Rock physics modelling of shale diagenesis

Anders Dræge, Morten Jakobsen and Tor Arne Johansen

Centre for Integrated Petroleum Research, University of Bergen and Department of Earth Science, University of Bergen,
Allegaten 41, 5007 Bergen, Norway

ABSTRACT: A model for estimating the effective anisotropic properties of

cemented shales is presented. The model is based on two mathematical methods for
estimation of effective properties of a composite medium; a self-consistent approxi-
mation and a differential effective medium model. In combination these theories
allow approximation of a shale with connected clay minerals and cement, and
disconnected pores and quartz grains, which can be compared with the conditions in
a real cemented shale. A strategy is also presented for estimation of stiffnesses in the
transition zone from mechanical compaction to chemical compaction dominated
diagenesis. Combining these theories with a shale compaction theory, enables
modelling of the effective elastic stiffnesses for shales from deposition and
mechanical compaction to deep burial and chemical compaction/cementing. Results
from the model were compared with velocity data from three wells, showing good
fit for velocity predictions, following the main velocity trends with increased
temperature and depth.
KEYWORDS: shale, diagenesis, anisotropy, velocity, cementation, rock physics

INTRODUCTION isotropic properties of cemented sandstones. They treated the

Shales play an important role in fluid flow and wave propaga- rock as an isotropic material with spherical matrix grains, in
tion because of their low permeability and anisotropic micro- contrast with the anisotropic shale model presented in this
structure. When they overlie hydrocarbon reservoirs, shales can paper.
be effective seals. Since shales make up most of the overburden The study of Bjørkum & Nadeau (1998) indicated that illite
of many of the worlds hydrocarbon-bearing reservoirs, quanti- precipitation in shales at temperatures of 80–100C reduces the
tative information about the velocity anisotropy of shales is already low permeability by several orders of magnitude. They
typically needed for the calibration of amplitude versus offset asserted that dissolution/precipitation processes can continue
(AVO) measurements, imaging and processing purposes. The mainly independently of fluid pressure and effective stress. The
model described in this paper provides a range of stiffnesses lower limit to which mechanical compaction can reduce the
from soft and compliant shales that are dominated by mech- permeability of clay-rich sediments can be estimated to approxi-
anical compaction, to hard and rigid shales dominated by mately 10
6 D (Chen & Nur 1994), but Bethke et al. (1988) had
chemical compaction/cementation. This paper refers to shale already shown that lower permeabilities were needed to main-
as a sedimentary rock with 50% or more clay minerals in the tain high overpressures over geological time. Thus, it is natural
framework. to relate further permeability reductions to chemical reactions
Cement in shales can consist of dissolved clay minerals that inside the shale, which is also supported by the observations of
precipitate as more stable clay minerals and quartz. Calcite Nadeau et al. (2002).
cement can precipitate at shallower depths. Cementation The cementation processes can be considered as a transition
stabilizes the rock and inhibits/ceases mechanical compaction. from ‘soft’ mudstones and shales to hard and rigid rocks.
Temperature is commonly the key factor in dissolution and Theories that describe effective shale properties as a function of
precipitation of clay minerals, since increased temperatures mechanical deformation (e.g. Ruud et al. 2003) are unlikely to
make some of the minerals unstable. At 60–80C smectite can provide a good prediction for the effective properties of
react to form illite and quartz (Bjørlykke & Aagaard 1992). cemented shales, since chemical processes and mechanical
Bjørlykke & Aagaard (1992) also reported that smectite can mechanisms are believed to have distinct effects on both
dissolve and precipitate as chlorite or corrensite, depending on porosity loss and grain–grain contacts. The transition from
the mineralogy. Bjørkum & Nadeau (1998) asserted that the mechanical compaction-dominated diagenesis in shales, to
dissolution/precipitation processes are linearly dependent on cementation-dominated diagenesis is illustrated in Figure 1a.
time, and exponentially dependent on temperature. A mineral The same figure also introduces another modelling problem:
reaction that exemplified this behaviour was the transformation the transition from one regime to another is unlikely to be
of kaolinite and K-feldspar to illite, which initiates at around abrupt. Some mechanical compaction is expected in the first
120C. Chemical reactions and cementation will take place in stage of cementation, since a certain amount of cement is
sandstones as well, though the initiation temperature and required to stabilize the rock sufficiently to resist the over-
reactants might differ. Dvorkin et al. (1994, 1999) and Dvorkin burden. Figure 1b demonstrates this by decomposing the
& Nur (1996) presented models for estimating the effective porosity loss in the transition zone (TZ) into porosity loss
Petroleum Geoscience, Vol. 12 2006, pp. 49–57 1354-0793/06/$15.00  2006 EAGE/Geological Society of London
50 A. Dræge et al.

Fig. 1. (a) Clay minerals are orientated randomly at deposition (1), but reorientate to a sub-parallel alignment and lose most of their porosity
due to mechanical compaction (2). Scattered cementation will create strong bonds in some of the grain contacts in the transition zone (3). The
cementation rate increases exponentially with temperature (depth), so a slow rate of burial in the cementation regime leads to lower degree of
cementation (4) than a rapid subsidence (5) at the same period of time. Do and Dcem denote the depths at the top and bottom of the transition
zone (TZ), respectively. (b) A hypothetical porosity–depth curve. In the TZ the sum of the effects of mechanical compaction (
comp) and
cementation (
cem) constitute the total porosity loss.
o,
o and
cem denote the initial porosity, the porosity at which cementation starts and
the porosity at which mechanical compaction ceases, respectively. The contribution to porosity loss decreases for the mechanical compaction
and increases for cementation when the depth in TZ increases.

Fig. 2. (a) Vp, Vs and the Vp/Vs ratio

with depth. Depths are in km; velocities
are in km s
1; (b) the Thomsen
anisotropy parameters. The stippled
lines are the elongations of the
compaction and cementation trends
into the TZ. The porosity in both (a)
and (b) follows the same pattern as in
Figure 1.

related to mechanical and chemical compaction. Figure 2 shows the rock is treated as though it consists of continuous phases of
depth trends for vertical P-wave velocity (Vp), vertical S-wave clay minerals in the framework and cement in the pore space. It
velocity (Vs) and Thomsen anisotropy parameters, obtained is assumed that when the cement has stabilized the shale
from the models presented in this paper. The results are for a enough to prevent further mechanical compaction, the cement
hypothetical shale which follows the porosity–depth pattern can be considered a connected phase. The term ‘inclusion’ is
shown in Figure 1b. Clearly, the evolution paths of the used on both silt-size quartz particles and fluid-filled pores. In
properties vary between the different stages of diagenesis. The the cementation model, these inclusions are treated as isolated
parameter–depth trends alter in the TZ for all parameters, but and non-communicating. The flattening of the constituents is
most significantly for the anisotropy parameters. defined by the pore aspect ratio,
which is the ratio between
This paper presents a model for estimation of the effect of the shortest and longest axis of a spheroid. Following Jakobsen
cementation on the effective elastic properties in a shale
et al. (2000, 2001), the cement and pore shapes are considered
(cementation model). A strategy for estimating the effective
to have the same aspect ratio as the clay minerals, which is
rock properties in the transition from mechanical compaction-
dominated to cementation-dominated diagenesis in shales is constant with increasing depth and cementation. The shape of
also introduced. First, some geometrical considerations about the cement minerals compared with matrix clay minerals is
cemented shales are presented, followed by a description of the decisive for the effect of cementation. In the cementation
cementation model. Then the focus shifts to the relation- model the same aspect ratio is assigned to the cement and
ship between mechanical and chemical compaction in the matrix clay minerals, although grain size and surface area might
TZ. Further, it is demonstrated how stiffnesses and vertical differ. This makes the cementation model less dependent on
velocities may vary as the shale loses its porosity by mechanical cement volume for a given porosity, as long as the shale is
compaction and cementation. The model predictions are tested pervasively cemented.
by comparison with observed vertical velocities from three The three-step procedure of Hornby et al. (1994) is applied,
wells. together with the theories of Jakobsen et al. (2000, 2001) to
estimate the effective stiffnesses of the cemented shale. The
shale is considered to be an aggregate of building blocks, where
CEMENTATION MODEL the clay minerals in each block are fully aligned, but the
Highly compacted shales have low porosity and permeability; orientations of the building blocks vary due to an orientation
however, severely overpressured shales can be undercompacted distribution function (ODF) that will be discussed later. Figure
for a given depth. Cement will entail further permeability 3 illustrates the five-step modelling procedure. The first step
reduction, which makes the pores practically unconnected (see makes use of the self-consistent approach (SCA) (Willis 1977).
Fig. 1). Thus, after a certain amount of cement has precipitated, The basic idea of SCA is that the composite medium itself is
 Modelling shale diagenesis 51

If a new phase is introduced in the composite material by

reducing the concentration of only one of the constituents,
leaving the others unchanged, a modified version of DEM,
denoted DEM*, can be applied (Jakobsen et al. 2000). The
change dC in the effective stiffness tensor C resulting from an
increase dv1, of one material and corresponding reduction dv2,
of material two can be written (Jakobsen et al. 2000):
dv1 dv2
dC = (C
C)Q1
(C
C)Q2 (1)
1
v1 1 1
v2 2
where Qi (i=1, 2) is a tensor given by:

Qi=[I+P(C)(Ci
C)]
1.

C1 is the stiffness tensor of the material whose concentration

increases with dv1, and C2 is correspondingly the stiffness
tensor of the material whose concentration decreases with dv2.
I is the identity tensor and vi denotes the volume concentration
of medium i. P is a fourth-rank tensor calculated from the
response of a single inclusion embedded in an unbounded
matrix of the effective material, and is discussed in Jakobsen
et al. (2000).
Step 3 uses equation (1) to calculate the effective stiffnesses
as fluid (porosity) is introduced to the building block as isolated
Fig. 3. Illustration of the five steps used to model the stiffnesses in inclusions at the cost of cement concentration only. This will
a cemented shale. SCA, DEM and DEM* are explained more imitate the reverse cementation process, where cement replaces
thoroughly in the text. k, µ and  denote bulk modulus, shear fluid in the pores. The operation requires that the resulting
modulus and density, respectively. The indices cl, cem, flu and q cement content in step 2 also includes the fluid concentration,
denote clay, cement, fluid and quartz, respectively. Vi (i=cl, cem, flu,
q) is the concentration of the different constituents. c11–c44 are the since the cement content is reduced further in step 3.
stiffnesses of the transversely isotropic medium. ODF, orientation The properties of one single building block of a quartz-free
distribution function. shale with connected clay minerals and cement and isolated
pores have now been estimated. Sedimentary materials are
commonly complex, with clay mineral orientations that deviate
embedded in a homogeneous background medium of yet from the fully aligned orientation. Step four assumes that the
unknown effective elastic properties. The scattering effect of a clay minerals are fully aligned locally, but that the alignment axis
wave that travels through this composite medium will be zero, varies from place to place. This is modelled by varying the
due to destructive interference, if the composite properties of orientations of the building blocks. If the effective properties
the background medium are those of the effective medium. A and the orientations of the building blocks are known, one can
consequence of modelling with the SCA model, is that a phase estimate the effective stiffnesses of an aggregate of building
represented by approximately 40% or more (shape dependent) blocks, which in this case constitute the cemented shale.
in a composite medium, will constitute a connected phase, Johansen et al. (2004) studied ODFs for the building blocks of
opposite to isolated inclusions of lower concentrations (Hornby shales. They found that the ODF resulting from vertical
et al. 1994; Jakobsen et al. 2000). mechanical compaction of an initially isotropic (randomly
Step two utilizes differential effective medium theory (DEM) orientated) pure shale can be described as:
(Nishizawa 1982) to estimate the effective properties of the
two-phase medium at the desired concentrations. The DEM 1 c2
Wc(U) = , (2)
approach utilizes the principle of porosity growth (Cheng 1978) 8 (cos (U) + c2sin2(U))3/2
2 2

to extend the results of Kuster & Toksöz (1974) to be valid at where U is the angle between the short axis of the penny-
high porosities. Mechanical interactions between inclusions are shaped clay platelets and the vertical axis, and c is the mech-
modelled by a gradual adding/removing of inclusions to/from anical compaction. c can be expressed by porosity due to
the background medium, followed by a calculation of the new mechanical compaction
t and critical porosity
0:
effective properties in the medium with new concentrations.
The background medium for each small alteration in inclusion 1

t
concentrations is represented by the effective properties of the c= . (3)
last calculation. When the desired concentrations of inclusions 1

0
are reached, the effective properties of the composite are
Critical porosity for shales can be expressed as a function of
obtained. As for the SCA model, this model accounts for the
quartz content (Ruud et al. 2003):
effect of interactions of the individual inclusions. But the DEM
model has the ability of preserving connectivity status when the 1
v
concentrations of the constituents change. Thus, connected
0 = , vc0.50, (4)
1
phases will remain connected when the concentrations change,
v
and vice versa for unconnected phases. New constituents added 0.80
to the composite media by using DEM, will be added as vq
where v = , v and vc denote the solid fractions of quartz
unconnected inclusions. Variation in the concentration of one vq+vc q
of the components will affect all the other concentrations. and clay in the shale, respectively (vq+vc = 1

) and 0.80 is the
52 A. Dræge et al.

critical porosity assigned to pure shale. Equation (3) includes

only porosity loss that is due to mechanical compaction, since
no further alignment of clay minerals is expected as a function
of chemical compaction. Estimation of
t will be discussed
later. The compaction is assumed to take place in the vertical
direction only, thus the compacted medium will be vertically
transversely isotropic.
Ruud et al. (2003) presented a quartz volume-dependent
ODF for shales that incorporates the ODF for shales in
equation (2). The ODF for a completely random and isotropic
medium can be written: Wi= 1 2 . The effect of (near-spherical)
8
quartz grains entails an increased randomness in the clay
mineral orientation, since the clay must drape around the larger
and rounder quartz grains. Thus, the ODF of a shale with
quartz inclusions can be approximated by a weighted average of
the ODF for pure shale, and the ODF of an isotropic medium:

W(U) = vqWi + vcWc(U). (5)

Fig. 4. Different paths from mechanical compaction- (K
(1)=0) to
cementation- (K
(n)=1) dominated shales. In this example Do and
vc and vq are normalized to the solid part of the rock volume. Dcem are 3.0 km and 3.2 km, respectively.
The Voigt (1928) average, which is the upper bound of
effective stiffnesses of a mixture, is applied to average over the
distribution of building blocks since the latter are considered to increase more rapidly near Dcem. An appropriate weight with
be cemented in the boundaries to neighbouring building blocks. these properties can be written:
A convenient averaging scheme was developed by Morris
(1969) and adapted by Sayers (1994). Thus, step four estimates K
= [0,dk,2dk,...,ndk]x
, (6)
the effective anisotropic stiffnesses of a shale with varying
orientations of the building blocks. where the array [0,dk,2dk,...,ndk] is linear and increases with dk
The fifth step applies DEM to add isolated quartz inclusions in each step. x
controls the local dependency of porosity loss
to the building blocks, as shown in the last step of Figure 3. on increasing cementation and must be a positive number.
Thus, the effective anisotropic stiffnesses for a shale with Figure 4 shows how the different values of x
lead to different
connected clay minerals and cement, isolated inclusions of paths from completely mechanical compaction-dominated
fluid and quartz, and with an estimated internal orientation (K
(1)=0) to completely cementation-dominated (K
(n)=1)
geometry, is obtained. shales, in the depth interval 3.0–3.2 km. High positive values of
x
entail an initial slow increase in cementation-related porosity
loss, while a value of x
=1 corresponds with a linear transition
ESTIMATION OF COMPACTION AND ODF IN between the influence of mechanical compaction and cemen-
THE TRANSITION ZONE tation on the porosity loss. x
is dependent on burial rate and
Equation (2) indicates that the level of mechanical compaction temperature–depth ratio, because these parameters affect the
(c) has great influence on the orientation of the building blocks. rate of cementation. dk can be chosen due to the desired
Mechanical compaction will still be effective in a limited depth resolution of the modelling. The lower the value of dk, the
interval after initiation of cementation. However, the influence smaller the intervals into which TZ is divided. It is crucial that
of cementation will gradually increase from the depth of initiat- the last element in K
equals one, thus the number of elements,
ing cementation, Do, to a depth Dcem, where the rock is strong n, must equal 1/dk, see equation (6). The depths D, that
enough to resist the pressure from the overburden. Then, correspond with the weights can be written as:
porosity loss will be dependent on cementation processes alone.
Let
o, 
comp and 
cem denote the porosity at depth Do, D = D0 + K
1⁄x (Dcem
D0). (7)
and the accumulated positive porosity change due to mech-
anical compaction and cementation in the TZ respectively, as Since the first and last weights are zero and one, respectively,
illustrated in Figure 1. The sum of 
comp and 
cem thus D will increase linearly from Do to Dcem in n steps. The
expresses the total porosity changes in the TZ. The porosity porosity loss due to cementation is now expressed as:
loss in shales is related to quartz content (e.g. Ramm &
Bjørlykke 1994), so the porosity loss should be estimated for 
cem = K

. (8)
different quartz contents.
The temperature increase with depth is normally considered Since 
=
cem+
comp:
as a linear function. Thus, since the rate of cement precipitation
in a shale is increasing exponentially with temperature, it will 
comp = (1
K
)
. (9)
also increase exponentially with depth, provided that all the
reactants are present at sufficient concentrations. The total porosity loss in TZ due to mechanical compaction,
A weight K
must be defined such that when multiplied with
t, then becomes:
the total porosity loss in the TZ, it gives the porosity loss that
is due to cementation alone. K
should reflect the exponential
t = (
0

0‘)
comp, (10)
increase of cement precipitation rate with increasing
temperature/depth, by letting the influence of cementation be which, combined with equation (3), provides the overall mech-
weak in the first stage of the TZ, while the influence should anical compaction c of the rock. Including the mechanical
 Modelling shale diagenesis 53

Table 1. Parameters used in the modelling will thus lead to a significant increase in the overall stiff-
nesses of the shale. Further cementation will still increase the
k µ   stiffnesses, but not at the same rate as the initiating cement.
GPa GPa g cm
3 The modelling of stiffnesses in the TZ is a four-step
Clay 1 21.0 7.0 2.60 0.15 procedure. The first step is to apply the shale compaction
Clay 2 25.0 9.0 2.55 0.15 model of Ruud et al. (2003) in the interval Do–Dcem to estimate
Quartz 37.0 44.0 2.65 1.00 the effective properties of a shale with no cement, but other-
Cement 30.0 12.0 2.62 0.15 wise the exact same component properties as the cemented
Brine 2.96 0.00 1.03 — shale. This modelling obtains the stiffnesses in a hypothetical
The properties k, µ,  and  denote the bulk modulus, shear modulus, density
case, where the cementation does not influence the stiffening of
and aspect ratio, respectively. the rock. The next step models the partly cemented shale in the
interval Do–Dcem as if it were independent of mechanical
compaction, by following the steps in the cementation model.
compaction that occurs in the TZ, includes the continuous Thus, we obtain the stiffnesses of a shale with a connected
alignment of the building blocks in the modelling. The align- cement phase (e.g. illite), and where porosity loss is due to
ment is expressed by the compaction-dependent ODFs in chemical compaction alone.
equations (2) and (5). The third step is to apply a weight on the same form as in
equation (6), which controls the ratio of the influence of
mechanical compaction and cementation on stiffnesses in the
MODELLING OF STIFFNESSES IN THE TZ. For estimation of stiffnesses, let Kc and xc correspond to
TRANSITION FROM MECHANICAL K
and x
in equation (6). Thus, Kc and xc control how fast the
COMPACTION-DOMINATED TO cementation will dominate the stiffness increase in a partly
CEMENTATION-DOMINATED REGIME cemented shale. Kc should reflect the strong effect the first
Before cementation the shales are soft and compliant. But, as cement has on the shale. The lower the positive value of xc, the
cementation initiates, the cement will bind the grains together faster the model will treat cementation as the dominating
in stronger bonds (Bjørlykke & Høeg 1997). The first cement mechanism for increasing stiffnesses (see Fig. 4). xc is related to

Fig. 5. Modelled vertical Vp and Vs

and Thomsen anisotropy parameters as
they vary with mechanical compaction-
and cementation-related porosity loss
and quartz concentration. Porosity is a
fraction of rock volume.
54 A. Dræge et al.

Fig. 6. Modelled velocity data (grey)

compared with log velocity data
recorded in three different wells (black).
The clay content is normalized to the
matrix volume, while the porosity is
normalized to the total rock. The log
recordings with less than 50% clay in
the matrix are excluded from the
figures and the modelling. In all wells
Do and Dcem are at 60C and 100C,
respectively.

x
since the amount of precipitated cement affects the influence The starting point in the modelling is a hypothetical shale where
of cementation on both porosity and stiffnesses. But unlike x
, the porosity is reduced from 20% to 15% (depth = Do) by
the value of xc is also dependent on the way the first cement mechanical compaction. The TZ is defined to occur from
precipitates. If all the initiating cement precipitates in the 15% to 12%. Reduction of the porosity from 15% to 12%
contact areas of the framework grains, or bridges the pores, the corresponds with the depth interval Do–Dcem, where mech-
stiffnesses will increase rapidly, and xc will have a lower value anical compaction and cementation processes both occur.
than if cement precipitated in the pore space between the Further porosity reduction below Dcem is considered to be
contacts. The weight (1
Kc) is assigned to the compaction- due to cementation processes alone. For simplicity, it is
related stiffnesses. Kc increases from 0 at Do to 1 at Dcem, assumed that the precipitation of cement accounts for half of
which makes the influence of compaction diminish through the porosity loss, and chemical compaction as a result of
TZ, and cease at Dcem. dissolved framework causes the other half, i.e. the shale is
The final step in this procedure is to use the stiffnesses for considered a closed dissolution/precipitation chemical system
the uncemented and cemented shale together with Kc, and for cementation.
apply the Voigt–Reuss–Hill (VRH) average (Hill 1952) to The values of x
and xc are related to 2 and 1/2, respectively.
estimate the effective rock properties for the depths in the TZ. The total mechanical compaction is modelled by using
The average tensor thus represents the overall stiffnesses of the equations (8)–(10) and (3) and defines the input to equation (2).
partly cemented shale, in the transition to a state where no Porosity reduction due to mechanical compaction dominates
further compaction occurs. the shallow part of the TZ, while cementation dominates the
deeper part.
RESULTS
Porosity loss and mechanical compaction Vertical velocities and elastic stiffnesses
The following examples model the influence of cement content Table 1 shows the parameters used in the modelling; the clay
on the effective properties in an anisotropic cemented shale. properties of clay 1 were used. Clay 1 parameters were adopted
from Tosaya (1982) and clay 2 parameters from Han et al.
(1986). Figure 5 shows modelled Vp and Vs for decreasing
Table 2. Results from modelling velocities in well 1, well 2 and well 3
porosities/increasing depth. As expected, the velocities increase
with increasing quartz content and decreasing porosity. The
Well 1 Well 2 Well 3
Vp/Vs Vp/Vs Vp/Vs
velocity increase is markedly larger in the TZ than in the
cementation regime. This reflects the choice of xc, which
Mechanical compaction 150.2/242.1 105.4/178.4 168.5/318.9 determines that the first cement is considered to strengthen the
Transition zone 98.3/110.1 168.7/217.3 114.1/93.5 contacts between the grains and/or bridge the pores. The effect
Cementation 172.7/159.4 126.7/111.0 130.4/132.2 of further cementation in an already cemented shale will be
Whole log 144.1/153.3 129.6/150.8 132.8/164.5
Percent deviation 4.4%/9.5% 4.1%/9.9% 4.2%/10.6%
smaller, as seen in Figure 5.
The modelled Thomsen’s anisotropy parameters are given in
Columns 2–4 show the absolute P- and S-wave velocity deviation from the the same figure. The  and  plots show that the anisotropy
log velocities. The overall Vp/Vs ratio in wells 1–3 deviated 7.8%, 9.6% and decreases with increasing quartz content in the shale. Quartz
16.8%, respectively. content is decisive for the anisotropy behaviour as cementation
 Modelling shale diagenesis 55

Fig. 7. Modelled effective properties of

a cemented shale with varying
constituent aspect ratio (
). Porosity is
a fraction of rock volume.

starts.  decreases most for shales with around 20% quartz porosities and clay contents from the logs are used as input in
content, while  decreases most for quartz-free shales, when the the modelling. Wells 1 and 3 are modelled with clay 1, while
shale enters the TZ. The plots show that the cementation clay 2 gives the best fit for well 2. To estimate cement
model predicts lower  dependency of quartz content than does concentrations, the porosity–depth curves for different quartz
the compaction model. This is partly due to the two different contents are first approached by second-order curve fitting.
ways these models handle the aspect ratio of the pores. The Theories shown previously are used to estimate how much of
compaction model of Ruud et al. (2003) suggests that  the porosity loss in the TZ is due to cementing. At depths
decreases continuously during mechanical compaction, while deeper than Dcem it is assumed that all porosity loss is due to
the cementation model uses a constant pore aspect ratio for all cementation processes and, as before, half of the porosity loss
porosities. In the cementation regime, the modelled  and  is dedicated to cement precipitation. Do and Dcem are set to
anisotropy decreases most for shales with low quartz content.
depths corresponding to 60C and 100C, respectively. Table 1
The -plot shows that the  values increase when cementa-
shows the properties of the constituents used in the modelling.
tion starts. The increase is most pronounced for shales with low
quartz contents. But with further cementation, the  values The same values for the TZ parameters (2.0 and 1/2) for x

decrease smoothly towards zero. All the constituents in Table 1 and xc, have been applied for all wells.
are considered isotropic as they are, but the shape () of the Figure 6 and Table 2 shows that the modelled Vp and Vs
grains and inclusions induce the anisotropy in the composite mimic the log-recorded velocities quite well. At some intervals
media. the velocities deviate 200–300 m s
1 on average from the log
velocities, and a few places even more. But all modelling
follows the main trends of the recorded velocities with
Modelling real data increased mechanical compaction and cementation. The largest
The model has been compared with vertical P- and S-wave deviations occur in the uncemented part, especially for Vs,
velocities of shales recorded in three wells. Anisotropy data which seems to be modelled too high. It has been observed
were unavailable, so there is no real control data to support previously (Ruud et al. 2003) that the mechanical compaction
predicted anisotropy–depth trends. The wells penetrate model has a tendency to predict too high S-wave velocities.
horizontal shales vertically. The brine properties, densities, Velocities in the TZ are modelled very well for all wells. The
56 A. Dræge et al.

Fig. 8. Different modelled paths

between mechanical compaction regime
and cementation regime, as a function
of xc. Outside the TZ (12–15%
porosity) all the graphs coincide.
Porosity is a fraction of rock volume.

cementation model also produces modelled velocities which model is not very sensitive to different concentrations of the
mainly coincide with the log data. cement and the cement properties, as long as the fluid porosity
is unchanged. Thus, these uncertainties are not crucial for
DISCUSSION model predictions. The shapes of the inclusions have a larger
influence on the result.
In order to study quartz cementation in sandstones, there exist
models for estimating cement volumes mainly as a function Figure 7 illustrates the modelled effective properties of a
of temperature. One such model has been presented by cemented shale with 20% quartz, and the material properties as
Walderhaug (1996). A similar model for shales and clay defined in Table 1. For the cementation model, the aspect ratio
cementation could be combined easily with the modelling of the pores, cement and clay mineral vary, while only the shape
strategy presented in this paper. Conversely, the theories for of the clay mineral varies in the compaction model. This is
compaction and cementation in the TZ can be applied in rock because the compaction model expresses the pore aspect ratio
physics modelling of cemented sandstones. as a function of mechanical compaction. Thus, the different
The chemical reaction rates in shales are dependent mainly approaches entail the discontinuities in the TZ. Figure 7 clearly
on temperature, and they appear to be less time dependent than shows that the degree of elongation of the constituents will
those in sandstones (Bjørkum & Nadeau 1998). Thus, infor- have great influence on the vertical velocities and the degree of
mation is needed about the burial history of the rock to model anisotropy in the rock. The cement and pores are considered to
the temperature history. Rocks that have been subjected to have the same shape as the clay minerals, while the quartz
uplift/erosion will display a more pronounced cementation grains are approximated by spherical inclusions. The modelling
than expected from rocks at the same depth/temperature that indicates that the stiffnesses of a cemented shale are more
have not been subjected to uplift/erosion. It is assumed that dependent on the shape of the components, than a shale that
there are no uplift effects in the real data analysis. has only been subjected to mechanical compaction.
It is hard to obtain numerical values of some of the xc implicitly contains information about cement distribution.
parameters in the cementation model. In this modelling Figure 8 shows how the modelled anisotropy and vertical
the cement properties have been defined to lie between the velocities will vary for different values of xc in the TZ. The
properties of clay and quartz. The cement volume is calculated shale used in the modelling consists of 20% quartz in the
from the porosity losses for different quartz contents. But the matrix, and the component properties are given in Table 1. By
 Modelling shale diagenesis 57

comparing the model to observed data, xc can be found and REFERENCES

analysed. The curves for the three lowest values of xc in
Figure 8, imply that a significant amount of the cement is Bethke, C.M., Harrison, W.J., Upson, C. & Altaner, S.P. 1988. Supercomputer
precipitated in the pore space between the grain contacts. Thus, analysis of sedimentary basins. Science, 239, 261–267.
the cement will contribute little to the overall rock stiffnesses, Bjørkum, P.A. & Nadeau, P.H. 1998. Temperature controlled porosity/
permeabillity reduction, fluid migration and petroleum exploration in
until the cement concentration becomes so large that cement sedimentary basins. Australian Petroleum Production and Exploration Association
bridges the pores or starts to precipitate in the grain contacts. Journal, 38, 453–465.
The curves with xc values higher than one indicate that the Bjørlykke, K. & Aagaard, P. 1992. Clay minerals in North Sea sandstones. In:
concentration and distribution of the initial cement precipitated Houseknecht, D.W. & Pittman, E.D. (eds) Origin, diagenesis, and petrophysics
are capable of increasing the rock stiffnesses from the start of clay minerals in sandstones. SEPM Special Publication, 47, 65–80.
of TZ. The effect of varying x
is much smaller, since the Bjørlykke, K. & Høeg, K. 1997. Effect of burial diagenesis on stresses,
compaction and fluid flow in sedimentary basins. Marine and Petroleum
fluid porosity is the same, regardless of x
, and only the Geology, 14, 267–276.
framework–cement ratio is affected. Chen, Q. & Nur, A. 1994. Critical concentration models for porous materials.
An assumption has been made that the mechanical compac- In: Corapcioglu, M.Y. (ed.) Advances in Porous Media, 2. Elsevier,
tion is the only mechanism that enhances the alignment of Amsterdam, 169–308.
framework grains for a given clay/quartz ratio. Pore geometry Cheng, C.H. 1978. Seismic velocities in porous rocks: Direct and inverse problems. PhD
thesis. Massachusetts Institute of Technology, Cambridge, Massachusetts,
is considered constant with depth, and does not influence USA.
grain nor cement alignment in the cementation model. But Dvorkin, J. & Nur, A. 1996. Elasticity of high porosity sandstones: Theory
cementation can influence the alignment in theory; if the for two North Sea data sets. Geophysics, 61, 1363–1370.
dissolved materials are not aligned and precipitate on aligned Dvorkin, J., Mavko, G. & Hezhu, Y. 1994. Effective properties of cemented
granular materials. Mechanics of Materials, 18, 351–366.
grains, or vice versa, the cementation will affect the alignment. Dvorkin, J., Berryman, J. & Nur, A. 1999. Elastic moduli of cemented sphere
This effect is assumed to be very small, and negligible for the packs. Mechanics of Materials, 31, 461–469.
results. Han, D., Nur, A. & Morgan, D. 1986. Effects of porosity and clay content on
Shales that are severely overpressured have anomalously high wave velocities in sandstones. Geophysics, 51, 2093–2105.
Hill, R. 1952. The elastic behavior of crystalline aggregate. Proceedings of the
porosities when entering the TZ. They need more cement to Physical Society of London, A65, 349–354.
stabilize the grain contacts than shales with lower porosities. Hornby, B.E., Schwartz, L.M. & Hudson, J. 1994. Anisotropic effective-
This increases the TZ interval for the high porosity over- medium modelling of the elastic properties of shales. Geophysics, 59,
pressured shale. The overpressure will not directly affect the 1570–1583.
precipitation rate, since cementation is practically independent Jakobsen, M., Hudson, J.A., Minshull, T.A. & Singh, S.C. 2000. Elastic
properties of hydrate-bearing sediments using effective medium theory.
of pressure (Bjørkum & Nadeau 1998). But uncemented Journal of Geophysical Research, 105 (B1), 561–577.
fractures that result from overpressure will entail lower Jakobsen, M., Johansen, T.A. & Ruud, B.O. 2001. Modelled velocity
stiffnesses in the rock. The model does not include fracture and reflectivity properties of anisotropic hydrated sediments. Journal of
effects and will thus predict too high rock stiffnesses where Computational Acoustics, 9, 1507–1522.
Johansen, T.A., Drottning, A., Lecomte, I. & Gjøystdal, H. 2002. An
these occur. approach to combined rock physics and seismic modelling of fluid
substitution. Geophysical Prospecting, 50, 119–137.
Johansen, T.A., Ruud, B.O. & Jakobsen, M. 2004. Effects of grain scale
CONCLUSIONS alignment on seismic anisotropy and reflectivity of shales. Geophysical
A strategy has been presented to predict velocities and Prospecting, 52, 133–149.
anisotropy for shales and silty shales from shallow to deep Kuster, G.T. & Toksöz, M.N. 1974. Velocity and attenuation of seismic
waves in two-phase media: Part I. Theoretical formulations. Geophysics, 39,
burial, in a basin with a given thermal history. Since the 587–606.
diagenetic processes vary according to burial depth and thermal Morris, P.R. 1969. Averaging fourth-rank tensors with weight functions.
history, the rock physics models must account for both Journal of Applied Physics, 40, 447–448.
mechanical and chemical compaction. A rock physics model Nadeau, P.H., Peacor, D.R., Yan, J. & Hiller, S. 2002. I/S precipitation in
pore space as the cause of geopressuring in Mesozoic mudstones,
that predicts effective properties of cemented shales has been Egersund Basin, Norwegian Continental Shelf. American Mineralogist, 87,
suggested. The model is combined with an existing model for 1580–1589.
uncemented shales to cover the whole range of diagenetic Nishizawa, O. 1982. Seismic velocity anisotropy in a medium containing
changes from shallow to deep burial/high temperature. oriented cracks – transversely isotropic case. Journal of Physics of the Earth,
30, 331–347.
During burial, the rock enters a depth/temperature interval Ramm, M. & Bjørlykke, K. 1994. Porosity/depth trends in reservoir
(transition zone) where both chemical and mechanical compac- sandstones: Assessing the quantitative effects of varying pore-pressure,
tion are active. Here, the shale compaction model and the shale temperature history and mineralogy, Norwegian Shelf data. Clay minerals,
cementation model are combined to produce effective rock 29, 475–490.
properties in the transition zone. Ruud, B.O., Jakobsen, M. & Johansen, T.A. 2003. Seismic properties of shales
during compaction. 73rd SEG Meeting, Expanded abstracts, Dallas, Texas,
The model showed consistent results when comparing 1294–1297.
vertical P- and S-wave velocities with logs from three wells. Sayers, C.M. 1994. The elastic anisotropy of shales. Journal of Geophysical
However, there are is no data control to evaluate the Research, 99 (B1), 767–774.
anisotropy–depth predictions. Results from the model can be Tosaya, C.A. 1982. Acoustical properties of clay-bearing rocks. PhD thesis. Stanford
applied readily in AVO and velocity analysis. The proposed University, USA.
strategy for estimating stiffnesses in the transition zone can be Voigt, W. 1928. Lehrbuch der Kristallphysik. Teubner, Leipzig.
Walderhaug, O. 1996. Kinetic modelling of quartz cementation and porosity
adopted easily in rock physics modelling of sandstones. loss in deeply buried sandstone reservoirs. AAPG Bulletin, 80, 731–745.
Willis, J.R. 1977. Bounds and self-consistent estimates for the overall
AD would like to thank The Norwegian Academy of Science and properties of anisotropic composites. Journal of the Mechanics and Physics of
Letters for financial support. Solids, 25, 185–202.

Received 25 April 2005; revised typescript accepted 30 September 2005.