Sunteți pe pagina 1din 10

Downloaded from http://pg.lyellcollection.

org/ at Stanford University on July 18, 2016

Definition and practical application of mudstone porosityeective stress


relationships
Yunlai Yang and Andrew C. Aplin
NRG, School of Civil Engineering and Geosciences, University of Newcastle, Newcastle upon Tyne NE1 7RU, UK

ABSTRACT: The relationship developed in soil mechanics between void ratio and
vertical effective stress is a simple but practical way of describing the one-
dimensional, mechanical compaction of fine-grained, clastic mudstones. The com-
pression coefficients (e100 and ) that define this relationship are strongly influenced
by grain size, which can be simply described by the sediments clay content. In this
paper, data both from the soil mechanics literature and from geological samples
from the North Sea and Gulf of Mexico are used to construct the relationship
between clay content and compression coefficients. The two datasets yield different
values for the coefficients, but the authors believe that the coefficients derived from
the large geological database should be used to describe geological compaction.
Regression of the geological data generates the following relationships between clay
content and compression coefficients:

e100 = 0.3024 + 1.6867clay + 1.9505clay2

 = 0.0407 + 0.2479clay + 0.3684clay2

There is excellent agreement between true porosities and porosities calculated


using the standard soil mechanics compaction equation with these coefficients. This
indicates that the mechanical compaction of natural mudstones to 40 MPa can be
adequately described using the soil mechanics approach. Practical application of the
work is a two-stage process which first involves the evaluation of clay content and,
thus, compression coefficients directly from wireline log data. These data can then be
used to (a) help define effective stressporosity inputs for basin models and (b)
estimate mudstone pore pressures. Examples of pore pressure estimates are shown
from west Africa, the North Sea and the Gulf of Mexico.
KEYWORDS: mud, mudstone, compaction, eective stress, pore pressure

INTRODUCTION e=  (3)
Mechanical compaction is an inevitable consequence of burial 1
and basin evolution. As such, a significant effort has been made .
to quantify the process and to incorporate a mathematical In these equations,  is porosity, e100 is the void ratio at
description within computer-based basin models (e.g. Athy 100 kPa effective stress and  is the slope of the linear relation
1930; Smith 1971; Sharp & Domenico 1976; Yukler et al. 1978;
between void ratio and the natural logarithm of vertical effective
Bethke 1985; Lerche 1990; Ungerer et al. 1990; Luo & Vasseur
1992; Hermanrud 1993; Schneider et al. 1993; Audet 1996). This stress. Effective stress ( v) is defined as the difference between
paper focuses on the mechanical compaction of fine-grained, total stress (v) and pore fluid pressure (u). The form of equation
clastic sediments: muds and mudstones (hereafter referred to (1) is such that void ratio is a linear function of the logarithmic
collectively as mudstones). A variety of empirical equations value of effective stress. This is shown in Figure 1, which
have been developed to describe the mechanical compaction of indicates the virgin compression line along which initial compac-
mudstones, one of which is the simple equation which is based tion takes place. Figure 1 also shows that because mudstone
on the effective stress principle established in soil mechanics deformation is a predominantly plastic process, the major part of
(e.g. Terzhagi 1943; Skempton 1970; Burland 1990): the deformation is not recovered when sediments are unloaded. If
an unloaded mudstone is reloaded, the sediment returns to the

e = e100  ln ~ !
 v
100
(1)
virgin compaction line, after which further increases in effective
stress drives the mudstone along the virgin line.
It is also important to note that equations (1) and (2) can be
 v = v  u (2) used to estimate pore pressure from mudstone porosity, if the
Petroleum Geoscience, Vol. 10 2004, pp. 153162 1354-0793/04/$15.00  2004 EAGE/Geological Society of London
Downloaded from http://pg.lyellcollection.org/ at Stanford University on July 18, 2016

154 Y. Yang & A. C. Aplin

Fig. 1. Soil mechanics-based description of the one-dimensional,


mechanical compaction of fine-grained clastic sediments.

compaction is on the virgin compression line and the values of Fig. 2. Relationship between clay content and void ratio at liquid
the compression coefficients in equation (1) are known (Alixant limit for fine-grained clastic sediments based on the datasets from
& Desbrandes 1991). This is useful since the low permeability this study and Skempton (1944).
of mudstones means that pore pressures cannot be measured
using conventional techniques such as the Repeat Formation
Tester (RFT).
The geological application of the porosityeffective relation-
ships encapsulated in equation (1) requires the resolution of
two problems. First, because the data used to develop equation
(1) are from low (<5 MPa) environments (Skempton 1970;
Burland 1990), one needs to establish the extent to which the
approach can be extended to the higher stress levels typical of
sedimentary basins (up to 40 MPa). The second and more
serious problem is how to derive the compression coefficients
e100 and  for extensive and often lithologically variable
packages of fine-grained clastic sediments. In civil engineering,
the compression coefficients can be readily measured on small
numbers of samples. This is impossible in sedimentary basins,
where fine-grained clastic sediments comprise 6070% of basin
fill.
The aims of this paper are, thus, to (a) explore the extent to Fig. 3. Particle size data for three North Sea wells, showing the good
which equation (1) can be used to describe the one- correlation between clay content (% particles <2 m diameter) and
dimensional, mechanical compaction of mudstones in sedimen- the percentage of particles <0.1 m and <10 m. These data suggest
tary basins and (b) determine the numerical values of the that clay content is a reasonable measure of the overall grain size
distribution of mudstones.
compression coefficients e100 and , based on a quantitative
description of mudstone lithology. Data from the soil mech-
anics literature suggest that compression coefficients are Mexico. Finally, it is shown how the constructed relationships
strongly correlated to the void ratio at liquid limit (Skempton can be used to evaluate overpressure directly from wireline logs.
1970; Burland 1990; Chandler 2000). Void ratio at liquid limit is It is important to bear in mind that relationships describing
not a geologically useful description of sediment lithology, but the change in porosity as a function of effective stress are only
it is known to be strongly related to the clay content of valid for mechanical compaction. At higher temperatures
sediments (Skempton 1944; Figure 2), where clay content is (above approximately 80100 C) the recrystallization of clay
defined as the percentage of particles less than 2 m in minerals such as smectite changes the microfabric of mud-
diameter. In addition, clay content appears to be a simple but stones (Ho et al. 1999; Aplin et al. 2003; Charpentier et al. 2003)
reasonably robust indicator of the overall grain size distribution and may lead to a loss of porosity which is independent of
of mudstones (Fig. 3). The approach here is, thus, to correlate effective stress (see also Bjrlykke 1998, 1999; Nadeau et al.
the compression coefficients e100 and  with clay content, which 2002). The aim of this paper is to construct relationships
can be either measured (e.g. Yang & Aplin 1997) or, pragmati- describing the mechanical compaction of mudstones and, to
cally, estimated from geophysical wireline logs (Yang et al. this end, the main samples used are ones that it is believed have
2004). not been significantly affected by chemical diagenesis.
In this paper, data both from the soil mechanics literature
and from geological samples are used to construct the relation-
ship between clay content and compression coefficients. First, RELATIONSHIP BETWEEN COMPRESSION
well-constrained soil mechanics data are used to establish the COEFFICIENTS AND CLAY CONTENT:
general format of the relationship. Most of these data cover low DERIVATION FROM SOIL MECHANICS
stress levels. Secondly, therefore, the coefficients are estimated LITERATURE
from a much larger dataset, covering a much larger stress range The compaction of fine-grained sediments has been extensively
(0.840 MPa) and collected from in situ, geologically compacted studied by the soil mechanics and geotechnical engineering
fine-grained clastic sediments from the North Sea and Gulf of communities. Important datasets have been presented by
Downloaded from http://pg.lyellcollection.org/ at Stanford University on July 18, 2016

Mechanical compaction of mudstones 155

Skempton (1970) and Burland (1990). However, these data are


not in a format that can be immediately used in a geological
context. First, many of the measurements were made on
remoulded sediments and, second, the sediments are described
in terms of void ratio at liquid limit (eL), rather than a more
geologically tangible description such as grain size or clay
content. The data must, therefore, be manipulated in order to
generate the required relationships between clay content and
compression coefficients for natural sediments.
Burlands (1990) key paper presents two datasets; one,
originating from Skemptons (1970) work, for natural sediments
and a second comprising 26 sediments which were reconsti-
tuted at water contents between 1 and 1.5 times that of the
liquid limit. Burland used the latter dataset to define what he
called the intrinsic compression properties of the sediments,
which are independent of the natural, depositional state of the
sediments. These were defined, as a function of eL, by constants
*
of intrinsic compressibility (e100 and * or C*c ):
*
e100 = 0.109 + 0.679eL2  0.89eL + 0.016eL3 r = 0.991 (4)

C*c = 0.256eL  0.04 r = 0.985 (5)


where C*c is the slope of the void ratioeffective stress relation-
Fig. 4. Relationship between the intrinsic compression line (ICL)
ship if effective stress is expressed in terms of log base 10. Cc is and sedimentation compression line (SCL) (from Burland 1990). The
related to  by C*c =*ln~10!, so that: ICL defines the compaction behaviour of reconstituted muds
whereas the SCL describes the compaction of natural muds.
* = 0.11eL  0.0174 (6)
*
e100 can also be correlated linearly with eL:
*
e100 = e100 + 1.836* (11)
*
e100 = 0.1387 + 0.5616eL r = 0.989 (7)
Burland (1990) normalized the intrinsic compression data by and
defining the void index (Ivo):
 = 1.088* (12)
e  e100
*
Ivo = (8)
C*c Equations (11) and (12) provide the link between the
As shown in Figure 4, which is reproduced from Burlands compaction of reconstituted and natural fine-grained, clastic
paper, the normalized compaction curves for the reconstituted sediments. However, the compression coefficients are related to
sediments form a quite tightly defined line known as the eL and not to clay content. The relationship between eL and clay
intrinsic compression line (ICL). However, using equations content can be derived from the dataset shown in Figure 2:
(4), (5) and (8) to calculate the void indices of the naturally
compacted sediments reported by Skempton (1970), Burland eL = 0.723 + 0.775clay + 3.174clay2 r = 0.987 (13)
(1990) showed that the compaction behaviour of reconstituted
sediments was distinct from that of natural sediments. The Combining equations (6), (7), (11), (12) and (13) one gets:
resulting sediment compression line (SCL) is also shown in
Figure 4. It is sub-parallel to the ICL and is roughly linear at
 >10 kPa (burial depth of c. 2 m), giving a linear regression e100 = 0.659 + 0.592clay + 2.424clay2 (14)
line of:
and
Ivo = 0.7974  0.4726ln 
100 ~ ! (9)
 = 0.0686 + 0.0937clay + 0.384clay2 (15)
By combining equations (8) and (9), the following relation-
ship can be derived that relates the void ratio and effective These are the relationships between compression coefficients
stress of natural sediments using the constants of intrinsic and clay content for natural, fine-grained clastic sediments
compressibility: derived from the Skempton (1970) and Burland (1990) datasets.
They are based on a relatively small number of data points for
*
e = e100 + 1.836*  1.088* ln 
100~ ! (10) sediments that have been generally subjected to stress levels less
than 5 MPa, corresponding to only a few hundred metres burial
This has the same structure as the fundamental compaction depth. The next section determines how these relationships
equation of soil mechanics (equation (1)). Combining the compare to those derived from data measured on natural
coefficients in equations (1) and (10) gives: mudstones in sedimentary basins.
Downloaded from http://pg.lyellcollection.org/ at Stanford University on July 18, 2016

156 Y. Yang & A. C. Aplin

RELATIONSHIP BETWEEN COMPRESSION of 8003500 m. Each data point represents 1 m of mud-


COEFFICIENTS AND CLAY CONTENT: stone.
DERIVATION FROM NATURAL MUDSTONES + Type 3 data were collected from natural samples. The
In this section, the relationship between , e100 and clay content dataset comprises about 170 samples from 18 wells, includ-
ing 60 from a single Gulf of Mexico well. The samples are
is constructed using data from 21 North Sea wells and one well
mainly cuttings, with some sidewall cores and conventional
from the Gulf of Mexico. This requires clay content, void ratio
cores.
and effective stress data, all of which are subject to uncertainty.
The strategy has been to construct relationships based on a very Particle size distributions (including clay content) and void
large dataset, comprising (1) a laboratory dataset of around 200 ratio for Type 3 samples were measured in the laboratory (Yang
samples for which the clay content and porosity were measured & Aplin 1997). Samples were gently disaggregated using a
and (2) a dataset of over 3600 samples for which the clay freezethaw technique (Yang & Aplin 1997) prior to particle
content and porosity were estimated from wireline logs using size measurement. The particle size was measured using either
calibrated artificial neural network (ANN) models (Yang et al. the sedimentation pipette method (British Standard 1377:
2004). The models predict clay content to 10%. Effective British Standards Institution 1990) or a Micromeritics Sedi-
graph 5000ET particle size analyser, which had been calibrated
stress was estimated using equation (2) and total stress is
against the standard pipette method. Void ratios were calcu-
evaluated from the density log (Aplin et al. 1995).
lated from measured grain densities (Gs) and dry bulk densities
As the focus here is on the purely mechanical compaction of (d):
non-source rocks, samples with more than 5% TOC and
samples which were lithified either as a result of carbonate Gs
cementation or clay mineral recrystallization were excluded. A e= 1 (17)
d
parameter, Cement Factor was developed to indicate the
degree of lithification. Cement factor is a qualitative measure of The grain densities were measured by the small pyknometer
sediment strength and is based on the empirical observation method (British Standard 1377: British Standards Institution
that, when disaggregating mudstones with a gentle freezethaw 1990). Dry bulk densities were calculated from dry weights and
method (Yang & Aplin 1997), some samples could not be volumes of samples. The bulk volume of a sample was
disaggregated. ANN models (Yang et al. 2004) were constructed measured using a mercury porosimeter.
in which wireline log data were related to the qualitative Effective stress was evaluated from RFT data in adjacent
strength of the samples. In the construction of the model the sands (as for Type 2 data). Laboratory-measured void ratios are
samples that could not be disaggregated were assigned a often different to those derived from the density log (Fig. 5).
Cement Factor of 1. Samples which were easily disaggregated Differences between log porosity and measured porosity may
were assigned a Cement Factor of 0. The output cement factor be due to expansion as a result of unloading and/or shrinkage
of the model is between 0 and 1. Samples with cement factors resulting from drying of samples at 105 C for the purpose of
greater than 0.1 were excluded from the dataset used here to measuring dry bulk densities. Measured porosities were, there-
construct the porosityeffective stress relationships in this fore, corrected to log porosity using relationships derived
paper. between log-derived and measured porosities:
Depending on the method used to evaluate pore pressure, For North Sea samples:
void ratio and clay content, the data are classified into three
types. log = meas. for meas.c0.2
+ Type 1 data were collected from shallow (<700 m) buried log =  0.2 + 1.071meas. + 0.291clay0.5/ln~h!
sections, with large proportions of sand. Due to the nature  0.022clay0.5meas. for meas.>0.2 (18)
of shallow burial and the predominance of coarse sediments,
pore pressures were assumed to be hydrostatic and effective For Gulf of Mexico samples:
stress was evaluated from equation (2) using the density log
to determine total vertical stress. Clay contents and grain log =  0.267 + 3.2meas. + 0.195clay0.5/ln~h!
densities were determined from wireline logs using ANN  0.218clay0.5meas. (19)
models (Yang et al. 2004) and void ratios (e) were calculated
using the density log and the grain density: where log is logged porosity, meas. is laboratory-measured
porosity, clay is clay content (%) and h is burial depth (m). The
Gs   comparison of measured porosity and corrected porosity is
e= (16)
  Gw shown in Figure 5.
In summary, 3847 data points were collected from 22 wells.
where Gs is grain density (g cm3),  is bulk density The data have the following ranges clay content: 890%;
(g cm3) and Gw is formation water density (g cm3). In effective stress: 0.840 MPa, corresponding to burial depths of
total, 1977 data points were collected from five North Sea 100 m to c. 3500 m; void ratio: 0.071.2 (porosity: 0.060.55);
wells. Each data point represents 1 m of mudstone. total organic carbon content: less than 5% for all samples and
+ Type 2 data are similar to Type 1 data except that pore less than 1% for most of the samples.
pressures were interpolated or extrapolated from RFT press- The complete dataset is shown in Figure 6 and illustrates the
ures measured in adjacent sand units. Pore pressures were wide range of void ratio (or porosity) at a single effective stress.
only inferred for mudstones within 20 m of RFT data and At 5 MPa, for example, the void ratio varies from 0.33 to 1.1,
no interpolation was carried out across very fine-grained equivalent to a porosity range of 0.250.52.
units, since these tend to be of especially low permeability Analysis of the compaction data in Burland (1990) (as
and may exert a significant influence on pore pressure represented by equations (14) and (15) showed that the rela-
profiles (Yang & Aplin 1998). This group comprises 1702 tionships between clay content and compression coefficients
data points from 11 North Sea wells covering burial depths have the general form:
Downloaded from http://pg.lyellcollection.org/ at Stanford University on July 18, 2016

Mechanical compaction of mudstones 157

Fig. 7. Comparison of laboratory measured and log-derived void


ratios with void ratios modelled using the equations derived in this
paper.

Substituting equations (20) and (21) into the basic compac-


tion model (equation (1)), obtains:

~ !
e = a + b clay + c clay2  ~f + g clay + h clay2!ln  (22)
100

The constants in equation (22) were derived by regression of


equation (22) using the dataset:

e100 = 0.3024 + 1.6867clay + 1.9505clay2 (23)


Fig. 5. Laboratory-measured porosity versus porosity derived from
wireline logs for (a) North Sea and (b) Gulf of Mexico mudstones.  = 0.0407 + 0.2479clay + 0.3684clay2 (24)
Differences between log porosity and measured porosity may be due
to expansion as a result of unloading and/or shrinkage resulting Figure 7 shows the excellent agreement between actual and
from drying of samples at 105 C for the purpose of measuring dry modelled void ratios. Most of the modelled values are within
densities. Measured porosities (filled circles) were, therefore, cor- 0.1 of the true values. The very high correlation coefficient
rected using equation (18) in order to bring them into line with the (r2=0.92) indicates that the compaction of natural mudstones to
log-derived data. Corrected data are shown as open circles. 40 MPa is adequately described by equation (1) and that
equations (23) and (24) accurately represent the relationship
between clay content and compression coefficients. Equations
(23) and (24) are, thus, critical: they show that if one can
determine the clay content of a fine-grained clastic sediment,
the mechanical compaction curve (relationship between void
ratio and effective stress) can be defined accurately.
Figure 8 shows the modelled and measured void ratio-
effective stress relationships for four groups of fine-grained
clastic sediment with different clay contents. Scatter around the
line reflects both uncertainty in the model and the fact that the
model lines represent a single value of clay content whereas
the data points have a 10% range of clay contents.
Figure 9 compares the compression curves derived from the
BurlandSkempton data (equations (14) and (15)) and from the
data presented in this study (equations (23) and (24)). At
stresses below 2 MPa and particularly for more fine-grained
material, the compaction trends derived from the two datasets
Fig. 6. Dataset used in the construction of the relationships between are similar. In general, however, the relationships are signifi-
clay content and compression coefficients for geologically com- cantly different; the data from this paper give lower porosities
pacted sediments. at higher effective stresses. Similarities and differences between
the two compaction trends reflect the nature of the two
datasets: the Skempton (1970) and Burland (1990) datasets are
e100 = a + bclay + cclay2 (20) dominated by low effective stress and high clay content
samples, whilst the current dataset effectively covers the full
 = f + gclay + hclay2 (21) range of clay content and much larger range of effective stress.
Since the study here has been very careful to exclude chemically
where a, b, c, f, g, h are constants to be determined. compacted sediments from the dataset, the differences between
Downloaded from http://pg.lyellcollection.org/ at Stanford University on July 18, 2016

158 Y. Yang & A. C. Aplin

Fig. 8. Void ratioeffective stress data


compared to the model relationship
derived in this paper (equation (22)).
Data in each figure are for samples
with ranges of clay content:
(a) 2535%; (b) 3545%; (c) 4555%;
(d) 5565%. Model curves are for
specific clay contents.
the two datasets do not reflect fundamental differences in the effective stress relationships developed in this paper (equations
nature of the compaction process in shallow and deep-burial (1), (23) and (24)). Although several other porosity or wireline-
settings. based approaches have been applied to the estimation of pore
Compression curves derived from the current dataset (Figs 8 pressure in mudstones (e.g. Hottman & Johnson 1965; Eaton
and 9) show that the finer the material, the higher is its initial 1975), the approach here differs in that it is based on an explicit
porosity but also the greater the slope (); finer-grained mud- physical principle and does not require local calibration to
stones, thus, lose porosity faster with increasing effective stress shallow and putatively hydrostatically pressured sections.
than do coarser-grained mudstones. As a result, the porosities The basic compaction equation (1) can be rearranged such
of sediments with different clay contents converge at high levels that effective stress can be evaluated if the void ratio (e) is
of effective stress. known and the compression coefficients e100 and  can be
EVALUATION OF PORE PRESSURE FROM WELL estimated from the clay content of the mudstone:
LOGS
e
This section shows how mudstone pore pressures can be
estimated directly from well log data using the porosity-
 = 100exp ~e
100

! (25)

Pore pressure is then the difference between total (vertical)


stress and (vertical) effective stress.
Compression coefficients are calculated from equations (23)
and (24) as a function of clay content, which can be estimated
from wireline log data (Yang et al. 2004). Total stress can be
calculated by integration of the density log, but very rarely does
the logging run start at the sedimentwater interface. The total
stress at the depth at which the density log run starts can either
be estimated using the method introduced by Aplin et al. (1995)
for normally pressured sections, or by using average, regional
stress trends. Here, using Aplin et al.s (1995) method, the total
stress trends for both the Gulf of Mexico and North Sea have
been established using data from ten North Sea and six Gulf of
Mexico wells:

v = 0.01799h + 9.9510  7h2 ~North Sea! (26)

v = 0.006074h + 4.4410  8h2 ~Gulf of Mexico! (27)

where h is true vertical depth from sea floor (metres for North
Sea, feet for Gulf of Mexico) and v is total vertical stress from
the sea floor (MPa). These equations are used to calculate the
vertical stress to the depth at which the density log starts its
Fig. 9. Comparison between the normal compaction trends derived
run.
from mudstone compaction data published by Burland (1990) Where possible, void ratios were calculated from the density
(equations (14) and (15)) and data from this study (equations (23) and log, using grain densities estimated from wireline logs using
(24)). ANN models (Yang et al. 2004). Where the density log was
Downloaded from http://pg.lyellcollection.org/ at Stanford University on July 18, 2016

Mechanical compaction of mudstones 159

Fig. 10. Estimation of pore pressure in


a North Sea well using the approach
developed in this paper. Cement factor,
lithology and clay content were
evaluated from wireline logs using
artificial neural networks (Yang et al.
2004). Cement factor is an indication
that mudstones are lithified, as a result
of either carbonate cementation or clay
mineral recrystallization. A lithology
value of one indicates the sediment is
predominantly carbonate. Porosity
estimates were made from either the
density log or the sonic log. The
modelled normal compaction curve
(thinner line) represents the porosity the
sediment would have if it were
normally pressured. True vertical depths
below the seabed are shown at the top
and bottom of the depth axis. Pore
pressures determined with the Repeat
Formation Tester are shown as circles.

Fig. 11. Estimation of pore pressure in


a North Sea well using the approach
developed in this paper. Cement factor
and clay content were evaluated from
wireline logs using artificial neural
networks (Yang et al. 2004). Cement
factor is an indication that mudstones
are lithified, as a result of either
carbonate cementation or clay mineral
recrystallization. A lithology value of
one indicates the sediment is
predominantly carbonate. Porosity
estimates were made from either the
density log or the sonic log. The
modelled normal compaction curve
(thinner line) represents the porosity the
sediment would have if it were
normally pressured. True vertical depths
below the seabed are shown at the top
and bottom of the depth axis. Pore
pressures determined with the Repeat
Formation Tester are shown as circles.

unavailable or of poor quality, void ratio was calculated from be lower and will only reflect the pore pressure resulting from
the sonic log, using Raiga-Clemenceau et al.s (1988) equation: mechanical compaction. Further, the approach is not applicable
to quantifying pore pressure in uplifted sediments, since the
c
=1 ~tt !
m
(28)
effective stress and pore pressure will have evolved in a
potentially complicated and unpredictable way. On the other
hand, the technique might supply some information on the
where tm is the matrix transit time (s ft1), t is the sonic amount of uplift based on the evaluated maximum effective
transit time (s ft1) and c is a coefficient. In this work, stress the sediments experienced prior to uplift.
equation (28) was calibrated for each well using porosities Figures 10, 11, 12 and 14 show four examples of pore
derived from the density log over well sections where the pressure profiles estimated directly from wireline logs using the
density log was clearly of high quality. The void ratiodepth procedure outlined above. The vast majority of these predic-
profiles shown in Figures 1013 are, thus, a mixture of the tions are for mudstones which were not used in the construc-
values derived from both the density and sonic logs. tion of the porosityeffective stress relationships developed
The essentially inelastic nature of mudstones means that here and, thus, represent a true test of the method. Each figure
porosity is only slightly restored in situations where effective shows (a) Cement factor and lithology type (0=non-carbonate,
stress decreases (Fig. 1). Equations (1) and (25) only describe 1=carbonate) evaluated using the constructed ANN models
virgin compaction and the effective stress estimated in this way (Yang et al. 2004), (b) the clay content derived from wireline
is the maximum that the sediment has ever experienced. If logs using calibrated ANN models (Yang et al. 2004), (c) log
the sediment is currently at its maximum effective stress, the porosity and the porosity the mudstones would have were they
estimated pore pressure should be correct, assuming that the hydrostatically pressured (normal compaction trend) and (d) the
model appropriately describes the compaction process. How- pore pressure estimated from the effective stressporosity
ever, if the pore pressure in the sediment has increased as a relationships outlined in this paper. Sands are seen on the clay
result of processes other than mechanical compaction, such as content logs as zones of very low clay content. Since the ANNs
lateral transfer (Yardley & Swarbrick 2000) or the cracking of were calibrated only with mudstone data, these apparent clay
oil to gas, then the pore pressure evaluated by this method will contents have no meaning. Equally, pore pressures calculated
Downloaded from http://pg.lyellcollection.org/ at Stanford University on July 18, 2016

160 Y. Yang & A. C. Aplin

Fig. 12. Estimation of pore pressure in


a Gulf of Mexico well using the
approach developed in this paper. Clay
contents were evaluated from wireline
logs using artificial neural networks
(Yang et al. 2004). Neither carbonates
nor carbonate-cemented or lithified
sediments were found in this section.
Porosity estimates were made from
either the density log or the sonic log.
The modelled normal compaction curve
(thinner line) represents the porosity the
sediment would have if it were
normally pressured. True vertical depths
below the seabed are shown at the top
and bottom of the depth axis. Pore
pressures determined with the Repeat
Formation Tester are shown as circles.
Pore pressures estimated from a
cross-plot of laboratory measured
permeability and effective stress are
shown as squares (see Fig. 13 for an
explanation of this technique) and are
lower than the RFT pore pressures. In
the upper part of the section, RFT pore
pressures are generally higher than pore
pressures estimated from mudstone
porosities. This could be due to natural
unloading of the sediments or could be
an indication that the RFT data are not
a true measure of pore pressure but
have been inflated by the pressure
exerted by the drilling mud
(supercharged).

for the sands using a mudstone compaction model have no are very similar to those measured in the sand units around
meaning and, thus, are not shown in Figures 1013. 2500 m. The sands are likely to be acting as a fluid drain for the
Figure 10 shows data for a North Sea well comprising a overlying mudstones.
relatively thick Tertiary mudstone section. Interspersed sands Figure 11 shows data for another North Sea well, in which a
and carbonates are seen by the logs as low porosity/low clay mud-rich Tertiary section overlies around 500 m of Cretaceous,
content layers, results which have no physical meaning. The Shetland Group mudstones. The log quality in the upper
majority of the sequence appears to be significantly overpres- Tertiary section is rather poor, but sands are generally revealed
sured, increasing downwards to about 1950 m. This is consist- by their relatively low clay contents. Below 2500 m, the
ent with the thick layer, between 1520 m and 1950 m, of sediments are dominated by Jurassic and Triassic sandstones,
extremely fine mudstones with clay contents between 70% and with many carbonate cemented layers. As in the previous
82%. Finer-grained mudstones have especially low permeabili- example, overpressures are highest in the fine-grained, lower
ties and are, thus, particularly susceptible to retaining high fluid Tertiary section. The most striking feature of these data is that
pressures over geological time (Yang & Aplin 1998; Dewhurst the modelled pore pressure throughout the Cretaceous mud-
et al. 1999). Below this unit, modelled overpressures decline and stone section is close to hydrostatic and lies on the hydrostatic
line in the 150 m section which directly overlies the reservoir
units. In stark contrast, the underlying sands are highly over-
pressured. If the effective stress in the mudstones was calcu-
lated using the current overburden stress and the pore pressure
measured in the sands, porosities would be expected to be
around 30%, compared to the actual values of 20%. The
implication of these data is that the mudstones, and presumably
the underlying sandstones, were fully compacted and, thus,
hydrostatically pressured until geologically recent times, at
which point the sands became overpressured by a mechanism
other than local mechanical compaction (e.g. lateral transfer
from the basin centre; Yardley & Swarbrick 2000). The reasons
for this dramatic change in pore pressure and the implications
for the petroleum system in terms of fluid flow and
potential seal failure remain areas for future work.
Figure 12 is a mud-rich Miocene section from the deep-water
Gulf of Mexico. The well is deviated so that the hydrostatic
Fig. 13. Estimation of maximum effective stress using permeability pressure is not linear with KB depth. Sand layers are indicated
data measured in the laboratory at a range of effective stresses. The
slope of the relationship changes as the sample is taken beyond the by their lower clay contents and higher logged porosities. In the
maximum effective stress it has experienced naturally. This is one of lower part of the section, the modelled pore pressures are very
the two samples from the Gulf of Mexico well shown in Figure 12. close to the measured pressures in adjacent sands. In contrast,
Downloaded from http://pg.lyellcollection.org/ at Stanford University on July 18, 2016

Mechanical compaction of mudstones 161

Fig. 14. Estimation of pore pressure in


a west African well using the approach
developed in this paper. Cement factor
and clay content were evaluated from
wireline logs using artificial neural
networks (Yang et al. 2004). Cement
factor is an indication that mudstones
are lithified, as a result of either
carbonate cementation or clay mineral
recrystallization. A lithology value of
one indicates the sediment is
predominantly carbonate.
Laboratory-measured clay contents are
shown as open circles. Porosity
estimates were made from either the
density log or the sonic log. The
modelled normal compaction curve
(thinner line) represents the porosity the
sediment would have if it were
normally pressured. True vertical depths
below the seabed are shown at the top
and bottom of the depth axis. Pore
pressures determined with the Repeat
Formation Tester are shown as circles.

the measured pressures in the upper section are higher than the and vertical effective stress. The compression coefficients e100
modelled pressures. There are three potential explanations for and  are strongly dependent on clay content, relationships
this apparent discrepancy. First, the RFT data are too high, as which have been quantified here using both data for natural
a result of supercharging during the test. This is difficult to mudstones from the North Sea and Gulf of Mexico, and data
confirm but it is notable that the RFT pressures are close to the for both natural and remoulded muds compacted to low levels
pressure exerted by the drilling mud. Secondly, the porosity of effective stress (Skempton 1970; Burland 1990).
effective stress model here is wrong and is underestimating the For practical purposes, the compression coefficients can be
pore pressure in this region. Third, and as in the second North evaluated from clay content data derived from wireline logs
Sea example, the sediments have been unloaded through an (Yang et al. 2004). The data can then be used both to define
increase in pore pressure due to a mechanism other than that more accurately the effective stress-porosity inputs for basin
related to local mechanical compaction. Some independent models and, using an explicit model, to estimate mudstone pore
evidence suggests that this is the most likely explanation. There pressures directly from both conventional wireline log data and
are two samples from this part of the section for which there data collected during drilling. Acquisition of data whilst drilling
are a series of laboratory permeability data measured at a range opens up the exciting possibility of real time estimation of
of effective stresses both below and above the current, in situ mudstone pore pressure.
value. When plotted on a log scale, permeabilityeffective stress
data define two lines of different slope, with the break in slope The work presented in this paper was supported by a NERC ROPA
representing the maximum effective stress to which the sample award, the EU FP4 and FP5 programmes, Norsk Hydro, BP and the
has been subjected in its geological history (Fig. 13). The GeoPOP consortium, comprising Amerada Hess, BG, BP, Chevron-
Texaco, ConocoPhillips, ExxonMobil, JNOC, Norsk Hydro, Shell,
maximum effective stress estimated by this technique is close to Statoil and Total, and precursors of those companies. Norsk Hydro
that suggested by the porosity data (Fig. 12), suggesting that the and BP kindly provided data and samples. The authors also thank the
pore pressures in the sands have, indeed, been inflated. reviewers for their positive and thoughtful critiques.
Figure 14 is for a well from offshore West Africa, an area
from which there is no calibration data for either the ANN REFERENCES
models or the compaction models. This well is, therefore, a test
for both models. Six major sandstone layers are indicated by Alixant, J.-L. & Desbrandes, R. 1991. Explicit pore pressure evaluation:
their low clay contents and modelled clay contents agree well concept and application. SPE Drilling Engineering, 6, 182188.
with the measured values (Fig. 14). Modelled pore pressures for Aplin, A.C., Yang, Y.L. & Hansen, S. 1995. Assessment of , the compres-
sion coefficient of mudstones and its relationship to detailed lithology.
mudstones suggest a generally hydrostatic upper section devel- Marine and Petroleum Geology, 12, 955963.
oping modest overpressure below 2.8 km. Below 2.8 km, the Aplin, A.C., Matenaar, I.F. & van der Pluijm, B. 2003. Influence of
mudstones appear to be more highly pressured than the sands, mechanical compaction and chemical diagenesis on the microfabric and
implying that the sandstones are acting as conduits for the fluid flow properties of Gulf of Mexico mudstones. Journal of Geochemical
Exploration, 78-79, 449451.
dewatering of adjacent muddy sections, both upwards and Athy, L.F. 1930. Density, porosity and compaction of sedimentary rocks.
downwards. Higher pore pressures in top seals are useful in that American Association of Petroleum Geologists Bulletin, 31, 241287.
they will help support longer petroleum columns than would Audet, M.D. 1996. Compaction and overpressuring in Pleistocene sediments
otherwise be possible. Note finally that the apparently high on the Luisian Shelf, Gulf of Mexico. Marine and Petroleum Geology, 13,
467474.
overpressure around 2850 m results from the high log porosity, Bethke, C.M. 1985. A numerical model of compaction driven ground water
which is due to the poor log quality in this section. flow and heat transfer and its application to the paleohydrology of
intracratonic sedimentary basins. Journal of Geophysical Research, 90,
68176828.
CONCLUSIONS Bjrlykke, K. 1998. Clay mineral diagenesis in sedimentary basins a key to
the prediction of rock properties. Examples from the North Sea Basin.
The one-dimensional, mechanical compaction of mudstones in Clay Minerals, 33(1), 1534.
sedimentary basins is adequately described by the simple Bjrlykke, K. 1999. Principle aspects of compaction and fluid flow in
relationship, developed in soil mechanics, between void ratio mudstones. In: Aplin, A.C., Fleet, A.J. & Macquaker, J.H.S. (eds) Muds and
Downloaded from http://pg.lyellcollection.org/ at Stanford University on July 18, 2016

162 Y. Yang & A. C. Aplin

Mudstones: Physical and Fluid-flow Properties. Geological Society, London, Raiga-Clemenceau, J., Martin, J.P. & Nicoletis, S. 1988. The concept of
Special Publications, 158, 2243. acoustic formation factor for more accurate porosity determination from
British Standards Institution 1990. British Standard Methods of test for Soils for civil sonic log data. The Log Analyst, 29(1), 5460.
engineering purposes, Part 2, Classification tests (BS 1377: Part 2: 1990). British Schneider, F., Burrus, J. & Wolf, S. 1993. Modelling overpressures by
Standard Institution, London. effective-stress/porosity relationships in low-permeability rocks: empirical
Burland, J.B. 1990. On the compressibility and shear strength of natural clays. artifice of physical reality? In: Dor, A.G. (ed.) Basin Modelling: Advances and
Geotechnique, 40, 329378. Applications. Norwegian Petroleum Society (NPF) Special Publication, 3.
Chandler, R.J. 2000. Clay Sediments in Depositional Basins: the Geotechnical Elsevier, Amsterdam, 333341.
Cycle. Quarterly Journal of Engineering Geology and Hydrogeology, 33, 739. Sharp, J.M. & Domenico, A. 1976. Energy transport in thick sequences of
Charpentier, D., Worden, R.H., Dillon, C.G. & Aplin, A.C. 2003. Fabric
compacting sediment. Geological Society of America Bulletin, 87, 390400.
development and the smectite to illite transition in Gulf of Mexico
mudstones: an image analysis approach. Journal of Geochemical Exploration, Skempton, A.W. 1944. Notes on the compressibility of clay. Quarterly Journal
78-79, 459463. of the Geological Society of London, 100, 119135.
Dewhurst, D.N., Yang, Y.L. & Aplin, A.C. 1999. Permeability and fluid flow Skempton, A.W. 1970. The consolidation of clays by gravitational
in natural mudstones. In: Aplin, A.C., Fleet, A.J. & Macquaker, J.H.S. (eds) compaction. Quarterly Journal of the Geological Society of London, 125, 373411.
Muds and Mudstones: Physical and Fluid-flow Properties. Geological Society, Smith, J.E. 1971. The dynamics of shale compaction and evolution of pore
London, Special Publications, 158, 2243. fluid pressure. Mathematical Geology, 3, 239263.
Eaton, B. A. 1975. The equation for geopressure prediction from well logs. Terzaghi, K. 1943. Theoretical Soil Mechanics. John Wiley, New York.
Society of Petroleum Engineers Paper No.5544. Ungerer, P., Burrus, J., Doligez, B., Chenet, J.Y. & Bessis, F. 1990. Basin
Hermanrud, C. 1993. Basin modeling techniques an overview. In: Dor, evaluation by integrated two-dimensional modeling of heat transfer,
A.G. (ed.) Basin Modelling: Advances and Applications. NPF Special fluid-flow, hydrocarbon generation and migration. American Association of
Publication, 3. Elsevier, Amsterdam, 134. Petroleum Geologists Bulletin, 74, 309355.
Ho, N.C., Peacor, D.R. & van der Pluijm, B.A. 1999. Preferred orientation of Yang, Y.L. & Aplin, A.C. 1997. A method for the disaggregation of
phyllosilicates in Gulf Coast mudstones and relation to the smectite-to-
mudstones. Sedimentology, 44, 559562.
illite transition. Clays and Clay Minerals, 47, 495504.
Yang, Y.L. & Aplin, A.C. 1998. Influence of lithology and effective stress on
Hottmann, C.E. & Johnson, R.K. 1965. Estimation of Formation Pressures
From Log-Derived Shale Properties. Journal of Petroleum Technology, 717722. the pore size distribution and modelled permeability of some mudstones
Lerche, I. 1990. Basin Analysis. Quantitative methods, 1, 2. Academic Press, San from the Norwegian margin. Marine and Petroleum Geology, 15, 163175.
Diego, California. Yang, Y.L., Aplin, A.C. & Larter, S.R. 2004. Quantitative assessment of
Luo, X. & Vasseur, G. 1992. Contributions of compaction and aquathermal mudstone lithology using geophysical wireline logs and artificial neural
pressure to geopressure and the influence of environmental conditions. networks. Petroleum Geoscience, 10, 145151.
American Association of Petroleum Geologists Bulletin, 76, 15501559. Yardley, G.S. & Swarbrick, R.E. 2000. Lateral transfer: a source of additional
Nadeau, P.H., Peacor, D.R. & Yan, J. 2002. et al. IS precipitation in pore overpressure? Marine and Petroleum Geology, 17(4), 523537.
space as the cause of geopressuring in Mesozoic mudstones, Egersund Yukler, M.A., Cornford, C. & Welte, D.H. 1978. One dimensional model to
Basin, Norwegian Continental Shelf. American Mineralogist, 87(11-12), simulate geomogic, hydrodynamic and thermodynamic development of a
15801589. sedimentary basin. Geologische Rundschau, 67, 960979.

Received 16 December 2002; revised typescript accepted 10 November 2003.

S-ar putea să vă placă și