39
Technological Developments Better, faster, easier to use software Grid cell arms race More integration between data and disciplines New Methods and Tools Grid creation (unstructured, easy gridding ) Small scale bedding impact on large scale flow Multi-point geostatistics Inversion Loops Discrete Fracture Networks (DFN),Geomechanics Increased use of digital & outcrop op analogues, a Potential for industry/academia collaboration 40
Limitations of Traditional Geostatistics ariance Semi-Va Lag Distance 1 2 3 2-point correlation is not enough to characterize connectivity Source: Caers 41
Multi-Point Geostatistics Offers a way of including more geology Training Images required Could be based on digital analogue/outcrop data Still depends on selection of an appropriate training image Training Image Final Model Tetzlaff et al 2005 42
The Future. Effective modeling in the future will require a blend of technology and process Technology will continue to evolve More detail = better models? Multi-point statistics, better use of seismic, better workflows, automated history match, no upscaling, etc. Process and smarter working practices may deliver greater benefits Better understanding of uncertainty Fit for Purpose models 43
The Future? Easy Gridding Input Validation Geological l Model Validation Inversion Loops Automatic History Match Input MPS / Digital Analog 44
Summary & Conclusions 3D Geoscience modeling will continue to be a widely used and indispensible technique Modeling techniques have implicit assumptions built in to them Know what they are and what affect they have Modeling methods are continuously evolving Only expert modelers may be able to keep up to date Do not be seduced by Nintendo GeoEngineering It is still good practice to: Think about the purpose of modeling (Fit for Purpose) Understand data, data validity, data limitations Define main assumptions and continue to challenge them 45
Let s Model It! 3D Geoscience Modeling Implications for Reserves Estimation and Field Development Planning Doug Peacock Gaffney, Cline & Associates Society of Petroleum Engineers Distinguished Lecturer Program www.spe.org/dl 46
References Worthington, P.F., 2004, The Effect of Scale on the Petrophysical Estimation of Intergranular Permeability: Petrophysics, vol 45, no 1 Caers, J., 2002, Stochastic inverse modeling under realistic prior model constraints using multiple-point geostatistics. Invited presentation for the IAM2002 Workshop on ""Quantifying uncertainty and multiscale phenomena in subsurface processes, Minneapolis, Minnesota, Jan 7-11 Tetzlaff et al, 2005, Application of multipoint geostatistics to honor multiple attribute constraints applied to a deepwater outcrop analog, Tanqua Karoo Basin, South Africa: SEG Expanded Abstracts vol 24, 1370, 47