You are on page 1of 8

FEATURE ARTICLE

ROCK PHYSICS FOR THE REST OF US –


AN INFORMAL DISCUSSION
By Jan Dewar - Scott Pickford, a Core Laboratories Company, Calgary

What is Rock Physics? Establishing relationships between seismic expression and


Rock Physics describes a reservoir rock by physical properties physical rock properties therefore requires 1) knowledge about the
such as porosity, rigidity, compressibility; properties that will affect elastic properties of the pore fluid and rock frame, and 2) models
how seismic waves physically travel through the rocks. The Rock for rock-fluid interactions. This is the domain of Rock Physics.
Physicist seeks to establish relations between these material prop-
erties and the observed seismic response, and to develop a predic- Rock Physics / Petrophysics: What’s the difference?
tive theory so that these properties may be detected seismically. Stated very simply, (and therefore with apologies to Rock
Physicists and to Petrophysicists):

Rock Physics... Petrophysics...

Rock Physics uses sonic logs, density logs, and also dipole Petrophysics uses all kinds of logs, core data and production
(shear velocity) logs if available. data; and integrates all pertinent information.

Rock Physics aims to establish P-wave velocity (Vp), S-wave Petrophysics aims at obtaining the physical properties such
velocity (Vs), density, and their relationships to elastic as porosity, saturation and permeability, which are related to
moduli κ (bulk modulus) and µ (rigidity Modulus), porosity, production parameters.
pore fluid, temperature, pressure, etc. for given lithologies
and fluid types.

Rock Physics talks about velocities and elastic parameters, Petrophysics is generally less concerned with seismic, and
because these are what link physical rock properties to seismic more concerned with using wellbore measurements to
expressions. contribute to reservoir description.

Rock Physics may use information provided by the Petrophysics can provide things like porosity, saturation, per-
Petrophysicist, such as shale volume, saturation levels, and meability, net pay, fluid contacts, shale volume, and reservoir
porosity in establishing relations between rock properties or zonation.
in performing fluid substitution analyses.

Rock Physics is the interest of Geophysicists (and maybe Petrophysics is the interest of Petroleum Engineers, Well Log
Physicists). Analysts, Core Analysts, Geologists and Geophysicists.

Why do we need Rock Physics? ■ how sensitive?


■ what increments of gas saturation can be detected
Accurate relations between rock properties and seismic
attributes can to put “flesh on the bones” of a seismic interpretation. seismically?
That is, Rock Physics allows the interpreter to put “rock properties • how robust would AVO seismic response be in situations of
together with seismic horizons.” (Peeters). Information about varying thickness?
porosity, pore-fill, and lithology becomes available to augment the • how robust would AVO seismic response be in situations of
seismic interpretation. varying porosity?
• can an optimal processing strategy be determined for process-
What can Rock Physics studies contribute? ing the seismic data for AVO analysis?
• what seismic characteristics or attributes may be useful to dis-
Typical Rock Physics studies will answer questions such as: tinguish gas-sand from shale?
• can porosity and saturation be obtained from (seismic) inter- • what is the influence of gas saturation on velocities, density
val velocity? and reflectivity in a particular play?
• what are the velocity-porosity relations in various lithologies? • what is the influence of clay content?
• is seismic AVO response sensitive to gas saturation levels in • what is an appropriate mudrock line value for further AVO
a particular play? analysis of a particular play?

Continued on Page 43
42 CSEG Recorder May, 2001
FEATURE ARTICLE Cont’d
ROCK PHYSICS FOR THE REST OF US – AN INFORMAL DISCUSSION
Continued from Page 42

Rock Physics in Practice: some examples


Example 1: Cross-plotting well log data
A typical task in a Rock Physics study is to calculate AVO
attributes from the well logs and cross-plot various attributes from
selected geologic units. These cross-plots are used to:
• understand how the rock properties are related
• determine the resolution of rock properties in various litholo-
gies in the area
• determine the sensitivity of various attributes to fluid effects
• contribute velocity constraints such as mudrock line values to
AVO analysis of seismic data.
• contribute to interpretation of attribute sections: which
attribute(s) are best for describing a given reservoir?

The following example is from a well log from western Canada.


Data points from various depth intervals are cross-plotted in vari-
ous attribute spaces. Note how data from various depth intervals
(lithologies) cluster. Note also how the clusters are separated more
easily in some attribute spaces than in others.

Figure 1: Crossplots of parameters calculated from wireline data. Such


crossplots indicate which attributes will be helpful to discriminate gas
sands in a particular play.

Continued on Page 44

May, 2001 CSEG Recorder 43


FEATURE ARTICLE Cont’d
ROCK PHYSICS FOR THE REST OF US – AN INFORMAL DISCUSSION
Continued from Page 43

Example 2: Crossplotting seismic data Impedance versus S-Impedance) crossplots as compared to the
This example illustrates how crossplots can be used to see which Lambda*Rho versus Mu*Rho crossplots. This set of crossplots illus-
attributes may show clear separation for the zone of interest. In this trates how, at least in this study area, good sand may be distin-
case, the zone of interest is an oil-bearing sand. guished from coal, shale, wet sand, and regional wet sand by
crossplotting AVO attributes or elastic rock properties, and that var-
On the following page are crossplots from AVO Attribute sec- ious attributes are available for crossplotting.
tions derived from pre-stack seismic data. The highlighted data
points are taken from data at well locations A, B, C, and D, over the One could then use this knowledge of crossplot cluster patterns
zone of interest (as interpreted on the AVO Attribute time sections). at known wells to further investigate cluster patterns at various
Note differences in separation of clusters in the Ip versus Is (P- other well locations, and at potential locations beyond.

Example 3: S- wave velocity prediction


Shear wave velocities are required for most AVO analyses.
However, the S-wave velocities are not commonly logged, and
often must be predicted from P-wave sonic logs. This is another
typical task of the Rock Physics study.

In this example from northern Mexico, the P-wave velocity


model was built by calibrating the interval velocities obtained from
NMO analysis of the seismic data with the sonic velocities. The
mud-rock line, estimated from the well’s P- and S- velocities, and
shown in the figure below, was used to scale and shift the P-wave
velocities to obtain the S-wave velocity field.

Figure 3: Vp vs Vs Graph. The S-wave velocity function was obtained


from the mud-rock line estimated from the well’s dipole sonic log. The
green points, which correspond to the gas sand were not included in the
linear fit.

Continued on Page 45
44 CSEG Recorder May, 2001
FEATURE ARTICLE Cont’d
ROCK PHYSICS FOR THE REST OF US – AN INFORMAL DISCUSSION
Continued from Page 44

Example 4: Fluid substitutions There are two main modeling algorithms: Zoeppritz and Elastic
Another practical application of Rock Physics is to perform fluid wave equation. The Zoeppritz equations modeling uses ray tracing
substitutions in well log data, to assess ‘what if’ possibilities. For and approximates the source signal as a plane wave propagating
example, the gas-charged zone of a well log can be edited to simu- through our earth model. The more computationally intensive
late the wet well case. Fluid substitutions should not be considered Elastic Wave Equation modeling provides a more sophisticated
to be trivial matters because fluid type plays an influence on many result by using the full wave equation to propagate the spherical
properties. For instance, one cannot simply lower the P-velocity to wavefront through the depth model. It accounts for peg-leg multi-
substitute gas for brine, the density will also be affected when gas ples, surface multiples, absorption, transmission losses, and con-
is substituted for brine. The main point here is that fluid substitu- verted waves.
tion studies cannot be approached casually.
The synthetic pre-stack gather can then be processed in the same
way as one would process any real pre-stack gathers. This includes
extracting AVO attributes and Lame’s parameters from modeled
synthetic gathers in just the same way as one extracts these from
real pre-stack data. In this way one can predict from the model
those attributes, if any, which will be useful AVO hydrocarbon
indicators for actual seismic data.

As mentioned earlier, one can vary reservoir properties such as


the fluid type, porosity, thickness, to study the resulting seismic
AVO response and investigate possible non-unique physical causes
of the observed seismic response. We can also study the sensitivity
Figure 4: In this example from a carbonate play, the effects of porosity of the seismic to acquisition parameters, signal to noise ratio, band-
and fluid content have been investigated. This particular example, width, and processing parameters.
shows that Lambda*Rho is decreased by gas, and the gas effect is greater
at low porosities. Mu*Rho may be increased or decreased, but is gener-
ally relatively unchanged by the presence of gas.

In addition to fluid substitutions, sensitivity to thickness of a


rock layer may be examined by perturbing the well logs to alter the
thickness, or even to remove the layer. Attributes can be calculated
from the modified wells to simulate seismic response to the various
thickness increments.

Similarly, porosity, saturation, or other physical properties


could be incrementally changed by editing the logs, then calculate
Vp, Vs, density, and elastic moduli relationships to investigate seis-
mic response. This is illustrated in the next example.

Example 5: 1.5D pre-stack seismic modeling


Reservoir rocks have physical material properties. Pre-stack Figure 5: Pre-Stack Synthetic gather. The depth model is a well from
seismic modeling means calculating the synthetic pre-stack seismic western Canada; the well data is shown here in two-way time.
gather that would result from these rock properties. Pre-stack seis-
mic modeling essentially presents similar information as does
Example 6: 2D pre-stack modeling
cross-plotting, but since synthetic seismic wiggle traces are output,
one has the added benefit of being able to view seismic expression Using horizon based interpolation, several well logs can be used
for the modeled earth properties, not just well logs or crossplots. to construct a 2D geologic earth model and generate an interpolat-
ed ‘2D Line’ of synthetic pre-stack gathers. One can stack this ‘line’
Synthetic seismic gathers are generated by calculating how the of modeled pre-stack gathers to study the stack response. One can
wavefield would pass through an earth model. The earth model is also extract pre-stack AVO attributes and assess the relative impor-
first constructed by using sonic, density, and shear wave logs to tance of attributes before investing time and effort in an AVO anal-
form a description of the subsurface depth. ysis of actual field seismic data.

Continued on Page 46

May, 2001 CSEG Recorder 45


FEATURE ARTICLE Cont’d
ROCK PHYSICS FOR THE REST OF US – AN INFORMAL DISCUSSION
Continued from Page 45

Zhijing Wang summarizes the tension between theoretical and


empirical approaches this way:

Most direct measurements are carried out either in the laborato-


ry or inside a borehole, whereas most theoretical calculations are
based on the Gassmann equation (Gassmann, 1951) because of its
simplicity and ease of use. Direct laboratory measurements are car-
ried out uncontrolled, simulated reservoir environments and pro-
vide accurate effects of pore fluids on seismic properties. Direct
borehole measurements, however, are often affected by uncontrol-
lable factors such as stress concentration, hole washout, mud inva-
sion/filtration, and saturation conditions. In both laboratory and
borehole measurements, the wave frequencies are higher than seis-
Figure 6: 2D Modeled (Lambda*Rho - Mu*Rho) Difference Section.
mic frequencies.
The pre-stack modeling is based on well log data and rock physics anal-
yses.
Theoretical calculations such as those using the Gassmann equa-
tion require input parameters that have to be directly measured in
the laboratory. In fact, some theories require input parameters that
Digging Deeper: Where does all this come from?
are often hard to obtain with a reasonable precision. On the other
Two approaches: Theoretical and Empirical hand, some theories, particularly the Gassmann equation, are based
on frequencies comparable to seismic wave frequencies. Therefore,
Equations that attempt to describe the relationships between a direct comparison of laboratory results with theoretical calcula-
seismic velocities and lithology, porosity, pore fluid, etc. are either tions often involves the dispersion problem, where dispersion
theoretical or empirical. Theoretical relationships start with underlying means that seismic velocities are functions of the wave frequency.
physical principles and (Wang, 2000, p 8)
attempt to propose a univer-
sal relationship (at least for The crux of all this is that there are a great number of relation-
the assumptions they ships between seismic velocities (and constituent elastic properties)
may be forced to make). and rock parameters that are all valid to some degree but not valid
One of the amazing always, and many that do not illuminate the physical principles
things about many theo- involved. The trick is to try to gain a fundamental understanding so
retical relationships in that, on a practical basis, the different relationships can be evaluat-
Physics is how well they ed for their applicability to solving specific problems.
can work. But they all
break down at some Gassman equations: an overview
point or in some particu-
Gassman’s (1951) equations provide a way to calculate the bulk
lar circumstances when
modulus of a fluid-saturated porous medium using the known
assumptions, sometimes
bulk moduli of the solid matrix, of the frame, and of the pore fluid.
hidden, are violated.
For a rock, the solid matrix consists of the rock-forming minerals,
the frame refers to the rock sample with empty pores (dry rock),
Empirical relationships
and the pore fluid can be a gas, oil, water, or a mixture. The equa-
are derived from
tions also express that the shear modulus is not affected by fluid
experiment. Physical properties of a suite of rock samples are mea-
saturation.
sured, analysed, graphed, and a mathematical function (often a lin-
ear equation of the form y = mx + b) is fit to the data points.
Assumptions in Gassman’s equations
Sometimes a linear fit is natural (the data points are linear for the
most part, but in some manner fall away from a linear trend, often “The rock-fluid system is so complicated that virtually all the
near the end points). Sometimes measured data points fit to non- theories for such a system have to make major assumptions to sim-
linear equations. Empirical relations most often work very well for plify the mathematics. “ (Wang 2000, pp 9,10)
the data they were derived from, but can be difficult to compare
from one research project to the next. With empirical relations one So what can be said about the uses and limitations of
must be careful about ascribing physical meaning to what are Gassmann’s equation?
essentially generic mathematical formulae. The Gassmann equation works reasonably well for rocks with
interconnected high aspect-ratio pores such as unconsolidated

Continued on Page 47
46 CSEG Recorder May, 2001
FEATURE ARTICLE Cont’d
ROCK PHYSICS FOR THE REST OF US – AN INFORMAL DISCUSSION
Continued from Page 46

The basic assumptions in the Gassmann equation. What does this mean?
(The devil is in the details)
1. The rock (both the matrix and the frame) is This common assumption ensures that the wavelength is long
macroscopically homogeneous and isotropic. compared to the grain and pore sizes. Most rock can generally
meet this assumption for seismic (20-200 Hz) to laboratory fre-
quencies (100 kHz - 1 MHz).
2. All the pores are interconnected or communicating. This implies that porosity and permeability are high. The rea-
son behind this assumption is to ensure full equilibrium of the
pore fluid flow induced by the passing wave can be attained
within the time frame of half a wave period. For seismicwaves,
only unconsolidated sands can approximately meet this
assumption because of the finite wavelength.
3. The pores are filled with a frictionless fluid (liquid, gas, or The viscosity of the saturating fluid is zero. The purpose of this
mixture) assumption again is to ensure full equilibrium of the pore fluid
flow. In reality, because all fluids have finite
viscosities and all waves have finite wavelengths, most
calculations using the Gassmann equation will violate this
assumption.
4. The rock-fluid system under study is closed (undrained). For a lab rock sample, this means that the rock-fluid system is
sealed so that no fluid can flow in or out of the rock’s
surface.
For a reservoir rock, the volume v which is under study must
be part of a much larger volume V, and be located far enough
from the surface of V that the passing seismic wave does not
cause any apprreciable flow through the surface of v.
5. When the rock is excited by a wave, the relative motion This key assumption is the essence of the Gassmann equation. It
between the fluid and the solid rock is negligibly small requires that wavelength be infinity (or the frequency be zero).
compared to the motion of the whole saturated rock itself. It is also perhaps the reason why the measured bulk modulus
or velocity are usually higher than those calculated by the
Gassmann equation. This is because at high frequencies,
relative motion between the solid matrix and pore fluid will
occur so that 6the waves are dispersive.
6. The pore fluid does not interact with the solid in a way that In reality, the pore fluid will interact with the rock’s solid
would soften or harden the frame. matrix to change the surface energy. When a rock is saturated
by a fluid, the fluid may either soften or harden the matrix.

clean sands and sandstones at high effective pressures. (“works In unconsolidated or poorly consolidated sands, the measured
reasonably well” means little difference exists between the shear modulus of the dry frame usually has high uncertainty
Gassmann-calculated and laboratory-measured seismic velocities). because of high shear wave attenuation. A 10% uncertainty in dry
frame shear modulus yields a 2% uncertainty in the Gassmann-cal-
Carbonate rocks have strong elastic frames and very different culated Vp.
pore systems compared to siliciclastic rocks. Because many pores
in carbonate rocks are not well connected, the Gassmann equation Biot
is in general inadequate for carbonate rocks. Gassman’s equations are not adequate to calculate frame mod-
uli in the high frequency range of laboratory data. Biot’s (1956) the-
For rocks with flat pores, cracks, or fractures, and rocks saturat- ory includes the entire frequency range up to the point where the
ed with high-viscosity fluids, Assumption 5 cannot be satisfied, so grain scattering becomes important and the rocks can no longer be
the Gassmann-calculated Vp is always less than the measured Vp. considered homogeneous. Gassmann’s equations are the low fre-
This is because such pore shapes do not allow room for the fluid to quency limit of Biot’s more general relationships. Note that for the
equilibriate in the half-wavelength time period required by the case of perfect coupling, Biot’s equations reduce to the zero-fre-
equation. quency case (Gassmann’s Assumption #5 is met).

At high pressures, the flat pores, cracks, or fractures are closed,


so then the Gassmann-calculated velocities agree better with the Continued on Page 48
measured values.

May, 2001 CSEG Recorder 47


FEATURE ARTICLE Cont’d
ROCK PHYSICS FOR THE REST OF US – AN INFORMAL DISCUSSION
Continued from Page 47

Further Refinements earth model with geological, petrophysical, and geophysical data in
Much of the ongoing research work in Rock Physics is to all grid-blocks will be available soon. This ‘unified model’ would
approach physical reality more closely. For example, the Biot equa- have the resolution of cores near the well bore; of logs in most other
tions say that the fluid must participate in the solid’s motion by vis- places; and will be used for both static and dynamic modeling.”
cous friction and inertial coupling, but it is known that fluid also (Peeters)
squirts out of pores when deformed by a passing seismic wave.
Traditionally, the Biot mechanism has been treated macroscopically, The challenge of calibration
and the squirt-flow mechanism at the individual pore level. Work Seismic sections that show rock and fluid properties will be
by Dvorak and Nur (1992) offers a model which treats both mecha- more meaningful if they can be calibrated with forward modeling
nisms as coupled processes and relates Vp and attenuation to from accurate rock properties measured at seismic frequencies.
macroscopic parameters: the Biot poroelastic constants, porosity, There is a huge difference between a 50 Hz seismic wavlet (approx-
permeability, fluid compressibility and viscosity, and a new micro- imately 40 m resolution) and a 10 kHz sonic wireline tool (20 cm).
scopic-scale parameter - a fundamental and measurable character- Which high frequency results can be extrapolated to the seismic
istic squirt-flow length. Such local flow models are representative of realm?
current work to extend Gassmann’s description to include more
realistic portrayals of rocks, and to relative poroelastic behavior to The lack of measured shear velocities
macroscopic measurable parameters such as permeability, porosity, Dipole logs are not commonly performed, and estimates of Vs
saturation, pore-fluid compressibility, density, and viscosity). carry uncertainty, particularly in unconsolidated sediments.
Please see ‘For Further Reading’ for some excellent papers in this
regard. The promise of the prize

Issues in Rock Physics Just a few of the problems facing the geoscientist are mentioned
above.There are plenty of issues and shortcomings with Rock
The challenge of scale Physics, as with any discipline handling complex physical process-
The scales at which Geophysics and Petrophysics work are very es. Nonetheless, the Rock Physics effort is progressing in a valid
different. Logs and cores give resolution less than 0.3 metres, while direction. The achievement of rock properties from seismic data,
seismic resolution is often no better than 15 metres. This may be integrated with wireline, petrophysical, and geologic knowledge is
expressed in terms of the frequency ranges used: a goal that is becoming more attainable. Rock physics draws togeth-
er the disciplines of Geophysics and Petrophysics, bringing the
• The range of seismic frequencies is typically considered to be
possibility of a unified 3-dimensional earth model within reach. R
20 - 200 Hz (and more realistically 10 - 80 Hz)

• Logging frequencies are around 10 kHz

• Laboratory frequencies have traditionally been 100 kHz - 1


MHz. Also note that in the lab, it has been nearly impossible Biography
to carry out wave propagation measurements at low (seismic Jan Dewar graduated from the
realm) frequencies as the sample length is required to be at University of Alberta in 1981 with a
least half a wavelength. For a rock with 4000m/s Vp, a 50Hz B.Sc. in Physics. Jan is currently work-
wave would require a 40m sample. However, recent promis- ing with Scott Pickford in Calgary,
ing advances in Rock Mechanics have succeeded in velocity with a special enthusiasm for commu-
measurements in the lab at the 100Hz range, which is very nicating technical concepts including
promising indeed (Lewis Lacy, Director of Geomechanics, AVO, Inversion, Modeling, VSP and
CoreLab Rock Mechanics Lab, personal communication). Transfer Filter processing, Rock
Physics, and just about anything else
The challenge of focus that can be puzzling to the average bear.
Advances in computing power mean that efforts are turning jdewar@scopica.com
from the traditional seismic task of getting an accurate image of
subsurface structures to extracting more and more rock property
information from the seismic wavelet. Computing power also ben-
efits Reservoir Modeling, and as more complex dynamic simula-
tions become possible, real integration of seismic and reservoir
characterization gets closer. Establishing accurate relations
between rock properties and acoustic parameters is becoming more
imperative. “If the trend in computering power continues, a 3-D

Continued on Page 49
48 CSEG Recorder May, 2001
FEATURE ARTICLE Cont’d
ROCK PHYSICS FOR THE REST OF US – AN INFORMAL DISCUSSION
Continued from Page 48

Acknowledgments http://www.geophysics.mines.edu/max/Resmod.html
I am indebted to my colleagues Yongyi Li, Alvaro Chaveste,
and Michael Burianyk for their assistance. Wang, Z., 2000, The Gassmann equation revisited: Comparing labo-
ratory data with Gassmann’s predictions, Seismic and Acoustic
Velocities in Reser voir Rocks, Vol. 3, Recent Developments, SEG
Reprint Series, pp 1-23.
References & Further Reading
Biot, 1956, Theory of propagation of elastic waves in a fluid saturated
porous solid, 1. Low frequency range, J. Acoust. Soc. Am., 28, 179-
191.

Castagna, J., 1993, AVO Analysis - Tutorial


and Review, Chapter one of Offset-dependent
reflectivity - theory and practice of AVO analy-
sis, Investigations in Geophysics No. 8, SEG.

Castagna, J.,Han, D., & Batzle, M.L., 1995,


4D/4C Processing
Issues in rock physics and implications for DHI
interpretation, The Leading Edge, August Whether you are looking for the natural fractures or variations
1995. of fluid properties from your producing fields,
CGG’s 4D/4C Processing is the answer.
Dvorkin, J., & Nur,A., 1993, Dynamic poroe-
lasticity: A unified model with the squirt and the
Radial / transverse energy
Biot mechanisms, Geophysics 58, 524-533. ratio time slice
-57100.0

Dvorkin, J., Nolen_Hoeksems, R. & Nur, A.,


1994, The squirt-flow mechanism: Macroscopic Determine the natural -57200.0

description, Geophysics, 59, 428-438. fracture orientation from


the shear wave analysis -57300.0
284°
of 4C seismic data. In
Gassmann,F., 1951, Elastic waves through a this example, S1 waves -57400.0 104°
packing of spheres: Geophysics, 16, 673-685. indicate a primary
fracture orientation of -57500.0
Fast S1 axis
104 degrees from North.
direction
Kelder, O., & Smeulders, D.M.J., 1997,
570000.0 570200.0 570400.0 confirmed
Observation of the Biot slow compressional wave on panoramic
in water-saturated Nivelsteiner sandstone, receiver stack.
Geophysics, 62, p 1794.
Time-Lapse multi-component seismic analysis provides a direct image of
Peeters, Ir. M., Physical Reservoir Models: the evolution of the fluid properties from the producing reservoir.
From Pictures to Properties, an address pre-
sented at the commencement of the Baker
Hughes Distinguished Chair of
Petrophysics and Borehole Geophysics,
Colorado School of Mines,
Initial Survey Time-Lapse Survey
Increase in amplitude in the Time-Lapse data
indicates gas coming out of solution. Also note
a separate compartment which appears to be
connected to the main reservoir. Amplitude difference map
over the reservoir interval

Processing and Reservoir Services

CGG Canada Services Ltd. #700 404-6th Avenue S.W., Calgary, Alberta Phone 266-1011 www.cgg.com

May, 2001 CSEG Recorder 49

You might also like