Sunteți pe pagina 1din 96

Mapping Natural Fractures Using 3D Seismic and Well Data: Application to a Shale Play

A. Ouenes

Prism Seismic March 2010

Acknowledgements

T. Anderson , B. Black and V. Stamp from the Rocky Mountain Oilfield Testing Center (RMOTC)

A. Aoues, M. Mediani, D. Klepacki and C. Baillie from Prism Seismic

Outline

Motivation CFM Methodology (Continuous Fracture Modeling) Application to Niobrara Shale Conclusions

Fractured Reservoirs

More than 60% of the world reservoirs are fractured The distribution of natural fractures is not random Fracture density depends on the COMBINED effect of structural setting, lithology, reservoir and flow unit thickness, faults, and many other drivers.

Data Types Available for Fracture Modeling


Well Data
Image logs Core measurements Well test Production Data

Geomechanical Data
Spatial distribution of stress and strain related parameters based on seismic interpretation and geomechanical modeling

Geometrical Data
Geometry of the reservoir (thickness, dip, curvature, slopes, etc.)

Seismic Data
Pre-stack attributes Post-stack attributes

Seismic Data in Fracture Modeling


Fault identification on P-wave data Zones near faults assumed to be fractured Azimuthal anisotropy Shear wave data P-wave data Additional (velocity and amplitude anisotropy) Cost ! Requires specialized acquisition and processing to record the required range of azimuths and process the data to detect azimuthal variations

Seismic Data in Fracture Modeling


We can use conventional (narrow azimuth) P-wave prestack and post-stack data in fracture modeling.

Data is readily available No additional cost Some acquisition systems (marine streamer) are still not fully capable of recording wide azimuth ranges

Seismic Data in Fracture Modeling


2 different objectives

Anisotropy methods Detect effects (velocity or amplitude anisotropy) caused by fractures Pre-stack and Post-stack P-wave methods Derive seismic attributes that provide direct or indirect information on rock properties. These reservoir properties are called Fracture Drivers since they drive the origin and intensity of fracturing.

Example of Fracture Drivers and their Seismic Proxy

Faulting is a major fracture driver. The best seismic attributes that provides accurate information on faults are those derived from volumetric curvature Rock hardness plays a major role in fracturing. The impedance derived from a seismic inversion provides a good indication on rock hardness and constitute a key fracture driver

The most negative curvature used as a fracture driver


Faults

Fracture density derived from FMI Logs

Impedance used as a fracture driver

Outline

Motivation CFM Methodology (Continuous Fracture Modeling) Application to Niobrara Shale Conclusions

Fractures are Present at Different Scales

L 1= 101 102 m L = 101 103 m


2

Modeling Fractured Reservoirs


To plan the drilling of new wells and to optimize the field recovery, a 3D geocellular model of fracture density, porosity and permeability is needed The gridblock sizes are: X and Y ~ 50 to 100m Z ~ 2 to 5m

Modeling Fractured Reservoirs with the CFM Approach

The Continuous Fracture Modeling (CFM) will be used to solve two problems:
Distribute the available wells logs (fracture density, permeability, etc..) in the 3D geocellular grid Distribute in 2D, hydrocarbon performance indicators (Cumulative Oil production, EUR, or any production proxy for the fracture density around the wells) to find un-drilled sweet spots

CFM Workflow
Seismic Data

Structural Interpretation in Time

Depth converted and resampled seismic attributes

Attribute Generation Acoustic Impedance Inversion Spectral Imaging Volumetric Curvature Seismically constrained geologic models Well Data Lithology Porosity

Stochastic 3D fracture model

Fracture Identification logs in wells

Outline

Motivation CFM Methodology (Continuous Fracture Modeling) Application to Niobrara Shale Conclusions

Importance of the Niobrara Shale

Large oil and gas resources are available in the Niobrara shale and remain unexploited (From Longman, Luneau, Landon, 1995)

Importance of the Niobrara Shale

Isopach map of the total Niobrara (From Longman, Luneau, Landon, 1995)

Objectives

The Niobrara Shale at Teapot Dome (WY) is used as an illustration for the application of the CFM approach The oil productivity of the Niobrara shale is estimated by using
Few core permeability measurements at one well Fracture density extracted from 3 image logs that show very limited fractures Cumulative oil production at 88 wells

Where is Teapot Dome ?

Geology of the Teapot Dome


Teapot Dome is an asymmetric
anticline spilled westward, with hinge slightly curved.

Nine productive formations.


Target Formation: Niobrara Age: Upper cretaceous

Niobrara shale produces oil at Teapot Dome

Average thickness of the


Niobrara formation is 450 feet ~ 138 m

Niobrara Oil Shale Project

Very limited data 20 wells have some logs which include sonic, gamma ray, and density. 8 wells have neutron porosity 3 wells have FMI image log data Only one core from one well has few permeability and porosity measurements The goal: build reliable reservoir models adapted to the available data at hand. The challenge: how to build reliable reservoir models when dealing with scarce well data, a common situation found in shale projects ? Prism Seismic software, CRYSTAL and REFRACT, and workflows are used to solve this problem.

Well Correlations showing GR and Sonic Logs

Top
NBRRws

NBRRsg

Niobrara Formation (Shale)


Bottom

CRLL

F1WC

Seismic Resolution Enhancement


Original Seismic Enhanced Seismic

Higher amplitudes in the 42.5 - 90 HZ frequency range in the enhanced data

Enhanced Seismic data has a better

vertical resolution compared to the original seismic data

Seismic Resolution Enhancement


Original Seismic Xline 54
NBRRws

Improvement in vertical resolution is observed in the enhanced amplitude data

Seismic Enhanced Xline 54


NBRRws

Well Tie & Synthetics


Well - 490251070900 Ricker

NBRRws

NBRRsg

CRLL

Ricker Frequency=40Hz Phase=180deg Length=100ms

Seismic Structural Interpretation

For the seismic interpretation of horizons and faults, the following seismic attributes were used simultaneously :
Enhanced Seismic (Horizons & Faults) Colored Inversion (Horizons & Faults) Volumetric Curvature Attributes (Faults)

Seismic Attributes Used in Structural Interpretation Colored Inversion

Enhanced seismic

Top and Bottom Niobrara Horizons


Top Bottom

Seismic Attributes

Used various algorithms available in CRYSTAL to compute multiple 3D seismic attributes. These algorithms are:
Volumetric curvature Spectral imaging Stochastic and Deterministic Seismic inversions

Volumetric Curvature Used for Fault Interpretation


Faults and potential fractured zones

0.1

0.07

C-Curvature Slice 630 ms


0 -0.03

Most Positive Curvature Slice 630 ms


- 0.07

Maximum Curvature Slice 630 ms

Inferred Faults @ Surface Vs Subsurface

C - Curvature

From Cooper et al, 2000

Deterministic vs Stochastic Impedance


Average Impedance 7500
N

GLI Inversion

Stochastic Inversion

Average Impedance 9000

Building a Structural Framework in the Time Domain The water tight Structural Framework model was built with 14 faults and 3 horizons:
NBRRws NBRRsg CRLL

3D Structural Framework Model in the Time Domain

Time Depth Conversion

The water tight structural framework is used to build a 3D Geocellular grid in the time domain All the seismic attributes are snapped to the 3D geocellular grid in time The 3D geocellular grid in time is depth converted using an average velocity computed at the top of the grid and an interval velocity cube computed from the well data All the reservoir modeling is done on the depth 3D geocellular grid

Sequential Geologic Modeling

Prism Seismic Sequential Geologic Modeling approach is used to estimate 3D models of reservoir properties that are constrained by multiple seismic attributes and geologic models This stochastic modeling approach uses a neural network which allows the user to build reliable 3D models even with very limited well data reduced to one well (core permeability example described in the next slides)

3D Reservoir Modeling

A neural network is used to find a relationship between well data and a large number of seismic attributes and any available geologic model This approach is applied in a sequential manner to:
Gamma ray logs available at 20 wells Density logs available at 20 wells Neutron porosity available at 8 wells Core permeability available at one well Fracture density available at 3 wells

3D Geocellular Grid
A 3D Geocellular grid with 1,217,520 cells was used for the 3D modeling The grid has Nx = 89 , Ny = 171, Nz = 80 cells in the 3 main directions The cell size is 67 m x 67 m x 1m There are 80 conformable layers each about 1 meter thick

Data Used in the 3D Modeling


Well data
20 wells have GR, Density and 8 wells have Neutron Porosity logs 3 wells have an FMI log Only one well has few core data measurements of Porosity and Permeability

Seismic data
- Seismic attributes : Impedance from various Post stack inversions, various attributes from volumetric curvature, and Spectral imaging attributes

Wells with Logs Used in the 3D Modeling

20 wells have GR, Density logs 8 wells have Neutron porosity logs

Well with Core Data Used for the 3D Modeling

Only one well 38-AX-10 has core data

Blocked logs of core porosity and Permeability at 38-AX-10

The core data intersects only 10 layers of the 3D geocellular Grid

Porosity

Permeability

Well with FMI Data Used for the 3D Modeling

3 wells: 25-1-X-14, 67-1-Tpx-10, 71-1-X-4 have FMI data

Fracture density logs from FMI data along X section showing the stochastic impedance 25-1-X-14 67-1-TpX-10 71-1-X-4

Seismic Attributes used in the 3D Modeling


Some seismic attributes used to constrain the geologic models :
Impedance cubes from various Seismic Inversions Most Negative Curvature Maximum Curvature 16% Energy Frequency 84 % Energy Frequency Decay Instantaneous Frequency 80 hz Envelope 44 % -16 Maximum Amplitude Total energy Tuning Frequency Upper Bandwidth

A Deterministic Impedance from Inversion

Maximum Curvature from Volumetric Curvature

Maximum Amplitude from Spectral Imaging

Tuning Frequency from Spectral Imaging

Upper Bandwidth from Spectral Imaging

1st Geologic Model: Gamma Ray

Input:
GR logs available at 20 wells Multiple seismic attributes derived from the seismic inversions, spectral imaging and volumetric curvature

Output: Multiple realizations of GR models

3 Realizations of the 3D GR model Shown at Layer 10

Average 3D Gamma Ray Model

2nd Geologic Model: Density

Input:
Density logs available at 20 wells Multiple seismic attributes derived from the seismic inversions, spectral imaging and volumetric curvature GR models derived in the previous step

Output: Multiple realizations of density models

3 Realizations of the 3D Density model Shown at Layer 10

Average 3D Density Model

2D Average map of Density in the Niobrara interval shown on the Crll horizon Low Density zones

Comparison between 2D Average Density map and Curvature attribute

Low density zones seem to be present around some faults and could be areas rich in calcite (see next slide)

About 15 ft of Sussex Sandstone U D dip-slip displacement

Courtesy of Brian Black, RMOTC

3rd Geologic Model: Neutron Porosity

Input:
Neutron Porosity logs available at 8 wells Multiple seismic attributes derived from the seismic inversions, spectral imaging and volumetric curvature GR models derived in the previous steps Density models derived in the previous step

Output: Multiple realizations of neutron porosity models

3 Realizations of the 3D Neutron Porosity model Shown at Layer 10

Average 3D Neutron Porosity Model

4th Geologic Model: Core Permeability

Input:
Core data available at 1 well Multiple seismic attributes derived from the seismic inversions, spectral imaging and volumetric curvature GR models derived in the previous steps Density models derived in the previous steps Neutron porosity models derived in the previous step

Output: Multiple realizations of core permeability models

Core Data in the 3D Geocellular grid

The core data intercepts only the top 10 layers of the 3D geocellular grid Each layer is about 1 meter thick

3 Realizations of the 3D Core permeability model Shown at Layer 10

Average 3D Core permeability Model

2D Average map of Core Permeability in the Niobrara interval shown on the Crll horizon

Comparison between Average Permeability map computed in the Niobrara Interval and Cumulative Shale Oil Production
Cumulative Shale Oil Production Average permeability

Cross section of Permeability model (Y section)

Cross section of Permeability model (X section)

5th Geologic Model: Fracture Density

Input:
Fracture density computed from FMI data available at 3 wells Multiple seismic attributes derived from the seismic inversions, spectral imaging and volumetric curvature GR models derived in the previous steps Density models derived in the previous steps Neutron porosity models derived in the previous steps Core permeability models derived in the previous step

Output: Multiple realizations of fracture density models

3 Realizations of the 3D Fracture Density model Shown at Layer 10

Average 3D Fracture Density Model

Log normal distribution

2D Average map of Fracture Density computed in the Niobrara interval shown on the Crll horizon

3D Fracture Density and Core Permeability Model at layer 1

3D Fracture Density and Core Permeability Model at layer 20

3D Fracture Density and Core Permeability Model at layer 40

3D Fracture Density and Core Permeability Model at layer 60

3D Fracture Density and Core Permeability Model at layer 80

Comparison between Average Fracture Density map in the Niobrara Interval and Cumulative Shale Oil Production
Cumulative Shale Oil Production Average Fracture Density

Observations from the 3D Modeling

Using the very limited core permeability and fracture density available at the wells, the derived 3D models seem to correlate with the cumulative oil production What if there are no core permeability or fracture density logs ?

2D Modeling of the Niobrara Productivity

The total cumulative oil production is used as a proxy for the fracture density around a well. Based on the cross sections of permeability, we will assume that most of the production comes from Niobrara reservoir and the contribution of the Steele is minimal The key drivers used in the 3D modeling will be averaged over the entire Niobrara reservoir and used as drivers for the 2D modeling

2D Drivers: Average Seismic Attribute Maps

Most Negative Curvature

Tuning Frequency

2D Drivers: Average Geologic Attribute Maps

Gamma Ray

2D Drivers: Average Geologic Attribute Maps

Porosity Permeability

Linking the Drivers to Cumulative Oil

Seismic and Geologic 2D Attribute Maps

ReFract Neural Net

Cumulative oil at the wells

Cumulative Oil 2D Maps

3 Realizations of Cumulative Oil Production

Average Predicted Cumulative Oil Production Map vs Actual Cumulative Oil at the Wells

Analysis of the 2D Niobrara Cumulative Oil

The derived model with its multiple realizations for the cumulative oil are available over the entire Teapot area. The models could be used to derive probability maps to encounter a location that could have 25,000 bbls cumulative oil

Probability to have 25 K bbls cumulative Oil

Conclusions

With very limited well data, and a narrow azimuth 3D seismic, both 3D and 2D reservoir models related to the Niobrara shale productivity could be derived. In 3D, very limited core permeability and image logs were used to derive 3D models constrained by a multitude of seismic and geologic attributes In 2D, average geologic and geophysical attributes derived from the 3D modeling effort, were used to map the Niobrara productivity represented by cumulative oil recovery

Conclusions

Advanced reservoir modeling technologies reveal un-drilled Niobrara Shale sweet spots in the Teapot Dome almost a century after Niobrara production started

Additional Reading
Jenkins, C., Ouenes, A., Zellou, A., Wingard, J.: Quantifying and predicting naturally fractured reservoir behavior with continuous fracture models, AAPG Bulletin, V. 93, No 11 (November 2009) Boerner, S, Gray, D., Todorovic-Marinic, D. Zellou, A., Schnerck, G., Employing Neural Networks to Integrate Seismic and Other Data for the Prediction of Fracture Intensity, SPE 84453 Christensen, S.A., Ebbe Dalgaard, T., Rosendal, A., Christensen, J.W., Robinson, G., Zellou, A., Royer, T: Seismically Driven Reservoir Characterization Using an Innovative Integrated Approach: Syd Arne Field, SPE 103282 Pinous, O., Sokolov E.P., Bahir, S.Y., Zellou, A., Robinson, G., Royer, T., Svikhnushin, N., Borisenok, D., Blank, A., Application of an integrated approach for the characterization of a naturally fractured reservoir in the West Siberian basement (example of Maloichskoe Field), SPE 102562

Additional Reading

Ross, J.G, Zellou, A., Klepacki, D.: Seismically Driven Fractured Reservoir Characterization Using an Integrated Approach Joanne Field UK, Paper Q048, EAGE Annual Meeting June 2009 Ouenes, A., Robinson, G., Balogh, D., Zellou, A., Umbsaar D., Jarraya, H., Boufares, T., Ayadi, L, Kacem R.: Seismically Driven Characterization, Simulation, and Underbalanced Drilling of Multiple Horizontal Boreholes in a Tight Fractured Quartzite Reservoir: Application to Sabria Field, Tunisia., SPE 112853 Bejaoui, R, Ben Salem, R., Ayat, H., Kooli, I., Balogh, D., Robinson, G., Royer, T., Boufares, T, Ouenes, A. Characterization and Simulation of a Complex Fractured Carbonate Field Offshore Tunisia, SPE 128417 Ouenes, A., Anderson, T., Klepacki, D., Zellou, A., Araktingi, U., Bachir, A., Boukhelf, D., Holmes, M., Black, B., Stamp, V.: Integrated Characterization and Simulation of the Fractured Tensleep Reservoir at Teapot Dome for CO2 Injection Design., SPE 132404

S-ar putea să vă placă și