Geological Modeling

Uncertainty Analysis

Serdar Kaya,
Please contact for any suggestion, comments or contribution [email protected]
Uncertainty can be defined as degree of
inaccuracy in a measurement or prediction
at a level of reasonable precision
Uncertainties are present in all data and more
imposed during the data process!

Seismic Data Water Saturation

Seismic Interpretation Fault Transmissibility
Time to Depth Conversion Fault Displacement
Well Markers Fractures
Reservoir Thickness Permeability
Compartments Porosity
Fluid contacts NTG
Fluid Characteristics
Facies Model

3
Develop an Approach
Gather Information
• Collect uncertainty information on data and data
processing

Decide approach to use

• Stochastic Multiple Realization
• Deterministic Scenario Base

Perform Analysis
• Statistical
• Quantitative
 Quantification of Uncertainty
• Accuracy : Systematic shifts due to environmental correction or
measurement tools/techniques
• Difference between Core porosity and Phie
• Variation in Modeling steps
• Difference between well log porosity and scaled-up porosity
• Difference between scaled-up porosity and modeled porosity
• Difference between realizations on seed number
• Difference between variogram range

• Precision:
Structural Uncertainty High and Low Case

High case Top Structure

Well
Low Case Top Structure Marker

Base Case
Uncertainty Analysis
Uncertainty in Top Structure

240 feet

FWL 40 feet

Structural uncertainty doesn’t affect the volumetric

 Porosity Measurement Statistic
Core Porosity Log Porosity

Statistics for Core Porosity

Name Type Min Max Delta N Mean Std Var Sum
Scaled-up Cont. 2.35 27.24 24.89 227 18.08 5.292 28.003 4104.0497
Well logs Cont. 2.35 28.62 26.26 592 18.855 5.157 26.59 11162.171
Statistics for Log Porosity
Scaled-up Cont. 0.68 29.91 29.23 441 17.8 6.37 40.57 7850.9
Well logs Cont. 0 30.72 30.72 2706 18.41 6.39 40.84 49812.24
 Uncertainty on Stochastic Modeling
Statistics of Porosity Realizations in Oil Pool
Min Max Mean

Realization #1 2.06 28.79 14.54

Realization #2 1.23 29.07 14.6

Realization #3 0.83 27.48 14.63 R8 has highest mean 14.99

Realization #4 1.16 27.61 14.33
R4 has minimum mean 14.33
Realization #5 0.74 27.54 14.42
Difference is 0.66
Realization #6 1.2 28.3 14.52

Realization #7 1.33 27.38 14.46

Realization #8 0.69 27.65 14.99

Realization #9 1.13 28.13 14.5

Porosity Measurement Comparison
Phie
Cpor

Porosity Statistics
Phie Cpor
Min 2.79 9.212 Difference in mean
Max 29.34 28.616
Delta 26.54 19.404 is 2.05 PU
Number of
Defined Values 296 92
Mean 18.36 20.41
Std. dev. 6.44 4.1544
Variance 41.54 17.259
Sum 5434.2 1877.7
Porosity Modeling
Result of Various Modeling Approach

Bulk shift for uncertainty in porosity

-/+ 2.05 PU

Bulk shift for uncertainty is applied on the best history match

performing porosity realization
 Porosity Model Uncertainty
Uncertainty Embedded Workflow

Log Porosity (316 wells)

(reconciled to OB corrected core porosity)
Scenario Based
Uncertainty
Scale-up porosity to 3D Model

Porosity Model by SGS Porosity Model by Krigging Krigging versus SGS

SGS with Collocated Krigging Model Krigging Model Krigging Model Trend
Co-Krigging to Porosity as 2D Trend without trend with modified
Map trend Correlation
coefficient

SGS with Data SGS with 5 SGS with Krigging with Data Krigging with 5 Krigging with 30
Range Variogram Km range 30 Km range Range Variogram km Variogram
Variogram
km Variogram

1 2 3 4 5 6
Seed number
 Parametric Dependencies
Conditional Modeling and
Permeability Uncertainty
Perm Modeling
Uncertainty Perm = a * Porosity b

Estimate
Permeability

data point
Poro Perm relationship
Porosity
Uncertainty Standard Variation
bounds

Input variable
Porosity
 Number of Cases to Define Distribution
1.2 1 Runs 4.5 10 Runs 8 50 Runs

4 7
1
1 Case 3.5 10 Cases 6 50 Cases
0.8 3
Frequency

Frequency

Frequency
5
2.5
0.6 4
2
3
0.4 1.5
1 2
0.2
0.5 1
0 0 0
0.85 0.9 0.95 1 1.05 0.85 0.9 0.95 1 1.05 0.85 0.9 0.95 1 1.05
Millions Millions Millions

Emission Emission Emission

60 500 Runs 600 5000 Runs 1200 10000 Runs

50 500 1000
500 Case 5000 Cases 10000 Cases
40 400 800

Frequency
Frequency
Frequency

30 300 600

20 200 400

10 100 200

0 0 0
0.85 0.9 0.95 1 1.05 0.85 0.9 0.95 1 1.05 0.85 0.9 0.95 1 1.05
Millions Millions Millions

Emission Emission Emission

 Characteristics
Uncertainty:

HC Pool Area
Connectivity
Productivity

Injectivity

Reservoir

Structure
STOOIP

Fluid
FWL
Reduce your Data:

Production Test/
uncertainty by PBU/PFO
Long Term Test
consciously (Prod/Inj)
Interference Test
developing Production Logging

data gathering Tracer test

Formation Pressure

program PVT data

3D Seismic
Well Coverage
Core data (whole
core analysis)
Log data
15
Model Upscaling – Lost Precision in Processing
Upscaled 13 and 23 Layers vs. Geomodel (47 Layers)

13 Layers
23 Layers 47 Layers

16
Permeability Model
Cloud Definition for Uncertainty Multiple Realizations

Density of the points

Extended limits
 Upper Reservoir (R1) Porous 1.2 Sub-zone

Extended Cloud Case Reduced Cloud Case

Permeability Permeability

Extended Cloud Case Reduced Cloud Case

Permeability Permeability
 Volumetric Uncertainty Analysis

Tornado Chart

Structure

Porosity

Water
Contact

Water
Saturation

STOIIP (MMSTB)
 Porosity Modeling
Uncertainty parameters

Final porosity model is obtained by history match process

Uncertainty cases are generated on various parameters:
Modeling algorithm: SGS versus Krigging
Trend
Variogram
Seed number
Bulk shift due to core porosity – log porosity difference
 Permeability Modeling
Uncertainty Embedded in Permeability Modeling Workflow

Permeability (10 wells)

(Distribution on Phie-K Cross Plot per sub-zone)
Scenario Based
Scale-up permeability to 3D Model Uncertainty

Cloud Transformation Cloud definition

GRFS with 2D GRFS with Porosity GRFS without any Trend Trend
Porosity Map Model Trend
Trend

Data range Variogram 5 and 30 Km range

variogram
Variogram

Seed number

1 2 3 1 2 3
Permeability Modeling
Uncertainty Parameters

Final permeability model will be decided after history match

Uncertainty cases are generated on various parameters
Modeling algorithm: SGS with various trends
Cloud definition
Variogram
Seed number
Bulk shift due to uncertainty in measurements
 Conclusion
• Uncertainty analysis starts with data processing and review
• Along the modeling work, analysis can be incorporated into
workflow
• Uncertainty parameters can be evaluated by stochastic or
deterministic methods
• Conscious data planning can reduce uncertainty