Scribd
Upload
8 views

Introduction to Digital Rock Physics And

Uploaded by

Ajendra Singh
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
8 views

Introduction to Digital Rock Physics And

Uploaded by

Ajendra Singh
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 22

Introduction to

Digital Rock Physics and Predictive Rock Properties

Mike Marsh, Ph.D. Shawn Zhang, Ph.D.


Carl Zeiss Microscopy DigiM Solutions

BP

Numerical Rocks (FEI-Lithicon)


Ingrain
What is Digital Rock Physics?

 Application of Numerical Modeling to rock properties

 Digital Rocks IS NOT Digital Image Analysis


- Porosity
- Digital petrography and granulometry
- Fracture mapping and analysis
- Bedding analysis
- Mineralogy
- Bioturbation analysis

 Digital Rocks IS simulation of rock physics


- Fluid transport (Stokes equations)
- Electrical conductivity (Ohm’s law)
- Diffusion (Ficke’s second law)
- …
Introduction to Numerical Modeling

Numerical modeling is a mainstay of engineering

 Computer-aided Engineering (CAE)

 General-purpose CAE Tools (not exhaustive)


- Ansys CFD
- Abaqus
- COMSOL Multiphysics
- Exa
- Nastran
- OpenFOAM
- …many, many more

Mainstay of Reservoir Engineering


 Dynamic Reservoir Modeling Tools

Nonlinear static analysis of a 3D structure subjected to plastic


deformations
Solving physics by simulation

 Construct a model
- Discretize a spatial domain

 Constrain with boundary conditions


- Interface behavior
- Initial values

 Use a compute engine and a numerical


scheme to integrate
differential equations for
physics of interest

 Iterate until some convergence


criterion is satisfied

Nonlinear static analysis of a 3D structure subjected to plastic


deformations
Solving physics by simulation

 Where does the model come from?


- Traditional CAE:
Engineer drafts a model
- Digital Rocks:
Model is derived from an image

 CAD-based
- Engineering draws a picture of the spatial domain
- Limited to simple human-engineered domains

 Image-to-simulation
- Directly image a material
- Natural materials
» Bones, rocks
- Manmade materials
» Foams, composites

- Segment image into discrete phases


(e.g. pore and matrix)
- Convert segmentation into spatial model Nonlinear static analysis of a 3D structure subjected to plastic
deformations
- Result is sensitive to imaging and
segmentation accuracy
(difficult to assess, but see DRP benchmarks paper)
Diverse Expertise Required for Digital Rocks

Geological
Image Numerical Interpretation
Imaging
Processing Modeling and
Application
Spatial discretization

 Unstructured polyhedral meshes


- Uniform or Adaptive Mesh Refinment

 Lattice models

 Pore network models

Autodesk website

exolete website
PoreSim website

7
Flow modeling - Geometry

Pi Po
Q

Pi, Po - inlet and outlet pressures


L
Q is the flow rate
L is the length of the porous medium
S is the cross section
Flow modeling – Darcy’s Law

Pi Po
Q

μ is the dynamic viscosity of the fluid

k is the permeability
Flow modeling – Darcy’s Law
Stokes equations can be solved to give you the pressure drop

Pi Po
Q

μ is the dynamic viscosity of the fluid

k is the permeability
Flow modeling – Darcy’s Law
Image-derived model captures pore space geometry

Pi Po
Q

μ is the dynamic viscosity of the fluid

k is the permeability
Flow modeling – Darcy’s Law
Boundary conditions constrain the simulation

Pi Po
Q

μ is the dynamic viscosity of the fluid

k is the permeability
13
Tradeoffs between Compute Time and Domain Size

Different strategies for encoding the spatial domain capture different

Tradeoff
 Size of the domain (more representative results)
 Precision of the geometry (more accurate results)

Bigger domain Precise geometry


Compromised geometry Smaller domain

Pore network modeling Lattice Boltzmann Irregular Mesh


iRock, Lithicon, PoreSim, IFP Ingrain, Exa Ansys/Fluent
COMSOL Multiphysics
Pixel FVM
GeoDict, Xlab-Hydro, Simpleware, Volume Graphic
Academic FVM solvers
Compute strategies

Can I have both: precise geometry AND big domains?


 Shared memory machines (expensive, but easy to program)
 Beowulf clusters (cheap, but difficult to program)
 GPU clusters (difficult to program, difficult to scale)

Linden et al. SCA 2014-057


Advantages and Disadvantages relative to lab
experiments

 Advantage: Speed
- Faster than lab experiments – consider SCAL on shales

 Disadvantage: REV
- Imaged and modeled volume may not be representative and far smaller than domain of
corresponding lab experiment

 Advantage: Perform experiments that are difficult to do


- No real-world apparatus required
- Repeat experiments under different boundary conditions
Landscape of DRP:
Academic Efforts

 ANU/UNSW (Digital Core Consortium)


 ETH
 IFP
 Imperial College (Consortium on Pore Scale Modeling)
 Heriot Watt
 LSU (PoreSim Consortium)
 Stony Brook University
 University of Wyoming
 University of of Texas
Business Models in Digital Rock Physics

DRP as part of a Research Contract

Research contracts may address an open problem related to reservoir


characterization or production. Sometimes, service providers may use DRP
simulation engines as part of their effort to fulfil such contracts

DRP: Simulation as a Service


Service providers may offer a DRP simulation service, whereby customers can
provide rocks which must be imaged (or existing images) and the service provider
will execute a DRP simulation and provide the results to the customer.

DRP: Simulation Software installed on End-user Computer


Software vendors may sell software licenses for applications that end-users can
install on their own hardware, whereby those end-users can execute the DRP
simulations at any time on any digital image.
Additional Reading

Digital Rocks: Developing an Emerging Technology Through to a Proven Capability


Deployed in the Business
Fredrich et al. 2014 (SPE-170752-MS)
SPE ATCE 2014

Digital rock physics benchmarks—Part I: Imaging and segmentation


Andrae et al. 2013
Computers and Geoscience

Digital rock physics benchmarks—Part II: Computing effective properties


Andrae et al. 2013
Computers and Geoscience

19
Advantages and Disadvantages relative to lab
experiments

 Advantage: Speed
- Faster than lab experiments – consider SCAL on shales

 Disadvantage: REV
- Imaged and modeled volume may not be representative and far smaller than domain of
corresponding lab experiment

 Advantage: Perform experiments that are difficult to do


- No real-world apparatus required
- Repeat experiments under different boundary conditions
Diverse Expertise Required for Digital Rocks

Geological
Image Numerical Interpretation
Imaging
Processing Modeling and
Application

You might also like