REE170609 DOI: 10.

2118/170609-PA Date: 11-May-16 Stage: Page: 1 Total Pages: 13

Imaging Radial Distribution of Water

Saturation and Porosity Near the Wellbore
by Joint Inversion of Sonic and
Resistivity Logging Data
S. Shetty, L. Liang, T. M. Habashy, V. Simoes, A. J. Boyd, B. K. Sinha, and S. Zeroug, Schlumberger; and
C. Sirju, T. N. Pritchard, and B. Glassborow, BG Group

Summary methods. When drilling with oil-based mud, additional uncer-

We present a work flow for joint inversion of sonic flexural-wave tainty arises from the varying amount of filtrate invasion and its
dispersion data and array-induction resistivity data acquired in a effect on log response. In such situations, using acoustic dipole
vertical well. The work flow estimates a pixel-based radial distri- flexural-wave and array-induction logs in the petrophysical work
bution of water saturation and porosity extending several feet flow provides an alternative method for evaluating porosity and
into the formation at each log depth. Radial changes in saturation water saturation. Shear and flexural waves are less influenced by
and porosity are caused by mud-filtrate invasion and mechanical fluid, and coupling the radial responses of the dipole flexural-
damage, respectively. The flexural-wave and array-induction wave and array-induction tools in a joint-inversion framework can
data have similar multiple investigation depths extending several provide porosity and water saturation from the near-wellbore
feet into the formation. Furthermore, flexural-wave data are sen- region to the far-field region.
sitive to porosity but have weak sensitivity to saturation, whereas Here, we introduce an inversion-based work flow for estimat-
induction data are sensitive to both porosity and saturation. Thus, ing water saturation and porosity from sonic flexural-wave disper-
integration of these data in a joint inversion can help to charac- sion and array-induction data acquired in a vertical well. The
terize the formation beyond the altered zone and reduce uncer- inversion solves for a pixel-based model of water saturation and
tainty of the interpretation. The work flow is validated on porosity that extends radially from the wellbore several feet into
synthetic data for several scenarios of near-wellbore alteration. the formation at each log depth. The inversion result is simultane-
The work flow is then applied to field data from an offshore well ously consistent with radial sensitivities of all the data. Petrophys-
drilled with oil-based mud in a gas-bearing clastic formation. ical transforms specific to the formation lithology map water
The results are compared with traditional interpretation and core saturation and porosity to elastic velocities and resistivity. These
analysis, demonstrating an efficient and accurate inversion-based latter properties are entered into electromagnetic and sonic tool-
work flow that can complement traditional formation evaluation response simulators for simulating the data in the inversion. The
in challenging conditions. mapping helps to reduce nonuniqueness by constraining the solu-
tion space of the inversion to satisfy both the petrophysical trans-
forms and the physical equations governing the tool response. The
Introduction mapping is coupled with physical bounds on the properties and
Water saturation and porosity of the formation are routinely esti- smoothness regularization to further reduce nonuniqueness.
mated by petrophysical interpretation of sonic, resistivity, and For the examples in this paper, the petrophysical transform
density logging data. Accurate estimation of formation properties applied to map elastic properties is an effective-medium rock-
requires that the interpretation account for near-wellbore altera- physics model. These models facilitate analysis of sonic data in
tion caused by mud-filtrate invasion or mechanical damage. formations with complex mineralogy consisting of multiple
Currently, this is achieved through separate inversions of multire- minerals and porosity types. Different porosity types may be
solution data that provide a parametric invasion profile for resis- characterized by their pore shape (Xu and White 1995; Xu and
tivity (Howard 2002), homogeneous formation density (Allioli Payne 2009), which is an additional output from the work flow.
et al. 1997), or radial distributions of elastic velocities (Burridge Relative changes in pore shape could demarcate intervals with
and Sinha 1996; Zeroug et al. 2006). However, water saturation round vugs or thin cracks, helping to inform completion and
and porosity derived from these inversions may not be mutually production decisions.
consistent because of different formation volumes investigated by In the following sections, we describe the input measurements,
the various measurements, restrictive formation parameteriza- forward-modeling framework, measurement sensitivities, and
tions, or nonuniqueness of the inverse problems. As shown in inversion work flow. The work flow is validated on synthetic data
prior work (Mallan et al. 2009; Gao et al. 2013), multiphysics relevant to wells drilled with oil-based mud in an oil-bearing pre-
inversions have the potential to improve the accuracy and reliabil- salt carbonate reservoir offshore Brazil, and a gas-bearing clastic
ity of the petrophysical interpretation. offshore East Africa. The work flow is applied to field data for the
In addition, petrophysical analysis for determining porosity gas-bearing clastic reservoir, and results are compared with tradi-
and water saturation can be challenging in the presence of light tional interpretation and core analysis. Finally, we review the
hydrocarbons because of their strong effect on density-neutron, assumptions underlying the work flow to enable a quality check
nuclear-magnetic-resonance, and sonic sensors. In new frontier of the results.
areas of offshore Brazil and East Africa, complex mineralogy
arising from the presence of feldspars, heavy minerals, and vary-
ing clay types further complicates traditional log-interpretation Measurements and Forward Modeling
The sonic measurements consist of dispersion data for the well-
bore-guided flexural wave excited by a dipole sonic logging tool
Copyright V
C 2016 Society of Petroleum Engineers
(Pistre et al. 2005). These data are sensitive to formation elastic
This paper (SPE 170609) was accepted for presentation at the SPE Annual Technical properties at radial depths of investigation (DOIs) ranging from
Conference and Exhibition, Amsterdam, 27–29 October 2014, and revised for publication.
Original manuscript received for review 15 January 2015. Revised manuscript received for one-half to three wellbore diameters into the formation. The
review 2 February 2016. Paper peer approved 5 February 2016. lower-frequency data have deeper DOIs than the higher-frequency

2016 SPE Reservoir Evaluation & Engineering 1

ID: jaganm Time: 14:16 I Path: S:/REE#/Vol00000/160028/Comp/APPFile/SA-REE#160028

 REE170609 DOI: 10.2118/170609-PA Date: 11-May-16 Stage: Page: 2 Total Pages: 13

Petrophysical Tool-response
transforms Rh (ri ) simulators
Pixel-based
formation model
Archie’s ri rBH EM Array-induction
Law Solver data
Sw (ri ),φ(ri )

Rh
ri

rBH
Vp(ri ), Vs(ri )

Acoustic
Rock ri rBH Guidance Dispersion data
Sw, φ, α Physics Condition

Vp, Vs, α
Assume isotropic, thick beds

Fig. 1—Forward-modeling framework with formation model linked to tool-response simulators through petrophysical transforms.

data (Burridge and Sinha 1996). The vertical resolution of the forward models to simulate the data, as illustrated in Fig. 1. The
data depends on the aperture of the receiver array, ranging from 6 formation elastic velocities and resistivity are isotropic in each
ft for the entire array to a maximum of 1 ft with multishot subar- pixel. The formation is vertically and azimuthally homogeneous
ray processing. For waves excited in a fluid-filled wellbore sur- within the union of the formation volume investigated by all chan-
rounded by a cylindrically layered elastic medium, the wellbore nels of both sonic and resistivity data.
acoustic wave spectrum can be shown to arise from the contribu-
tion of a few poles and branch lines in the complex wave-number Petrophysical Transforms. The petrophysical transforms are
plane. The wavefield associated with each pole corresponds to a chosen depending on the formation lithology. For the examples in
wellbore-guided wave mode. The poles are defined by the zeros this paper, we apply Archie’s law to compute formation resistiv-
of the determinant of the so-called acoustic guidance matrix, with ity, and the Xu-Payne effective-medium rock-physics model (Xu
the latter determined from the boundary conditions satisfied by and Payne 2009) to compute formation elastic properties. The
the wavefield at each layer interface. The poles may be computed Xu-Payne model has been benchmarked on formations with mul-
with a root-finding algorithm (e.g., Newton-Raphson algorithm), tiple minerals and pore types such as shaly sandstones and carbo-
and the dispersion curves for various wellbore-guided modes are nates. To compute effective elastic moduli from this model,
obtained by tracking the poles over the frequency range of interest moduli of the dry-rock frame are first computed from Kuster-
(Tang and Cheng 2004). Toksoz and differential-effective-medium models, followed by
The resistivity measurements consist of induction data Gassmann fluid substitution to compute moduli of the fluid-
acquired with a triaxial array-induction tool. This tool has two saturated rock. Pores in the clay that contain bound water are
source frequencies of 13 and 26 kHz, and an array of eight included in the dry-frame calculation rather than in the fluid-
receivers spaced from 9 to 72 in. from the source (Rosthal et al. substitution calculation. Effective bulk modulus of the pore-fluid
2003). Here, we include responses for only the axial source and mixture is computed from Wood’s law. We test the option of
axial receivers at a single frequency of 26 kHz. This subset of applying an empirical pore-fluid mixing law if gas is present in
data can also be acquired with earlier versions of the tool and is the pores (see Appendix B).
sensitive to formation resistivity at DOIs ranging from 1 to 7 ft The inputs to the Xu-Payne model consist of elastic moduli,
and vertical resolution ranging from 1 to 4 ft (Barber et al. 2004). densities, and volume fractions of the individual mineral and
The resistivity data are simulated with a semianalytical electro- fluid components of the rock. An additional input is the pore
magnetic solver for cylindrically layered media on the basis of the shape, defined as the ratio of the minor axis to the major axis of a
numerical mode-matching method (Chew et al. 1991). For all spheroidal pore, which is specified for each porosity type in the
results in this paper, data from only five channels are used in the rock. The pore orientations have a uniform random distribution
inversion, corresponding to the response from the axial source for an isotropic medium. Increasing (or decreasing) the pore as-
and axial receivers for 26 kHz at 15-, 21-, 27-, 39-, and 72-in. pect ratio has the impact of stiffening (or softening) the rock.
spacing. The shorter-spacing 9- and 12-in. channels are excluded Consequently, compliant microporosity can be characterized by
because they are more sensitive to wellbore effects that are lower-aspect-ratio pores, whereas stiffer macroporosity by
ignored in the forward modeling. Benchmarking the inversion higher-aspect-ratio pores. Here, we define a single, effective as-
without the additional channels helps to illustrate the performance pect ratio, denoted by a, for all pores in the matrix. This effective
in more-challenging scenarios relevant for field data. aspect ratio is estimated at each logging depth in the work flow,
described later, and would be indicative of the main porosity type
Formation Parameterization. At each logging depth, the forma- at a particular depth.
tion is described by a 1D radial model as in Fig. 1, with properties
defined in discrete bins or pixels. The water saturation and porosity Measurement Sensitivities. To design a more-robust inversion,
in a pixel at radius ri are denoted by Sw ðri Þ and /ðri Þ, respectively. it is helpful to understand the sensitivity of the measurements to
The outermost pixel is unbounded in radial extent and represents formation parameters. Fig. 2 illustrates the sensitivity of the data
the virgin far-field formation, with water saturation S w and poros- to porosity and water saturation for a radially homogeneous for-
ity /. The wellbore radius is denoted by rBH . The pixel-based mation. The data were simulated with the framework in Fig. 1.
parameterization is more flexible than a model-based parameter- The upper and lower rows contain sensitivity for a gas-bearing
ization—for example, step or annulus models—which requires sandstone and an oil-bearing carbonate, respectively. The parame-
prior knowledge of the nature of the near-wellbore alteration. ters for each formation are in Tables 1 and 2 and are based on
Petrophysical transforms map water saturation and porosity in interpretation of real log data. The gas-bearing formation has
each pixel to formation resistivity, compressional-wave velocity, higher porosity and thinner pores to create a slower formation,
shear-wave velocity, and density, denoted by Rh ðri Þ; Vp ðri Þ; whereas the oil-bearing carbonate is a fast formation with lower
Vs ðri Þ; qðri Þ, respectively. The radial models of resistivity and porosity and rounder pores. For both formations, flexural-wave
elastic properties are entered into the electromagnetic and sonic dispersion data are sensitive to porosity but have much weaker

2 2016 SPE Reservoir Evaluation & Engineering

ID: jaganm Time: 14:16 I Path: S:/REE#/Vol00000/160028/Comp/APPFile/SA-REE#160028

 REE170609 DOI: 10.2118/170609-PA Date: 11-May-16 Stage: Page: 3 Total Pages: 13

400 400 102 102

Sw = 25%
φ = 29 PU
380 Sw = 50%
φ = 32 PU

App. Resistivity (Ω.m)

Slowness (μ-sec/ft)

Slowness (μ-sec/ft)

App. Resistivity (Ω.m)

360 380 Sw = 75%
φ = 35 PU Sw = 100%
340
360 101 101
320
300 340
Sw = 25%
280 φ = 29 PU
Sw = 50%
φ = 32 PU 320
260 Sw = 75% 100 100
φ = 35 PU Sw = 100%
240 300
0 2 4 6 8 10 1 2 3 4 5
0 2 4 6 8 10 1 2 3 4 5
Frequency (kHz) Channel Index
Frequency (kHz) Channel Index
220 220 103 103
φ = 10 PU
200 200 φ = 15 PU
Slowness (μ-sec/ft)

Slowness (μ-sec/ft)

App. Resistivity (Ω.m)

App. Resistivity (Ω.m)

φ = 20 PU
180 180 102 102
Sw = 25%
Sw = 50%
160 Sw = 75%
160
140 140 101
φ = 10 PU 101
Sw = 25%
120 φ = 15 PU 120 Sw = 50%
φ = 20 PU Sw = 75%
100 100 100
0 2 4 6 8 10 0 2 4 6 8 10 100
1 2 3 4 5 1 2 3 4 5
Frequency (kHz) Frequency (kHz) Channel Index Channel Index

Fig. 2—Sensitivity of flexural-wave dispersion data and array-induction data to formation porosity and water saturation. Each row
contains sensitivities for a different reference formation. The reference formation for the upper row is a slow gas-bearing sand-
stone with Sw 5 50%, / 5 0.35, a 5 0.05. The reference formation for the lower row is a fast oil-bearing carbonate with Sw 5 50%,
/ 5 0.15, a 5 0.1. Parameters for the formation are in Tables 1 and 2. In each row, the two leftmost panels contain dispersion data,
and the next two panels contain induction data. In all figures, the Channel Index 1—5 for the resistivity data corresponds to 26 kHz
at 15-, 21-, 27-, 39-, and 72-in. spacing.

sensitivity to water saturation. The latter effect is because the flex- Preprocessing. Induction data are rotated to the well coordi-
ural wave is mainly sensitive to formation shear-wave slowness, nate system and corrected for wellbore effects. The sonic dipole
which is only weakly affected by fluid substitution through changes waveforms are processed with the extended Prony method to
in fluid density. The fluid effect is seen to be stronger for the slow extract dispersion data, followed by a labeling algorithm to isolate
gas-bearing formation, but is still relatively weak in comparison to the flexural-wave mode. Compressional- and shear-wave velocity
the porosity effect. In contrast, induction data are sensitive to both logs are obtained by semblance-time-coherence (STC) processing
porosity and water saturation as dictated by Archie’s law. of the waveforms. The sonic data can be generated with multishot
subarray processing to produce data at vertical resolution similar
Inversion Work Flow. The complementary sensitivity of sonic to that of the resistivity data. Archie parameters and a radially ho-
and resistivity data to water saturation motivates the following mogeneous porosity and water saturation are derived by tradi-
work flow. Starting from a radially homogeneous formation, we tional petrophysical interpretation of triple-combo logs and core
estimate radial distribution of porosity from flexural-wave data data whenever available. The pore aspect ratio at each log depth
alone. This porosity distribution is held fixed while estimating ra- is estimated by minimization of the L2-norm relative error
dial distribution of water saturation from induction data alone. between the compressional- and shear-wave velocities from the
Finally, both porosity and water saturation are refined simultane- rock-physics model and the velocities from STC processing. Min-
ously from all the data. From tests on noisy synthetic data, we eralogy at each log depth is assumed known from analysis of a
find that building an initial model by estimating a subset of pa- geochemical log.
rameters from a subset of more-sensitive measurements improves • Step 1. A first estimate of porosity /ðri Þ is obtained from
robustness of the inversion (i.e., reduces sensitivity to choice of flexural-wave dispersion data by minimizing a cost function
initial model) in comparison to directly estimating all the parame- defined as the L2-norm relative error between the measured
ters from all the data. The work flow is applied at each logging and simulated dispersion data. The dispersion data are
depth with a sliding window. When applied to an interval of log robustly and efficiently simulated by linearizing the root
data, the results are displayed as images with radial distribution of finder about the radially homogeneous initial model. The po-
water saturation and porosity at each log depth. The details of the rosity for the linearization can also be obtained from the
work flow are as follows. low-frequency dispersion data (typically less than 3 kHz) by

Parameter Gas-Bearing Sandstone Oil-Bearing Carbonate

Borehole diameter 12.25 in. 12.25 in.
3 3
Mud density 1.00 g/cm 1.00 g/cm
Mud slowness 265 µs/ft 205 µs/ft
Archie parameter, m 1.8 2.2
Archie parameter, n 1.9 3.4
Formation-water resistivity, Rw 0.11 Ω⋅m 0.03 Ω⋅m

Pore aspect ratio, α 0.05 0.1

Mineralogy 100% Quartz 75% Calcite, 25% Quartz

Table 1—Parameters for the synthetic-data examples.

2016 SPE Reservoir Evaluation & Engineering 3

ID: jaganm Time: 14:16 I Path: S:/REE#/Vol00000/160028/Comp/APPFile/SA-REE#160028

 REE170609 DOI: 10.2118/170609-PA Date: 11-May-16 Stage: Page: 4 Total Pages: 13

3
Component Bulk Modulus (GPa) Shear Modulus (GPa) Density (g/cm )
Quartz 37.0 44.0 2.65
Calcite 76.7 32.3 2.71
N-Feldspar 61.5 29.6 2.61
K-Feldspar 29.6 27.6 2.57
Dolomite 82.5 43.8 2.87
Illite 15.3 11.5 2.79
Chlorite 6.4 30.0 3.01
Smectite 4.5 10.1 2.78
Water 2.7 - 1.05
Oil 1.8 - 0.75
Gas 0.02 - 0.20

Table 2—Elastic moduli and density for mineral and fluid components.

minimizing the guidance-mismatch cost function (Braunisch water saturation. The corresponding radial distribution of oil satu-
et al. 2004). The guidance-mismatch cost function measures ration is fixed from the constraint that the 100% fluid saturation is
the degree to which dispersion data satisfy the boundary composed of water, oil, and gas saturations. From the figure, the
conditions for a wellbore-guided wave. The mathematical flushed zone near the wellbore is seen to contain only oil (from
formulation is in Appendix A. the filtrate) and water, whereas the far field contains only gas and
• Step 2. A first estimate of the water saturation Sw ðri Þ is water. The transition zone between the flushed zone and far-field
obtained from array-induction data by minimizing a cost contains all three fluid phases and has structure similar to that of
function defined as the L2-norm relative error between the the water-saturation distribution.
measured and simulated induction data. The porosity /ðri Þ For processing gas-bearing formations, the inversion requires
is held fixed from the previous step. prior knowledge of the radial distribution of gas or oil saturation
• Step 3. Porosity and water saturation from Steps 2 and 3 are because there is insufficient data sensitivity to estimate these reli-
simultaneously refined by minimizing a cost function ably in the inversion. Therefore, here, we assume that the forma-
defined as the L2-norm relative error of both array induction tion contains only oil (from the filtrate) and water in the flushed
and flexural-wave dispersion data. This step is more impor- zone, and only gas and water in the far field, and impose a struc-
tant for slower formations in which the porosity from Step 1 tural similarity constraint with the water saturation to infer the oil
may be less accurate because of fluid effect on the flexural- and gas saturations in the transition zone. Note that errors in the
wave data. assumed oil or gas saturations in the altered zone have small
In all three steps, the L2-norm data-misfit cost function is mini- impact on the porosity reconstruction because of the relatively
mized, as described in Habashy and Abubakar (2004), with weak fluid effect on the flexural wave. As seen from the upper
Gauss-Newton optimization, adaptive smoothness regularization, row of Fig. 3, the inversion reconstructs the true water saturation
box constraints /ðri Þ 2 ½/min ; /max , and Sw ðri Þ 2 ½0; 1. The first and porosity models starting from a radially homogeneous initial
two steps typically converge in approximately 10 iterations to a model. The good agreement between the measured and simulated
relative accuracy of 1% in formation parameters, whereas the final data for the final model confirms the quality of the reconstruction.
step converges in two to three iterations. The corresponding models of formation elastic properties and re-
sistivity are in the lower row of Fig. 3. The formation resistivity
Synthetic-Data Examples decreases radially from the wellbore, consistent with the radial
increase in water saturation. The density also decreases radially
Here, we validate the inversion work flow on noisy synthetic data because of the invasion of higher density mud-filtrate into lower-
for several examples of relevance to real logging scenarios. For density gas-zone. The shear-wave slowness shows a small radial
all examples, the wellbore diameter, mud density, mud slowness, decrease because of the decrease in density. It is interesting to
mineral composition, Archie parameters, and pore aspect ratio are observe that the formation compressional-wave slowness is
assumed known. The formation model is discretized into 22 radial strongly sensitive to the presence of filtrate invasion in the gas
pixels. The inner 21 pixels each have a radial extent of rBH =3 zone, and also has a nonmonotonic shape because of the different
(approximately 2 in.) and represent the altered zone, whereas the relative impacts of density and bulk-modulus variations on the
outermost pixel represents the radially unbounded far field. Both compressional-wave speed.
sonic and resistivity data are corrupted with 3% random additive The second example has mechanical alteration leading to an
noise. We take /min ¼ 0:01 and /max ¼ 0:5 as the physically real- increased porosity of 40 p.u. at the wellbore that linearly
istic range for porosity. decreases to the far-field porosity of 35 p.u. The alteration zone
extends approximately 0.75 ft (or 1.5 rBH ) from the wellbore. The
Slow Gas-Bearing Sandstone Drilled With Oil-Based Mud. water-saturation model is similar to the previous example. As
The parameters for the formation are in Tables 1 and 2 and are seen in the upper row of Fig. 4, the inversion recovers the true
relevant to a well drilled offshore East Africa. The first example far-field water saturation and porosity, and also provides good
has oil-based-mud-filtrate invasion with a monotonic ramp shape reconstructions of the alteration zones. The good agreement
in a formation with homogeneous porosity of 35 p.u. This sce- between the measured and simulated data for the final model con-
nario can occur near the gas/water transition zone and is chosen to firms the quality of the reconstruction. The corresponding forma-
illustrate the capabilities of the inversion in reconstructing a small tion resistivity and elastic properties are in the lower row of Fig.
contrast in water saturation in the hydrocarbon zone. The filtrate 4. In this case, there is a large variation in formation shear-wave
invades up to 1 ft (or 2.0 rBH ) into the formation. The true radial slowness because of the mechanical damage, and the combined
distribution of gas saturation is presented in the same plot as the effect of mechanical damage and fluid substitution creates

4 2016 SPE Reservoir Evaluation & Engineering

ID: jaganm Time: 14:16 I Path: S:/REE#/Vol00000/160028/Comp/APPFile/SA-REE#160028

 REE170609 DOI: 10.2118/170609-PA Date: 11-May-16 Stage: Page: 5 Total Pages: 13

40 1 10
True model True model 400 Messured
Inversion (initial) Inversion (initial) 9 Inversion (initial)

Slowness (μ-sec/ft)

App. Resisivity (Ω.m)

0.8 Inversion (final)

Water Saturation
Inversion (final) Inversion (final)

Porosity (PU)
35 Gas saturation 8
0.6 350
7
0.4
30 300
Messured 6
0.2
Inversion (initial)
250 Inversion (final)
250 0 51
0.25 0.5 0.75 1 1.25 1.5 0 0.25 0.5 0.75 1 1.25 1.5 0 2 4 6 8 10 2 3 4 5
Radial Distance (m) Radial Distance (m) Frequency (kHz) Channel Index

360 220 2.2 102

Compressional (μ-sec/ft)
True model
340 Inversion (initial) 210
Inversion (final)
Shear (μ-sec/ft)

2.1

Density (g/cc)
320 200

Resisivity (Ω.m)
300 190 2 101
280 180
1.9
260 170

240 160 1.8 100

0 0.25 0.5 0.75 1 1.25 1.5 0 0.25 0.5 0.75 1 1.25 1.5 0 0.25 0.5 0.75 1 1.25 1.5 0 0.25 0.5 0.75 1 1.25 1.5
Radial Distance (m) Radial Distance (m) Radial Distance (m) Radial Distance (m)

Fig. 3—Inversion results for oil-based-mud-filtrate invasion into a slow gas-bearing sandstone formation. In the upper row, panels
from left show reconstructed porosity, water saturation, flexural-dispersion data, and array-induction data. In the lower row, the
panels from left show corresponding formation shear-wave slowness, compressional-wave slowness, density, and resistivity.

complex nonmonotonic distribution of formation compressional- saturation and porosity, while also providing good reconstruc-
wave slowness and density. tions of the alteration zones. The corresponding formation resis-
tivity and elastic properties are in the lower row of Fig. 6. The
mechanical alteration affects all the elastic properties because of
Fast Oil-Bearing Carbonate Drilled With Oil-Based Mud. the porosity variation.
The parameters for the fast oil-bearing carbonate formation are in
Tables 1 and 2 and are relevant to a well drilled offshore Brazil.
The first example has oil-basd-mud-filtrate invasion with a mono- Field-Data Example
tonic ramp shape into a formation with homogeneous porosity of Traditional Interpretation. We analyze a 50-m interval of data
20 p.u. In the upper row of Fig. 5, the inversion reconstructs the for a well drilled with oil-based mud offshore East Africa. Data
true water-saturation and porosity models starting from a radially were acquired with a triaxial induction service, an acoustic scan-
homogeneous initial model. The corresponding resistivity and ning platform, and an integrated wireline-logging tool. The Archie
elastic properties are in the lower row of Fig. 5. As expected, parameters, wellbore diameter, and mud properties are as for the
there is a strong variation in formation resistivity because of the gas-bearing sandstone synthetic example in Table 1. Traditional
resistivity contrast between oil and water, but a very weak effect interpretation of density, neutron, and resistivity logs shows that
on the elastic properties of the formation because of the similar the interval contains gas with water saturation of 10 to 40% and
elastic properties of oil and water. porosity of 10 to 30 p.u. Fig. 7 contains tracks with mineralogy,
The second example has migration of mud fines into the rock fast/slow shear-wave slowness logs, ultrasonic wellbore-image
leading to reduced porosity of 15 p.u. at the wellbore that linearly log, and shallow/deep resistivity logs. The mineralogy is mainly
increases to the far-field porosity of 20 p.u. The alteration zone quartz, sodium and potassium feldspars, and small amounts of
extends approximately 0.75 ft (or 1.5 rBH ) from the wellbore. shale. The elastic velocities and density for each mineral compo-
The mud-filtrate invasion is similar to the previous example. In nent from a database are in Table 2. Separation between fast and
Fig. 6 (upper row), the inversion recovers the true far-field water slow shear-wave slowness indicates azimuthal anisotropy over the

40 1 500 10
True model True model Messured
Inversion (initial) Inversion (initial) 9 Inversion (initial)
0.8
Slowness (μ-sec/ft)

App. Resisivity (Ω.m)

Water Saturation

Inversion (final) Inversion (final) 450 Inversion (final)

Porosity (PU)

Gas saturation 8
35 0.6
400 7
0.4
30
0.2 350 Messured 6
Inversion (initial)
Inversion (final)
0 300
250 0.25 0.5 0.75 1 1.25 1.5 0 0.25 0.5 0.75 1 1.25 1.5 0 2 4 6 8 10 51 2 3 4 5
Radial Distance (m) Radial Distance (m) Frequency (kHz) Channel Index

420 240 2.1 102

Compressional (μ-sec/ft)

True model
400 Inversion (initial) 230
Inversion (final)
Shear (μ-sec/ft)

2
Density (g/cc)

Resisivity (Ω.m)

380 220

360 210 1.9 101

340 200
1.8
320 190

300 0 0.25 0.5 0.75 1 1.25 1.5 180 0 0.25 0.5 0.75 1 1.25 1.5 1.7 100
0 0.25 0.5 0.75 1 1.25 1.5 0 0.25 0.5 0.75 1 1.25 1.5
Radial Distance (m) Radial Distance (m) Radial Distance (m) Radial Distance (m)

Fig. 4—Inversion results for oil-based-mud-filtrate invasion into a mechanically damaged slow gas-bearing sandstone formation.
In the upper row, panels from left show reconstructed porosity, water saturation, flexural-dispersion data, and array-induction
data. In the lower row, the panels from left show corresponding formation shear-wave slowness, compressional-wave slowness,
density, and resistivity.

2016 SPE Reservoir Evaluation & Engineering 5

ID: jaganm Time: 14:16 I Path: S:/REE#/Vol00000/160028/Comp/APPFile/SA-REE#160028

 REE170609 DOI: 10.2118/170609-PA Date: 11-May-16 Stage: Page: 6 Total Pages: 13

30 1 220 100
True model True model Measured
Inversion (initial) Inversion (initial)
200 Inversion (initial)

Slowness (μ-sec/ft)
Inversion (final) Inversion (final)

App. Resisivity (Ω.m)

0.75

Water Saturation
Inversion (final)
Porosity (PU)
25 180 50

0.5 160
20 140
0.25 Measured
120 Inversion (initial)
Inversion (final)
15 0 100 10
0 0.25 0.5 0.75 1 1.25 1.5 0 0.25 0.5 0.75 1 1.25 1.5 0 2 4 6 8 10 1 2 3 4 5
Radial Distance (m) Radial Distance (m) Frequency (kHz) Channel Index
200 100 2.5 103

Compressional (μ-sec/ft)
True model
180 Inversion (initial) 90 2.45
Shear (μ-sec/ft)

Inversion (final)

Resisivity (Ω.m)
Density (g/cc)
160 80 2.4
140 70 2.35 102
120 60 2.3
100 50 2.25
80 40 101
0 0.25 0.5 0.75 1 1.25 1.5 0 0.25 0.5 0.75 1 1.25 1.5 2.20 0.25 0.5 0.75 1 1.25 1.5 0 0.25 0.5 0.75 1 1.25 1.5
Radial Distance (m) Radial Distance (m) Radial Distance (m) Radial Distance (m)

Fig. 5—Inversion results for oil-based-mud-filtrate invasion into a fast oil-bearing carbonate formation. In the upper row, panels
from left show reconstructed porosity, water saturation, flexural-dispersion data, and array-induction data. In the lower row, the
panels from left show corresponding formation shear-wave slowness, compressional-wave slowness, density, and resistivity.

entire interval. The ultrasonic image indicates vertical fractures Fig. 8 contains the simulated data for the initial and final for-
near X190 m and X210 m, and crossover between fast and slow mation models, which are compared with the measured data. The
flexural dispersion indicates stress-induced anisotropy at other convergence of the inversion is confirmed by good reconstruction
depths. Separation between shallow and deep resistivity in the of all the measured data. Note that the resistivity logs in the tradi-
interval from X180 to X200 m indicates radial variation of forma- tional interpretation, labeled AT10 and AT90 in Fig. 7, are
tion resistivity. This may be caused by invading oil-based-mud fil- obtained by post-processing the raw, wellbore-corrected induction
trate displacing formation water at the wellbore or by radial logs to produce depth-corrected and vertical resolution-matched
variation in formation porosity from mechanical damage. The logs at different radial depths (Barber and Rosthal 1991). The
interval has negligible resistivity anisotropy, as ascertained from present inversion seeks to match the raw, wellbore-corrected
traditional anisotropy processing of data from the triaxial induc- induction logs rather than the logs from post-processing.
tion service. Fig. 9 contains inversion results presented as images with radial
and vertical distribution of porosity and saturation near the wellbore.
Inversion Results. The inversion work flow was applied to the The images were created by stacking 1D radial solutions from the
previous interval with the flexural-wave data from the fast direc- inversion at each log depth. The initial image contains the radially
tion. We have tested the inversion on both fast and slow flexural- homogeneous porosity and deep water saturation from traditional
wave data and find that data for the fast direction produce far-field interpretation. The final saturation image from the inversion has
results that are more consistent with traditional interpretation. The reduced water saturation near the wellbore at several depths, such as
fast shear is often used for interpretation of matrix properties from X180 m to X200 m, which indicates oil-based-mud-filtrate
under the assumption that it primarily responds to the rock matrix invasion. Elevated porosity near the wellbore is evident from the
and pores, whereas the slow shear also responds to fractures and final porosity image at several depths, such as near X180 m,
stress-induced cracks (Prioul et al. 2007). X190 m, and X210 m, which indicates mechanical damage.

25 1 220 100
True model True model Measured
Inversion (initial) Inversion (initial) 200 Inversion (initial)
Slowness (μ-sec/ft)

App. Resisivity (Ω.m)

Inversion (final) 0.75 Inversion (final)

Water Saturation

Inversion (final)
50
Porosity (PU)

20 180
0.5 160

15 140
0.25 Measured
120 Inversion (initial)
Inversion (final)
100 0 100 10
0.25 0.5 0.75 1 1.25 1.5 0 0.25 0.5 0.75 1 1.25 1.5 0 2 4 6 8 10 1 2 3 4 5
Radial Distance (m) Radial Distance (m) Frequency (kHz) Channel Index

200 100 2.5 103

Compressional (μ-sec/ft)

True model
180 Inversion (initial) 90 2.45
Inversion (final)
Shear (μ-sec/ft)

Density (g/cc)

160 80
Resisivity (Ω.m)

2.4
140 70 2.35 102
120 60 2.3
100 50
2.25
80 40
0 0.25 0.5 0.75 1 1.25 1.5 0 0.25 0.5 0.75 1 1.25 1.5 2.2 101
0 0.25 0.5 0.75 1 1.25 1.5 0 0.25 0.5 0.75 1 1.25 1.5
Radial Distance (m) Radial Distance (m)
Radial Distance (m) Radial Distance (m)

Fig. 6—Inversion results for oil-basd-mud-filtrate invasion into a mechanically damaged fast oil-bearing carbonate formation. In
the upper row, panels from left show reconstructed porosity, water saturation, flexural-dispersion data, and array-induction data.
In the lower row, the panels from left show corresponding formation shear-wave slowness, compressional-wave slowness, den-
sity, and resistivity.

6 2016 SPE Reservoir Evaluation & Engineering

ID: jaganm Time: 14:16 I Path: S:/REE#/Vol00000/160028/Comp/APPFile/SA-REE#160028

 REE170609 DOI: 10.2118/170609-PA Date: 11-May-16 Stage: Page: 7 Total Pages: 13

ELAN_dry_weights RHOZ
0.50 5.50 1.85 g/cm3 2.71
ELAN_dry_weights PEF8
0 1 2 12

Dry_shale_QE_Sur
Dry_Weight_Quart:
Dry_Weight_Dolorr
Dry_Weight_K-Fek
Dry_Weight_N-Fek
WBI256 Reserved NPHI AT90 DSTC_FAST
0.00 320.00 0.5 m3/m3 0 0.2 Ω.m 200 200 µs/ft 140

MD UBI_AMP_DYN MRP_UHR AT10 DSTC_Slow

(m) .
0 dB 2.60+02 0.5 m3/m3 0 0.2 Ω m 2000 200 µs/ft 140
1:200

×190

×200

×210

×220

×230

×240

Fig. 7—Traditional interpretation. Tracks from left to right are mineralogy (green, shale; yellow, quartz; pink, feldspar), ultrasonic
image, density/photoelectric/neutron porosity/magnetic-resonance porosity, shallow/deep resistivity (AT10/AT90), and fast/slow
shear.

Comparison of Inversion Results and Traditional Fig. 12 contains a comparison of the inversion far-field water
Interpretation. Tracks in Fig. 10 contain far-field porosity, far- saturation and porosity with traditional interpretation and core
field water saturation, and pore aspect ratio from the inversion. analysis. The core porosities were corrected for net overburden
The far-field water saturation and porosity are generally in good pressure, and the saturation-height modeling was based on mer-
agreement with traditional interpretation. The pore aspect ratio is cury-injection capillary pressure data converted to gas/brine reser-
relatively uniform, approximately 0.07, over the interval. This voir conditions. Both inversion and traditional interpretation are
value is consistent with other studies (Xu and White 1995). Inter- in good agreement with core data. Near X217 m, both traditional
vals with more shale have a lower pore aspect ratio. porosity from the density log and far-field porosity from the inver-
Tracks in Fig. 11 contain the far-field shear- and compres- sion significantly underestimate core porosity, most likely because
sional-wave slowness from the inversion compared with STC of core sampling missing the tight intervals indicated by the large
logs, and the far-field resistivity from the inversion compared increase in photoelectric factor (PEF) between X215 and X217 m.
with the deep resistivity log. The far-field properties from the Water saturation from both the interpretation and the inversion at
inversion are generally in good agreement with traditional proc- these depths is therefore not reliable.
essing. The far-field water saturation from the inversion is system-
atically lower than traditional water saturation in intervals with
more shale content (near X210 m, X220 m, X230 m, and Assumptions and Limitations of Work Flow
235 m). Because the far-field porosity from the inversion and The pore aspect ratio estimated in the work flow preprocessing
the traditional porosity are close in these intervals, the difference may be biased in the presence of radial alteration. This bias can
in water saturation is primarily because of difference between far- be reduced by solving for the pore aspect ratio in the inversion
field resistivity from the inversion and the AT90 log. and adding additional measurements, such as density, to decouple

2016 SPE Reservoir Evaluation & Engineering 7

ID: jaganm Time: 14:17 I Path: S:/REE#/Vol00000/160028/Comp/APPFile/SA-REE#160028

 REE170609 DOI: 10.2118/170609-PA Date: 11-May-16 Stage: Page: 8 Total Pages: 13

15 in., Re(ZZ) 21 in., Re(ZZ) 27 in., Re(ZZ) 39 in., Re(ZZ) 72 in., Re(ZZ) Depth ×183.2 m

260

Slowness (μ-sec/ft)
× 180
Messured
220
Inversion (initial)

Inversion (final)

× 190 180
Messured
Inversion (step 1)
Inversion (final)
140
1 2 3 4 5 6
× 200 Depth ×200.6 m

260

Slowness (μ-sec/ft)
Depth (m)

× 210
220

180
Messured
× 220 znversion (step 1)
Inversion (final)
140
1 2 3 4 5 6

Depth ×225.9 m
× 230
260

Slowness (μ-sec/ft)
× 240 220

180 Messured
Inversion (step 1)
× 250 Inversion (final)
140
20 200 20 200 20 200 20 200 20 200 1 2 3 4 5 6
App. Res. (Ω.m) App. Res. (Ω.m) App. Res. (Ω.m) App. Res. (Ω.m) App. Res. (Ω.m) Frequency (kHz)

Fig. 8—Reconstruction of data by inversion. First five tracks show array-induction data from the triaxial induction service for each
channel used in the inversion. The induction logs [labeled 15 in., Re (ZZ) and others] are the rotated, borehole-corrected apparent
resistivity logs for each coil spacing using the axial source and axial receiver. The last track contains flexural-dispersion data from
the acoustic scanning platform at a few depths.

Porosity (Initial) Porosity (Final) Water Saturation (Initial) Water Saturation (Final)

×180

×190

×200
Depth (m)

×210

×220

×230

×240

×250
0 0.5 1 1.5 0 0.5 1 1.5 0 0.5 1 1.5 0 0.5 1 1.5
Radial Distance (m) Radial Distance (m) Radial Distance (m) Radial Distance (m)

0.1 0.2 0.3 0.1 0.2 0.3 0 0.1 0.2 0.3 0 0.1 0.2 0.3

Fig. 9—Images with radial and vertical distribution of water saturation and porosity from inversion.

8 2016 SPE Reservoir Evaluation & Engineering

ID: jaganm Time: 14:17 I Path: S:/REE#/Vol00000/160028/Comp/APPFile/SA-REE#160028

 REE170609 DOI: 10.2118/170609-PA Date: 11-May-16 Stage: Page: 9 Total Pages: 13

Porosity (Total) Aspect Ratio Water Saturation

ELAN ELAN ELAN
×180 Inversion (Far-Field) Inversion Inversion (Far-Field)

×190

×200
Depth (m)

×210

×220

×230

×240

×250
10 20 30 0 0.1 0.2 0.3 0 50 100
Percent Percent

Fig. 10—Far-field porosity, water saturation, and pore aspect ratio from inversion. Far-field properties are compared with tradi-
tional interpretation.

DTSH DTCO Rh
DSTC STC AT90
×180 Inversion (Far-Field) Inversion (Far-Field) Inversion (Far-Field)

×190

×200
Depth (m)

×210

×220

×230

×240

×250
100 120 140 160 180 200 40 60 80 100 120 140 2 20 200
μs/ft μs/ft Ω.m

Fig. 11—Formation shear slowness (DTSH), compressional slowness (DTCO), and resistivity (Rh). Far-field properties from the iIn-
version are compared with slownesses from STC processing, and deep resistivity log (AT90) from induction-data processing.

2016 SPE Reservoir Evaluation & Engineering 9

ID: jaganm Time: 14:17 I Path: S:/REE#/Vol00000/160028/Comp/APPFile/SA-REE#160028

 REE170609 DOI: 10.2118/170609-PA Date: 11-May-16 Stage: Page: 10 Total Pages: 13

ELAN_dry_weights MRP_UHR
POR_QE SW_AR
0.50 5.50 0.5 m3/m3 0

ELAN_dry_weights NPHI POR_FAR SW_inv

0 1 0.5 m3/m3 0 0.5 v/v 0 1 m3/m3 0

Dry_shale_QE_Sur
Dry_Weight_Quart:
Dry_Weight_Dolorr
Dry_Weight_K-Fek
Dry_Weight_N-Fek
Heated Reserved PEF8 POR_QE SW_AR_AT90

0.00 320.00 2 12 0.5 v/v 0 1 v/v 0

MD UBI_AMP_DYN RHOZ CPOR SW_MICP

(m)
1:200 0 dB 2.6E+02 1.85 g/cm3 2.71 0.5 m3/m3 0 1 v/v 0

×190

×200

×210

×220

×230

×240

Fig. 12—Comparison of inversion far-field properties with traditional interpretation, core porosity, and saturation-height data. Tra-
ditional porosity and water saturation are labeled PHI_QE and SW_AR_AT90. Inversion far-field porosity and water saturation are
labeled POR_FAR and SW_INV. Core porosity and water saturation are labeled CPOR and SW_MICP.

the effect of porosity and pore aspect ratio on the sonic measure- the resistivity data. However, sonic data and ultrasonic images
ments. The pore aspect ratio may also be biased by errors in the indicate an effect of anisotropy arising from low relative dip beds,
tabulated elastic velocities for the mineral components. Velocities fractures, and stress variation. This anisotropy may lead to a false
of minerals such as quartz are standard, whereas velocities of signature of radial alteration of porosity when interpreted with an
clays are variable and difficult to measure in a laboratory. isotropic model. For the earlier results, applying the inversion
The formation model and petrophysical transforms assume an work flow on flexural-wave data from the fast direction produced
isotropic formation, which would have to be verified from the far-field porosity consistent with both traditional porosity and
data before application of the method. For the field-data example, core data. This suggests that, at least for the current data set,
traditional interpretation indicates negligible anisotropy effect on uncertainties in minerology and core analysis are too large to

10 2016 SPE Reservoir Evaluation & Engineering

ID: jaganm Time: 14:18 I Path: S:/REE#/Vol00000/160028/Comp/APPFile/SA-REE#160028

 REE170609 DOI: 10.2118/170609-PA Date: 11-May-16 Stage: Page: 11 Total Pages: 13

resolve errors in the modeling arising from ignoring anisotropy Rh ¼ formation resistivity
and using data from the fast direction in the inversion work flow. Rw ¼ formation-water resistivity
The formation model in the inversion assumes that properties So ¼ oil saturation
are vertically and azimuthally homogeneous within the volume of Sw ¼ far-field water saturation
investigation of the various channels of the sonic and resistivity Sw ¼ water saturation
data. Therefore, any 2D or 3D variation of the formation proper- vi ¼ velocity of ith point in dispersion data
ties within this volume would lead to bias in the inversion results. Vp ¼ formation compressional-wave velocity
In particular, sufficiently thin layering can introduce a vertical Vs ¼ formation shear-wave velocity
transverse anisotropy effect on the dispersion data, creating a false a ¼ effective pore aspect ratio
signature of reduced porosity near the wellbore. /min ¼ lower bound on porosity for inversion
The rock-physics model assumes a linear isotropic formation /max ¼ upper bound on porosity for inversion
and therefore may not be valid in the mechanically damaged zone / ¼ far-field porosity
because of nonlinear deformation or presence of oriented fractures / ¼ porosity
induced by drilling. Hence, the porosity in the damaged zone xi ¼ frequency of ith point in dispersion data
inferred from the rock-physics model may not be accurate. How- q ¼ formation density
ever, if the damaged rock has smaller bulk and shear moduli rela-
tive to the far-field rock (Plona et al. 1998), the porosity in the
damaged zone would be increased relative to the far-field porosity
and serve as a qualitative indicator of mechanical damage. Simi- References
larly, presence of fines in the pore space may locally increase either Allioli, F., Faivre, O., Jammes, L. et al. 1997. A New Approach to Com-
the bulk density or stiffness of the formation, leading to reduced puting Formation Density and Pe Free of Mudcake Effects. Presented
porosity relative to the true far-field porosity. If the fines effect is at the SPWLA 38th Annual Logging Symposium, Houston, USA,
mainly on the bulk density, then the corresponding reduction in po- 15–18 June. SPWLA-1997-K.
rosity relative to the far-field porosity would be compatible with Barber, T. D. and Rosthal, R. A. 1991. Using a Multiarray Induction Tool
the parameterization and modeling framework presented here. In to Achieve High-Resolution Logs With Minimum Environmental
that case, integration of the density log in the framework would Effects. Presented at the SPE Annual Technical Conference and Exhi-
help to better constrain formation properties near the wellbore. bition, Dallas, USA, 6–9 October. SPE-22725-MS. http://dx.doi.org/
10.2118/22725-MS.
Barber, T., Anderson, B., Abubakar, A. et al. 2004. Determining Formation
Conclusions Resistivity Anisotropy in the Presence of Invasion. Presented at the SPE
A new fully automated inversion work flow is developed for Annual Technical Conference and Exhibition, Houston, USA, 26–29
imaging radial distribution of water saturation and porosity near September. SPE-90526-MS. http://dx.doi.org/10.2118/90526-MS.
the wellbore by joint inversion of sonic flexural wave and array- Braunisch, H., Habashy, T. M., Sinha, B. K. et al. 2004. Inversion of Guided-
induction resistivity. The solution space of the inversion is con- Wave Dispersion Data With Application to Borehole Acoustics. J.
strained by the measurement physics honoring the radial sensitiv- Acoust. Soc. Am. 115 (1): 269–279. http://dx.doi.org/10.1121/1.1625683.
ities of both sonic and resistivity data, as well as petrophysical Brie, A., Pampuri, F., Marsala, A. F. et al. 1995. Shear Sonic Interpretation
transforms specific to the formation lithology. Physical bounds on in Gas-Bearing Sands. Presented at the SPE Annual Technical Confer-
the properties and smoothness regularization further constrain the ence and Exhibition, Dallas, USA, 23–25 October. SPE-30595-MS.
space of feasible solutions. The inversion is pixel-based and can http://dx.doi.org/10.2118/30595-MS.
therefore handle complex and arbitrary invasion geometries. Burridge, R. and Sinha, B. K. 1996. Inversion for Formation Shear Modu-
Moreover, integrating multiphysics measurements improves accu- lus and Radial Depth of Investigation Using Borehole Flexural Waves.
racy and reduces uncertainty. Presented at the SEG Annual Meeting, Denver, USA, 10–15 Novem-
The output from the work flow consists of images that map ra- ber. SEG-1996-0158.
dial distribution of water saturation and porosity, several feet Chew, W. C., Nie., N., Liu, Q-H. et al. 1991. An Efficient Solution for the
from the wellbore at each log depth, consistent with the DOIs of Response of Electrical Well-Logging Tools in a Complex Environ-
the induction and flexural-wave data. The images characterize ment. IEEE Transactions on Geoscience and Remote Sensing 29 (2):
alteration from filtrate invasion or mechanical damage, and pro- 308–313. http://dx.doi.org/10.1109/36.73673.
vide logs of far-field properties for petrophysical analysis. The Gao, G., Abubakar, A., and Habashy, T. M. 2013. Borehole Petrophysical
work flow also provides a log of pore shape that could help to Imaging Using Induction and Acoustic Measurements. Petrophysics
identify relative changes in porosity type. 54 (4): 383–394.
The work flow was validated on synthetic data for different Habashy, T. M. and Abubakar. A. 2004. A General Framework for Con-
near-wellbore alteration scenarios, and tested on field data from straint Minimization for the Inversion of Electromagnetic Measure-
an offshore well drilled with oil-based mud in a gas-bearing for- ments. Progress in Electromagnetics Research 46: 265–312. http://
mation. The results are in good agreement with independent core dx.doi.org/10.2528/PIER03100702.
analysis and traditional interpretation. Future versions of the work Howard Jr., A. Q. 2002. A New Invasion Model for Resistivity Interpreta-
flow will attempt to extend the inversion framework to address tion. The Log Analyst 33 (2): 96–110.
anisotropy and thin layering, and to integrate additional logging Mallan, R. K., Ma, J., Torres-Verdin, C. et al. 2009. Joint Radial Inversion
measurements to constrain the interpretation. of Resistivity and Sonic Logs to Estimate In-Situ Petrophysical and
Elastic Properties of Formations. Presented at the SPE Annual Techni-
cal Conference and Exhibition, New Orleans, USA, 4–7 October. SPE-
Nomenclature 124624-MS. http://dx.doi.org/10.2118/124624-MS.
CF ¼ guidance mismatch cost function Pistre, V., Kinoshita, T., Endo, T. et al. 2005. A Modular Wireline Sonic
Di ¼ determinant of guidance-condition matrix at (vi , xi ) Tool for Measurements of 3D (Azimuthal, Radial, and Axial) Forma-
Di ¼ root-mean-square magnitude of Di tion Acoustic Properties. Presented at the SPWLA 46th Annual Log-
Kf ¼ effective bulk modulus of pore fluid ging Symposium, New Orleans, USA, 26–29 June.
Kg ¼ bulk modulus of gas Plona, T. J., Winkler, K. W., Sinha, B. K. et al. 1998. Measurement of
Kw ¼ bulk modulus of water Stress Direction and Mechanical Damage Around Stressed Boreholes
Ko ¼ bulk modulus of oil Using Dipole and Microsonic. Presented at the SPE/ISRM Rock
m ¼ Archie porosity exponent Mechanics in Petroleum Engineering, Trondheim, Norway, 8–10 July.
n ¼ Archie saturation exponent SPE-47234-MS. http://dx.doi.org/10.2118/47234-MS.
ri ¼ location of ith pixel in formation model Prioul, R., Signer, C., Boyd, A. et al. 2007. Discrimination of Fracture and
rBH ¼ radius of wellbore Stress Effects Using Image and Sonic Logs for Hydraulic Fracturing

2016 SPE Reservoir Evaluation & Engineering 11

ID: jaganm Time: 14:18 I Path: S:/REE#/Vol00000/160028/Comp/APPFile/SA-REE#160028

 REE170609 DOI: 10.2118/170609-PA Date: 11-May-16 Stage: Page: 12 Total Pages: 13

Design. The Leading Edge 26 (9): 1134–1139. http://dx.doi.org/ Schlumberger in 2009. Shetty’s research interests include fluid
10.1190/1.2780783. mechanics, borehole acoustics, rock physics, scientific com-
Rosthal, R., Barber, T., Bonner, S. et al. 2003. Field Test of an Experimental puting, and reservoir characterization. He holds a PhD degree
Fully-Triaxial Induction Tool. Presented at the SPWLA 44th Annual in mechanical engineering from the University of California at
Logging Symposium, Galveston, Texas, 22–25 June. SPWLA-2003-QQ. Berkeley. Shetty is a member of SPE, Society of Professional
Well Log Analysts (SPWLA), and American Physical Society.
Tang, X. M. and Cheng, A. 2004. Quantitative Borehole Acoustic Meth-
ods, Vol. 24. Gulf Professional Publishing. Lin Liang is a principal research scientist and program man-
Xu, S., and White, R. E. 1995. A New Velocity Model for Clay-Sand Mix- ager at SDR. He works on multiphysics modeling and inversion
tures. Geophysical Prospecting 43: 91–118. http://dx.doi.org/10.1111/ for hydrocarbon exploration and reservoir characterization, as
j.1365-2478.1995.tb00126.x. well as reservoir simulation and history-matching problems.
Liang holds a BS degree in fluid machinery and fluid engineer-
Xu, S., and Payne, M. A. 2009. Modeling Elastic Properties in Carbonate ing from Tsinghua University and a PhD degree in environmen-
Rocks. The Leading Edge 28 (1): 66–74. http://dx.doi.org/10.1190/ tal science from Peking University. He joined Schlumberger in
1.3064148. 2002. Liang is a member of SPE and the Society of Exploration
Zeroug, S., Valero, H. P., Bose, S. et al. 2006. Monopole Radial Profiling Geophysicists (SEG).
of Compressional Slowness. Presented at the SEG Annual Meeting,
Tarek M. Habashy is currently a Schlumberger fellow and the
New Orleans, USA, 1–6 October. SEG-2006-0354. managing director of SDR. He holds a PhD degree from the
Massachusetts Institute of Technology (MIT) in electrical engi-
Appendix A neering. Habashy joined SDR in 1983, where he has held a
number of scientific and managerial positions. In his current
The guidance mismatch cost function CFð/Þ is formulated as position, he oversees the research activities at SDR with a total
population of 150 þ scientists and engineers. Before joining
XlogDi  logeDi  SDR, Habashy was a visiting research associate at MIT.
CFð/Þ ¼  ; . . . . . . . . . . . . . . . ðA-1Þ Throughout his career, he has conducted research on elec-
 logeD 
i i tromagnetic waves and fields, inverse scattering theory, multi-
physics inversion, antenna theory and design, dielectric and
where Di  Dðxi ; vi ; /Þ is the magnitude of the determinant of resistivity logging tools and techniques, mixed boundary value
problems, and numerical analysis. Habashy is a member of
the acoustic guidance condition matrix evaluated at each velocity-
the editorial boards of Inverse Problems, Wave Motion, and
frequency point ðxi ; vi Þ in the measured dispersion data, e is a the Journal of Electromagnetic Waves and Applications, and
small user-chosen parameter (set to 10–6 for the examples in this is a member of the advisory committee board for the book se-
paper), and Di are assumed to be always greater than machine pre- ries Progress in Electromagnetic Research. He is a former edi-
cision so that the logarithm function is well-defined. The Di are tor of Radio Science. Habashy is a fellow of the Institute of
defined as is the root-mean-square magnitude of Di over the po- Physics and Institute of Electrical and Electronics Engineers
rosity range of interest: (IEEE), and a full member of Commission B of the International
Union of Radio Science. He holds 47 US patents and has pub-
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi lished 10 book chapters, more than 150 scientific articles in
u
u /ð max refereed journals, and more than 150 conference-proceed-
u 1 ings papers.
u
Di  t jDðxi ; vi ; /Þj2 d/: . . . . . . ðA-2Þ
ð/max  /min Þ Vanessa Simoes graduated in computing and applied math-
/min
ematics from University of Sao Paulo in 2006; she earned an
MS degree and a PhD degree in mathematics from Instituto
de Matemática Pura a Aplicada in 2009 and 2013, respec-
The guidance-mismatch cost function has a global minimum tively. Simoes joined Schlumberger in 2013 as a research asso-
when Di ¼ eDi ; that is, when Di  Di . The normalization by eDi ciate in Brazil Research and Geoengineering Center (BRGC),
is required to account for the scaling of the determinant magni- Rio de Janeiro, where she works on the multisensor inversion
tude. For example, the matrix eI, where I is the identity matrix, is project and in acoustics processing. Her technical interests
not rank-deficient, but has an arbitrarily small determinant magni- include sonic, rock physics, and carbonates characterization.
tude. Thus matrix scaling should be taken into account if determi- Austin Boyd is a research director at Schlumberger BRGC.
nant magnitude is used to identify rank deficiency. In addition, Before that, he was Petrophysics Program manager at SDR,
the logarithm function is useful because Di may vary over several and he has held a variety of positions with Schlumberger in
orders over the frequency range of interest. We minimize the field operations, product development, research, and engi-
guidance-mismatch cost function with a Brent 1D line-search. neering. Boyd joined Schlumberger in 1981 as a field engineer
after graduating from Dalhousie University with a degree in
electrical engineering.
Appendix B
Bikash K. Sinha is a scientific adviser and program manager
In Brie et al. (1995), an empirical law is proposed for computing for Acoustics Research at SDR. His current research interests
the bulk modulus of a mixture of pore fluids containing gas. This include borehole sonic data analysis for the characterization
law is proposed in conjunction with Gassmann fluid substitution of formation anisotropy, fractures, and stresses. He is also
for interpretation of sonic log data from gas-bearing formations. involved in the development of quartz pressure sensors for
The proposed law predictions of data from partially saturated soft high-temperature and high-pressure applications. Sinha has
gas sands are better than predictions of Wood’s law. On the basis authored or coauthored more than 200 technical papers and
of those findings, we use the following equation to compute the holds 43 US patents. An IEEE Fellow, Sinha received the 1993
effective bulk modulus Kf of the pore-fluid mixture: outstanding-paper award for the innovative design and de-
velopment of a quartz pressure sensor published in the IEEE
Transactions on ultrasonics, ferroelectrics, and frequency
Kf ¼ ðKw  Kg ÞSew þ ðKo  Kg ÞSeo þ Kg ; . . . . . . . . . ðB-1Þ control (UFFC). He is also a recipient of the 2015 IEEE-UFFC
Achievement and 2012 IEEE Cady Awards for theoretical
where Kw ; Ko ; Kg are the bulk modulus of water, oil, and gas, advancements and innovations with regard to the thickness-
respectively; So is the oil saturation; and e is a fitting parameter. shear and surface-acoustic-wave-based quartz pressure sen-
sors, as well as contributions to the borehole sonic technology
For e ¼ 1, Eq. B-1 reduces to the serial law, and for larger e the
in the oil and gas industry. Sinha holds a BTech (Hons.) degree
predictions approach Wood’s law. Here, we use e ¼ 3, which was from the Indian Institute of Technology, Kharagpur, and an
proposed as yielding the best fit to data acquired from gas sands. MASc degree from the University of Toronto, both in mechani-
cal engineering, and a PhD degree in applied mechanics
Sushil Shetty is a senior research scientist at Schlumberger-Doll from Rensselaer Polytechnic Institute, Troy, New York. He is a
Research (SDR), Cambridge, Massachusetts, USA. He joined member of SPE, SEG, and SPWLA.

12 2016 SPE Reservoir Evaluation & Engineering

ID: jaganm Time: 14:18 I Path: S:/REE#/Vol00000/160028/Comp/APPFile/SA-REE#160028

 REE170609 DOI: 10.2118/170609-PA Date: 11-May-16 Stage: Page: 13 Total Pages: 13

Smaine Zeroug is a research director at SDR. He has been with Tim Pritchard is the Global Head of Petrophysics with BG Group
Schlumberger for 22 years. Zeroug’s current research interests (subsidiary of Royal Dutch Shell). His technical interests include
revolve around developing advanced multiphysics measure- thin-bed clastics and complex carbonates formation evalua-
ments and interpretation products for oilfield reservoir characteri- tion, unconventional formations (i.e., tight gas, shale gas, and
zation, development, and production. He holds a PhD degree in coalbed methane), and geomechanics.
electrophysics from Polytechnic Institute of New York University.
Brent Glassborow is an energy-industry consultant focused on
Clive Sirju is the Petrophysics Discipline lead for Nexen Petro- Australasian new ventures and technical-program manage-
leum UK Limited. Before joining Nexen, he was a consultant pet- ment. Before returning to Australia, he was the Subsurface Tech-
rophysicist at the BG Group, based in Reading, UK, where he nology manager at BG Group, based in Brazil and managing
managed the petrophysics function for the East Africa region. the global-research functional teams for geology, geophysics,
Sirju’s interests include data-acquisition planning, core-to-log reservoir engineering, and petrophysics, including thematic
integration, and new technology applications that can pro- portfolios specific to carbonate and tight reservoirs. Previous to
vide material benefit to the industry. He also managed a num- this, Glassborow held various portfolio-management and tech-
ber of technology projects focused on delivering functional nical positions in Brazil, the UK, and Australia. He holds a BSc
excellence to the business. Sirju’s research interests include for- (Honors) degree in petroleum geology and geophysics from the
ward-modeling techniques to predict reservoir productivity. Australian School of Petroleum (University of Adelaide).

2016 SPE Reservoir Evaluation & Engineering 13

ID: jaganm Time: 14:18 I Path: S:/REE#/Vol00000/160028/Comp/APPFile/SA-REE#160028