You are on page 1of 14

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

LIFTING THE FOG OF CONFUSION SURROUNDING CLAY AND


SHALE IN PETROPHYSICS
Paul Spooner – Senergy

Copyright 2014, held jointly by the Society of Petrophysicists and Well Log include a relationship between volume of shale
Analysts (SPWLA) and the submitting authors.
This paper was prepared for presentation at the SPWLA 55th Annual Logging and either volume of clay or volume of silt.
Symposium held in Abu Dhabi, United Arab Emirates, May 18-22, 2014.

It is very important not to confuse clay and shale


ABSTRACT models as this simple error can lead to many
problems: misunderstanding the difference
Consideration of the distinction between rocks and between effective porosity and total porosity; the
minerals is of vital importance in the petrophysical erroneous use of structural, laminated and
task of determining porosity and hence, water dispersed clay models when really it is a change in
saturation. Given the significance of this task it is the clay content of the shale; treating silt as a
surprising how much confusion there is across the mineral when it is a grain size; shale computed
industry over this issue. The confusion between with zero porosity; other low quality reservoir
clay and shale is the most common, to the extent layers, such as siltstones, computed with zero
that many books, papers, training courses and porosity leading to them being modelled as zero
software products still do not differentiate, or permeability baffles, when in reality they can store
explain these clearly or correctly. hydrocarbons and be considered reservoir when
they are high enough above the free water level.
Shale is a rock, typically defined as an indurated,
finely laminated, sedimentary rock, composed This paper seeks to lift the fog of confusion.
primarily of clay, mud and silt. The important
feature to note is that this definition does not
describe the mineralogy but rather the grain size. INTRODUCTION
In this definition, clay refers to clay sized
particles, i.e. < 1/256 mm. The term „shale‟ is used to describe a finely
laminated sedimentary rock composed primarily of
Whilst clay can refer to grain size it can also refer clay, mud and silt sized particles. Different
to clay minerals, and it is the dual meaning of the technical disciplines, such as soil scientists,
word clay that is at the heart of the confusion in geotechnical engineers, colloidal chemists etc, use
the industry. Clay minerals are a group of hydrous different definitions for these particle sizes.
aluminium silicates with a sheet-like structure Geologists typically use the Wentworth (1922)
(phyllosilicates), which adsorb water on their grade scale which defined clay sized particles as <
surfaces. It is these clay minerals that we are 1/256mm (3.9 microns, often simplified to 4
concerned about when determining porosity and microns) and silt sized particles as > 1/256mm and
water saturation. < 1/16mm (62.5 microns, often simplified to 60
microns). Silt is not defined by mineralogy at all, it
In shale, most of the clay sized particles are is just finer grained matrix material, e.g. in a
composed of clay minerals. When computing shaley-sand the silt is finer than „very fine sand‟.
porosity we need to account for the clay minerals, Mud is not specifically defined as a particle size
for example when their density is different from by Wentworth (1922) but is often used as a term to
the matrix density. When computing water describe a mix of silt and clay sized particles, for
saturation we need to account for the excess example Folk (1954) defined mud as having a
conductivity due to the clay minerals. In both clay:silt ratio between 1:2 and 2:1.
cases it is the volume of clay minerals we need to
correct for, not the volume of shale. Most, but not all, of the clay sized particles in
shale are composed of clay minerals, a group of
Shale models are technically correct when they hydrous aluminium silicates with a sheet-like

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

structure (Phyllosilicates) which adsorb water on


their surfaces. Clay sized particles that are not clay
minerals are often termed mud, although this is not
a strict definition.

Clay minerals form in the presence of water,


commonly by the weathering of feldspars and
micas, and include the Kaolinite group, the
Semctite group which includes Montmorillonite,
the Illite group and the Chlorite group, although
some authors consider the Cholrite group to be a
separate group of Phyllosilicates and not true
clays. Clay minerals may be allogenic, created
elsewhere and then deposited at the time of
sedimentation, or they may be authogenic, created
in-situ through diagenesis.

The significant petrophysical features of shale are:

1. shale is a rock defined in terms of particle size and


structure, not mineralogy
2. shale has effective porosity associated with non- Fig.1 Typical descriptions of shale distribution
clay mineral grains, primarily the silt
3. the ratio of clay and silt sized particles in shale can „Structural‟ shale is typically described as grains
vary considerably such that shale can be sub-dived of sand replaced with grains of shale. However,
into claystone, mudstone and siltstone this description is problematic as shale is a mix of
silt and clay sized particles. If the „shale‟ is
Whilst most geologists and petrophysicists in the allogenic, i.e. sourced from elsewhere and
industry are familiar with the difference between deposited with the sand, then to replace a larger
clay and shale, there has been much confusion sand grain with a „grain of shale‟ it is most likely a
over the use of the terms Vshale and Vclay, to the „clast‟ which itself is a mix of silt and clay
extent that many use the terms interchangeably. containing some porosity.
This is problematic because using these volumes
incorrectly can lead to the miscalculation of „Dispersed‟ shale is typically described as
effective porosity and hence, water saturation. confined to the pore space such that is it is grain
This paper explores common mistakes and coating or pore filling, which effectively means
recommends an approach to avoid these errors. the „shale‟ is authigenic. If this was the in-situ
conversion of minerals such as feldspar to clay,
SHALE DISTRIBUTION MODELS then the shale is clay mineral. Quartz overgrowths
and precipitated carbonate cementation are also
Shale is frequently found in discrete beds, much authigenic, but they are matrix. Only if both
thicker than the vertical resolution of the logs, processes occur does the description of „dispersed
where it can be easily identified based on its shale‟ make sense.
particular log responses. However, it is also very
often present in shaley-sand formations in which In the case of „Laminated‟ shale there are discrete
case it needs to be accounted for when computing laminations of sand and shale, where the shale
porosity and water saturation. In shaley-sand laminations are below the vertical resolution of the
formations, shale distribution models (Figure 1 logs. If they were thicker they would be resolved
and Figure 2) are frequently used to illustrate the by the logs as individual „beds‟ and the
influence of shale on porosity. interpretation would be less problematic.

The shale distribution is often termed „Structural‟, The volume of shale, Vshale, and the distribution
„Dispersed‟ or „Laminated‟ (Figure 1). of the shale is typically considered to influence the
2
SPWLA 55th Annual Logging Symposium, May 18-22, 2014

relationship between total porosity, t, and as it cannot be 100% de-watered, this condition
effective porosity, e (Figure 2). only exists after oven drying core samples when
all the clay bound water has been driven off. From
equations [1] and [2] we can state

t = e + Vclay*tclay [3]

Note that the relationship between t and e is a


function of Vclay and not the volume of shale,
Vshale. This definition of e follows the rock
model used in NMR interpretation of shaley-sands,
Coates, Xiao and Prammer, (1999), and makes no
reference to whether porosity is connected or
disconnected. This is a reasonable definition to use
as the standard porosity tools, density and neutron,
and even NMR tools respond to porosity without
discriminating its connectivity.

„Clean sand‟ (Figure 2) is considered to consist of


a volume of matrix grains and porosity. There is
no clay bound water so

t = e [4]

The „typical‟ explanation for „Structural shale‟


Fig.2 Typical description of how shale distribution (Figure 2) is that shale replaces sand, on a grain
affects porosity for grain basis, so there is no change in porosity
and equation [4] still applies. However, from
To consider the validity of the porosity equation [3] this can only be true if the volume of
relationships, as a function of Vshale and the clay bound water is zero, or more specifically if
distribution of that shale, we should first consider Vclay is zero as tclay cannot be zero in-situ.
what is meant by t and e. In this paper, t is Hence, the „Structural shale‟ model assumes that
considered to be the sum of all the fluid filled pore shale has no porosity, shale has no clay bound
space, which is equal to the sum of e and the water, and that all of the shale is matrix with no
volume of clay bound water, CBW. clay minerals, i.e. the shale is 100% silt and 0%
clay mineral. This is clean sandstone with some
t = e + CBW [1] grain sizes < 1/16mm, which could more correctly
be called siltstone.
The volume of clay bound water is the volume of
water adsorbed, or ionically bound, to the clay The „typical‟ explanation for „Dispersed shale‟
minerals, and/or trapped in nano-pores within the (Figure 2) is that shale replaces pore space so the
clay structure. The volume of clay bound water is porosity is reduced by the volume of shale.
determined from the volume of the clay minerals,
Vclay, and the apparent total porosity of the clay t = e + Vshale [5]
minerals, i.e. how much water it holds per volume
of clay, tclay. Hence, the „Dispersed shale‟ model assumes that
shale has porosity, shale has clay bound water, and
CBW = Vclay*tclay [2] from comparison with equation [1], that all of the
shale is wet clay, i.e. the shale is 0% silt and 100%
Note that Vclay is the volume of clay minerals and clay mineral.
not the volume of clay sized particles. If tclay is
zero the clay is „dry‟ clay and if tclay is non-zero The „porosity model for „Laminated shale‟ (Figure
the clay is „wet‟ clay. Dry clay cannot exist in-situ 2) shows that the shale replaces both matrix and
3
SPWLA 55th Annual Logging Symposium, May 18-22, 2014

pore volume. Nsal is the neutron formation salinity correction


Nhc is the apparent hydrocarbon hydrogen index
t = e + Vshale*tshale [6]
By using two equations, [9] and [10], in a
Hence, the „Laminated shale‟ model assumes that deterministic solution we can solve two variables,
shale has porosity, shale has clay bound water, typically e and ma so that we account for
and, by comparison with equation [3] it assumes variable matrix mineralogy.
that clay and shale are the same.
Equations [9] and [10] solve e using the mineral
Therefore, the typical descriptions of porosity in volume of wet clay, Vclay, not the rock volume of
terms of Vshale (Figure 2) use an inconsistent shale, Vshale.
model for shale.
100% wet clay rarely exists in-situ. Shale typically
Porosity should be expressed in terms of Vclay, a has some silt content as well so the clay content
mineral term, and not Vshale, a rock term, as the may be 40% to 80%. This makes the identification
bulk density response is a linear volumetric sum of of the wet clay point difficult as it lies outside the
the minerals and fluids in the rock. For a dual range of measured data. However, by recognizing
mineral system consisting of „matrix‟ and „clay‟, trends in the data we can reduce the uncertainty in
such as shale, the bulk density, b, can be determining the wet clay point.
expressed as
NEUTRON DENSITY CROSSPLOT TRENDS
b = Vma*ma + Vclay*cl + e*fl [7] 
Neutron density crossplots can be used to show the
Vma is the volume of matrix given by 1-Vcl-e difference between clay and shale, and to pick the
ma is the matrix density wet clay point required to compute e from
cl is the wet clay density equations [9] and [10].
fl is the density of the fluid in e given by
In a „typical‟ shaley-sand we see two trends in the
fl = Sxo*w + (1-Sxo)*hc [8] data forming an „L‟ or ‟V‟ shape (Figure 3).

Sxo is the flushed zone water saturation, which the


density tool is assumed to be reading
w is the density of the water in the flushed zone,
which may be a mix of WBM filtrate, formation
water or injection water
hc is the apparent hydrocarbon density, corrected
for electron density.

e can then be solved by re-arranging equations


[7] and [8].

[9]

Similarly, we can use the neutron response


equation to solve e.

[10]
Fig.3 Neutron density crossplot trends in shaley-
sands
N is the neutron porosity log (in limestone units)
Nclay is the neutron wet clay response
The left arm (Figure 3) is the shaley-sand trend
Nma is the neutron matrix correction
with the cleanest sands at the upper left. In this
Exfact is the neutron excavation factor
4
SPWLA 55th Annual Logging Symposium, May 18-22, 2014

example there are light hydrocarbons so the data


extends beyond the clean sand matrix line. The
shale point at the lower right of the shaley-sand
trend represents the shale distributed with the
sand. The shaley-sand trend is the linear
volumetric mix of clean sand and shale, i.e. with
50% shale by volume the data will plot half way
along this trend.

The right arm (Figure 3) is the shale trend with the


shale point at the left end and the wet clay point at
the right end. The shale point is at the intersection
of the shaley-sand trend and the shale trend. The
shale trend is important as it provides the means to
identify the wet clay point.

The mineral content of shale is a mix of matrix


and wet clay. The „shale line‟ (Figure 4) extends
from the matrix point (100% silt, 0% wet clay, 0%
porosity) to the wet clay point (100% wet clay, 0% Fig.5 Neutron density crossplot CSR and Isilt
silt). The location of the shale point on the shale
line (Figure 4) depends on the clay:silt ratio of the On a typical neutron density crossplot the shale
shale. If the shale is 30% silt and 70% wet clay the point is easier to identify than the wet clay point,
shale point is 70% along the shale line from the but the wet clay point is required to compute e.
silt point. This reflects the general classification of The important feature of the shale line is that when
shale as siltstone, mudstone or claystone. it has been identified we can extrapolate from the
matrix point, through the shale point to the wet
clay point, and hence we can solve e.

Fig.4 Neutron density crossplot Shale Line

The clay:silt ratio of the shale can be expressed in Fig.6 Neutron density crossplot shaley-sand trends for
terms of clay, i.e. 0.7, often termed the clay shale shales with different clay:silt ratios
ratio or CSR (Figure 5). Alternatively the clay:silt
ratio of the shale can be expressed in terms of silt, If different shaley-sand sequences contain shale
i.e. 0.3, often termed the „silt index‟ or Isilt. with different clay:silt ratios, then the shaley-sand
5
SPWLA 55th Annual Logging Symposium, May 18-22, 2014

trends observed on the neutron density crossplot the clean matrix line, from dry clay at the bottom,
will differ (Figure 6). the highest density, through the wet clay point to
the water point at 100% porosity. The distance
If the shale has high clay content, the shaley-sand along the clay line from the dry clay point to the
trend goes from the clean sand point toward the wet clay point is tclay, and can be determined
claystone region of the shale line. The angle from a simple density porosity equation The dry
between the shaley-sand trend and the shale trend clay point is a function of the clay mineralogy, and
becomes very obtuse and the shale point is much while it cannot be estimated precisely from the log
closer to the wet clay point. data, it can be determined from core analysis or
estimated by matching the computed porosity to
If the shale has similar clay and silt content then overburden corrected core porosity.
the shaley-sand trend goes toward the mudstone
region of the shale line. The angle between the A compaction trend may also be apparent on the
trends becomes close to 90deg and the shale point neutron density crossplot (Figure 8). As the clays
is close to the midpoint of the shale line. de-water, tclay will be reduced and the wet clay
point will move down the clay line toward the dry
If the shale is very silty the shaley-sand trend goes clay point, which is a fixed point based on the clay
toward the siltstone region of the shale line. The mineralogy.
angle between trends becomes very acute and the
shale point is much closer to the matrix point. Silt Unfortunately the shaley-sand and shale trends are
is not a mineral, it is matrix with grain sizes < not always clear on neutron density crossplots.
1/16mm, so the same principles apply and e can This can be due to a variety of reasons.
be determined from equations [9] and [10] using
variable ma to account for changes in If the clay:silt ratio of the shale is changing
mineralogy. If the matrix mineralogy does not fit throughout the shaley-sand sequence then the data
the standard sand-lime-dolo model then any two will not have one clear shaley-sand trend, it can
minerals can be used to solve variable ma, e.g. appear to be more of a „blob‟. However, the shale
quartz and feldspar or quartz and mica. line should still be apparent at the lower edge of
the „blob‟ and the wet clay point extrapolated.

If the shaley-sand sequence is in compaction or the


clay mineralogy is changing, then the wet clay
point will be moving and the data will again
appear to be more of a „blob‟. The interpretation
can be zoned to account for the changing wet clay
point, or, depending on the software being used,
the wet clay log values can be allowed to
continuously vary as a function of TVD or burial
depth.

If the shale beds in a laminated shaley-sand


sequence are thinner than the log resolution, the
shale point and the shale line cannot be identified
as the logs „see‟ a volumetric mix of sand and
shale. In such cases the surrounding shale beds are
often used as they will show a shale trend. The
validity of this relies on the commonly used
assumption that the shale in the surrounding beds
Fig.8 Neutron density crossplot Clay Line and is the same as the shale in the shaley-sand
Compaction trend However, this assumption is very often false e.g.
when the nearby shale is from a completely
A clay line can be constructed on the neutron different depositional environment.
density crossplot (Figure 8) in the same manner as
6
SPWLA 55th Annual Logging Symposium, May 18-22, 2014

VCLAY AND VSHALE Vshale can be determined by linear interpolation,


parallel to the clean line, where Vshale is 0%, to
Vclay can be determined by linear interpolation, the shale point where Vshale is 100% (Figure 11).
parallel to the clean line, where Vclay is 0%, to the
wet clay point where Vclay is 100% (Figure 9).

Fig.11 Neutron density crossplot Vshale with


adjusted clean line in light hydrocarbon zone
Fig.9 Neutron density crossplot Vclay

Light hydrocarbons lead to the underestimation of


Vclay unless „corrected‟ for. A simple adjustment
to the slope of the clean line in the hydrocarbon
zone is a pragmatic solution (Figure 10).

Fig.10 Neutron density crossplot Vclay with


adjusted clean line in light hydrocarbon zone Fig.12 Neutron density crossplot Vshale and Vclay
7
SPWLA 55th Annual Logging Symposium, May 18-22, 2014

Figure 12 shows a comparison plot of Vclay and important, not only because it is used for the
Vshale determined using neutron density crossplot computation of e, but because in shaley-sand a
methodology. Note that in this example, Vshale is water saturation equation that accounts for the
approximately 1 in the shale while Vclay is excess conductivity of the clays should be used.
approximately 0.7. The excess conductivity is not a function of the silt
in the shale, but the clays, so Vclay should be
The clay shale ratio, CSR, can now be defined as used. For example, the Dual Water equation,
the ratio of Vclay to Vshale. Coates, Xiao and Prammer, (1999), is expressed in
terms of clay bound water saturation, Swb.
CSR = Vclay/Vshale [11]
Swb = CBW/t = Vclay*tclay/t [12]
Vshale and Vclay are often confused as the shale
point and wet clay point are frequently picked Care must be taken as not all software products are
incorrectly. Consideration of the trends on the thorough, and many confuse the terms Vshale and
neutron density crossplot allows the shale point Vclay. If Vshale is used in lieu of Vclay then e
and wet clay point to be clearly identified. tends to be computed too low. This is one of the
main reasons why many interpretations of shale
If the software product used for the volumetric show zero e, when in fact most shales have non-
interpretation solves the porosity equations, [9] zero e, as shown by numerous NMR logs.
and [10], in terms of Vclay and the wet clay point,
then the user is required to make the appropriate Organic shale reservoirs were overlooked for
„picks‟ to generate a Vclay curve (Figure 10). many years, not just because they were tight, but
because they were routinely interpreted with zero
If the software product used for the volumetric e. Confusing Vshale and Vclay can also lead to
interpretation solves the porosity in terms of siltstones easily being misinterpreted and e being
Vshale and the shale point, then the user is computed as too low.
required to make the appropriate „picks‟ to
generate a Vshale curve (Figure 11). Once the computed e reaches zero, whether it is
a siltstone or a shaley-sand, then any permeability
However, as porosity equations, [9] and [10], are derived from that porosity will also be zero. This
expressed in terms in clay and not shale then can then have an effect on the dynamic model as
Vclay and the wet clay point are still required, and zero permeability layers act as baffles in the
therefore a parameter defining the clay:silt ratio, reservoir simulation. Hydrocarbon storage,
such as CSR or Isilt, is still required. The shale STOIIP, may also be negatively impacted, as
point can then be extrapolated to the wet clay modelled with zero e and zero permeability they
point, based on the shale line, and Vclay can be will not be able to store hydrocarbons at all. In
determined from Vshale using equation [11]. reality though, if high enough above the free water
level they can store hydrocarbons.
Computing e using CSR and Vshale is
volumetrically correct in the shaley-sand trend, but GR VSHALE AND VCLAY INDICATOR
for data beyond the shale point this is no longer
true. For data in the shale trend (Figure 3) CSR is The Gamma Ray log, GR, is very commonly used
now a variable as there is increasing clay content for the determination of Vshale and Vclay as the
within the shale. If a constant CSR is used then three main sources of radioactivity in the earth‟s
Vclay will be „clipped‟ and e will be crust, potassium-40, the thorium series and the
overestimated. uranium series tend to be concentrated in clays and
shales.
This paper recommends using a Vclay workflow,
not only because that is valid in the shales as well Clean rocks, i.e. with no shale or clay, rarely have
as the shaley-sands, but also because that GR log reading zero due to trace minerals and
encourages the interpreter to focus on the wet clay dissolved salts in the pore fluids. A value can be
point. If the focus is on the shale point then CSR is „picked‟ to represent clean matrix, the „GRclean‟
often neglected and left as „default‟. Vclay is more line in the Gamma Ray track (Figure 13), and a
8
SPWLA 55th Annual Logging Symposium, May 18-22, 2014

value can be „picked‟ to represent the shale, the There are many non-linear GR relationships
„GRshale‟ line in the Gamma Ray track (Figure commonly used in the industry, such as Clavier
13). A simple linear interpolation from GRclean to (1971), Larionov (1969) (Young Rock and Old
GRshale can then be used to determine Vshale. Rock), and Steiber (1971) (Figure 14). There are
no physical theorems or models for these non-
Similarly, a value can be „picked‟ to represent linear relationships, they were all empirically
100% clay, the „GRclay‟ line in the Gamma Ray derived to address the observation that linear GR
track (Figure 13), and Vclay determined. overestimated Vshale. In all cases the non-linear
relationships reduce the computed Vshale, and
have the greatest effect in the middle of the range.

Fig.13 Vshale and Vclay derived from GR Fig.14 Non linear GR relationships

As with the neutron density crossplot it is much The basis for the non-linearity is often considered
easier to „pick‟ GRshale than GRclay, as 100% to be due to changes in the age of the rocks, clay
clay rarely exists in-situ. Hence many interpreters mineralogy, porosity, CSR, and to some extent
prefer to work with Vshale, and historically GR is how the clays are distributed. However, it is not
typically referred to as a Vshale indicator. that the GR response itself is non-linear, as it is a
linear volumetric sum of the gamma rays emitted
However, when using a GRshale „pick‟ to by all the radioactive elements in the measured
determine Vshale an assumption must still be volume.
made regarding the parameter CSR or Isilt, as
Vclay is still required to compute e. If using a Due to the half life of the radioactive elements, the
GRclay „pick‟ to determine Vclay, then the same older the rock becomes then the lower the GR
assumption is made but this time it is the „picking‟ count rate, but by making an appropriate „pick‟ for
of GRclay to give a particular Vclay in the shale. It GRclay this effect is accounted for.
is the same interpretation problem working with
Vclay or Vshale just expressed in different ways, If the clay mineralogy or tclay changes then
picking GRclay working with Vclay or selecting GRclay should be zoned accordingly. If GRclay
CSR, or Isilt, working with Vshale. has been picked in surrounding shale that does not

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

have the same clay as that in the shaley-sand „correct‟ it.


sequence, then it is not applicable to the shaley-
sand and Vclay will be wrong, especially if a With appropriate zoning of GRclean and GRclay,
significantly, younger, overlying shale was used. a liner relationship should provide a valid Vclay,
and indeed, using equation [2] with an estimate of
If the porosity is changing then GRclean may be tclay, a comparison with the CBW derived from
affected but that will depend on how much of the NMR often shows this to be the case.
radioactivity is in the matrix minerals and how Furthermore, the Vclay computed from neutron
much is in the dissolved salts. GRclean should be density crossplot is a linear interpolation and it too
zoned and adjusted accordingly. tends to match a linear Vclay from GR, if the GR
picks have been made appropriately.
If the CSR is changing, but the clay mineralogy
and tclay are consistent, then the GRclay „pick‟ THIN BED MODELS
will be consistent and Vclay will not be affected,
but GRshale will be affected as there will be more Thin bed models, where the beds are below the
or less clay in the shale. Hence, using Vclay is vertical resolution of the logging tools, based on
more reliable. However, GRclean may also need to the „structural‟, dispersed‟ and „laminated‟ shale
be adjusted with varying silt content, as that may models (Figure 1 and Figure 2) are commonly
also reflect a change in porosity. used, Thomas and Steiber (1975), Juhasz (1986).
These rely on the identification of shale
If there is a change in clay distribution then this distribution trends in a crossplot of Vshale against
can affect the GR count rates, but this effect is porosity, sometimes using GR as a proxy for
usually relatively small when considered alongside Vshale.
the other possible reasons for variation and the
uncertainty in the GR log measurement. For If the porosity was computed from neutron and
example, consider clean sand with t of 0.3 density, then the Vshale porosity crossplot is
(Vmatrix of 0.7) and GRclean of 20 API, essentially the neutron density crossplot with the
assuming zero GR contribution from the fluids, clean matrix line rotated to the Y-axis and the
and a shaley-sand consisting of the same clean shale line rotated to the X-axis, (Figure 15). It is
sand with a Vclay of 0.2 and GRclay of 120 API. also common to assume the same trend lines
indicating the „shale distribution‟
If the clay is dispersed then the GR log will be
20 + (0.2*120) = 44 API.

If the clay is structural then the GR log will be


(20*0.5/0.7) + (0.2*120) API = 38.3 API.

If the clay is laminated then the GR log will be


(20*0.56/0.7) + (0.2*120) = 40 API.

Another possibility for the non-linear


relationships, and potentially much more
significant, is the confusion of Vshale and Vclay.
The overestimation of the linear Vshale is
typically „verified‟ by comparison to core data,
either by comparison of the resulting porosity or
by comparison of the clay content determined
from X-Ray Diffraction, XRD. In both cases it is
Vclay that is required for the comparison and yet
frequently it is Vshale that is used. If Vclay is
determined for direct comparison to the core, but a
GRshale „pick‟ was used, then it will overestimate Fig.15 Neutron density crossplot rotated to Vshale
Vclay and a non-liner relationship might appear to  crossplot with „shale distribution‟ trend lines
10
SPWLA 55th Annual Logging Symposium, May 18-22, 2014

The basic tool equations for the density and


neutron, [9] and [10], are expressed in terms of
clay, not shale, so any trend lines should be
defined in terms of clay distribution not shale
distribution. Equations [9] and [10] do not contain
terms for laminated, dispersed or structural clay,
so the logs do not specifically contain any
information about the clay distribution. The tool
responses are functions of the mineral and fluid
volumes and not the clay distribution. Therefore,
the porosity computed from [9] and [10] cannot be
affected by the clay distribution unless that
distribution affects the mineral or fluid volumes.

Juhasz (1986) showed how „shale distribution‟


affects the mineral and fluid volumes, and
expressed the trend lines in terms of t and e,
where e was defined by equation [6] with the
assumption that “shale-water does not contribute
to effective porosity”. Unless the shale is 100%
clay this assumption is invalid as capillary bound
Fig.17 Vclay e crossplot with „structural clay‟,
water associated with silt contributes to e. To be „dispersed clay‟ and „laminated clay‟ overlay
consistent with definitions of clay and shale, e
should be expressed in terms of Vclay, equation The trend lines shown on the crossplots (Figure 16
[3], „clay‟ substituted for „shale‟ throughout the and Figure 17) are defined by considering a
model, and the overlays for structural clay volumetric mix clean of sand with porosity max,
(Vstruc), laminated clay (Vlam) and dispersed clay with structural clay, dispersed clay, or laminated
(Vdisp) plotted on Vclay against t (Figure 16) or clay, where in each case the clay has the same
Vclay against e (Figure 17). properties, i.e. the same mineralogy and tclay.

Structural clay is considered to „replace‟ sand


grains. On the t Vclay crossplot (Figure 16), t
increases from max by the volume of clay bound
water as Vclay (=Vstruc) increases.

t = max + Vstruc*tclay [12]

On the e Vclay crossplot (Figure 17) increasing


Vclay (=Vstruc) makes no difference to e.

e = max [13]

Dispersed clay is considered to „replace‟ porosity.


On the t Vclay crossplot (Figure 16), t reduces
from max by the volume of clay but also
increases t by the volume of clay bound water.

t = max - Vdisp*(1-tclay) [14]

On the e Vclay crossplot (Figure 17) e reduces


Fig.16 Vclay t crossplot with „structural clay‟, from max by the volume of clay.
„dispersed clay‟ and „laminated clay‟ overlay

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

e = max - Vdisp [15] and t or Vclay and e, a solution can found for
Vlam, Vstruc, Vdisp and the porosity in the sand
Laminated clay is considered to „replace‟ both lamina.
sand grains and porosity. On the t Vclay
crossplot (Figure 16), t is converted from max Whilst this method is mathematically valid, it
to tclay as Vclay (=Vlam) increases from 0 to 1. relies on some basic assumptions that should be
considered.
t = max – Vlam*(max – tclay) [16]
The constructs, both the overlays and the
On the e Vclay crossplot (Figure 17), e reduces equations used to solve them, are only appropriate
from max to 0 as Vclay (=Vlam) increases from when using Vclay and not Vshale, unless CSR or
0 to 1. Isilt is used to convert Vshale into Vclay.

e = max - Vlam*max [17] The equations are derived from Vclay and
porosity, either t or e, and not density and
A parallelogram can be constructed based on the neutron logs. It is assumed that the computed
structural and dispersed trends with the lower porosity is the correct volumetric average
corner of the parallelogram defined as the point on accounting for the mineralogy and clays, equations
the dispersed trend where Vclay = max. This is [9] and [10]. A simple average of density porosity
the maximum volume of dispersed clay that can and neutron porosity will probably yield the wrong
exist, when all of the clean sand porosity has been porosity.
filled with dispersed clay.
By comparison of the porosity Vclay crossplots
If e is solved from equations [9] and [10] then (Figure 16 and Figure 17) with the neutron density
Vlam can be solved from equation [17] by crossplot (Figure 15) the same structural,
assuming max, which is typically obtained by laminated and dispersed trends can apparently be
extrapolating the shaley-sand trend on the e identified in the neutron density data. However,
Vclay crossplot (Figure 17) back to Vclay = 0. such trends could easily be due to changes in the
Similarly, Vlam can be solved from equation [16] clay:silt ratio rather than the distribution of the
by assuming max and tclay. clay. One of the stated assumptions by Thomas
and Steiber (1975), was that the clay:silt ratio for
The parallelogram can then be interpolated in two the shale must stay constant, which of course it
halves, the upper half for a mix of structural and often does not.
laminated clay and the lower half for a mix of
laminated and dispersed clay, (Figure 18). The model can also be interpreted wrongly when
the clay:silt ratio is constant, if the wrong
assumption is made about the nature of the shaley-
sand. For example, if the shaley-sand is very silty
with low Vclay, then this trend could easily be
misinterpreted as dispersed clay. Even when Vclay
and e have been correctly determined the
interpreter has to rely on other sources of data,
such as cuttings, core data or image logs, to
determine whether the sand lamina contain
dispersed clay or silt, in which case this model is
not appropriate. This will obviously affect the
computation of water saturation in the sand
lamina, but it is also important not to confuse silt with
Fig.18 Thin bed models solve laminated and structural dispersed clay as dispersed clays, such as Illite in pore
clay or laminated and dispersed clay throats, generally have a significant effect on
permeability. There may also be scaling errors
With the assumption that Vclay is the sum of all introduced as there is a significant difference in
the clays, then for any pair of data points, Vclay the vertical resolution of the density and neutron
12
SPWLA 55th Annual Logging Symposium, May 18-22, 2014

tools and the core data used to validate the models. NMR IN THIN BEDS

The distribution of clays in shaley-sands is not Shaley-sands and thin beds in particular, are an
usually as simplistic as that assumed by the model. ideal situation for NMR logs as they can provide
For example, for a given Vclay, a mix of structural t and e directly from the logs without having to
clay and dispersed clay can give the same result as make assumptions or calculate matrix properties.
laminated clay, and this combination is not solved by In thin beds, the standard NMR cutoff
the model. The observed trend in the data often follows interpretation principles still apply. If a zone can
the „laminated‟ trend, which is not surprising as that be identified as being at irreducible water
represents a mix of everything which is probably the saturation, ideally from testing, and it is water wet,
closest to reality in most cases. then a simple integration of the free fluid porosity
provides the hydrocarbon pore volume directly,
Very commonly, the deviations from the laminated i.e. Sw equals the bound fluid divided by t.
trend fall in the laminated-dispersed region of the plot,
which as discussed may be due to a change in clay silt If a zone is at irreducible water saturation, and
ratio silt, toward more silt, but it may also be due to there are no conductive minerals, then the
uncertainty in the computed porosity or Vclay as well. resistivity is primarily responding to the bound
fluids, both the clay bound water and the capillary
The definition of structural clay is also somewhat bound water. Hence, a simple overlay plot, or a
problematic, even when considered in the sense of crossplot, of the resistivity and bound fluid can be
„an equivalent volume of clay replacing a volume used to identify zones at irreducible water
of matrix‟, rather than „grain replacement‟, as it saturation
does not consider the source of the clay.
The key to utilizing NMR logs in this way is not to
If the clay is allogenic then there is no try and resolve the thin beds by reducing stacking,
„replacement‟ in the geologic sense of the word, as but to try and improve the accuracy of the porosity
the clay was deposited with the sand, but in the measurement, by increasing stacking to increase
volumetric sense it does occupy a gross volume SNR.
that could have been matrix, as in shale models
(Figure 2) with the word „clay‟ substituted for CONCLUSIONS
„shale‟. In this case, the structural model, [12] and
[13], are consistent. Historically Vshale and Vclay have frequently
been confused and are often considered to be the
However, if the clay is authigenic it is not usually same thing. They are not. Shale is a rock term
a complete „replacement‟, in the geologic sense of while clay is both a grain size and a mineral term.
the word. This can cause an increase in porosity,
for example feldspar may not fully convert to clay In Petrophysics it is the mineral term that is
and there can be extra voids and „fractures‟ within needed to solve porosity and water saturation, so
the mineral fragments. Other processes occurring we need to use Vclay. Vshale can be used in the
during diagenesis, such as the precipitation of shaley-sand trend but only if a CSR or Isilt
cementation, decrease the porosity. Hence, for parameter is utilized.
authigenic clays, the structural model, [12] and
[13], may be wrong. Computations of Vclay from GR logs should be
valid when using a linear relationship, as long
Due to all of these issues, the use of these thin-bed appropriate picks are made.
models is very subjective. With knowledge of the
correct answer, for example from core data, they Shale distribution models are ill posed and should
can provide robust results, but they can easily lead be expressed in terms of Vclay.
to misinterpretation, very often when silt is not
correctly interpreted as matrix and confused with In thin-beds it may be more reliable to use NMR
dispersed clay. logs, but the classical method can be used if it is
well constrained by matching to core data or
image logs.
13
SPWLA 55th Annual Logging Symposium, May 18-22, 2014

ACKNOWLEGEMENTS years in the field as a Wireline Engineer, then latterly in


the office on integrated projects. During this time he
Derek Crombie, Senergy Technical & Product built up his Petrophysical knowledge and in particular
Development Manager and Frank Whitehead, his expertise on NMR interpretation and core
Senergy IP Software Architect for numerous integration. More recently, Paul has been engaged
discussions and invaluable help concerning the issues along with others in the development and delivery of
raised in this paper. the Petrophysics and Formation Evaluation MSc course
from Aberdeen University.
REFERENCES

Coates, G.R., Xiao, L., Prammer, M.G., 1999, NMR


Logging Principles and Applications, ISBN 0-9679026-
0-6, Chapter 7, 135-137

Folk, R.L., 1954. The distinction between grain size


and mineral composition in sedimentary rock
nomenclature. Journal of Geology 62 (4), 344-359

Juhasz, I., 1986, Assessment of the Distribution of


Shale, Porosity and hydrocarbon Saturation in Shaly
Sands, Tenth European Formation Evaluation
Symposium, April 22-25, 1986, Paper AA

Larionov, V.V., 1969, Radiometry of boreholes,


NEDRA, Moscow.

Thomas, E.C., Steiber, S.J., 1975, The Distribution of


Shale in Sandstones and its effect upon Porosity,
SPWLA Sixteenth Annual Logging Symposium, June
4-7, 1975, Paper T

Wentworth, C.K., 1922, Scale of Grade and Class


Terms for Clastic Sediments, Jour. Geol., Vol.30 July-
Aug 1922, 377-392

ABOUT THE AUTHOR

Paul is currently IP Product Champion for Senergy,


providing support, training and helping with the
development of Senergy's IP software product. Prior to
joining Senergy Software Paul worked on many
complex petrophysical and integrated projects within
the consultancy department of Senergy GB Ltd, and
formerly Production Geoscience Ltd. Paul spent many
14

You might also like