You are on page 1of 21

SPE-175111-MS

Fracture Detection Using the Conductivity Triaxiality Index


Gong Li Wang, Aria Abubakar, Tom Barber, and Samer Alatrach Schlumberger

Copyright 2015, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE Annual Technical Conference and Exhibition held in Houston, Texas, USA, 28 –30 September 2015.

This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents
of the paper have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect
any position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written
consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may
not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.

Abstract
Over the years, triaxial induction tools have found applications in not only determining formation
resistivity, anisotropy, and dip, but also in detecting fractures. Fracture detection techniques provide
useful information in identifying fractured reservoirs because fractures provide both space for hydrocar-
bon storage and also the channel for fluid flow. When fractures are developed in a formation, the
formation can exhibit triaxially anisotropic conductivity if the fractures are filled with a fluid with a
significantly different conductivity than that of the formation. This is particularly true for transversely
isotropic formations. In an isotropic sedimentary formation, fractures can also cause the formation to be
triaxially anisotropic if the geometric properties, orientation and porosity of fractures are varied from
place to place in the fractured zone.
In this paper, we present a new method that uses the information of the conductivity tensor to detect
fractures. The new method indicates the presence of contrasting fluid-filled fractures by the difference in
three conductivity components of the tensor. This difference is called triaxiality in the paper. The
conductivity tensor is found using an inversion approach. In addition to three conductivity components ␴x,
␴y, and ␴z, the inversion also provides three Euler angles ␥, ␣, and ␤. Multiple initials for Euler angles
are used to ensure that a global minimum is obtained. The inversion is followed by a rotation operation
on the inverted conductivity tensor to eliminate the ambiguity effect of the principal coordinate system.
The new Euler angles are found by solving a minimization problem.
We have applied the new method to both synthetic and field examples. Synthetic examples show that
the rotation operation is indispensable to remove the ambiguity effect of the principal coordinate system.
Furthermore, it is shown that the triaxiality is indeed responsive to the presence of fracture. Previous
processing of the field example has showed that fractures exist in many zones. The triaxiality index
derived from the new method is found in a fairly good agreement with the fracture index from a previous
method.

Introduction
Conventional induction tools use only coaxial transmitter and receiver coils to measure formation
resistivity. Data acquired with such tools are generally used to infer formation resistivity with an
assumption that the formation is isotropic. Triaxial induction tools are a new generation of induction tools
that have been developed since the 1990s. A fully triaxial induction tool consists of three orthogonally
2 SPE-175111-MS

collocated transmitter coils and receiver coils (Barber, et al. 2004), as shown in Fig. 1. Such a
configuration allows for measuring a full apparent conductivity tensor, making it possible to extract more
information of formations. One popular application of triaxial induction data is to find horizontal and
vertical resistivity as well as dip of a formation assuming the formation is transverse isotropic (TI).

Figure 1—Configuration of a triaxial induction tool with mutually orthogonal and collocated transmitter and receiver coils.

Transverse isotropic formations are the most common type of formation because, in general, sedimen-
tary formations have a layered structure or are intrinsically anisotropic. However, when fractures are
developed in a formation, the resistivity can be different in all three orthogonal directions, leading to
triaxial anisotropy in the formation (also called biaxial anisotropy using the jargon of the optical society).
Fractures seen in field are generally either drilling induced or naturally formed. Drilling-induced fractures
tend to be oriented at a high angle relative to the tool plane due to the orientation of stress field in the
near-borehole region. In other words, the fracture planes are nearly parallel to the borehole axis. Most
natural fractures are caused by ancient earth stresses, and therefore do not necessarily intersect the
borehole at a high angle. The triaxial anisotropy can be found in not only TI-anisotropic formations, such
as shales and laminated sands. It can also happen in fairly isotropic formations, such as carbonates, if the
geometric properties of fracture, such as length, width, and opening, as well as the orientation and
porosity, are varied from place to place in the fractured zone. The presence of fractures on its own does
not necessarily lead to significant change of resistivity. It does so only when fractures are filled with
materials with a significantly different resistivity from that of the ambient formation. This is often the case
with drilling-induced fractures because they are filled with mud or mud filtrate that generally has different
resistivity than the formation. Natural fractures can be cemented or filled with formation fluid if they do
not intersect with the borehole. In the latter circumstance, the filling fluid often shows a significant
resistivity contrast with the formation.
Many works have been published to model microstructures or fractures with anisotropic conductivity
(Torquato 1991; Moskow et al. 1999; Davydycheva et al. 2003; Habashy and Abubakar 2007; Sævik et
al. 2012). These works show that, in principle, fracture parameters such as opening, orientation and
porosity can be related to a conductivity tensor, or, specifically, its six parameters. The six parameters are
three conductivity components ␴x, ␴y, and ␴z, and three Euler angles ␥, ␣, and ␤. Georgi et al. (2008) give
a fracture model assuming the fractures are perpendicular to the bedding plane. Accordingly, a workflow
SPE-175111-MS 3

is developed to find three conductivity components and fracture orientation as well as fracture porosity.
This perpendicular assumption, however, can be violated when formations are dipping relative to the
borehole in the presence of drilling-induced fractures. As far as natural fractures are concerned, there are
different types of fractures that may be developed in the formation, not all of them aligned with the
principal stresses. Moreover, different types of fractures can exist simultaneously in the formation. Shown
in Fig. 2 are natural fractures that may be developed in the normal-faulting stress regime, as well as a
schematic illustration of a drilling-induced fracture in a dipping formation (Thompson 2000). Yin (2010)
presents a general fracture model in which the fracture can be oriented in an arbitrary direction. To our
knowledge, no results have been reported regarding the application of this general model. Assuming that
the ambient formation is TI-anisotropic and homogeneous, it can be described with four parameters:
horizontal conductivity, vertical conductivity, and anisotropy dip and azimuth. On the other hand, a single
infinitely extending fracture can also be described with four parameters: fracture opening, conductivity,
dip, and azimuth. In total, eight parameters are needed to describe such a single fracture model.
Considering that there are only six parameters in the conductivity tensor, it is not sufficient to solve for
all eight parameters. For the problem to be solvable, either the fracture model needs to be simplified or
additional information should be provided. For example, with the knowledge of resistivity and orientation
of the ambient formation, the problem becomes overdetermined and in principle can be solved with the
nonlinear least squares method.

Figure 2—Natural fractures and drilling-induced fractures. On the left are the three types of fractures of the normal-faulting stress
regime: joints (blue), faults (green), and stylolites. On the right is the drilling-induced fracture in a formation with dipping anisotropy.

In this paper, we will not attempt to determine fracture orientation, opening and porosity. Considering
its scope and importance, this topic deserves to be covered in a separate paper. We will concentrate on
using three conductivity components to indicate the presence of fracture. To this end, we introduce a new
index, triaxiality, which measures how far the triaxial conductivity model is from a TI model. The
triaxiality index is derived from the conductivity tensor found with an inversion approach. It is a function
of geometric parameters and filling fluid properties of fractures. In general, the larger the triaxiality, the
higher the possibility of fractures is in the formation.
It is noted that the inversion is only part of our new processing method. There are mainly two
challenges for this problem. One challenge is with the forward problem, or, specifically, the evaluation of
2D integrals for apparent conductivity. Unlike previous methods, we solve the problem directly in the tool
coordinate system so that the accuracy and efficiency is ensured for all dip angles. The other challenge
is with the nonuniqueness of Euler angles. Tang (2008) used the Gauss-Newton method for the inversion
without discussion of the nonuniqueness. Rabinovich et al. (2008) and Georgi et al. (2008) first limited
4 SPE-175111-MS

Euler angles to small ranges, then used a two-step approach for the inversion. They first found optimal
Euler angles by solving a large number of three-unknown problems. Once the Euler angles are determined,
another three-unknown problem is solved for three conductivity components. In our processing, we solve
for the six parameters simultaneously. Considering the cyclic property of Euler angles and data noise,
potentially there may exist local minimums for the inversion. We use multiple initials for Euler angles to
ensure that the inversion is not trapped in a local minimum. Once the conductivity tensor is found, a
rotation operation is performed on it to remove the ambiguity effect of the principal coordinate system.
We then apply our new processing method to synthetic examples to validate its capability and
understand its performance and behavior. We show that the rotation operation is dispensable to have a
proper interpretation. We also use the new processing to process a field dataset, and comparison with
previous processing results is provided.

Triaxially Anisotropic Conductivity Model


Previous works have shown that in a fractured formation, its conductivity can be triaxially anisotropic.
These works include, but are not limited to, those of Torquato (1991), Moskow et al. (1999), Davydycheva
et al. (2003), Habashy and Abubakar (2007), and Sævik et al. (2012), to name a few. Based on the results
of previous works, in general, the conductivity of a fractured formation should be described with a
second-order tensor:
(1)

There are two coordinate systems used in Eq. 1: the tool coordinate system and the principal coordinate
system. Matrix is defined in the tool coordinate system. Scalars ␴X, ␴y, and ␴z are three conductivity
components in the x-, y-, and z-directions of the principal coordinate system, respectively. Matrices ,
, and are three elemental rotation matrices associated with Euler angles ␥, ␣, and ␤, respectively,
which are given by
(2)

(3)

(4)

Theoretically, three Euler angles each can take any value over the range of [0,360°]. However, using
symmetric properties of , the ranges of ␥, ␣, and ␤ can be reduced to [0,180°], [0,90°], and [0,360°],
respectively. By so doing, the nonuniqueness of Euler angles is eliminated for a given order of three
conductivity components. Fig. 3 describes the rotation between the principal coordinate system and the
tool coordinate system. It is noted that there are more than one type of rotation operations that can rotate
the tool coordinate system to the principal coordinate system. The current selection as shown in Eqs. 2-4
is called z -y=-z⬙ rotation in the literature. This type of rotation is compatible with the commonly used
SPE-175111-MS 5

convention of rotation for a TI formation. It is readily shown that when ␴x ⫽ ␴y, becomes redundant,
and and reduce to nothing but the two elemental rotations associated with the anisotropy dip ␣ and
azimuth ␤ of the formation. Similarly, when ␴x ⫽ ␴y, ␣ and ␤ designate, respectively, the dip and azimuth
of the xy-plane of the principal coordinate system.

Figure 3—Rotation between the principal coordinate system and the tool coordinate system. Here, xyz is the tool coordinate system;
XYZ is the principal coordinate system. The green line y= is the line of node.

There are a plethora of works for electromagnetic fields in triaxially anisotropic media. As far as
analytical methods for homogeneous media are concerned, one widely used method is to first find the
electromagnetic field in the principal coordinate system, and then apply rotation operation to find the field
in the tool coordinate system if the latter is not aligned with the former (Nekut 1994; Gianzero et al. 2002;
Yuan et al. 2010). In contrast, we choose to solve the problem directly in the tool coordinate system. In
so doing, the rotation is performed on the conductivity tensor before the computation instead of on the
field after the computation. The advantage of solving the problem in this manner is that the integrand of
the resultant 2D integral is more benign for large ␣ than in the principal coordinate system. To evaluate
the integral, we convert the integration into polar coordinate system over (k␳,␸). Gaussian quadrature is
then applied to the integration in the azimuthal direction first and then to the one over k␳, sequentially. The
integration over k␳ is similar to Sommerfeld integrals, except that there are no Bessel functions in our
formulation. Here, we use a technique commonly used for the evaluation of Sommerfeld integrals (Chang
and Mei 1981). The details of our new forward algorithm can be found in the Appendix.
Processing Method
The data acquired by a fully triaxial induction tool is a full apparent conductivity tensor. Given the data,
the goal of processing is to solve for three conductivity components and three Euler angles. We formulate
the problem as an optimization problem for which the cost function is given by
(5)

(6)
6 SPE-175111-MS

In the above, and are the measured and simulated apparent conductivity, respectively. We use
the Gauss-Newton method to solve this problem. The regularization parameter ␭ is determined using the
multiplicative regularization technique (Habashy and Abubakar 2004).
Because of the noise in data and the cyclic property of apparent conductivity as a function of Euler
angles, caution needs to be exercised on the selection of initials for Euler angles to avoid the iteration
process being trapped at a local minimum. We use multiple initials for Euler angles to address this
problem. The initials are selected in such a way that they are uniformly distributed in the model space to
cover all potential local minimums. The initials that we use in this paper are: 45° and 135° for ␥; 45° for
␥; and 45°, 135°, 225°, and 315° for ␤.
One problem that has not been discussed before is the ambiguity of the principal coordinate system.
It can be shown that there are six principal coordinate systems, each of which is associated with a
permutation of ␴x, ␴y, and ␴z. All of the six permutations give the same in the tool coordinate system.
The six principal coordinate systems are depicted in Fig. 4. Because the principal coordinate system is not
specified before inversion, the solution can fall into any of the six coordinate systems. For the results from
different depths and different runs to be comparable, three conductivity components from inversion should
be rearranged in the same order. We choose to rearrange them in a descending order. In other words, after
rearranging, ␴x ⱖ ␴y ⱖ ␴z. Such a choice is compatible with the selection of the rotation definition in Eq.
1. Obviously, when ␴x ⫽ ␴y, the formation becomes TI-anisotropic, and ␴x and ␴y become ␴h in a natural
manner.

Figure 4 —Ambiguity of the principal coordinate system. The same diagonal conductivity tensor ⴝ diag{␴1,␴2,␴3} has six different
forms corresponding to the six feasible orientations of the three principal axes. In addition to ⴝ diag{␴1,␴2,␴3}, the other five forms
are ⴝ diag{␴1,␴3,␴2} ⴝ diag{␴2,␴1,␴3}, ⴝ diag{␴2,␴3,␴1}, ⴝ diag{␴3,␴1,␴2}, and ⴝ diag{␴3,␴2,␴1}, respectively.

Rearranging ␴x, ␴y, and ␴z is straightforward, but finding new Euler angles corresponding to the new
order is not. Different methods exist for finding new Euler angles. In this paper, the new Euler angles are
found by solving the following minimization problem:
(7)
SPE-175111-MS 7

In the above, ␴INV is a triplet of ␴INVx, ␴INVy, and ␴INVz found from inversion; eINV ⫽ (␥INV, ␣INV,
␤ ) is a triplet of three Euler angles from inversion; ␴P is also a triplet of ␴INVx, ␴INVy, and ␴INVx, but
INV

rearranged in the descending order; eP ⫽ (␥p, ␣p, ␤p) is a triplet of new Euler angles corresponding to
triplet ␴P. Rearranging three conductivity components and finding the corresponding Euler angles is, in
fact, a rotation operation. It rotates the principal coordinate system directly from the inversion to a new
principal coordinate system such that three conductivity components are in a descending order.
From the reconstructed ␴INVx, ␴INVy, and ␴INVz, the triaxiality index ␹ can be derived:
(8)

Where ␴max and ␴min are respectively the maximum and minimum values of three inverted conduc-
tivity components, and ␴mid is the one between ␴max and ␴min. Obviously, the larger the triaxiality index,
the larger the difference is between three conductivity components. When ␹ ⫽ 0, at least two components
are equal to each other, so the formation will become TI-anisotropic. Therefore, the smaller the triaxiality
index, the more similar the formation is to a TI formation; the larger the triaxiality index, the more
different the formation is from a TI formation.
Most sedimentary formations are TI-anisotropic either because of the prevalence of layered structure
or the intrinsic anisotropy. Therefore, when factures are developed in the formation, the formation can
exhibit triaxial anisotropy if the filling fluid has a different resistivity than the formation. Theoretically,
for a single fracture to cause triaxial anisotropy, it must cut the formation at an angle. When the fracture
plane is nearly parallel to the lamination plane of formation, it does not cause large triaxiality. The vertical
conductivity of the formation may still be changed significantly, though. Therefore, we anticipate that the
triaxiality index will give a more clear indication when the fracture cuts the formation at a high angle.
Synthetic Examples
In this section, we apply the new processing method to two synthetic examples to understand its behavior
and assess its performance. We demonstrate the importance of multiple initials and the rotation processing
with the results in a synthetic triaxial anisotropic model. As a more practical testing, we then show the
results in a TI-anisotropic formation traversed by a single dipping fracture.
Example 1. In this example, the formation is homogeneous but triaxially anisotropic. Its three
conductivity components and Euler angles are given in the first row of Table 1. The second row contains
the same conductivity components, but in a descending order, and the corresponding Euler angles. We use
the forward method introduced in the Appendix to generate the synthetic apparent conductivity of a
triaxial induction tool with a spacing of 39 in. (distance from the transmitter to the main receiver). The
apparent conductivity is then contaminated with a 5-mS/m noise with a uniform distribution before being
fed to the inversion.

Table 1—Conductivities and Euler angles of a homogeneous formation with triaxial anisotropy
␴x (mS/m) ␴y (mS/m) ␴z (mS/m) ␥ ␣ ␤

100 500 200 108° 72° 54°


500 200 100 161.1° 72.9° 149.7°

The inverted conductivity components and Euler angles for different initial values of Euler angles are
summarized in Table 2. The initials for three conductivity components are all set to 10 mS/m and are kept
the same for all cases. Also shown in the rightmost column of Table 2 are the data misfits for all cases.
All the cases except for the last one converge to the same level of misfit. It appears that the last case does
8 SPE-175111-MS

not converge to the global minimum. It clearly shows the importance of using multiple initials for the
inversion. Except for the last case, all cases give the same set of conductivity, with the only difference
being in the order of conductivity components. However, their Euler angles are quite different. Compared
to the first row of Table 1, which is the model used to generate the synthetic data, it seems that only case
2 give the right answer. However, as mentioned previously, the seemingly wrong results are due to the
ambiguity of the principal coordinate system.

Table 2—Initial values for Euler angles, inverted conductivity and Euler angles, and data misfit
␴INVx ␴INVy ␴INVz
Case ␥0 ␣0 ␤0 (mS/m) (mS/m) (mS/m) ␥INV ␣INV ␤INV Misfit (%)

1 45° 45° 45° 490.9 102.3 204.3 18.91° 75.72° 47.73° 0.4
2 135° 45° 135° 102.3 490.9 204.3 108.9° 75.72° 47.73° 0.4
3 45° 45° 225° 204.3 490.9 102.3 74.94° 71.70° 142.6° 0.4
4 135° 45° 315° 490.9 204.3 102.3 164.9° 71.70° 142.6° 0.4
5 45° 45° 45° 102.3 204.3 490.9 38.15° 23.54° 282.0° 0.4
6 135° 45° 135° 204.3 102.3 490.9 128.2° 23.54° 282.0° 0.4
7 45° 45° 225° 204.3 102.3 490.9 128.1° 23.54° 282.0° 0.4
8 135° 45° 315° 113.8 184.7 481.8 179.2° 21.61° 298.4° 1.12

Once the inverted conductivity components of all cases are rearranged in the descending order, all the
Euler angles become identical immediately. The rearranged results are summarized in Table 3. Note that
case 4 does not need to be rearranged because its conductivity components are already in the descending
order. Compared to the rearranged true model in the second row of Table 1, the relative errors of three
conductivity components for cases 1-7 are approximately 1.8%, 2.2%, and 2.3%, respectively, and those
of Euler angles are 2.4%, 1.7%, and 4.7%, respectively. Considering the level of noise in the data, such
levels of errors are reasonable. Not surprisingly, the relative errors of the last case are higher, with the
largest error observed for ␤. It is approximately 21%, a large difference indeed. The other parameters do
not digress that much from their true values, though. We emphasize that, in practice, this set of results will
be dropped anyway because of the relatively large error, and hence they will not affect the final results.

Table 3—Inverted conductivity and Euler angles after rearranging in a descending order
Case ␴Px (mS/m) ␴Py (mS/m) ␴Pz (mS/m) ␥P ␣P ␤P

1 490.9 204.3 102.3 164.9° 71.70° 142.6°


2 490.9 204.3 102.3 164.9° 71.70° 142.6°
3 490.9 204.3 102.3 164.9° 71.70° 142.6°
4 Rearrangement of conductivities is not necessary.
5 490.9 204.3 102.3 164.9° 71.70° 142.6°
6 490.9 204.3 102.3 164.9° 71.70° 142.6°
7 490.9 204.3 102.3 164.9° 71.70° 142.6°
8 481.8 184.7 113.8 0.3101° 68.40° 117.6°

Example 2. In this example, the background formation is assumed to be TI-anisotropic and is traversed
by a single dipping fracture. To illustrate how the fracture dip affects the inversion result, we assume the
formation anisotropy is horizontal (i.e., its dip is 0°). When the dip vanishes, the anisotropy azimuth has
no effect on the tool response. Therefore, we set both the dip and azimuth of anisotropy to 0° in the
forward model. For fracture properties, we assume the fracture resistivity is isotropic because in most
realistic cases the filling material is fluid. This assumption allows for the use of four parameters for a
SPE-175111-MS 9

single fracture model: fracture conductivity ␴F, opening ␦F, dip ␣F, and azimuth ␤F, as illustrated
schematically in Fig. 5. Because the anisotropy orientation is not aligned with that of the fracture plane,
this model cannot be simulated with the commonly-used TI model. We assume that the fracture is
infinitely extending in the formation. This allows us to use the crossbed model for synthetic data
generation (Wang et al. 2014).

Figure 5—A TI-anisotropic formation traversed by a single dipping fracture. Here, the fracture azimuth ␤F is 0°. The anisotropy of the
formation is horizontal; therefore, its dip and azimuth are both set to 0°.

We first consider a fracture filled with a highly resistive material. This situation can happen for drilling
induced fractures in an oil-base mud well. Table 4 gives the model parameters for two cases used in the
study. Note that the difference of the two cases is only in fracture dip. The fracture azimuth may affect
Euler angle ␤ or the azimuth of the xy-plane of the principal coordinate system, but not the triaxiality
index. We are more interested in the latter; therefore, the fracture azimuth is set to 0° in both models. We
apply our new processing to the two cases. Their results are plotted in tracks 1-4, and tracks 5-8,
respectively, in Fig. 6. For each case, the leftmost track is the triaxiality index, followed by three
resistivity logs (reciprocals of ␴Px, ␴Py, and ␴Pz, respectively), and then the dip log of the xy-plane and the
␥ log.

Table 4 —Parameters of formation and fracture for a single resistive fracture model
Case ␴h (mS/m) ␴v (mS/m) ␴F (mS/m) ␦F (in) ␣F ␤F

1 1000 500 1 0.1 30° 0°


2 1000 500 1 0.1 60° 0°
10 SPE-175111-MS

Figure 6 —Triaxiality processing in the presence of a single resistive fracture dipping at 30° (tracks 1-4) and 60° (tracks 5-8).

The last track is data misfit recorded when the inversion stops, so it reflects how well the inverted
model fits the data. The triaxiality index shows that when the dip is 30°, the tool begins to see the fracture
when the tool is approximately 5 ft away from the fracture (measured by the distance from the record point
of the tool and the intersection of fracture plane and borehole axis). In contrast, when dip is 60°, the tool
begins to see the fracture when the tool is as far as 10 ft away. The dips of the xy-plane or Euler angle
␣ for the two cases are 20° and 52°, respectively, when the tool is right opposite to the fracture (depth is
0 ft). Theoretically, ␣ is affected by both formation anisotropy dip and the fracture dip; therefore, it is a
tradeoff between the two dips. In this case, because of the high resistivity of fracture, ␣ is dominated by
the fracture dip. It is also seen that Euler angle ␤ (shown as the orientation of tadpoles) clearly reflects
the fracture orientation. This is expected because the formation anisotropy is horizontal. Otherwise, it
would be affected by both the fracture and formation anisotropy azimuth. In principle, the accuracy of
Euler angle ␥ is determined by the contrast of ␴x and ␴y. When ␴x and ␴y differs significantly, the inverted
␥ becomes more stable if the fracture does not change rapidly with depth. This stable feature can be seen
when the tool is close to the fracture. Otherwise, as is shown that when the tool is 2-5 ft away from the
fracture, ␥ log begins to exhibit rapid variation with depth. Apparently, the data misfits are relatively high
near the facture. This is not surprising because in the triaxially anisotropic model, it is actually assumed
that many parallel fractures are uniformly distributed in the whole space. Here, because only one single
fracture is present, the triaxially anisotropic model should not be expected to capture all the details of this
model.
The second scenario is when the fracture is filled with a highly conductive material. This situation can
occur when the fracture is filled with either the filtrate of water-base mud or salty formation water,
assuming the formation is relatively resistive. The parameters in the study are given in Table 5. Here, we
assume the fracture conductivity is 104 mS/m. As in the above example, we change only the dip of fracture
and study its effect on the triaxiality index. The results for two dips are plotted in Fig. 7. In comparison
to the triaxiality index for the resistive fracture (Fig. 6), the triaxiality index for the conductive fracture
SPE-175111-MS 11

gives a more clear indication. However, the dip logs show a relatively low reading. The Euler angle ␤ right
opposite to the fracture is approximately 0°, reflecting correctly the fracture orientation. However, at some
depths near the fracture, ␤ is slightly off 0°. Nonetheless, overall Euler angle ␤ is still a good estimation
of fracture orientation. The difference between three resistivity logs reflect the change of resistivity in the
fractured zone, but the high amplitude of Rz appears to be counterintuitive. In addition to the model
mismatch between the triaxially anisotropic model and a single fracture model, the stronger nonlinearity
of the tool response to conductive fracture is probably another reason for the large Rz reading.

Table 5—Parameters of formation and fracture for a single conductive fracture model
Case ␴h (mS/m) ␴v (mS/m) ␴F (mS/m) ␦F (in) ␣F ␤F

1 100 50 10000 0.2 30° 0°


2 100 50 10000 0.2 6° 0°

Figure 7—Triaxiality processing in the presence of a single conductive fracture dipping at 30°(tracks 1– 4) and at 60° (tracks 5– 8).

Yet another phenomenon worth mentioning is that the triaxiality index is more affected by the
conductive fracture than the resistive fracture when the tool is far away from the fracture. It is seen that
when the dip of fracture is 30°, the triaxiality index begins to tick up when the tool is approximately 10
ft away. Comparatively, when the dip of fracture is 60°, it begins to sense the effect when the tool is as
far as 25 ft away. However, we emphasize that this should not be interpreted as that the triaxiality index
can be used at such a large distance. In field conditions, a variety of noise can exist in field data, and such
small levels of signal are easily swamped in noise, and hence cannot be used as a reliable indicator for
fracture presence. Moreover, the actual fractures are always finite in terms of both length and width, hence
do not affect the triaxiality index in exactly the same way.
The above results suggest that the triaxiality index is indeed able to indicate the presence of a fracture,
regardless of whether it is resistive or conductive. In general, the larger the fracture dip, the larger the
12 SPE-175111-MS

triaxiality index is. The triaxiality index is more affected by the conductive fracture, especially when the
fracture dip is large. In addition, the nonlinearity effect of the tool response appears to be stronger for the
conductive fracture. In this study, we have been concentrated on the effect of dip and the resistivity of
filling material. We have not discussed the effect of fracture length, width, and porosity, even though they
all can affect the triaxiality index and six parameters in certain way. Considering the scope and the
importance, these problems are better discussed in greater detail in a separate paper, and therefore are
covered in this paper.

Field Example
Fig. 8 shows the results with the new processing method in a south Texas gas shale well drilled with oil-
base mud. The 39-in. data of the triaxial induction tool were used as field data to the inversion. The
triaxiality index, three resistivity logs, and Euler angles are plotted in track 3, 4, and 5, respectively. As
in the synthetic case, because ␣ and ␤ are the dip and azimuth of the xy-plane of the principal coordinate
system, they are plotted together as tadpoles as for formation dips. The 39-in. vertical fracture index (VFI)
computed with the vertical fracture detection method (Wu et al. 2013) is shown as reference in track 2.
Plotted in track 6 are the handpicked dips from borehole resistivity image logs. The array induction logs
are in track 7. The last track on the right is the overlay of the misfit logs from the triaxiality inversion and
a zero-D inversion. Gamma ray, hole diameter, hole deviation, and hole azimuth are shown in the first
track.
SPE-175111-MS 13

Figure 8 —Triaxiality processing in a South Texas gas shale well drilled with oil-base mud.

High gamma ray readings with little variation over this 400-ft zone indicates that the formation is a
fairly uniform shale. However, large excursions of array induction logs in track 7 suggest that there is
something abnormal happening in the formation. Obviously, the large excursions correlate quite well with
14 SPE-175111-MS

the change of VFI log. In fact, previous study has shown that these excursions are very probably due to
fractures (Wu et al. 2013). In general, the triaxiality index and the VFI log match quite well with each
other in the entire zone. The only zone showing large difference is at 188 ft. However, handpicked dips
at the same depth clearly show that there are fractures in the formation. Note that handpicked dips show
up at the depths where the triaxiality index exhibits high values, (i.e., 108.5 ft, 188 ft, and 267 ft). On the
other hand, at many places there are no indications of fractures on handpicked dips (e.g., at 160 ft, 220
ft, and 350 ft). However, both VFI and triaxiality index indicate there are some abnormalities in the
formation. This situation can happen when fractures are so small that the human eye cannot distinguish
them from other features or when fractures are behind the borehole wall and do not intersect with the
borehole.
Three resistivity and Euler angle logs also provide useful information about fractures. Note that the
formation dip is low (⬍4°), so fracture orientation can be estimated from resistivity and Euler angle logs.
Moreover, the hole deviation is small (⬍3°), therefore the apparent dip is basically identical to the true
dip, and no conversion operation is needed. This enables us to compare the dip estimation with
handpicked dips (defined in the earth coordinate system) directly. Note that at 188 ft, Rz is more than 10
times larger than Rx and Ry and Rx and Ry are similar to Rh and Rv in fracture-free zones (i.e., at 300 ft).
So we can reasonably infer that the current in the z-direction crosses the fracture in a direction nearly
perpendicular to the fracture plane. Therefore, the dip of the xy-plane is nothing but the fracture dip in this
case. Obviously, both the xy-plane and the handpicked fracture are inclined towards the north-west
direction, and both indicate a nearly vertical fracture. The good agreement between handpicked dips and
inverted dips is also seen at 108.5 ft and 267 ft. Table 6 contains numeric values of the handpicked dip
and the inverted dips at the three depths. At some depths (e.g. 127 ft), handpicks dips shows the orientation
of fracture is in the opposite direction of those in the vicinity. Considering the dip is nearly 90°, the
fracture in fact is nearly the same as the one oriented in the opposite direction.

Table 6 —Handpicked dips and dips from the triaxiality processing.


Depth (ft) DIP_Handpicked AZI_Handpicked DIP_INV AZMNV

108. 5 79.1° 307.8° 85.4° 309.6°


188 77.3° 300. 8° 80.8° 303.1°
267 81.7° 305.1° 87.2° 311.3°

The accuracy of the ␥ log, plotted in red in track 5, as discussed before, relies on the contrast of Rx and
Ry. In fracture-free zones (i.e., at 20 ft, 300 ft, and 390 ft), Rx is nearly equal to Ry, so the ␥ log becomes
jittery, suggesting there is a high level of uncertainty in ␥. Generally, in fractured zones, because the
contrast of Rx and Ry is relatively large, the ␥ log becomes more stable. However, it may still change
rapidly if the orientation of fractures changes rapidly with depth. The ␥ value is important for inferring
fracture orientation if the principal coordinate system is not aligned with the tool coordinate system. The
last track shows the misfit logs from both the triaxiality processing and a zero-D inversion. In the zero-D
inversion, a homogeneous TI-anisotropic formation is used. Therefore, it is expected that such a simplified
model may not work well in triaxially anisotropic formations. In the zones with large triaxiality, the misfit
from the triaxiality processing is lower than that of the zero-D inversion. Obviously, by using the triaxially
anisotropic mode, the inverted model indeed fits the data better than using the TI-anisotropic model.
Conclusions
In this paper, we have introduced a new method using the triaxiality index for fracture detection. The
triaxiality index is a measure of the difference between three conductivity components of the formation
SPE-175111-MS 15

conductivity tensor. The new method is based on the fact that the presence of a fracture can cause large
triaxial anisotropy in conductivity if the resistivity contrast between the filling fluid and formation is
significantly different.
We have applied the new method to both the synthetic and field examples. The results of the synthetic
examples demonstrate that the triaxiality index is indeed able to detect a fracture regardless of the type
of filling material. In general, the larger the fracture dip, the larger the triaxiality index is, suggesting that
the new method is more sensitive to high-angle fractures. The results show that the effect of a resistive
fracture is more localized than that of a conductive fracture. Moreover, the nonlinearity of the triaxial
induction tool appears to be stronger for a conductive fracture than for a resistive facture. As a result, Rz,
the largest resistivity of three resistivity components, can be significantly higher than both the formation
and fracture resistivity for a conductive fracture.
The results of the new processing method in the field example compare well with those of the vertical
fracture detection method. Similar to the vertical fracture index, the triaxiality index is also responsive to
fractures not intersecting with the borehole. Since the relative dip of anisotropy is low, the inverted Euler
angles can be used to estimate fracture orientation. Comparison with the handpicked dips shows that the
estimation using Euler angles works reasonably well. In realistic situations, however, the formation
anisotropy can be tilted relative to the tool plane. As a result, the fracture plane may be oriented in any
direction relative to the orientation of anisotropy. For such a complex situation, a more sophisticated
model is needed to extract the fracture orientation from inverted conductivity tensor and Euler angles.
We have developed a new forward model and a new processing method to invert for three conductivity
components and Euler angles. In the forward model, we solve the problem in the tool coordinate system
instead of the principal coordinate system. By so doing, the integrand decays exponentially with the
increase of wavenumber k␳, making the integration efficient and accurate for all values of Euler angle ␣.
Besides, by using symmetric relationships of formation conductivity tensor, the ranges of Euler angles are
reduced significantly. More importantly, the nonuniqueness of Euler angles is thus eliminated for a given
order of conductivity components.
We have found that there are six principal coordinate systems that can be used to describe the
conductivity tensor. All of the six coordinate systems give the same conductivity in the tool coordinate
system. A rotation operation must be performed after the inversion to have a proper interpretation of
inverted conductivity and Euler angles. The rotation is performed by first rearranging three conductivity
components in a given order, then finding the new Euler angles by solving a minimization problem.
Currently it takes about 2 to 3 hours to process a 100-ft-long interval in a desktop or laptop computer.
The main reason for the slowness is the need to evaluate 2D integrals in each call to the forward solver.
Such a speed is not quite practical. The direct solution is to replace the forward solver with a table.
Considering there are only six parameters in the forward model, building such a table is not a formidable
task. It can be built in a reasonable time frame in desktop or laptop computers using high-performance
computing techniques. We anticipate that the processing can be performed at real-time once the table
look-up method substitutes for the rigorous forward solver.

Symbols and Nomenclature


field magnitude determined by the source
eINV a triplet consisting of three inverted Euler angles
eP a triplet of new Euler angles corresponding to ␴P
f frequency of the operation current
and type I and type II mode for the electric field
i the imaginary unit, i ⫽
ks a vector of transverse wave numbers
16 SPE-175111-MS

kx, ky and kz wave numbers in the x-, y- and z-directions of the tool coordinate system
and wave numbers of type I and type II modes in the z-direction, respectively
kp radial wave number, or the radial coordinate in the polar coordinate system
rs and r=s transverse coordinates of the field and source points, respectively
sx source term of a magnetic dipole in the x-direction in the wave number domain
sy source term of a magnetic dipole in the y-direction in the wave number domain
sz source term of a magnetic dipole in the z-direction in the wave number domain
x a vector of unknowns in the inversion, consisting of three conductivity components ␴x, ␴y,
and ␴z, and three Euler angles ␥, o and ␤
z and z= vertical coordinates of the field and source points, respectively
C cost function of the inversion problem
E electric field
H magnetic field
៮k operator ␦⫻␦⫻ in the wave number domain
a 2x2 diagonal matrix consisting of two wave numbers in the z-direction
M magnetic current source
Rx formation resistivity in the x-direction of the principal coordinate system
Ry formation resistivity in the y-direction of the principal coordinate system
Rz formation resistivity in the z-direction of the principal coordinate system
, and elemental rotation matrices for Euler angles ␥, ␣ and ␤, respectively
sensitivity matrix of the triaxial induction tool
␣F and ␤F fracture dip and azimuth, respectively
␥, ␣ and ␤ Euler angles
␥ ,␣
INV INV
and ␤INV Euler angles obtained with the inversion
␥ , ␣ and ␤P new Euler angles corresponding to ␴P
p P

␦F fracture opening
␥ regularization parameter of the inversion
⫾ ⫾
␭ I and ␭ II coefficients associated with type I and type II modes from the decomposition
␮0 vacuum magnetic permeability
formation conductivity tensor in the tool coordinate system
␴INV a triplet consisting of three inverted conductivity components
␴ P
a triplet of three inverted conductivity components rearranged in the descending order
a simulated apparent conductivity tensor
measured apparent conductivity tensor
␴F fracture conductivity
␴h and ␴v horizontal and vertical conductivity of the background formation
␴x formation conductivity in the x-direction of the principal coordinate system
␴y formation conductivity in the y-direction of the principal coordinate system
␴z formation conductivity in the z-direction of the principal coordinate system
␴zz elemental conductivity at the lower right corner of formation conductivity tensor ␴ in the
tool coordinate system
␴max maximum value of three inverted conductivity components
␴min minimum value of three inverted conductivity components
␴mid intermediate value of three inverted conductivity components
␴xINV ␴x in the principal coordinate system obtained with the inversion
␴y INV
␴y in the principal coordinate system obtained with the inversion
␴zINV ␴z in the principal coordinate system obtained with the inversion
␴x P
␴x after rearranging three conductivity components in the descending order
SPE-175111-MS 17

␴yP ␴y after rearranging three conductivity components in the descending order


␴zP ␴z after rearranging three conductivity components in the descending order
␸ polar angle in the polar coordinate system
␹ triaxiality index
␻ angular frequency of the operation current

Acknowledgments
The authors would like to thank the oil company for kindly allowing us to use their data in this paper.

References
Barber, T., Anderson, B., Abubakar, A. et alet al. 2004. Determining Formation Resistivity Anisotropy
in the Presence of Invasion. Presented at the SPE Annual Technical Conference and Exhibition,
Houston, Texas, USA, 26-29 September. SPE-90526-MS. http://dx.doi.org/10.2118/90526-MS.
Chang, S. K. and Mei, K. K. 1981. Multipole Expansion Technique for Electromagnetic Scattering by
Buried Objects. Electromagnetics 1: 73–89. http://dx.doi.org/10.1080/02726348108915125.
Davydycheva, S., Druskin, V., and Habashy, T. 2003. An Efficient Finite-Difference Scheme for
Electromagnetic Logging in 3D Anisotropic Inhomogeneous Media. Geophysics 68(5): 1525–
1536. http://dx.doi.org/10.1190/L1620626.
Georgi, D., Schön, J., and Rabinovich, M. 2008. Biaxial Anisotropy: Its Occurrence and Measurement
with Multi-Component Induction Tools. Presented at the SPE Annual Technical Conference and
Exhibition, Denver, Colorado, USA, 21-24 September. SPE-114739-MS. http://dx.doi.org/
10.2118/114739-MS.
Gianzero, S., Kennedy, D., Gao, L. et alet al. 2002. The Response of a Triaxial Induction Sonde in a
Biaxial Anisotropic Medium. Petrophysics 43(3): 172–184.
Habashy, T. M. and Abubakar, A. 2004. A General Framework for Constraint Minimization for the
Inversion of Electromagnetic Measurements. Progress in Electromagnetic Research 46: 265–312.
http://dx.doi.org/10.2528/pier03100702.
Habashy, T. and Abubakar, A. 2007. A Generalized Material Averaging Formulation for Modelling
of Electromagnetic Fields. J. of Electromagnetic Waves and Applications 21(9): 1145–1159.
http://www.tandfonline.com/doi/abs/10.1163/156939307794731123.
Huang, Y. and Lee, J. K. 2011. Dyadic Green’s Functions for Unbounded and Two-Layered General
Anisotropic Media. Progress in Electromagnetics Research B. 30: 27–46. http://dx.doi.org/
10.2528/pierb11031306.
Moskow, S., Druskin, V., Habashy, T. et alet al. 1999. A Finite Difference Scheme for Elliptic
Equations with Rough Coefficients Using a Cartesian Grid Nonconforming to Interfaces. SIAM
Journal of Numerical Analysis 36(2): 442–464. http://dx.doi.org/10.1137/s0036142997318541.
Nekut, A. G. 1994. Anisotropy Induction Logging. Geophysics 59(3): 345–350. http://dx.doi.org/
10.1190/L1443596.
Rabinovich, M., Mogilatov, V. S., and Tabarovsky, L. A. 2008. Processing of Multi-Component
Induction Measurements in a Biaxially Anisotropic Formation. US Patent Application No.
20080215243A1.
Sævik, P. N., Berre, I., Jakobsen, M. et alet al. 2012. Electrical Conductivity of Fractured Media: A
Computational Study of the Self-Consistent Method. Presented at SEG International Exposition
and 82th Annual Meeting, Las Vegas, Nevada, 4-9 November. http://dx.doi.org/10.1190/
segam2012-1026.1.
Tang, Y. 2008. Modeling and Inversion of Multicomponent Induction Logs in a Biaxial Anisotropic
Formation. PhD thesis, University of Houston, Houston, Texas (May 2008).
18 SPE-175111-MS

Thompson, L. 2000. Atlas of Borehole Imagery, AAPG Datapages Discovery Series 4, American
Association of Petroleum Geologists, Tulsa.
Torquato, S. 1991. Random Heterogeneous Media: Microstructure and Improved Bounds on Effective
Properties. Appl. Mech. Rev. 44(2): 36 –76. http://dx.doi.org/10.1115/1.3119494.
Wang, G., Barber, T., Wu, P. et alet al. 2014. Triaxial Induction Tool Response in Dipping and
Crossbedded Formations. Presented at SEG International Exposition and 84th Annual Meeting,
Denver, Colorado, 26-31 October. http://dx.doi.org/10.1190/segam2014-1406.1.
Wang, G. L., Wu, P., Barber, T. et alet al. 2012, Triaxial Induction Applications in Difficult and
Unconventional Formations. Presented at SPWLA 53rd Annual Logging Symposium, Cartagena,
Columbia, 16-20 June. Paper G.
Wang, H., Barber, T., Morriss, C. et alet al. 2006, Determining Anisotropic Formation Resistivity at
Any Relative Dip Using a Multiarray Triaxial Induction Tool. Presented at the SPE Annual
Technical Conference and Exhibition, San Antonio, Texas, 24-27 September. SPE-103113-MS.
http://dx.doi.org/10.2118/103113-MS.
Wu, P., Barber, T., Wang, G. L. 2013, Fracture Characterization Using Triaxial Induction Tools.
Presented at SPWLA 54th Annual Logging Symposium, New Orleans, Louisiana, 22-26 June.
Paper CC.
Yin, H. 2010. Method for Reservoir Fracture and Cross Beds Detection Using Tri-Axial/Multi-
Component Resistivity Anisotropy. US Patent Application No. 20100230095A1.
Yuan, N., Nie, X.C., Liu, R. et alet al. 2010. Simulation of Full Responses of a Triaxial Induction Tool
in a Homogeneous Biaxial Anisotropic Formation. Geophysics 75(2): E101-E114. http://dx.do-
i.org/10.1190/L3336959.
SPE-175111-MS 19

Appendix A
Response of a Triaxial Induction Tool in a Homogeneous and Triaxially Anisotropic Formation

The frequency domain vector-wave equation for electric field E in an anisotropic medium is
(A.1)

In Eq. A.1, the source is assumed to be a magnetic current, designated by M; the formation is assumed nonmagnetic, hence,
␮ ⫽ ␮0; ␻ ⫽ 2␲f is the angular frequency of the operation current, with f being its frequency. The time dependence is e⫺i␻t
where i is the imaginary unit, . The conductivity of a homogeneous formation is given in Eqs. 1-4.
Applying the Fourier transform on both sides of Eq. A.1, and then using the residual theorem, it is found that the electric
field E can be written as:
(A.2)

where rs ⫽ x ⫹ yŷ and r=s ⫽ x= ⫹ y=ŷ are the transverse coordinates of the field and source points, respectively in the
tool coordinate system; z and z= are their axial coordinates, respectively; ks ⫽ kx ⫹ kyŷ designates transverse wave numbers;
consists of two axial wave numbers, where subscripts I and II denote type I and type II modes,
respectively. k⫾z,I and k⫾z,I can be found by solving the dispersion equation as given by
(A.3)

where
(A.4)

The plus and minus signs designate upgoing and downgoing waves, respectively; b⫾␩,0 is the field magnitude determined
by the source,
(A.5)

In Eq. A.5, and are the source terms in the spectral domain. Here, ␩ stands for the source

orientation, ␩ ⫽ x, y, z; It is readily found that


(A.6)

(A.7)

(A.8)

They correspond to a magnetic dipole directed in x-, y-, and z-direction, respectively in the tool coordinate system. Unit
vectors K៮ ⫾I and K៮ ⫾II are the type I and type II modes of the electric field, and ␭⫾I and ␭⫾II are their associated coefficients.
The two wave modes K៮ ⫾I and K៮ ⫾II can be found by the following decompositions (Huang and Lee 2011):
(A.9)

(A.10)

The two associated coefficients ␭⫾I and ␭⫾II in Eq. A.5 are given by
20 SPE-175111-MS

(A.11)

(A.12)

where
(A.13)

In the above, ␴zz is an element of ␴ ៮ in the tool coordinate system, ␴zz ⫽ (␴


៮ )i⫽3, j⫽3.
The 2D integral in Eq. A.2 can be evaluated in the polar coordinate system over (k␳,␸). Here, k␳ is the radial wavenumber,
and ␸ is the polar angle. For our application, rs ⫽ rs= ⫽ 0 because both transmitter and receiver are situated on the z-axis. As
a result, the integrand decays exponentially with the increase of k␳, a property amenable to numerical integration. Once the
expression of electric field is found, the magnetic field can be found by
(A.14)

Eq. A.14 gives the magnetic field in the tool coordinate system. For a triaxial induction tool consisting of one transmitter
and one receiver, the apparent conductivity is given by
(A.15)

where S៮ is a constant sensitivity-factor matrix (Wang, et. al. 2006) and o is a symbol for a Hadamard product, or
element-wise matrix product. For a realistic triaxial induction tool, ˆH is a superposition of the magnetic field at the main and
the bucking receivers.
Author Biographies
Gong Li Wang has worked on induction modeling, processing, and interpretation since he joined Schlumberger in 2008.
Before joining Schlumberger, he worked with the University of Petroleum, East China, on electromagnetic well logging from
1993 to 2002. From 2002-2008, he worked with the Center for Computational Electromagnetics and Electromagnetics
Laboratory of the University of Illinois at Urbana-Champaign and the Center for Petroleum and Geosystems Engineering of
the University of Texas at Austin. He received a bachelor’s degree in geophysical well logging and a PhD degree in applied
geophysics from the University of Petroleum, East China, in 1993 and 2001, respectively. In 2001, he was awarded the Second
Prize of the Award for Technological Innovation by China National Petroleum Corporation. His research interests have been
in wave propagation in inhomogeneous media, numerical modeling methods, and inverse problems.
Aria Abubakar received an MSc degree (cum laude) in electrical engineering and a PhD degree (cum laude) in technical
sciences, both from the Delft University of Technology, in 1997 and 2000, respectively. From September 2000 until February
2003, he was a researcher with the Laboratory of Electromagnetic Research and Section of Applied Geophysics, Delft
University of Technology. He joined Schlumberger-Doll Research Center in 2003 and his last position there was a Scientific
Advisor and the Program Manager of Multi-Physics Modeling and Inversion program. Since 2013, he has been the
Interpretation Engineering Manager and Scientific Advisor at Schlumberger Houston Formation Evaluation in Sugar Land,
Texas. At present, his main research activities include solving forward and inverse problems in acoustics, electromagnetics, and
elastodynamics. He is currently the Associate Editor of Geophysics. He was the 2014 SEG North America Honorary Lecturer.
He holds 14 US patents/patent applications and has published 1 book, 4 book chapters, over 75 scientific articles in refereed
journals, 155 conference proceedings papers, and 50 conference abstracts. He has also presented over 300 invited and
contributed talks in international conferences and institutes/universities.
Tom Barber has worked on induction modeling, array design, and environmental corrections since he joined Schlumberger
Well Services in Houston in 1978. There he developed the first commercial signal—processing algorithm for induction tools,
phasor processing, and the answer products for the array induction tool family. Together with Barbara Anderson, he developed
the concept and the methods for using modeling as an integral part of log interpretation. He proposed the concept for a
SPE-175111-MS 21

multiarray triaxial induction tool that became the Rt Scanner* triaxial induction service, and coordinated its physics
development. Author of numerous papers and holder of 32 patents, he was awarded the SPWLA Distinguished Technical
Achievement Award in 1993 for significant contributions in electromagnetic logging, and the SPWLA Gold Medal for
Technical Achievement in 2011. He previously worked on magnetic susceptibility logging measurements at Schlumberger-Doll
Research Center, Ridgefield, Connecticut. Before joining Schlumberger he worked at the NASA Marshall Flight Center,
Huntsville, Alabama, and as a postdoc at Brookhaven National Laboratory, Upton, NY. Tom did undergraduate work in physics
at Vanderbilt University, Nashville, Tennessee, and did graduate work in low- temperature physics at the University of Georgia,
Athens, Georgia.
Samer Alatrach was Principal Petrophysicist with Schlumberger Consulting Services in Denver, Colorado. He received
his BS degree in Civil Engineering from Kuwait University in 1984. Between 1991 and 1996, Samer worked for Schlumberger
Oilfield Services as a wireline logging engineer. He evaluated wells in the Middle East, India and United States. Samer has
been with Schlumberger Consulting Services since September 2004. During the past 10 years, he has worked on many
consulting projects (both domestically and internationally) for many different clients. He has expert knowledge on openhole
log analysis in tight gas sands and shale gas evaluation as well. Samer has provided petrophysical analysis for many large
integrated field studies on unconventional gas reservoirs for reserves evaluation and field optimization. He is currently the
Petrophysics/Acoustics Domain champion with Schlumberger Wireline for North America offshore.,

* Mark of Schlumberger

You might also like