Module 5 Property Modeling
Petrel Workflow Tools Introduction
l e r t e P o t o r t n I
Surfaces and Data edit
Stratigraphic Modeling
3D Grid Construction: Structural Modeling
e c o a i f r d e t u t n S I
Pillar Gridding
Fault Modeling 3D Grid Construction Structural Framework
3D Grid Construction Structural Gridding
Property Modeling
Well Log Upscale
Facies & Petrophysical Modeling
Make Horizons Zones & Layering Make contacts & Volume Calculation r o t i d E w o l f k r o W
Facies Modeling Objectives
General Property Modeling Workflow Discuss Different Facies Modeling Techniques Deterministic techniques Stochastic techniques. Learn How to use Common Settings: Set filters Learn How to use Zone Settings: Define zones Learn How to use different Algorithms Sequential Indicator Simulation Object Modeling Fluvial channel General object modeling Interactive Modeling.
– –
– – – – –
Property Modeling General Workflow
Less data More uncertainty Stochastic
Estimation
Pixel based
Interpolation
Object based
Addressed
Deterministic
More data Less uncertainty
Stochastic vs. Deterministic Modeling Methods
Stochastic
Deterministic
Random (Seed number)
It is unlikely due to unpredictable factors.
It generates different equiprobable results for different seed numbers.
It generates the same result for a given set of initial conditions.
Variable states are described by probability distributions.
Variable states are described by unique values.
It does not need upscaled cells: Unconditional modeling.
Need upscaled cells; needs more data.
Allows more complexity and variability in the model; can help assess uncertainty.
Faster to run.
Algorithms Covered in the Course Stochastic methods Pixel based technique Sequential Indicator Simulation algorithm
Object-based technique Object modeling algorithm
Distributes the property using the Allows you to populate a discrete histogram. Directional settings, property model with different such as variogram and trends, also bodies of various geometries, are honored. facies types, rules, and fractions.
Deterministic method Direct addressing technique Interactive modeling drawing
Allows you to paint facies directly on the 3D model.
Facies Modeling Dialog Box Two main modeling settings buttons are available: (Common and Zone settings).
Common Settings Defines general settings for the grid properties to be made for all zones.
Zone Settings Defines settings for individual zones (captured from Models pane > Zone filter folder).
Common Settings Use filter: Should be chosen only if a filtered part of the grid is to be modeled. Ensure that all cells get a value: If there is no input data, all cells will be populated by averaging surrounding cells. Local model update: Updates the model inside a region, inside a property, or around a well Number of realizations: When running Uncertainty analysis, multiple realizations are made with the same input data. Overwrite: Will overwrite the previous realizations with same suffix number.
Zone Settings Zone: Click to activate zonation. Choose a zone to model from drop-down list. Facies: If conditioning to a previous facies model, click the Facies button. Lock: Leave zone unchanged; unlock to activate zone settings. Method: Set the appropriate method from the drop-down list for the zone to be modeled.
Create a Sequential Indicator Simulation Property Model (1) SIS is a pixel-based modeling algorithm, using upscaled cells as the basis for fraction of facies types to be modeled. The variogram constrains the distribution and connectedness of each facies. 1. Set an upscaled property: (U) as suffix. 2. Choose the zone to model and unlock it. 3. Set SIS as the Method for one zone. 4. Choose the facies from the template. Click the Blue arrow to insert them into the model.
Create a Sequential Indicator Simulation Property Model (2)
5. Variogram (2 methods): Specify Range, Nugget and Type manually. Click Get a variogram from Data Analysis • •
6. Fraction (3 methods): Use Global fraction from Upscaled cells. Use probabilities (property/trend). Use attribute probability curves or vertical proportion curves from Data analysis. • • •
Variogram: Quantifies Spatial Continuity of the Data There are many variogram types that can be fit into the data. Petrel provides three options of prominent types: exponential, spherical, and gaussian variograms. You need three directions: Two in the horizontal (major and minor ) and one in the vertical direction. The range points the distance from which above, the spatial dependence is set to randomness. The azimuth is the rotation angle of the major range.
Variogram is calculated in 3 directions
Major
•
Vertical
•
•
Variogram & parameters Sill
Range
Nugget 1
2
Separation
3 4 5 distance (lag)
Create a Fluvial Channel Model (1): Facies Bodies The Object modeling method uses upscaled cells as a basis for the fraction of facies types to be modeled. The objects follow a strict geometry, distribution, and trend defined by the user. 1. Set an upscaled property: (U) as suffix. 2. Set the zone to model and unlock it. 3. Set Object modeling as the Method to use. 4. Click the Fluvial channels icon to insert a channel body. 5. Choose facies properties to match Channel and Levee. 6. Fraction (2 methods): Use fraction of Channels and Levees from upscaled cells. (Gray field is not editable.) Enter a fraction. (The white field is editable.) •
•
Create a Fluvial Channel Model (2): Geometry
Layout: Specify Orientation, Amplitude and Wavelength.
Note: Drift applies randomness to each parameter.
Channel: Specify the width and thickness of the channel. Thickness can be in distance units or as a fraction of the width.
Levee: Levees are the wing shaped deposits on the side of the channel. Specify width and thickness (smaller than channel).
Create a Fluvial Channel Model (3): Trends and Probabilities
Use volume probability: Use a function Use a surface Use a 3D probability property (usually a seismic attribute). • • •
Use Channel trends: Flow lines are digitized polygons used as fairways for the channels to follow Source points are indications of paleoheighs/provenance; where channels begin. •
•
Create a Fluvial Channel Model (4): Background Background facies After the channel is defined, choose a background facies. This is distributed wherever channels are not placed. •
•
Background can be undefined, a single facies type, or a previously generated property.
Create a General Object Model: Facies Bodies The General object modeling approach creates standalone objects following a strict geometry defined by the user. 1. Set an upscaled property: (U) as suffix. 2. Set the zone to model and unlock it. 3. Set Object modeling as the Method for the zone. 4. Click the Add a new geometric body button. (Ellipse geometry is chosen by default.) 5. Choose the facies type you want your body to have. 6. Fraction (2 methods): Use fraction of upscaled cells. Enter a fraction (white field = editable). •
•
Perform Interactive Modeling: (Draw Facies) Interactive drawing of facies types that are not easily modeled. Tip: Use Simbox view and make a copy of the property.
Brush type
Facies type
Radius Height
Note: Irreversible process: This overwrites all other facies, including upscaled cell values. No undo!
Profile
EXERCISE Facies Modeling
Extra: Object Modeling: Fluvial Channels Result
No drift applied (0)
Drift applied (>0, <1)
Facies Modeling Methods: Overview (1) Deterministic Estimation
Learning system
Direct Addressing
Indicator Kriging
Asign values
Interactive
Discrete distribution of the property honoring the predefined histogram
Choose from Allows you to paint undefined, constant, facies directly on the other property, surface 3D model. and vertical functions.
Artificial Neural Net Uses the classification model made in the Train Estimation model.
Facies Modeling Methods: Overview (2) Deterministic Estimation Indicator Kriging
Learning system
Direct Addressing Asign values
Interactive
Artificial Neural Net
Facies Modeling Methods: Overview (3) Stochastic Pixel based
Object based
Sequential Indicator Simulation
Truncated Gaussian Simulation
Truncated Gaussian Simulation with trends
Multi-point Facies Simulation
Object Modeling
Distributes the property using a histogram. Directional settings (e.g., variogram and extensional trends), also are honored.
Used mostly with carbonates where facies are known to be sequential. It deals with large amounts of input data, such as global fractions and trends.
Distributes the facies based on a transition between facies and trend direction. Trends are converted into probabilities to then run TGS.
The variogram is replaced by a training image giving both the facies and the relative position to each other, describing the spatial correlation from one-tomultiple points.
Allows to populate a discrete facies model with different bodies of various geometries, facies and fraction.
Facies Modeling Methods: Overview (4) Stochastic Pixel based Sequential Indicator Simulation
Truncated Gaussian Simulation
Truncated Gaussian Simulation with trends
Object based Multi-point Facies Simulation
Object Modeling
Object Modeling Adaptive Channel Modeling
Petrel 2008.1: modified to honor the channel-levee association with substantial well control over several layers (cross-layer). Uses sequential Gaussian simulation. Better to use than traditional object modeling techniques in situations with large numbers of well constraints and honors channel connectivity. In Petrel 2009.1, you condition the model to a 3D seismic probability.
A
C B
Object Modeling: Adaptive Channels 1. Property and zone selection a. Make sure to pick the correct property; must be upscaled, i.e., have (U) as suffix. b. Select Object Modeling as the method for one zone. 2. Facies body: a. Click the Adaptive channels icon to insert a channel body. b. Choose facies properties to match. c. Use the fraction of the upscaled cells or enter a value
Multi-Point Facies Simulation
Developed by Schlumberger Research (Boston) and introduced to the Facies modeling process for Petrel 2009.1. Honors well, seismic, and probability data.
It can model complex geological features and connectivity. It efficiently generates multi-million cell grids. A geological conceptual model is needed to build a pattern that will capture the probabilities and distribution of the facies. This training image subtitutes the variogram.