Measurements

A) Geophysical Observation

B) Assign Local Physical Properties

Fluid Substitution and Upscaling

Acknowledgements

References

We thank the Stanford University Global Climate and Energy Project for funding this work. Steve Brown was funded by DOE SBIR grant number DE-FG02-05ER84207.

Garbozi, E.J., and J.G. Berryman, 2001. Elastic moduli of a material containing composite inclusions: Effective medium theory and finite element computations, Mechanics of Materials, 33, 455-470.

Torp, T.A., and J. Gale, 2004. Demonstrating storage of CO2 in geological reservoirs: The Sleipner and SACS projects, Energy, 29, 1361-1364.

Wang, Z., M.E. Cates, and R.T. Langan, 1998, Seismic Monitoring of a CO2 flood in a carbonate reservoir: A rock physics study. Geophysics, 63, 5, 1604-1617.

In order to simplify the rest of the analysis, we used cluster analysis of the physical properties measurements to identify petrophysical facies within the Berea block. The identified facies are then treated as distinct regions within the sample which have identical physical properties. For this example we simplified the Berea sample from a continuous distribution of properties to an assemblage of 5 petrophysical units.

CO2 saturation was then simulated by building a capillary

pressure model from the porosity and permeability distributions, and progressively saturating the sample with CO

2 and water as a function of pore pressure.

Paul Hagin1 and Stephen Brown2

1Department of Geophysics, Stanford University, Stanford, CA 94305-2215, *hagin@stanford.edu 2NER, Inc., 331 Olcott Drive. White River Junction, VT, 05001

At Sleipner, CO2 is injected in the

Utsira Formation, a massive, high po-rosity sandstone aquifer with shaley interbeds.4-D seismic was selected for moni-toring and verification, with the hope of being able to track CO

2 dispersion.

While 4-D seismic successfully lo-cated CO

2 as it moved through the

aquifer, the volume or saturation of CO

2 at any given location was unde-

tectable (Torp and Gale, 2004).

At McElroy, a low-porosity carbonate is overlain by a higher-porosity car-bonate. Velocity changes following CO

2 are predicted to be <1%, and un-

detectable using 4-D seismic.However, the predicted changes in AVO parameters A and B are >10%, and easily detectable (Wang et al. 1998).

Defining the Problem

Proposing a Solution

The simple example from the McElroy Field shows that for very stiif rocks such as carbonates, the changes in velocity due to CO

2 injection are too small to be de-

tected, yet the changes in AVO response are significant.

Gassmann’s equations assume that fluid substitution in-volves perfectly mixed homogeneous fluids. However, the case of undissolved gas in water or oil represents a perfectly unmixed fluid, and results in patchy saturation. Our modeling suggests that dry frame heterogeneties can also cause patchy saturation.

Interestingly, laboratory measurements on Berea sand-stone, which is assumed to be homogeneous and iso-tropic, predict a saturation response which falls between the homogeneous and patchy end-members. For the case of supercritical CO

2, this results in an AVO re-

sponse that monotonically decreases with saturation.

Geophysical Observation: The geological in-jection target site is characterized using seismic, well-logs, core samples, and other sources of data.

Assign Local Physical Properties: Physical properties such as porosity, permeability, and stiffness are defined for each stratigraphic and flow unit (facies).

Fluid Substitution: The initial pore fluid satura-tion state is updated to reflect the saturation state following a given volume of injected CO

2, and the

associated change in physical properties is mod-eled.

Upscaling: Volume averaging of the local veloci-ties within individual petrophysical facies pro-duces a description at length scales correspond-ing to field seismic data.

AVO Calculation: the response of the principal interfaces can now be calculated.

Workflow for Prediction of AVO Response to variable CO2 Saturation

Standard geophysical methods for monitoring CO2 injection have severe

limitations in many cases. In particular, while the presence of CO2 can be

identified with seismic monitoring, it is difficult to quantify changes in CO2

saturation in most reservoirs. We hypothesize that time-lapse AVO may be more sensitive to changes in CO

2 saturation than traditional 4-D seismic.

We present an example using Berea sandstone that combines laboratory and numerical experiments to show that AVO can successfully track chang-es in CO

2 saturation. C) Fluid Substitution

GCEP Research Symposium 2007, Oct. 1-3, 2007

We have developed a multi-probe physical properties scanner, which allows for millimeter-scale mapping of petrophysical properties on a slabbed sample of rock or soil.

The scanner probe allows for the following physical properties to be measured: gas permeability, P- and S-wave velocities, and complex electrical impedance.

Using this device we have characterized the heterogeneity and anisotropy of a finely-bedded block of Berea sandstone. Berea sandstone has long been regarded as a labotory standard of homogeneous, isotropic behavior.

However, we find that permeability, resistivity, and velocity exhibit complex heterogeneity at a variety of scales, which we describe in the next section. Permeability varies by a factor of 3, velocity varies by ~10%, and resistivity varies by ~20%.

AVO Monitoring of CO2 Sequestration: A Benchtop Modeling StudyIntroduction

4-D seismic monitoring at Sleipner tracks the location of CO

2 in the

aquifer, but not changes in CO2

saturation. Saturation changes are currently modeled using a fluid flow simulator.

AutoScan II physical properties scanning table

Distribution of physical properties of Berea sandstone asmeasured in the laboratory. Using these data, we define the petrophysical faces through cluster analysis.

D) Upscaling

Gassmann fluid substitution for Berea sandstone as a function of CO2

saturation. Pore fluid properties were set a 3 different conditions, to simu-late CO

2 as both a gas and a supercritical liquid. The velocities have been

upscaled from the heterogeneous distribution described in B in two differ-ent ways. The first technique is based on simple volume averaging and is shown in the dashed lines. The second technique (solid lines) respects the anisotropy and heterogeneity in the sample, as described below.

We compute effective elastic properties of the Berea block at the scale of seismic resolution through a finite-element simulation. The advantage of this method is that the underlying spatial correlation of the heterogeneity is preserved as a general form of anisotropy in the upscaled elastic stiffness tensor.

We use the NIST elastic FEM program elas3d (Garboczi and Berryman, 2001) to compute the complete elastic stiffness tensor of the Berea block. This is done by applying strain boundary conditions and computing the cor-responding stresses. This is done in turn for a sequence of normal and shear strain boundary conditions to determine each component of the stiff-ness tensor. The results are restricted to the orthorhombic symmetry class to simplify the AVO calculations.

AVO Calculations

Sleipner Field

McElroy Field

0.05

0.055

0.06

0.065

0.07

0.075

0.08

0.085

0.09

0 5 10 15 20 25

Theta (degrees)

P-w

ave

Refl

ect

ion

Co

eff

icie

nt Pre CO_2 Flood

Post CO_2 Flood

Predicted AVO response of the McElroy Field following CO

2 injec-

tion equal to 60% saturation.

Rpp

(θ) = A + Bsin2θ + C[tan2θ + sin2θ]

Vp Vs ρ1 11

Vp Vs ρ2 2 2

θ

(A)

(C)

(E)

(D)

(B)

Geophysical Observation

Sw=0

Sw=1

Upscaling

Interpret Rock Types

Assign Local Physical Properties

Fluid Substitution

AVO Calculation

iVsiVp

ρi

While a large block sample of Berea sandstone looks relatively homogeneous, air permeability measurementsshow both vertical anisotropy and spatial heterogeneity.

Patchy Saturation End Member

Effect of Patchy Saturation

Vp/Vs

resistivity

velocity

permeability

Permeability189.032 704.706

Res

istiv

ity16

8.91

190.

407

Permeability189.032 704.706

Com

pres

sion

al V

eloc

ity21

49.2

924

47.3

8

Compressional Velocity2149.29 2447.38

Res

istiv

ity16

8.91

190.

407

Permeability189.032 704.706

Vp/V

s1.

4333

71.

4873

clusters

Conclusions

Methodology for predicting saturation from permeabilityand porosity measurements. Note that heterogeneities seen in permeability and porosity produce larger varia-bility in saturation.

Schematic representation of the strain boundary conditionsapplied to the Berea sandstone block in order to calculate the upscaled elastic stiffness tensor.

Initial State

AVO response for upscaled Berea sandstone assuming that CO

2 is a gas and the cap rock is a typical shale.

AVO response for upscaled Berea sandstone assuming that CO

2 is supercritical and the cap rock is a typical shale.