SPWLA 55th Annual Logging Symposium, May 18-22, 2014

PETROPHYSICS FOR ROCK PHYSICS: WHAT REALLY MATTERS AT

SEISMIC SCALE?
Jonathan Hall1 and Erick Alvarez 2,
1
Abu Dhabi Company for Onshore Oil Operations, U.A.E. 2Senergy (G.B.) Limited (now at Shell U.K.)

Copyright 2014, held jointly by the Society of Petrophysicists and Well Log to use the seismic to spatially predict such rock
Analysts (SPWLA) and the submitting authors.
This paper was prepared for presentation at the SPWLA 55th Annual Logging property changes.
Symposium held in Abu Dhabi, United Arab Emirates, May 18-22, 2014.
In this paper, we investigate the impact the petrophysics
ABSTRACT input has into the results of common rock physics
applications such as shear wave velocity prediction,
Rock Physics analysis is a key stage in seismic based fluid substitution, stress characterization and
reservoir characterization studies, which attempt to anisotropy. These we demonstrate through case studies,
determine the most plausible 3D distribution of the both in siliciclastic and carbonate reservoirs.
reservoir and the fluids they contain. Rock physics uses
the results of a petrophysical evaluation as input, We also show that, for property up-scaling, the
together with the results of laboratory measurements, common practice of smoothing and blocking does not
where available, and combines these using both respect the physics of wave propagation and therefore
theoretical and empirical relations to understand how can lead to significant errors in the interpretation of
the seismic signal responds to changes in the reservoir rock properties at seismic scale. Even though the errors
properties. The input petrophysical properties are often seem small if visually inspected, we demonstrate that
used as if they were implicit and not derived properties these produce a large cumulative error that produces the
and are rarely questioned or reviewed. In reality, these incorrect seismic response. We propose a slight tweak
properties are estimated empirically or semi-empirically in the implementation of Backus average (Backus,
and, therefore, carry intrinsic uncertainty. A common 1962) that allows resolving this problem. We conclude
misunderstanding, among Geophysicists, is that the that the minor and major solid and fluid fractions
calculated in petrophysics, even though they are below CC
small scale changes in mineral and fluid fractions have
little or no impact at seismic scale. As a result they the seismic tuning thickness, can still cause major
often accept a petrophysical evaluation that is over- effects in our ability to model and hence interpret
simplified for use in seismic applications. Furthermore, subsurface seismic data. This applies, not only in the
when a mismatch is found between the observed data reservoir section but also to the underburden and
and the Rock Physics predictions, the usual conclusion overburden.
is that this is due to uncertainties in the theories or the
seismic data rather than a problem with the input PETROPHYSICS BEYOND PETROPHYSICAL
petrophysical evaluation. VOLUMES

INTRODUCTION The traditional deliverables from petrophysicists have

been porosity, saturations, nett reservoir thickness and
Rock physics establishes a relationship between the pay intervals. This ‘one dimensional’ view of
seismic signal and the petrophysical properties in the petrophysics is understandable as the petrophysicist’s
subsurface. This connection allows us to understand work is largely borehole centric and boreholes are, after
and model the petrophysical and geometrical properties all, one dimensional. All but one of the terms that go
which give rise to the seismic signal. We then use this into the hydrocarbon pore volume equation for
to “extract” information from the seismic data about the volumetric determination, if we include the shrinkage
character of the reservoir in three dimensions to map factor, are petrophysics derived terms at the wellbore
extensions to existing reservoirs; locate unaccessed pay although structural and property mapping extend these
away from well control; resolve seal integrity issues; to three dimensions.
and guide optimum placement of wells in complex
reservoirs. Rock physics requires a knowledge and Specialist petrophysical skills have developed around
understanding of geophysics, petrophysics, tools with data array measurements. This includes
geomechanics, and the causes of the distribution of borehole geology from dipmeter and image logs, pore
fluids in the subsurface. By modeling we build an geometry evaluation from nuclear magnetic resonance
understanding of the sensitivity of seismic to these logging and full waveform acoustic logging for fracture
various effects and we assess whether it will be possible and stress studies. The tools and the skills required to
process and interpret them lead to a fuller description of

1
SPWLA 55th Annual Logging Symposium, May 18-22, 2014

the bedding geometry, pore geometry, fracturing and to petrophysicists. The difference in spatial coverage of
geo-mechanical properties of the reservoir, again at the quantitative seismic applications, compared to wellbore
well bore. The scale of the features and fabrics detected centered traditional petrophysics requires of us that
with these technologies extend outwards from the where multiple wells are to be incorporated into seismic
wellbore into the reservoir at gigascopic bedding and reservoir characterization, we should ensure that the
fracture network modeling scale to microscopic pore interpretations are consistent with the vertical and
size distribution scale. lateral variability in the reservoir.

Some rock physics models are entirely empirical and One way to do this is to conduct this interpretation by
locally calibrated so rock physicists still only required linear (and non-linear) inversion of the logs with a
scalar compressional and shear wave velocities plus matrix of tool response equations; suitably
bulk density. These may be obtained either from parameterized, and selected mineral and fluids
measured values or from the results of invested effort in assumed, or known, to be present. These solver
predicting these from empirical or theoretic models. We techniques, sometimes, perhaps inaccurately, referred to
will show that, whilst some models for rock physics as “stochastic” techniques, are well known and widely
shear wave estimation and fluid substitution do not available in today’s software offerings. They ensure
require multi-scale, multidimensional, data types, consistency in evaluation because, if the geology is
others, require rock fabric parameters such as bedding laterally continuous, a single model with a common set
anisotropy, pore shape and stress anisotropy by virtue of log inputs will capture real changes in mineralogy,
of the variability of the compliance of the variance pore fluid type and volume and porosity from well to well.
types present. In some oil producing provinces, such as in the
European North Sea, significant unconformities and
Further, an understanding of trends such as burial tectonic movement leads to significant vertical
compaction trends, sorting, porosity at the onset of rock variations in sedimentary reservoir rock type and cover
consolidation, pore shape and grain contact geometries and lateral dislocations in target intervals. Again
are sometimes explicitly required for model judicious use of solver techniques will ensure
parameterization. The parameterization of these models representative and consistent results required for rock
represents a technical challenge and career opportunity physics modeling. CC

Fig. 1. Shows an example from the Onshore Northern Mexico a) Linear inversion model reconstruction. b) complex
mineralogy and stacked thin bedded porous/cemented zones; each well below seismic tuning thickness, can be
evaluated with log inversion and still used in rock physics modeling and upscaling to seismic scale. c) stacked
cemented and oil charged silts below seismic resolution.

2
 SPWLA 55th Annual Logging Symposium, May 18-22, 2014

PETROPHYSICS FOR ROCK PHYSICS IN using Tozaya’s equation (Tosaya, 1982), with the
SILICICLASTIC RESERVOIRS coefficients shown in equation 1, showed that these
differences are not negligible and have an impact in the
Sequential deterministic workflows for siliciclastic Vp/Vs ratio and hence on the AVO behavior (Figure 3
reservoirs, and those traditionally characterized by; and 4)
shale volume determination, followed by calculation of
effective porosity and then water saturation, are the Vs = 3.7 − 7φt − 2Vclay
common deliverables picked up by the rock physicist as (1)
inputs for modeling. These may be inadequate or even
inappropriate descriptions of the rock for this purpose. Vs is the secondary (shear) wave velocity, and the other
Firstly, shale is a rock fabric description comprising of two terms represent total porosity and clay volume
a variable but significant fraction of clay minerals with fraction.
variable chemically bound ‘surface’ water. As Truman
et al. (1989) note, the average clay content in most
shales is between 30% and 55% bulk volume. Since
shales are not all clay, they can contain significant
amounts of dynamically effective porosity or at least
storage capacity by the petrophysicist's
definition. Further, shale volume is often computed
with the gamma ray alone or as the minimum of
multiple indicators. Whilst this may be adequate to
distinguish cap rock and under-burden from non
reservoir, it may not offer accurate estimations of the
volume of the clay fraction, nor does it distinguish
between clay distributed as structural grains, pore lining
fill or in discrete laminations which is important for
rock physics modeling. A more complete description of CC
shaly siliciclastic reservoirs would be to estimate all
mineral volumes, calculate total porosity and what
proportion of this is chemically bound (and capillary
bound as will be illustrated below) and to describe the
distribution of the clay fraction potential seismic
anisotropy that might result.

SHEAR WAVE VELOCITY PREDICTION !

Simple comparison between rock physics modeling and Fig. 2. Track 3 and 4 shows the reconstruction of the
subsequent AVO fluids characterization using results of shear wave velocity using the linear inversion lo
deterministic and linear inversion petrophysics analysis results as input to Rock Physics modeling.
workflows. To demonstrate the importance of the Visually the differences seem negligible but the in the
petrophysical evaluation in rock physics applications, modeled seismic signal they make a huge difference.
an exercise was performed in a gas well located in a Also it is better to QC shear wave reconstruction using
deltaic front depositional setting, containing significant crossplots than depth plots.
amounts of clay minerals and calcite cement, figure 2.
Since the presence of these minerals is way beyond Figures 3a and 3b provide a comparison between the
seismic resolution, it was generally considered that crossplots of the acquired compressional and shear
these effects would not have any impact in the seismic wave velocities and those modeled using the Tosaya
data. Two petrophysical evaluations were performed, equation above with either the results of ‘deterministic
first using sequential deterministic petrophysics or linear inversion petrophysical workflows. Figure 3c
modeling and linear inversion, ‘stochastic’ modeling shows the actual seismic gather for this West African
for rock physics application using a simple empirical data set plus the synthetic amplitude versus offset
relationship. The results of these evaluations are shown produced from the acoustic log dataset, and that from
in Figure 2, the differences between the two evaluations the application of the Tosaya formulation using as
is too small, even for the human eye to distinguish. inputs the results of the deterministic and linear solver
However, when performing a shear wave reconstruction workflows.

3
SPWLA 55th Annual Logging Symposium, May 18-22, 2014

Fig. 3a. Sequential deterministic petrophysics workflow. Comparison between shear (secondary) and compressional
(primary) wave velocity cross plot from measured acoustic logs (left) and the results of applying the calculation of
porosity and clay volume fraction and the Castagna (1995), Tosaya (1982) relationship above (right).

CC

Fig. 3b. Linear inversion (stochastic) petrophysics workflow. Comparison between shear (secondary) and
compressional (primary) wave velocity cross plot from measured acoustic (left) logs and results of applying the
calculation of porosity and clay volume fraction and the Castagna (1995), Tosaya (1982) relationship above (right).

Fig. 3c. The modeled seismic responses, purple lines using the linear solver results reproduces better the AVO
response than that from the deterministic petrophysics The dark blue line is the output seismic gather reflectivity
versus angle of incidence and the light blue dotted line represents the results from the two term Shuey
approximation (1985) to Zoeppritz equations (1919) comprise the AVO characterization workflow.

4
 SPWLA 55th Annual Logging Symposium, May 18-22, 2014

Extension of the test to commonly applied rock physics did not perform well in either case for this example.
models. To validate this surprising result, additional These results have the following implications:
shear wave predictions were performed using the most
common published methods, including Krieff (1990), 1) The petrophysics input has as much impact on the
Xu and White (1994, 1995), Greenberg and Castagna seismic modeling as the rock physics model used,
(1993) and Hall and Alvarez (2010). Formulations of 2) The linear solver model seems to be more appropriate
these models are in Appendix I. In these examples, the for generation of inputs to petrophysics for rock
best results obtained were, firstly, using the linear physics applications,
solver model petrophysical results rather than the 3) Depth plots should not be used to evaluate the results
deterministic ones and, secondly, using the Hall and of shear wave reconstruction as this can be
Alvarez model (2010) over the other published models. misleading,
Surprisingly, for the authors, the Xu and White method 4) Crossplots and AVO modeling are better tools to
evaluate the accuracy of the rock physics results.

CC

Fig. 4a. The AVO gradient analysis for various models using shear wave velocity, Vs, predicted from the results of
the deterministic petrophysics work flow.

Fig. 4b. The AVO gradient analysis for various models using shear wave velocity, Vs, predicted from the results of
the linear solver petrophysics workflow.
5
SPWLA 55th Annual Logging Symposium, May 18-22, 2014

Fig. 5. µρ versus λρ crossplots for sequential deterministic and linear solver inputs to rock physics modeling. Water
saturation on the Z-axis coded; warm colors low saturations to cold colors high water saturation.

FLUIDS IDENTIFICATION The reason for the differences are small-scale changes
in the way porosity, clay content and clay bound water
Using Lambda,λ Mu,µ, Rho, ρ, fluid identification. are calculated during the optimized process, leading to
Another exercise was performed, this time in a North small but important differences in the calculation of CC
Sea well located in a turbidite reservoir with relatively water saturation. From a physical perspective, the
clean sands. A legacy petrophysical evaluation with optimized petrophysics makes more sense, as λρ should
only shale fraction, effective porosity and water increases with increasing water saturation, whereas the
saturation was available. As part of our exercise, a legacy petrophysics misses part of the rocks with pay.
linear solver was used to calculate dry clay, quartz, clay This is of utter importance as this type of analysis is
bound water, total and effective porosity. Good quality part of AVO feasibility studies. If we had used the
measured compressional and shear sonic logs exist as legacy petrophysics, we would have classified AVO as
well as density. unfeasible (as ambiguity exists in λρ), however, the
optimized petrophysics shows that AVO would be
Using these logs, Lame impedances were calculated feasible to characterize the field.
(λρ and µρ) to test if the hydrocarbon saturated sands
could be separated in crossplot space. The results are Identify trends in your data: In Tertiary siliciclastic
shown in Figure 5. The X axis is λρ (which is basins where sedimentation continues today as it has for
proportional to the incompressibility and the Y axis tens of millions of years, such as the West Siberian
is µρ is proportional to the resistance to deformation In onshore oil province, a better description of clay
this domain we seek for the more compressible rocks fraction volume can be achieved using density-neutron
(low values of  λρ),  which can indicate the presence of log combination. Further, using these logs and
hydrocarbons, whereas   µρ   is almost independent of modeling the compaction trend implied it is possible to
fluid content. The left hand side crossplot is from the determine the sorting coefficient, or a proxy for this,
legacy petrophysics (deterministic) and the right hand which is important in some rock physics effective
side plot is with the new water saturation log from the medium models.
linear solver petrophysics.
Figure 6 illustrates such a modeling exercise. 6 wells
The areas inside the red circles in the legacy were evaluated specifically ensuring consistency in the
petrophysics plot appear as water saturated, whereas in petrophysical modeling workflow so that trends in
the optimized petrophysics they are included as low petrophysical properties,; porosity, clay/sand fraction
hydrocarbon saturated sands. and saturation could be compared using the Lame
impedance crossplots.

6
 SPWLA 55th Annual Logging Symposium, May 18-22, 2014

Fig. 6 The value of ensuring consistency. These crossplot demonstrate the results of µρ versus λρ modeling of a
West Siberian field from Alvarez et al 2009. Six wells were evaluated. Multi-well crossplots for two and for six
wells are shown. Z axis warm colors are sand and cold colors represent shale in a sand silt shale oil province. In this
case clay fraction was determined from density-neutron crossplot alone incorporating a compaction trend for the wet
clay points and subsequent multi-mineral and fluids evaluation used a linear solver model.

FLUID SUBSTITUTION IN THE PRESENCE OF effective medium theories are commonly used to
STRESS VARIATIONS: 4D APPLICATIONS calculate the effective elastic moduli of the solid as well
as the effective bulk modulus of the fluid components.
CC
Petro-elastic modeling (PEM) is a key element in 4D This means that we are in a similar situation as with the
seismic analysis. The correct interpretation of 4D shear wave reconstruction, where the results depend
seismic data depends completely on the appropriate completely on the input mineral and fluid volumes (i.e.
calibration of the rock and fluid physics parameters. petrophysics). Amini and Alvarez (2014) showed that
Fluid-substitution and stress-sensitivity analysis are the errors in the petrophysics input can cause uncertainties
main elements of PEM and although theoretical models in the rock frame characterization or rock stress
exist, these are mostly applicable to sand/shale systems sensitivity (which defines the balance between the
and their application in the presence of other effects of pressure and fluid saturation signals) and this
mineralogy depends completely on the quality of the can cause over-estimations of water flooding signals of
petrophysical interpretation. The reason for this is that up to 55% and up to 30% for gas breakout.
the Gassmann (1956) model assumes one solid and one
fluid component; therefore,
Softening Hardening Softening Hardening
Pressure  build-­‐up   Pressure  depletion Pressure  build-­‐up   Pressure  depletion
(GPa)

(GPa)

increase  in  
increase  in  

increase  in  
initial  effective  stress   )
increase  in   initial  effective  stress   )

(MPa) (MPa)
Fig. 7. Bulk and Shear Modulus variations with effective stress. After Amini and Alvarez, (2014) and Alvarez and
Macbeth, (2013).

7
SPWLA 55th Annual Logging Symposium, May 18-22, 2014

Impact of stress on 4D saturation monitoring. A further This is extremely difficult to characterize for rock
consequence of this observation can be illustrated in physics applications and, in consequence, models that
Figure 7 above. ADCO is planning a number of time explicitly describe pore shape and pore compliance
lapse micro-seismic and time lapse 3D VSP surveys in have been difficult to parameterize (Baechle, 2014).
support of EOR pilots and in preparation for these, the Baechle acknowledges that in many experimental and
switching from peripheral water injection to pattern numerical studies, aspect ratios are, a priori, assigned.
drive. Preparatory studies have concentrated on The current work demonstrates potential
systematic measurement of pressures at both injector parameterization of models that capture the pore shape
and producer locations. In some cases it was found that and compliance variability that results.
the pressure around the injectors neither equilibrated
with the pressure sumps near producers and in some Anselmetti and Eberli (1993), Eberli et al. (2003)
cases attained pressures higher than the initial pre- describe a relationship between pore type and velocity,
production pressure. This work suggests that attempts at where samples containing moldic and intraparticle
seismic saturation change monitoring might be masked porosity have a higher velocity than samples containing
by stress effects at lower effective stresses. micromoldic porosity and microporosity. Microporosity
is important in most Middle East reservoirs.
PETROPHYSICS FOR ROCK PHYSICS IN
CARBONATE RESERVOIRS - ANISOTROPY For Carbonates the pore shape and connectivity is a
property strongly affecting elastic properties of the rock
Elastic moduli of non-fractured carbonate rocks are (Kuster and Toksöz, 1974, Xu and Payne, 2009). In
affected by dry and saturated frame moduli and the pore contrast granular siliciclastic rocks are often adequately
shape and their connectivity. Carbonate reservoirs characterized through complete mineral and fluid
exhibit seismic anisotropy and all scales from aligned description, critical porosity, sorting and cementation
microcracks to fractures and faults and even in the caused by grain-grain contact (compaction) or void
absence of dislocations may exhibit shear anisotropy filling cements (Hall and Alvarez, 2010).
arising from the far field stress field.
To investigate this in 2011 ADCO contracted with a
We illustrate this with two examples, one from a CC
provider of DRP technology services to both generate
Middle East giant Barremian age reservoir where x-ray micro-tomographic images and compute DRP-
observed microfractures in SEM images are based rock physical properties for several core plug and
characterized using pore shape extraction of micro-CT micro plug samples from an early Cretaceous-age
scan images. The resulting histogram of pore aspect supergiant carbonate reservoir in the Middle East.
ratios was used to parameterize the Xu and Payne
(2009) Differential Effective Medium, D.E.M. and Investigating an alternative to finite element modeling:
results compared to Voigt and to Hashin-Shtrikman pore feature extraction. At the same time ADCO
Bounds. The pore space characterized suggests the undertook work with one of its international
presence of compliant non-equant pore cracks. shareholders, Total, an examination of image voxel
segmentation methods and found these wanting
Pore shape variations and the anisotropy that arises is (Chacko et al, 2012). To better understand the
the result of the sedimentary environment and complexity of the pore framework and network in the
allochems, burial compactions and early and late stage study carbonate samples relative to the simplifying
diagenesis. Foucks et al., (2013) offer an explanation assumptions in the Gassmann equation; viz.
for the origin of these microstructures, including the homogenous grain and pore distribution with
diagenetic development of micro-rhombic calcite, from connectivity of all pores, the authors extracted from
low-magnesium carbonate muds which are so pervasive MCT images pore attributes that influence flow,
in the producing reservoirs of the Middle East and are petrophysical, and elastic properties. These include pore
the cause of low resistivity contrast and low resistivity shape, size, orientation, throat size distribution,
pay. These textures are responsible also or stress connectivity, tortuosity, and relative volumes of
anisotropy due to the compliance of the pore space connected and isolated pores. These pore attributes
between the crystals. This is illustrated in the high- enable the use of effective medium rock physics
resolution scanning electron microscope images below. models.

8
 SPWLA 55th Annual Logging Symposium, May 18-22, 2014

Modeled  pore  
geometry

Original  pore  
geometry Sphere-­‐based  pore  
geometry  modeling

Division  of  model  into  discrete  

Geometry  restoration pore  bodies
Extraction  of  pore  
body  attributes

Fig. 8. The pore aspect ratios for each Micro CT samples were based upon a pore-matrix improved segmentation
and an assessment of the pore morphology.

Heterogeneity in pore shapes is not explicitly expressed need for advanced rock physics models to predict
in many commonly used rock physics models that elastic properties. CC
predict elastic moduli. The Gassmann Equation for
example assumes homogeneous grain and pore Pore Volume as Function of Aspect ratio
distribution, and connectivity of all pores. Our pore
attribute extraction quantified the complexity of 0.08

carbonate pore systems in terms of pore size, aspect 0.07

0.06
ratio, orientation, and connectivity. To accommodate
Pore Volume

0.05
this complexity, we evaluated the Xu-Payne 0.04
Pore Volume per Aspect
Differential Effective Medium (DEM) rock physics 0.03
ratio

model (Xu and Payne 2009) developed for carbonate 0.02

rocks. This is based on the Kuster and Toksöz inclusion 0.01

model (Kuster and Toksöz, 1974) that assumes dilute 0

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
pore concentrations of pore inclusions of varying aspect
Aspect ratio
ratio.

The authors calculated effective dry rock medium Fig. 9. Distribution of relative pore volumes as a
moduli in an iterative manner by adding the volume of function of their aspect ratios. Flat pores are seen to
pores with a given aspect ratio during each iteration. predominate in this micro-plug sample.
The pore aspect ratios for each Micro-CT samples were
based upon a pore-matrix improved segmentation and These observations of heterogeneity in pore aspect ratio
feature extraction, Figure 9 and 11. Figure 9 shows the and pore connectivity have implications for rock
distribution of pore aspect ratios from one micro-plug property modeling. Advanced rock physics models that
sample. The majority of pores are seen to have ‘flat’ or take this heterogeneity into account are required for
compliant, rather than equant or stiff geometry, which modeling reservoir static and dynamic properties.
reduces the bulk moduli of rocks in which they Micropores for example are likely to contain patchy
predominate. Several pore attributes extracted from our saturation under reservoir conditions, which would
segmentation process: pore size, aspect ratio, require an appropriate fluid mixing law when
orientation, and connectedness, indicate the complexity computing elastic moduli for hydrocarbon saturated
yet ordered nature of carbonate pore systems and the samples. Figure 10 illustrates pervasive microporosity.

9
SPWLA 55th Annual Logging Symposium, May 18-22, 2014

CC

Fig. 10 (a). Details of the micropore structure were also Fig 10 (b). The same three samples. Black bar is now
investigated with scanning electron microscope images just 1 micro-meter. At core plug scale these samples
from samples from a Barremian reservoir in the United now appear homogeneous but the micro-rhombic
Arab Emirates (above). Black bar equals 100 micro- calcite crystal contacts, in fact, define an anisotropic
meter. At core plug scale these samples appear medium for acoustic and other transport media. If the
heterogeneous. Images courtesy of Total Oil, Abu process of micro-rhombic calcite crystallization
Dhabi/Pau. continues the pore spaces and pore size reduces.

10
 SPWLA 55th Annual Logging Symposium, May 18-22, 2014

computed ones, the author’s D.E.M. computation from

extracted pore shapes; both from micro-CT images, and
laboratory measured values from the same samples, the
authors found that the measured values showed lower
P-wave velocities than those computed from micro-CT
images by either technique. This was as a consequence
of the lack of resolution available in the images at pore
scale and the inability to segment micro-pores at this
scale. Although promising, there is still a long way to
go in the capture of relevant micro-CT images for rock
physics modeling and other transport property
investigations. There is a need for better focusing and
wide-angle image capture at plug scale and above.

STRESS ANISOTROPY AND SHEAR WAVE

Fig. 11. 3-D segmented model pores. SPLITTING

These observations of heterogeneity in pore aspect ratio Rotation of the Ultrasonic imager throughout the
and pore connectivity have implications for rock Middle and Upper Kurrachine dolomite in the AS-109
property modeling. Advanced rock physics models that well Ash Shaer, Syria offered a surprising opportunity
take this heterogeneity into account are required for to use inline dipole shear slowness in a rotating tool
modeling reservoir static and dynamic properties. string to characterize stress anisotropy (Nur 1971, Hall
Micro-pores for example are likely to contain patchy et al. 2010).
saturation under reservoir conditions, which would
require an appropriate fluid mixing law when
computing elastic moduli for hydrocarbon saturated
samples. CC

Results of D.E.M. computations for bulk modulus are

shown in Figure 12. Included in the plot are commonly
referenced theoretical elastic bounds: Hashin-Shtrikman
and Voigt (Mavko et al, 2003), for a calcite matrix Dry
Frame. The DRP sample values are seen to be on trend
with, but having smaller than the bound values,
reflecting the predominantly flatter pore aspect ratios.

Bulk modulus: Xu-Payne vs Theoretical Bounds

90
80
Dry Frame (Calcite) Bulk

70
H-S Upper Bound
60
Modulus

50 Voight Bound
40
30 Xu-Payne (DRP
samples)
20
10
0
0

0.4

0.8
0.08
0.16
0.24
0.32

0.48
0.56
0.64
0.72

0.88
0.96

Porosity

Fig. 12. Results of Xu-Payne DEM computations of

bulk modulus for 7 DRP samples. The data are seen to
be on trend, but with smaller than the bound values,
reflecting flatter pore aspect ratios.

Whilst the Chacko et al. (2012) results are encouraging,

when the authors compared P-wave velocity versus Fig. 13. Ultrasonic imager tool rotation in AS-109
porosity; the service company digital rock physics Upper Kurrachine UKD unit.

11
SPWLA 55th Annual Logging Symposium, May 18-22, 2014

With the tool rotating the authors were able to be The actual XX and YY dipole difference was calculated
assured that any observed acoustic anisotropy was and compared with the model. The results were found
probably the result of far field stress anisotropy or the to be offset by about 90 degrees: a best fit was found to
presence of sub visible scale microfracturing and not a coincide with a modeled σHmax of 105 degrees (Figure
borehole or tool geometry artifact. 14 and 15).

Figure 13 shows the tool rotation (track 6) is continuous

and periodic throughout logging of the Middle and
Upper Kurrachine. The tool is rotating smoothly
throughout the UKD formation at between 32 and 17
metres per rotation except between 1420-1450 metres.

The tool was repeatedly rotating through the stress field

many times over the logged open hole section, so we
assume that where shear anisotropy is developed
significantly, the XX and YY inline dipole shear should
record either the fast and show shear velocity whenever
the tool’s signal polarisation for each dipole is aligned
with σHmax/σhmin.

A value for the anisotropy magnitude should be

approximated by the difference in computed shear
slowness log at these points and the tool orientation
should provide a crude delimit of the fast and slow
shear orientation over the open-hole section (Figure
14).
CC

Fig. 15 Inline dipole shear slowness in a rotating tool

string. Modelled dipole inline shear XX/YY differences
plotted against tool azimuth.

Synthetic results curves (theoretical XX and YY

curves) were forward modelled using nominal fast and
slow end point values were proportionally assigned to
XX_syn / YY_syn as a function of the tool azimuth
(P1AZ_MAST). The difference between the theoretical
curves was then plotted as a measure of theoretical
anisotropy. The initial fast shear orientation was
modelled as predicted from the UBI geomechanical
data, DTSH fast = 20 degrees.
The observed anisotropy (DIFF_DTS5_DTS6 track 3
Figure 15) is plotted against the modelled anisotropy
(MOD_DIFF) and reaches a maximum in the UKD3
with values of 3-10 µs/ft) and generally decreases
upwards.
Shear anisotropy in Ash Shear 109 is observed and is
Fig. 14. Observed dipole inline shear XX/YY
believed to associate with the regional stress field of the
differences plotted against tool azimuth. Z-axis is also
Palmyra trough and not with observable fluid filled
azimuth.
fractures.

12
 SPWLA 55th Annual Logging Symposium, May 18-22, 2014

ABOUT SEISMIC RESOLUTION AND This tuning thickness is considered the limit of
DETECTABILITY resolution in seismic data, this implies that if we wish
to follow (pick) a reflector (horizon), the layer
Seismic resolution is defined in terms of the minimum producing the impedance contrast should have a
separation required between two layers so that they thickness larger than the tuning thickness so that the
produce a clearly differentiated seismic response (figure signal is not affected by the signal of the surrounding
16). Mathematically, this minimum thickness also layers.
called “tuning thickness” is given by equation 2:
The concept of resolution (tuning) should not confused
λw with the concept of detectability (Yilmaz, 2000), and as
tuning = ; (2) shown in figure 17, seismic data is able to detect
4 changes even when we are below the tuning thickness,
where:
i.e. the change is still visible, but being below tuning
v implies that although detectable it is affected by the
λw =
f surroundings.

Figure 18 shows an example of a 3 metres thick gas

Here, λ is the wave length (the w is added to avoid
ω

bearing layer, for which AVO modeling has been

confusion with the λ mentioned before, which is the performed using the in-situ and Gassmann fluid
incompressibility), v is the p wave velocity and f is the substituted logs. The resultant AVO curves show that
seismic frequency. the fluid change is still visible, even though the tuning
thickness in this field is about 19 metres.

CC

Fig. 17. Example of a 3m. gas layer, for which AVO

modeling has been performed using the in-situ and
Fig. 16. ‘Tuning thickness’ effects modified from Gassmann fluid substituted logs. The resultant AVO
Widdess (1973). curves show that the fluid change is still visible, even
. though the tuning thickness in this field is about 19m.

13
SPWLA 55th Annual Logging Symposium, May 18-22, 2014

THE UPSCALING ISSUE using the sonic log to compute the wavelength at a
given frequency (i.e. the frequency at which we want to
Upscaling well log data to the seismic scale remains a upscale) This essentially generates a “wavelength log”
challenge in the oil industry. Running averages and that can be used to define a running average window
blocking are common practice for this matter, even with a variable size (according to the wavelength) that
though these do not represent valid methods from a will allow upscaling the logs accurately (Figure 18).
geophysics perspective as they are not consistent with
the seismic ray paths. Unfortunately, this method is not properly implemented
in the majority of the commercial packages, which still
Backus, (1962) proposed a ray tracing theory that require the user to input a fixed thickness which
allows computing the combined seismic velocity for a essentially invalidates Backus average and converts it
sequence of layers of velocity, V, and compressional into a normal average. The only parameter that should
modulus M, arranged parallel to the seismic be required as input would be the frequency at which
propagation and for wavelengths much larger than the we wish to upscale.
layer thicknesses (i.e. below tuning), this is defined as
follows in equation 3:

M ( K + 4 / 3µ
VBackus = = (3)
ρ ρ

where 1
K + 4 / 3µ =
fi
∑ ( K + 4 / 3µ )
k i

CC
and ρ = ∑ fi ρi
i

here, f , are the volume fractions of the layers, which

are essentially the layer thicknesses T1 to Ti divided by
the total layer thickness Tt (Figure 17)
Fig. 18 Sequential Backus average applied to well logs.
After Lindsay and Koughnet, 2001.

As shown in figure 20, the correct implementation of

Backus average produces identical seismic responses as
using the full scale logs. Figure 21 shows a comparison
of the “true” Backus average with a running average
using a constant window. Again, even though visually
the results seem very similar, the crossplots show that
only the Backus average preserves the correct spatial
location of the hydrocarbon bearing rocks (black
circle). The running average results do not preserve
this spatial arrangement at seismic frequencies.

Fig. 17. Representation of Backus average ray tracing These results have an important implication. This
theory suggests that all the small scale changes need to be
included for Backus averaging in order to capture the
This implementation unfortunately requires that the hydrocarbon bearing rock signature. This explains the
layer thicknesses are known, which is only true in case sensitivity of the modeled seismic to the petrophysics
of layer based modelling, but for well log upscaling is input even though these small scale changes are
not ideal. Lindsay & Koughnet (2001) proposed a considerably below the tuning thickness. They produce
slightly different implementation of Backus average cumulative responses that, with seismic, the interpreter
called Sequential Backus average. Although not
explicitly defined in their paper, this method implies
14
 SPWLA 55th Annual Logging Symposium, May 18-22, 2014

could detect even though we are not able to resolve them.

Fig. 20. Correct implementation of Backus average produces identical seismic responses as using the full scale logs.
At seismic frequencies, Backus theory provides reliable results even in complex petrophysical situations.

CC

Fig. 21. Comparison of the “true” Backus average with a running average using a constant window.

15
 SPWLA 55th Annual Logging Symposium, May 18-22, 2014

CONCLUSIONS AND RECOMMENDATIONS NOMENCLATURE

The aim of this publication that the petrophysics b = Biot poro-elastic term (Bouteca and Sarda).
community will recognize and address the challenges E = Young’s Modulus.
and opportunities that petrophysics for rock physics fi = ith Volume fraction.
applications requires.
Ksat = Bulk Modulus of the saturated rock.
It has been shown that the small scale variations in Kdry = Bulk Modulus of the dry rock.
K0 = Bulk Modulus of the matrix.
mineralogy, fluid content, porosity, pore shape, etc,
Kfl = Bulk Modulus of the fluids.
have a great impact in our ability to model seismic data
Kφ = Dry pore stiffness.
(and hence in the quantitative interpretation).
M = Compressional (or P) Modulus.
Regardless of the tuning thickness, during the rock
Meff = Effective Compressional Modulus
physics modeling stage it is important to have as much
MVoigt = Compressional Modulus Voigt bound.
detail as possible so that we can investigate what
MReuss = Compressional Modulus Reuss bound.
factors affect the seismic signal and how much. We
Ti = ith layer thickness.
have shown that small variations in the petrophysics
Tt = Total layer thickness.
interpretation, even when they are visually negligible
Vclay = Volume fraction of clay minerals
can generate bias in the rock physics interpretations,
increasing the risk of either missing pay or drilling bad α = Biot poro-elastic constant.
wells. λ = Lame parameter.
λω = acoustic/ seismic wavelength.
Beyond petrophysical volumes, rock physics requires a µ = Rigidity Modulus.
fuller description of the matrix and pore fabric that µsat = Rigidity Modulus of the saturated rock.
seismic waves traverse both reservoir intervals and the µdry = Rigidity Modulus of the dry rock.
over and under burden. Specifically, the authors call φ = Porosity.
upon petrophysicists to embrace the need for a fuller ρ = Bulk Density.
and richer description of petrophysics and to capture,
σHmax = Maximum horizontal stress magnitude CC
where possible, the following features and textures:
σhmin = Minimum horizontal stress magnitude
1. Describe matrix mineralogy, including clays and the σeff = Nett effective stress magnitude
fluids they contain, sediment sorting and cementation υ = Poisson's Ratio.
style,
2. Be consistent in multi-well applications in the choice
of zonal parameters for matrix and fluids and to
investigate where possible the use of trends; varying
parameters with depth in siliciclastic sequences
unaffected by major unconformities,
3. Describe intrinsic anisotropy; being the variation of
acoustic velocities and moduli with direction of the
measurement,
4. Determine if anisotropy is vertically transverse,
(VTI) i.e. layered or laminated horizontal bedding, is
horizontally transverse (HTI) by virtue of in situ
stress anisotropy, micro-fractures or fractures,
5. Investigate any azimuthal variations in travel time
and any shear wave splitting that might result,
6. Determine if pore aspect ratios, and presence of
microporosity is a determining factor in acoustic and
elastic velocities,
7. Investigate if upscaling log and core moduli will
mask or preserve segmentation of pore, fluid or stress
characteristics.

16
 SPWLA 55th Annual Logging Symposium, May 18-22, 2014

REFERENCES M.L., 2003, Factors controlling elastic properties in

carbonate sediments and rocks: The Leading Edge,
Alvarez, E., MacBeth, C., 2014, An insightful 22(7), 665-660.
parameterization for the flatlanders interpretation of
Gassmann, F., 1951, Uber die elastizität poröser
time-lapsed seismic data, Geophysical Prospecting, 62,
medien: Vierteljahrsschrift der Naturforschenden
No 1. Jan 2014, 75 - 96.
Gesellschaft, 96, 1-23
Alvarez, E., Hernandez, J., Hall, J., Meradi, H.,
Siccardi, O., and Mollicone, P., 2009, Rock physics and Greenberg, M.L., and Castagna, J.P., 1992, Shear wave
seismic characterization of thin clastic reservoirs, velocity estimation of seismic waves in imperfectly
EAGE - International Conference and Exhibition, elastic rock: J. Geophys. Res., 73, 3917-3913.
Tyumen 2009, 2nd March, 2009. Hall, J., and Alvarez, E., 2010, Overcoming the
Limitations of Rock Physics Modelling in Porous Rock
Amini, H. and Alvarez, E., 2014. Calibration of the with Complex Mineralogy: SPWLA Annual
petro-elastic model (PEM) for 4D seismic studies in Symposium, Perth Aus. Paper TT
multimineral rocks, EAGE/FESM Joint Regional
Conference, Petrophysics meets Geoscience, 17-18 Feb Hall, J.C., Gutteridge, P., Hamdoun, L., Miller, M.,
Kuala Lumpur, Malaysia. Stott, I., 2010. Petrophysical description and assessment
of storage and flow capacity in the Triassic Kurrachine
Anselmetti, F.S. and Eberli, G.P., 1993. Controls on Dolomite reservoirs of Ash Shaer Field, Palmyra, Syria:
sonic velocity in carbonate rocks: in R.C. Liebermann, SPWLA 51st Annual Logging Symposium, June 19-23,
ed, J. Pure and Applied Geophysics, 141, No 2/3, 287- 2010, Perth Australia, Paper DD.
323.
Hashin, Z. and Shtrikman,S., 1963. A variational
Backus, G.E., 1962. Long-wave elastic anisotropy approach to the theory of effective magnetic
produced by horizontal layering: J Geophys. Res., 67, permeability of multiphase materials: J Appl. Phys., 33,
4427-4440. 3125-3131.
CC
Baechle, G.T., Eberli, G.P., 2014. Comparison of rock Kuster, G.T. and Toksöz, M.N., 1974. Velocity and
physics models to predict rock properties in carbonate attenuation of seismic waves in two-phase media.
reservoirs: EAGE/FESM Joint Regional Conference, Geophys., 39, 587-618.
Petrophysics meets Geoscience, 17-18 Feb Kuala
Lumpur, Malaysia. Krieff, M., Garat, J., Stellingwerff, J and Ventre, J.,
1990. A petrophysical interpretation using the velocities
Biot, M.A., 1941, "General Theory of Three of P and S waves (full waveform sonic): The Log
Dimensional Consolidation: J. Appl.Phys., 15, 155-164. Analyst, 31. 355-369.

Biot, M.A., 1956, Theory of propagation of elastic Kumar, M., and Han, D., 2005, Pore shape effect on
waves in a fluid saturated porous solid. II Higher elastic properties of carbonate rocks: 75th Annual
frequency range: J. Acoust. Soc. Amer., 28, 168-178. International Meeting, SEG, Expanded Abstracts, 1477-
1480.
Bouteca, M.J., and Bary, D., 1993, An experimental
study of the contribution of the pore volume change to Lindsay, P. and Van Koughnet, R., 2001. Sequential
the production within the framework of Biot’s theory: Backus averaging: Upscaling well logs to seismic
Society of Core Analysts, SCA 1993, paper N° 9319. wavelengths: The Leading Edge, 20(2), p188-191.

Bouteca, M.J., and Sarda, J.P., 1991, Rock mechanics Loucks, R.G., Lucia, F.J., Waite, L.E., 2013. Origin and
contribution to the determination of fluid flow description of the micropore network within the Lower
properties: in Second European Core Analysis Cretaceous Stuart City trend tight gas limestone
Symposium, Eurocas II. reservoir in Pawnee Field in South Texas: GCAGS
Journal, 2, 29-41.
Chacko, S., Keskes, N., Hall, J.C., Noyer, S., 2012,
Textural Analysis of MCT Images: Measurements Laurent, J., Bouteca, M.J., Sarda, J.P. and Bary, D.,
Beyond Rock Properties, SPE 161383, ADIPEC, Abu 1993, Pore-pressure influence in the poroelastic
Dhabi 11–14 Nov. 2012. behaviour of rocks: Experimental studies and results,
SPEFE, June 1993, SPE 20922.
Eberli, G.P., Baechle, G.T., Anselmetti, F.S. and Incze,
17
SPWLA 55th Annual Logging Symposium, May 18-22, 2014

Mavko, G, Mukerji, T., and Dvorkin, J., 2003, in: The ABOUT THE AUTHORS
Rock Physics Handbook 2nd edition: Tools for seismic
analysis in porous media: Cambridge University Press. Jonathan Hall is currently the Petrophysics Expert at the
Abu Dhabi Company for Onshore Oil Operations,
MacBeth, C., 2004, A classification for the pressure- ADCO. Jonathan has worked as staff and consultant
sensitivity properties of a sandstone rock frame: petrophysicist in/to British Petroleum, British Gas,
Geophysics, 69, 497–510. Agip International, Pemex, Schlumberger, Qatar
Petroleum, Suncor (formerly Petro-Canada) and Dong
Nur, A. 1971, Effect of stress on velocity anisotropy in Energy E&P. He has also held the position of Head of
rocks with cracks: Journal of Geophysical Research, Petrophysics for Scott Pickford (then a Core
76, 2022-2034. Laboratories company) and Senergy. Jonathan
graduated in Geology from the University of London,
Nur, A. Mavko, G, Dvorkin, J and Gal, D., 1995, Kings College and studied Mineral Exploration at the
Critical porosity: The key to relating physical properties Royal School of Mines, Imperial College. Jonathan is
to porosity in rocks: 65th Ann. Int. Meeting, Soc. Expl. currently studying for an MSc. in Petroleum Geology at
Geophys., 878. Royal Holloway College, University of London.
Shuey, R. T., 1985, A Simplification of the Zoeppritz
Equations: Geophysics 50 (9): 609–614 Erick Alvarez is a Rock Physicist with 14 years
industry experience. Erick specializes in analysis,
Tosaya, C.A., 1982, Acoustical properties of clay- modeling and estimation of elastic properties, as well as
bearing rocks: Ph.D. dissertation, Stanford University, in Rock Physics/AVO based rock and fluid
California. characterization. Erick graduated from the Instituto
Politecnico Nacional (IPN) in Mexico and started his
Truman, R.B., Howard, W.E., and Luffel, D.L., Shale career as geophysicist at Core Laboratories. He worked
porosity—Its impact on well log modeling and as Reservoir Geophysicist for Paradigm and
interpretation: SPWLA 30th Annual Logging Schlumberger. At the time of the submission of this
Symposium, June 11-14, 1989. abstract, Erick worked as the Rock Physics technical
coordinator in Senergy (G.B.) Ltd. Erick recently took a CC

Wang, Z., 2001, The Gassmann equation revisited: position in Shell U.K. He is also completing his
external PhD research student and part of the Edinburgh
comparing laboratory data with Gassmann’s
Time Lapse Project at Heriot-Watt University,
Predictions, Seismic and Acoustic Velocities in
investigating 4D Seismic Analysis.
Reservoir Rocks: Geophysics reprint series No. 19.
Widess, M.B., 1973, How thin is thin bed geophysics, ACKNOWLEDGEMENTS
38, Geophysics, 1176-1180.
The authors would like to thank and acknowledge the
Xu, S. and White, R.E., 1994. A new physical model numerous colleagues and clients, too numerous to name
for shear-wave velocity prediction: In Expanded all, with whom they have jointly worked together
Abstract, 56th Eur. Assoc. Expl. Geoscientists Meet. during more than a decade in Mexico, Malaysia,
Tech., Exhib., Vienna, p117. Russia, Iran, West Africa, Europe. However, Jaume
Xu,, S., and White, R.E., 1995. A new velocity model Hernandez and Hamed Amini especially deserve our
gratitude.
for clay-sand mixture: Geopys. Prosp., 43, 91-118.
Xu, S. and Payne, M.A., 2009, Modeling elastic Special thanks go to Dr. Noomane Keskes of Total and
properties in carbonate rocks: The Leading Edge, his Sismage team in Abu Dhabi for stimulating
January 2009, 28, No 1, p 66-74. discussion crossing discipline and cultural boundaries.
Especially, our former collaborator and colleague now
Yilmaz,O. and Doherty, S.M., 2000, Seismic Data moved on to new pastures, Dr. Soman Chacko, for his
Analysis: Processing, Inversion and Interpretation of faith in micro-CT imaging as a possible solution to
Seismic Data: SEG Books, 2027 pages. overcome the practical limitations of the D.E.M. model.
Zoeppritz, K., 1919, Erdbebenwellen VII. VIIb. Über Thanks for recent discussions with Dr. Mark Sams of
Reflexion und Durchgang seismischer Wellen durch Ikon during his one day course in Malaysia which
Unstetigkeitsflächen. Nachrichten von der Königlichen convinced one of the authors that there were some
Gesellschaft der Wissenschaften zu Göttingen: observations herein that may be universal.
Mathematisch-physikalische Klasse, 66-84.

18
 SPWLA 55th Annual Logging Symposium, May 18-22, 2014

APPENDIX According to Nur, the bulk and rigidity moduli of dry

rocks can be expressed as:
Gassmann´s Equations (1951). In general, Gassman
equations try to predict the effective bulk modulus of a ⎛ φ ⎞
fluid saturated rock from the modulus of the fluids, K dry = K 0 ⎜⎜1 − ⎟⎟ (2)
matrix and the dry rock. The general form of Gassman ⎝ φc ⎠
equations is given by:

K dry K fl ⎛ φ ⎞
K sat µdry = µ0 ⎜⎜1 − ⎟
= + ; (1) ⎝ φc ⎟⎠
K 0 − K sat K 0 − K dry φ ( K 0 − K fl )
This model can be used in combination with Gassmann
µ sat = µdry (1951) equations as well as with Effective Medium
Theory to determine the effective moduli of the rock.
The disadvantage of this model is that a value of critical
Where: porosity is needed and this is empiric, so the result is
Ksat = Bulk Modulus of the saturated rock. also empirical.
Kdry = Bulk Modulus of the dry rock.
K0 = Bulk Modulus of the matrix
Kfl = Bulk Modulus of the fluids. Hall and Alvarez model 2010. Using the measurements
φ = Porosity taken by Boutecca and Sarda (1991) and Laurent et al.
µsat = Rigidity Modulus of the saturated rock. (1992) they suggested a non-linear function to relate the
µdry = Rigidity Modulus of the dry rock. biot poro-elastic term (Gertsma, 1957) to porosity and
critical porosity, then they used this function to define a
mixing law that allows performing fluid substitution in
1 1 φ the presence of complex mineralogy using the Voigt
= + (2) CC
K sat K0 K 0 K fl and Reuss iso-stress and iso-strain concepts. The
Kφ + functional form is as follows:
K 0 − K fl
M eff ≈ (1− α ) M Voigt + α M Reuss (1)
Where: Kφ = Dry pore stiffness.
Where
Gassmann equations are limited to low frequencies 1 φ
(<100 Hz) so that they are most commonly used in α≈ ln + 1
seismic data. At higher frequencies such as measured a φC
by sonic and ultrasonic logging and core analysis
devices, Gassman equations may become unstable, Here, Meff, is the effective compressional elastic moduli
although this is also a function of the rock type. (K, λ or M), MVoigt and MReuss are, Voigt and Reuss
Notwithstanding, because of the numerous parameters averages of the elastic moduli weighted for all minerals
that are involved in the formulation, in practice the use and fluids fractions and the value of the empirical
of Gassmann equations may be really difficult to constant a is generally around 4 (Corrigendum. The
implement. A further disadvantage is that the authors acknowledge a typographic error on page 3 in
formulation only allows a single solid phase and a the original 2010 publication, where they quoted an
single fluid phase, so it is necessary to combine this to incorrect value of 0.25).
be combined with another theory such as Effective
Medium Theory for multiphase and multifluid
implementation. Hashin-Shtrikman Bounds. Based on the same
assumptions of effective medium theory, the Hashin-
Nur’s Critical Porosity Model (1995). Nur proposed a Shtrikman bounds describe the range in which can be
model for determining elastic moduli K and Mu of dry located the effective moduli K or µ of a mixture of
rocks using the critical porosity of the rock which can grains and pores:
be defined as the maximum porosity of the rock before
falling apart, so this value of critical porosity separate
the mechanic and acoustical behavior of the rock.

19
SPWLA 55th Annual Logging Symposium, May 18-22, 2014

f2 the rock with fluids is considered the same as the one of

K HS ± = µ1 + (1) the dry rock (as in Gassmann equations) all the effect of
( K 2 − K1 ) + f1 ( K1 + 43 µ1 ) −1
−1
fluids would be considered a direct effect of K
overestimating the real effect of the fluids in the rock
experiencing compressive stress.
f2
µ HS ± = µ1 + Tosaya (1982) and Castagna’s Empirical Relations
2 f ( K + 2µ )
( µ 2 − µ1 ) −1 + 1 1 4 1 (1985). Based on log measurements, Castagna, et al
5µ1 ( K1 + 3 µ1 ) determined empirical relations relating compressional
velocity, Vp, shear velocity, Vs, and, φ, porosity under
The upper and lower bounds are calculated by water-saturated conditions. For ‘Mudrock’, they found:
interchanging which mineral is called 1 and which is
termed 2. The Hashin-Shtrikman Bounds have the VP ( Km / s) = 1.36 + 1.16VS (1)
advantage of being narrower than Voigt and Reuss, but
also they have the disadvantage of the interdependence
of K and Mu, i.e. one needs to know both the K and Mu and for shaly sands:
of the individual phases in order to estimate the bounds
of the effective elastic moduli. VP ( Km / s ) = 5.81 + 9.42φ − 2.21Vclay (2)

Kuster and Toksöz Equations (1974). Try to predict the

effective elastic moduli K*KT and µ*KT (and as a VS ( Km / s ) = 3.89 − 7.07φ − 2.04Vclay
consequence seismic velocities VP and VS) of a
multiphase medium considering different inclusion where;
shapes, from their volume concentrations and their
individual elastic moduli. The general form of Kuster Vclay is the volumetric fraction of clay minerals.
and Toksöz equations is:
CC
N These equations are commonly used for determining
* (K m + µm )
4
( K KT − Km ) *
3
= ∑ xi ( K i − K m ) P mi (1) Vp-Vs relations empirically, but they do not have
( K KT + µ m ) i =1
4
3 theoretical justification, so their application must be for
the geographical area of their formulation or the
(µm + ζ m ) N regression parameters be again optimized for a new
*
( µ KT − µm ) *
= ∑ xi ( µ i − µ m )Q mi area of application.
( µ KT + ζ m ) i =1
Xu and White Model (1995). Xu and White combined
Where : Kuster and Toksöz equations with Effective Medium
Theory to develop a theoretical model for estimating
velocities and density in shaly dry sandstones from the
µ (9 K + 8µ ) clay and quartz properties and the porosities associated
ζ =
6( K + 2 µ ) with sand and clay fractions.

The coefficients Pmi and Qmi describes the effect of an This model can be used to predict shear wave velocities
inclusion material i in a background medium m and the from known mineral properties, measured porosity, clay
terms Ki, Km, µi,, µm describe the individual moduli for content and measured compressional velocity Vp.
the different inclusion shapes with volume
concentrations xi. Inclusions with different material
properties or different shapes require separate terms in
the summation.

These equations can be used only at high frequencies

(~1x106 Hz) and have the disadvantage that it is
necessary that both K and µ of the individual phases are
known in order to estimate the effective elastic moduli.
of the rock. Reviewing Hooke’s law there is an
interdependence between K and µ and, if the rigidity of
20
 SPWLA 55th Annual Logging Symposium, May 18-22, 2014

1 ⎛ 1 − φ − Vclay ⎞ 1 V 1
= ⎜⎜ ⎟⎟ + clay (1)
VP 0 ⎝ 1 − φ ⎠ VP quartz 1 − φ VP clay

1 ⎛ 1 − φ − Vclay ⎞ 1 V 1
= ⎜⎜ ⎟⎟ + clay
VS 0 ⎝ 1 − φ ⎠ VS quartz 1 − φ VS clay

⎛ 1 − φ − Vclay ⎞ V
ρ 0 = ⎜⎜ ⎟⎟ ρ + clay ρ clay
⎝ 1 − φ ⎠ 1−φ

Xu and White model has the advantage of being

relatively easy to implement, the principal disadvantage
is that implicitly considers a linear Vp/Vs ratio, which
obviously is not accurate. Aside from this, the model
does not consider the influence of fluids in the
velocities and density terms, so is necessary to
complement the above with the Gassmann or Alvarez
and Hall here presented to model their impact.

Xu and Payne Model (2009). Xu and Payne extended

the Zu and White (1995) model to predict
acoustic/seismic velocities in carbonate rocks. This
model included two more pore types (shapes): rounded CC
pores and microcracks to predict the VP-VS relationships
for carbonates but also siliciclastics.

Using a pore-type inversion method based upon work

by Kumar and Han (2005), they quantify the effect of
microcracks on Gassmann fluid substitution. Their
results show that pore fluid in microcracks is relatively
un-relaxed, at high log data ultrasonic frequencies. This
they attribute to a local low permeability for these
cracks. A characteristic of the Xu-Payne model is that
individual pore components can be included into the
model such that they are either isolated, or, in perfect
fluid connectivity with the remaining pore space.
Further the model allows an investigation of the effect
on elastic properties including anisotropy of fluid
communication among matrix pores and fractures.

21