You are on page 1of 25

SPWLA 31st Annual Lagging Sympsium, June 24-27, 19

,---

Response of 2-MHz LWD Resistivity and Wireliie Induction Tools


in Dipping Beds and Larninat ed Format ions

Barbara Anderson, Stephen Bonner, Martin G. Lriling and Richard Rosthal


Schlumberger

Abstract

The rezponses of wireline induction and 2-MHz resistivity tools for logging
while drilling (LWD) are compared in dipping formations. The differences in mea-
surement physics, depths of investigation and vertical resolutions between induction
and 2-MHz tools lead to different responses in steeply dipped beds. Some new and
intereeting effects are presented that are characteristic of all 2-MHz LWD tools.
These effects have been observed on field logs and have also been reproduced with
computer modeling. In particular:
- In thin (2 ft – 10 ft) beds, dlp and shoulder effects can cause separations
between phsse shift and attenuation resktivity curves and affect the readhgs
for all 2-MHz tools. The amount of separation is a function of dip angle,
rezistivity level and resistivity contraat.
- At dipping bed boundaries, 2-MHz logs can show horns caused by electric
polarization of the interface. The horns are more prominent on phase shift
resistivity logs.
- Resistivity anisotropy causes a separation of 2-MHz phase shift and attenuation
resistivity curves in laminated or shaly formations logged at an angle.
A thorough understandkg of the effect of dlp improves interpretation of 2-MHz
tools.

Modeling Code for 2-MHz Tools in Dipping Beds

Electromagnetic propagation tools operating at 2 MHz are often used to provide


resistivity data when logging while drilling. These tools are frequently employed
in directional drilling where hydrocarbon-bearing zones and other thinly bedded or
laminated formations are encountered at steep relative dlp angles. The response of
2-MHz resistivity tools depends on the relative dlp angle and can show effects that
are not observed in vertical boreholes.
An accurate dipping bed modeling code is necessary to understand the un-
derlying physics of the effect of dip, and to quantitatively evaluate the resistivity
in thin dipping beds. Such codes already exist for induction tools 1~2*3. One of
these codesl hss been adapted for modeling the response of 2-MHz resistivity tools.
The model includes the relative formation dielectric constant, q, which mediates
the propagation of the electromagnetic wave. If Et is not explicitly specified, the

1
SPWLA 31st Annual Logging Symposium, Jun. 24-27, 1990

code uses an eatimate4 derived for the Compensated Dual Reziztivity (CDR*) tool,
namely, q = 108.5 u0”36.
The dipping bed model approximates the loop dipole antennzz of the actual
tool by magnetic point dipoles. The dipole moment is decomposed into two compo-
nents: one parallel and the other perpendicular to the bed boundaries. The fields of
these components are completely decoupled and thus are separately evaluated. The
point dipole approximation neglects the radial structure of the tool: specifically, the
mandrel with the antenna recesses, the borehole and possible invazion are ignored.
Thus, thw code cannot model borehole or invasion effects.
Because of the large differences between the LWD tool and the point-dipole
model, the code must be carefully qualified before it is used in thin bed modeling.
The dipping bed code has been verified at 0° dip by comparison with finite element
method (FEM) results in multiple beds. For d]p angles from 0° to 90°, the program
hss successfully reproduced experimental test tank meazurements. These results
will be dezcribed in detail in the following section. The agreement between the
point dipole approximation and the true tool response justifies the use of the code
for analyzing 2-MHz tool response in dipping thin beds and laminated formations.
As with the induction model, the 2-MHz dipping bed code can model tool
response to an arbitrary number of thin beds. The code haz been successfully used
to compute 2000 ft of synthetic log spanning sz many as 150 beds. The code hzs
also been applied to model CDR response to l-in. laminations over a 10-ft interval.
These modeled lamination results show a significant separation between the phase
shift and attenuation resistivity curves when dip is present. Such a separation h=
been observed in shales on several logs in highly deviated wells in the Gulf of Mexico.
The 2-MHz dipping bed code is now included in the electromagnetic modeling
package ELMOD *6, which also simulates laterolog and induction tool responses.
The consolidation of these codes allows a dkect comparison of induction and 2-MHz
LWD resistivity tools in identical formations, and thus simplifies the integration
of LWD and wireline formation evaluation. Modeling can be performed before,
during or after drilling. Before drilling, modeling based on offset well data allows a
predictive evaluation of anticipated formations. Data from vertical wells can be used
to predict log responses in highly deviated wells. While drilling, modeling based
on real-time logging data can improve the accuracy of early formation evaluation.
This early formation evaluation is useful in assessing the need for certain wireline
services. After drilling, modeling can aid the interpretation of strange or unexpected
log responses from both LWD and wireline tools.

* Mark of Schlumberger

2
SPWLA 31st Annual Lagging Syqwsium, June 24-27, 1990

Vflldation of the Code

The point dipole approximation must be carefully tested and validated for
2-MHz tools. For the induction tool, a point dipole approximation of the antennas
is permissible because the diameters of induction tool coils are small compared to
the coil spacing and the wavelength. However, 2-MHz LWD resistivity tools are
built around a drill collar with a diameter of at least 6.5 in. comparable to the
receiver-te-receiver spacing (6.0 in. for the CDR tool). Furthermore, the higher
operating frequency causes a much smaller skin depth compared to induction: at
1 ohm-m, the 2-MHz:skin depth is 14 in. compared to 140 in. for induction.
At O“ dip, the vertical (thin bed) response of 2-MHz resistivity tools is essen-
tially controlled by the axial antenna spacing. Thus, the point dipole approximation
should be able to reproduce the thin bed response of models that include exact an-
tenna dimensions. The performance of the dipping bed code has been compared
to results from an FEM code’. The FEM program models the antennae ss loops
wound around a finite-sized uniform mandrel centered in a borehole. The surround-
ing formation includes an arbitrary number of thin beds with arbitrary invaaion.
Figure 1 compares previously modeled” FEM results for the CDR tool in a
series of thin beds to the same thin bed sequence modeled with the dipping bed
code at zero dip. Results are shown for both phase shift resistivity (RPS *) and
attenuation resistivity (RAD ● ). The dipping bed code results agree very well with
the FEM results. This agreement confirms that the vertical response of 2-MHz tools
is controlled by the axial antenna spacing, and that the radial structure of the tools
has little effect on the th]n bed response. Therefore, the point dipole approximation
of the antennss models thin bed effects with a high degree of accuracy. Figure 1
also illustrates the excellent thin bed response of the 2-MHz tools, especially in
conductive beds.
For nonzero dip, there is no other modeling code available against which the
dipping bed code can be compared. Therefore, we performed test tank experiments
with the CDR tool and compared the experimentally observed effects to the dipping
bed code predictions. For the experiments, the test tank was filled with salt water.
The CDR tool was lowered into the tank at various anglee, thus ‘logging” the air-
water interface. The results for 1 ohm-m are shown in Figure 2. For a vertical CDR
tool, experimental and modeling results agree. Thk agreement further confirms the
accuracy of the dipping bed code.
At 14° dip, modeled and experimental results essentially agree. Compared
to the vertical log, there is a small dMference between the curves. This difference
becomes more visible for the 27° and 47° dip log. Specifically, the experimentally
observed curves appear more smeared out over the air-water interface than the
theoretical predictions.

* Mark of Schlumberger

3
SPWLA 31st Annual Logging Syqmsiwn, June 24.27, 1990

This small discrepancy is caused by the point d]pole approximations of the


modeled antennas. A point-like dipole crossw the air-water interface at one specific
depth, while the loop antenna with a dkrneter d crosses the air-water interface over
the vertical distance 6Z = d sin(~dip). For ddip = 27° ad an antenna dhrneter
of 6.5 in., thk dktance becomes 6Z s 3 in. The vertical separation between the
experimental and modeled curves corresponds to this distance.
The dipping bed code models the vertical resolution of the 2-MHz tools with
an accuracy equal to the tool diameter for any dipping angle. For dip angles of
61”, 76” and 88”, the vertical differences between theory and experiment grow ss
d sin(OdiP). Hence, the vertical difference between theory and experiment never
exceeds the diameter of the CDR tool.
The experimental results confirm the qualitative change in thin bed response
for the 2-MHz tools with increasing dlp angle. From 0“ to 47° dip, the logged
curves gradually smear out a little, but essentially retain their shape. The stability
of the thin bed response for small to moderate dip angles shows that the 2-MHz
LWD tools are quite insensitive to dip for zmall dlp angles up to about 45°. For dip
angles above 60”, both phase shift and attenuation measurements show a different
effect at the air-water boundary. The measurements tend to decrease below the
mymptotic ~luez in air. The point dipole eesumptiOn overestimates the amount
of thk excursion. Thk unusual effect is caused by the charge build-up at the air-
water interface. It has caused prominent horns in numerous field logs. The charge
build-up will be discussed in detail in a later section.

The Modeling of Thin Bed Sequences

The thm bed response of 2-MHz LWD resistivity took+strongly depends on the
background reaistivity. In conductive formations, the vertical resolution of LWD
tools is primarily determined by the receiver spacing. As the formation resistivity
increases, the vertical response broadens es the 2-MHz wavelength becomes larger.
Thk trend is illustrated in the modeled response to the Oklahoma lls sequence,
shown in Figure 3. The Oklahoma II sequence, with its large variety of thin resistive
and conductive beds, is an ideal testing ground for studying the behavior of 2-MHz
tools in thks beds with varying d]p.
Figure 3 shows CDR response to the Oklahoma II sequence without dip. Thk
figure illustrates the good qualitative vertical resolution of the 2-MHz tools: the
phsz.e shift resistivity ItPS identifies all but the lest th]n bed (between 148 h and
150 ft). Ri is reached in all conductive and most resistive thin beds. In a few
resistive beds, RPs falls short of Rt; however, the discrepancy is small enough to
allow for bed th]ckness corrections 4. For thk example, bed thickness corrections are
straightforward because all bed boundaries” are clearly identified by the crozs-overs
of the two resistivitiee.
Figure 4 shows a computed CDR log of the Oklahoma II sequence at a 45° dip

4
SPWLA31s[ hnud Logging Symposium. June 24-27, 19

angle. The depth shown in the figure is the true depth AZL, whkh is related to
the depth shown in the log Aze through the dlp angle Odip: AZ.L = Aze . co%(edi~).
There are only minor dHferencee between the 45” dipped log and the vertical log
confining that the thin bed response of the 2-MHz tools is fairly immune to dip
up to 45°. Figure 5 shows the Oklahoma II sequence for 60° dip.

Field Log Examples of Dipping Thin Beds


Figure 6a shows a field log from the Gulf of Mexico. There are two resistive
beds that are clearly seen by the CDR, Compensated Density Neutron (CDN*)
and gamma ray logs which are poorly delineated by the wire~me reeistivity logs.
Although the wireline induction and SFL* logs practically overlay in the resistive
beds indicating no invesion, the RPS and RAD curves show considerable separation.
Without invasion in the resistive beds, dip may cause the separation of the CDR
resistivitiea and the low resolution of the wireline logs. The dipping bed model was
used to match the field log and confirm thm hypothesis.
The inputs for the modeling code are the d]p angle and the bed thlcknessee
and resistivitiee. No dlpmeter weE available on this well, and a dkctional survey
indicated a deviation angle of 63°. However, consistency checks with an induction
dlp correction algorithm provided an actual apparent dip angle closer to 70”. R*
values are derived from the RPs curve, whkh is the hlgheet resolution resistivity
log in thk conductive environment. The inflection points of the CDR and CDN
curves were used to determine the bed boundary locations.
Figure 6b shows the modeling results for the induction and CDR responses in
thk formation. The bed boundaries and resistivities modeled are indicated by the
thin, straight Rt line. The resistive beds appear to be 10 ft and 7 ft thick on the
logs at 70° dip; however, their actual thicknesses are approximately 4 ft and 3 ft,
respectively.
The main features of the field log are reproduced by the computed log: the RPs
and RAD curves show a significant separation, and the two resistive beds appear
to be a lees resistive si?gle bed on the dual induction log. The steep dlp angle
causes a considerable effect on the induction logs. The main difference between the
computed and field logs are the horns on the computed logs which do not appear
on the field log. The size of the horns would be diminished if the dipping bed code
took the presence of the borehole into account 1°.
In addition to 70° dip, the induction and CDR responses were modeled for this
thin bed sequence at 0° and 40” dip, and for higher resistivity levels and contrests,
in order to differentiate dip and shoulder effects. These results are shown in Figure
7. (In order to match the same bed boundary locations that occur on the field
log, for 0° dip, bed thicknesses of 10 ft and 7 ft are modeled, and for 40° dip, bed
thicknesses of 8 ft and 5.5 ft are modeled.) For R. = 0.2 ohm-m, the CDR curves

,-- * Mark of Schlumberger

5
SPWLA 31s1 Annual lagging .synpsium, June 24-27, 199+3

separate only slightly at 0° dlp due to shoulder effect. Both dual induction curves
differentiate the two resistive thin beds and read Rt in the center of these beds.
For 40° dip, the resolution of the wireline logs has deteriorated a little, while the
CDR resolution remains about the same es for 0° dip. Thw example again confirms
that the 2-MHz tools are quite insensitive to dlp below 45°. The 70° dip log for
Ra = 0.2 ohm-m is reproduced from Figure 6b for completeness. In thk conductive
environment, the 2-MHz tools provide a good quantitative resistivity log even in
the presence of dip.
The influence of resistivity level on dlp effect is illustrated in the second row
of results in Figure 7. The same dipping thin bed configurations are shown with a
ten-fold incresae in resistivity level. The 0° d]p induction log shows no separation
between the IDPH and IMPH curves because the Phasor* algorithm hes totally
corrected both logs for shoulder effect. The RPS and RAD curves separate slightly
due to the larger shoulder effects at the higher resistivity level’~”. For 40° dip, the
separation between RPS and RAJJ is more pronounced at this higher resistivity level.
TKIs separation increes.es even further at 70° where RPS is no longer reading Rt in
the resistive beds due to the decreased vertical resolution at the higher rezistivity
level. The slight separation between IMPH and IDPH at 70” which wss observed
in the conductive sequence above increases a little more at this higher resistivity
level.
The bottom two rows in Figure 7 show modeled CDR and induction logs with
Rt in the thin beds increased by an additional factor of ten. Because the induction
curves overlap the CDR curves, they are plotted separately. The high resistivity
incresses the shoulder effects in the thin beds for all three dip angles. Since RPS hss
better vertical resolution than RAD, RPS reads closer to R~ than RAD, and a larger
separation between the two curves results. This separation is further incremed by
dip.
This model study leads to the following conclusions about dip and shoulder
effect on induction and CDR logs:
I) Separations between RAD and RPS can be caused by combinations of dlp and
shoulder effect, with RPS reading closer to Rt.
2) Separations between wireline IDPH and IMPH may be caused by dlp even
when Phasor processing can completely correct for shoulder effect at 0° dip.
3) Shoulder effects are greater at high resistivity than at low resistivity.
4) Dip effect extends over a greater dk.tance at high resietivity, but horning is
more pronounced at low resistivity.
5) Dip and shoulder effect are greater at Klgh reeistivity contrests than at low
contrssts.

“ Mark of Schlumberger

6
SPWLA 31st Annual Lagging Sympmium,June2627, 19

Figure 8 shows a field log from the Gulf of Mexico together with a modeled
reproduction. A resistive thin bed with shale shoulders at 0.8 ohm-m is logged with
the CDR tool at 72° dip. From the field log, a sequence of thin beds with suitably
chosen Rt values is estimated and modeled with the dipping bed code. The modeled
results are shown above the field log. The modeled shale-sand transitions are too
sharp, showing small horns. The point-dipole approximation enhances these horns
which appear smeared out on the log due to the finite size of the CDR tool and the
presence of the boreholel”.

‘I’he Polarization of a Dipped Bed Boundary

Figure 5 shows a modeled CDR log in the Oklahoma II sequence for 60° dip.
Compared to the O“ or 45° dlp log, this log has developed horns at some bed
boundaries. These horns are caused by polarizations at the boundaries. The height
of a horn depends on the resistivity contrsst. A small resistivity contrsat between
adjacent beds does not produce a horn. Conversely, a large resistivity contrset
generates very prominent horns.
The polarization is caused by a discontinuity of the electric field crossing the bed
boundary. In the absence of dip, for both 2-MHz and induction tools, the induced
electric fields and their currents are parallel to the bed boundaries. However, when
a dipping bed boundary is encountered, the induced currents have to cross the
bed boundary. The currents must be the same for a given current loop because the
electrons or ions cannot simply vanish. The boundary separates two beds of different
resistivity. Thus, to obey Ohm’s law, the electric field must be different in the two
beds. The normal component of the electric field hsz a discontinuity across the
boundary w~lch causes a polarization or a charge build-up at the boundaryg~lo,~1.
The polarization layer oscillates with the same frequency se the originating
wave. Thus, the oscillating charge acts like a secondary transmitter. As the re-
ceivers move pzat the boundary, the received signal strength from this secondary
transmitter suddenly becomes very strong, and the received signal is distorted into
a “polarization-hornnt Since the horn is located dkectly at the bed boundary, it
is a h]gh quality bed boundary indicator for large resistivity contrasts in steeply
dipped beds.
The polarization of the dipped bed boundary depends on the reaistivity con-
trazt and reaietivity level. Howardl 1 showed that the magnitude of the polarization
is proportional to ~ sinz (Od;P). Thk dependence is ihstrated in Figure
9. Thk figure shows the CDR response to a single bed boundary with a 2:1 or a
10:1 resietivity contrast, and resiztivity levels of <2.0 ohm-m and <20.0 ohm-m.
All figurez show that the vertical response of the 2-MHz tools is fairly stable for

t The horning effect in the presence of dlp was first explained as a consequence
.- of bed boundary polarization, or charge build-up by Allen Q. Howardg, 11.

7
SPWLA31 st Annual Jagging Symposium, June 2427, 1990

dip angles up to 45°. For larger angles, the polarization-horn develops. The horn is
much more prominent for the 10:1 contrasts compared to the 2:1 contrasts. Further-
more, the horns are more prominent in conductive than in resistive environments.
These observations explain the presence of the polarization-horns in the modeled
log examples (Figures 6 4-z8).
The polarization-horn at a steeply dipped interface hzz been observed for the
CDR tool in the test tank experiments (Figure 2). At 0° dip, the electric field is
oriented parallel to the water surface and thus cannot polarize the water surface.
Even for dip angles up to 47” the polarization is too weak to cause a horn. However,
at a dlp angle of 61°, the horn starts developing. The horn corresponds to the
phsse shift or attenuation falling below the saymptotic value in air far from the bed
boundary. The horn is always located within 6 in. of the air-water interface.
Further remits show the growth of t~ls horn at 76° and 88”. At these high
anglea and the infinite resistivity contrast, the effects of the point dipole approxi-
mation become more visible. The experimental data are more spread out than the
theory prediction, and the observed size of the horns is smaller than the prediction.
The difference in vertical spreading is caused by the point dipole assumption in the
model. The point dipole antenna in the model has a single intersection point with
the water surface while the actual dipped loop antenna crosses the water surface
over an extended interval.
The point dipole antenna in the model overestimates the size of the polariza-
tion-horn. IrJthe model, the surface charge layer exists right up to the point source,
wh]le the actual loop antenna is dktributed and separated from the polarized water
surface by an insulating layer. Therefore, the peaks and valleys in the experimental
results are shallower than the model prediction. The log examples (Figures 6 & 8)
show a similar overestimate of the horn in the model compared to the field data.
The polarization-horn hss been observed on steeply dipped field logs. Figure
10a shows a log example from the North Sea. This log has an interesting history.
Before the CDR tool was run, we wanted to confirm that it could identify the
interface between a 1 ohm-m shale and a 1000 ohm-m gas sand in a 72” deviated
well. This interface was modeled with ELMOD, and the computed RPs log clearly
showed the polarization-horn (Figure 10b). Thus, the bed boundary identification
could be assured even though the gas sand resistivity fell outside the specified
resistivity range of the CDR tool (RPS = 0.2. . . 200 ohm-m)4t. The log shows
the horn, and also shows that RP.s still provides a quantitative resistivity reading
at 700 ohm-m. Figure lla shows another North Sea log example. The CDR tool
logged the interface between a shale and a hydrocarbon bearing sand at 75”. The
polarization-horn is clearly present and ELMOD confirmed thk horn (Figure llb).

t For the specified resistivity range of the CDR tool, the measurement accuracy
is guaranteed as the larger of 3% or 0.5 mS\m. Above 200 ohm-m, the CDR tool
is operating, but the accuracy for RPs may be reduced.

8
SPWLA 31st Annual lagging Synpxium, June 2&27, 199

,,-

Vertical Response Functions in Dipping Beds

Dipping thm bed effects can be effectively described by dipping response func-
tions for 2-MHz toolse analogous to the concept of response functions for induction
tools, known es geometrical factor theory 12. The rwpome function describes the
relative amount that a specific volume of formation around a tool contributes to
the total measurement. Specifically, the response function essigns each volume ele-
ment around the tool a weight, such that the weighted average of the conductivityies
in the corresponding volume elements yields the total apparent conductivity. For
azimuthally symmetric mandrel tools, the response function depends in a fair]y com-
plex manner on the radial and axial distance from a reference point. Frequently,
th~ complex dependence is simplified by reducing the two-dimensional function into
two one-dmensional functions. For induction tools, this provides the radial geomet-
rical factor, which indicates the depth of investigation, and the vertical geometrical
fcator, which indlcatea the vertical (or thin bed) resolution. Thk concept can be
readily extended to 2-MHz propagation tools 6. In contrzat to induction tools, the
rad]al and vertical response of 2-MHs propagation tools varies considerably with
the background conductivity dt.
At 0° dip, the vertical response functions are derived from the modeled log of
a step profile

ut(z) = b + e(z – ZO)AU (1)

with a background conductivity D, a small conductivity contrzst AU and the sign


function O(z) = +1 for z greater or smaller than zero. The response function is
calculated from this small-step resistivity profile

gx(z,q . _ (2)

where the apparent conductivity U.(z) is given by the convolution integral


m
0. (z) = dz’gz(z – z’, O)ot(z’) . (3)
/
—ccl
Thk method provides a good approximation as long sa the conductivity contrast is
small: Au/a <1.0. In thk limit, the response function does not change appreciably
between the two resistivity zones.
The computation of vertical response functions can be extended to dipping
response timctions. The dipping bed code is used to model a synthetic log for a

t B~kground conductivity is an “equivalent” conductivity that represents the


-. average conductivity around the tool.

9
SPWLA 31st AIUIUalLagging Symposium, lune 24-27, 1990

step profile (Equation 1) at various dlp angles. The results are shown in Figure 12.
AU reaponze fimctions are plotted versus true depth ZL. The reaponze functions
change considerably es a function of angle. With increasing dip angle, the central
response lobe becomee lower and broader. At large angles, the central lobe splits
in two. Thk+ split becomez apparent above 30” dlp for RAD and above 00° for
RP9. Qualitatively, the rezponse shows the same behavior for both resistivity levels
although all the responses are considerably more spread out at the higher level
because of the longer wavelength.

Laminated Formations and Reeistivity Anisotropy 13

The dipping bed code can model an arbitrary number of beds and dip angles.
It hes been used to study tool response in laminated formations. Figure 13 shows
computed Rp$ and RAD curves in l-in. laminated formations consisting ofi

1) 25% sand & 75% shale


2) 50% sand & 50% shale
3) 75% sand & 25% shale
with Eaand = 25 ohm-m and ~~hale = 1 ohm-m. The modeled dip angles are 0° to
90°. The comparison in Figure 13 shows that above 40” dip, RP$ is significantly
higher than RAD for all three sand-shale resistivity distributions. The RPS curves
show larger oscillations than the RAD curves because of the higher vertical reso-
lution of the phaee shift measurement. These oscillations would be damped if the
presence of a borehole wae taken into account in the dipping bed model.
For induction tools, it has been demonstrated12’14 that ILD and ILM read the
volume average of the formation conductivityies at 0° dip. At various dip angles,
induction response without skin effect can be approximated 15from the 0° dip value
using the relationship

q ~A2cos2(OdiP) + Shz(odip)
Ue.+ip = (4)
A

with the dip angle ed,~, and the anisotropy coefficient A. Thk coefficient is defined
ESfunction of the horizontal conductivity Uh, and the vertical conductivity u.

(5)

Thk approximation is inadequate for 2-MHz propagation tools. While the RAD
curves in Figure 13 closely agree with the above induction approximations, the RP.S
curves depart significantly and become systematically higher at large dlp angles.
Whenever the lamination layers are much thhner than the receiver spacing, the
tool response can be approximated by the response in a homogeneous anisotropic

10
sPvmA31st.hud bggtig Sympsiutn. Jun. 24-27919

medium. The fields of a point magnetic dipole in an anisotropic medhm are given
by Moran and Gianzero M. Thae exPr=sion~ apply to the response of induction or
2-MHz propagation tools. Figure 14 compares the averaged CDR resistivities in the
laminated zones of Figure 13 to the apparent resistivities computed in equivalent
homogeneous anieotropic media. The good agreement of the two sets of curves
in Figure 14 indicates that larnination anisotropy and microscopic homogeneous
anisotropy are equivalent and cause similar separations between the RPS and RAD
curves. Thus, the previous modeling of laminated formations is also relevant to the
modeling of shales that typically are anisotropic.
Effective values for Oh and u. used in homogeneous anisotropic media calcula-
tions are derived from lamination conductivities and thlckneeses. If the laminated
med]um consists of two materials with conductivities al and aZ, and volume frac-
tions a and 1 – a, then the anisotropic conductivities are given by

Uh=CZC71+(l-(Y)U~ (6)

1 U1U2
(7)
‘“=a R1+(l–a)Rz=auz+ (1–a)ul “

For a laminated formation, the effective vertical conductivity will always be


,-,
km than the effective horizontal conductivity. The horizontal conductivity controls
the current traveling within the beddkrg planes. This current sees an effective
conductivity equal to the average of the conductivities which is analogous to parallel
resistors. On the other hand, the vertical conductivity controls the current traveling
perpendicular to the bedding planes. This current sees a resistivity equal to the
average resistivity whkh is analogous to series resistors.
Figure 15a shows the apparent CDR resistivities ss a function of dlp for forma-
tions in which Uh/Uv = 2. This anisotropy is a typical contrast for actual logging
situations. Curves are computed for four reeistivities Rh: 1, 10, 100 and 1000 ohm-
m. As expected, the RPS and RAD curves do not separate at small dip angles. Both
curves read the horizontal conductivity of the formation. As the dlp angle increases,
both readings become more resistive and separate. This separation becomes sizable
at dlp angles exceeding 45”. In all ces.es, RPs reads more resistive than RAD. The
separation bet ween the two resist ivitiess ystematically increases with increasing dlp
angle. At large dlp angles, the readings deviate more from the horizontal resistivit y
as the formation resistivity decreases.
A comparison of the CDR response (Figure 15a) with induction response (Fig-
ure 15b) in the same anisotropic formation shows that induction tools are aIso
affected by anieotropy, although to a far lesser degree. For induction, the appar-
ent conductivity derived from both the R and X signals is shown in Figure 15b.
The normal “boosting” algorithm has been applied to the R-signal to obtain the
apparent conductivity. A similar boosting algorithm wse derived for the x-signal,

11
SPW 31st Annual Logging Symposium, June 24-27, 1990

whkh is analogous to the CDR resistivity transform. There is negligible separation


between the induction R and X signals as a result of formation anisotropy and dip.
Anisotropy causes a separation between laterolog and induction curves in non-
invaded formations. The Iaterolog tools usually read higher resistivity than induc-
tion. This separation is caused by basic differences in the response characteristics of
the two tools: induction tools are most sensitive to horizontal conductivity, whereas
laterolog tools respond more to the vertical rezistivity16,17.
The previous modeling hsa shown that at large dlp angles, similar separa-
tions between the phase shift and attenuation resiativities of the 2-MHz tools can
be attributed to anisotropy. In particular, a separation between the two 2-MHz
reeistivitiee does not always indicate invasion.
Figure 16a shows a field log from the Gulf Coazt where shale anisotropy is
thought to cause the separations between the CDR resistivity curves in the central
portion of the log. The deviation angle in this well is 72°. The mud was fresh, so
borehole effects were minimal at this resistivity level. The region where the curve
separation occurs is fairly thick, and the attenuation curve is flat. Therefore, dip or
shoulder effects can be eliminated as a cause of the separation. The gamma ray log
is quite high over the entire interval indicating shale. In addition, the gamma ray
log becomes perceptively higher where the two resistivity curves separate, which
indicates a change in the shale characteristics.
Figure 16b shows the computed CDR response to a shale with an anisotropic
central region. The dipping bed code was used to model a central laminated section,
with lamination conductivities derived from Ok and u. using equations (6) and
(7). As noted above, the central region could have also been modeled by using
an anisotropic thin bed code. For the central region, the horizontal and vertical
shale conductivities are determined m: (7k = 0.5 S/m and UV = 1.5 S/m. These
valuea were obtained by recursively inverting the anisotropy response expressions
and solving for the log values.
The good agreement between the field log and the computed log indicates that
shale anisotropy is a likely cause of the reeistivity curve separations on this log.
It would be helpful to have a dual induction-SFL log in this case, because either
separations between the induct ion curves and SFL would confirm anisotropy, or
separations between the medium and deep induction curves would suggest invaaion.
Regardless of the outcome in this case, the previous modeling has demonstrated
that rezistivity anisotropy must now be considered in addition to invasion, shoulder
beds, borehole effects and dielectric effects when interpreting 2-MHz attenuation
and phaae shift resistivity logs. The effect of anisotropy together with dip cause
both apparent resistivities to differ from the readings in a vertical well and to
separate from each other.

12
SPWLA 31st Armual Logging Syqmsium, June 24-27.19

Conclusions

A computer program has been developed for modeling the response of 2-MHz
LWD resistivity tools in dipping beds. Thw program generatee simulated logs in an
arbitrary number of dipping th]n beds at any relative dip. It has been verified by
comparison with FEM mode~mg in the limiting caee of 0° dip. Modeling predictions
also agree with the results of test tank experiments where a 2-MHs LWD tool loge
an air-water interface at various anglee.
Thw program and a similar program for induction tools wee used to analyze and
explain some intereeting effects observed on field logs. It wee shown that dip effect
and shoulder effect can cause separations between LWD phsse sh:ft and attenuation
resistivity curves. Aniiotropy in laminated or shaly formations can cause deviations
from Rt and separations between curves ailxting the phase shift resistivity more
strong] y than the attenuation reaistivity. Modeling hea verified that horns on
2-MHz logs at steeply dipping bed boundaries are caused by electric polarization at
the dipping interface. These effects are preeent for all 2-MHz LWD tools.

Acknowledgments

First of all, we thank our clients who kindly permitted us to use sections of
their logs se examples in the paper. We thank David Best for his extensive help with
finding and processing field log examples. We also thank Craig Kienitz for providing
Gulf of Mexico log exarnplee and David Dudlyke and Jim White for providing North
Sea log examples. Finally, we thank Tom Barber and Allen Howard for dkmssions
during which they shared their insights into the charge build-up of dipped bed
boundaries.

References

1 B. Anderson, K. A. Safinya, T. Habsshy, “Effects of Dipping Beds on the


Response of Induction Tools”, SPE 61st Annual Technical Conference, New
Orleans (October, 1986), paper SPE 15488.

2 R. H. Hardman & L. C. Shen, “Theory of Induction Sonde in Dipping Beds”,


Geophysics, Vol. 51, No. 3 (March, 1986), pp. 800-809.

3 S. C. Gianzero & S. M. Su, “The Response of an Induction Dipmeter and


Standard Induction Tools to Dipping Beds”, to appear in Geophysics.

4 B. Clark et al., “A New Resistivity Measurement for FEWD”, Transactions of


the SPWLA 29th Annual Logging Symposium, San Antonio, June 5-8, 1988,
Paper A.
5 B. Anderson, T. D. Barber, J. Singer, T. Broussard, “ELMOD–Putting Electro-
,-. magnetic Modeling to Work to Improve Resistivity Log Interpretation”, ‘hns-

13
SPWLA 31st Annual Logging Sympsiw June 24-27,1990

actions of the SP WLA 30th Annual Logging Symposium, Denver, June 11-14,
1989, Paper M.
6 D. F. Allen & M. G. Liiling, “Integration of Wireline Resistivity Data with
Dual Depth of Investigation 2-MHz MWD Resistivity Data” Transactions of
the SP WLA 30th Annual Logging Symposium, Denver, June 11-14,1989, Paper
c.
7 S. Chang & B, Anderson, ‘A Finite Element Method for Electromagnetic Well
Logging in Cylindrical Boreholes”, Transactions of the IEEE Antennas and
Propagation Society International Symposium & URS1 Mlo Science Meeting,
San Jose, Calif., June 26-30, 1989.
8 T. D. Barber, “Introduction to the Digital Induction Tool”, JPT (September
1985), 1699-1706.
9 T. D. Barber & A. Q. Howard, “Correcting the Induction Log for Dip Effect”,
SPE 64th Annual Technkal Conference, San Antonio (October 1989), paper
SPE 19607.

10 A. Q. Howard, private communication.

11 A. Q. Howard & W. C. Chew, “A Variational Model of Induction Logging in


a Dipping Bed Environment”, IGARSS Symposium, Vancouver (July 10-14,
1989), Seesion F8, paper 5.

12 H. G. Doll, “Introduction to Induction Logging and Applications to Logging


of Wells Drilled with Oil-based Mud”, Petroleum Transactions, AIME (June
1949) , 148-162.
13 J. Tittman, ‘Formation Anisotropy. Reckoning With its Effects”, The Oilfield
Review, Vol. 2, No. 1 (January 1990), pp. 16-23.
14 B. Anderson, “The Analysis of Some Unsolved Induction Interpretation Prob-
lems Using Computer Modeling”, The Log Analyst, Vol. 27, No. 5 (Sept.-Ott.,
1986) , pp. 60-73.
15 J. H. Moran& S. C. Gianzero, “Effects of Formation Anisotropy on Reactivity
Logging Measurements”, Geophysics, Vol. 44, No. 7 (July 1979), pp. 1266-86.

16 R. Chemali, S. C. Gianzero, S. M. Su, “The Effect of Shale Anisotropy on Fo-


cused Resistivity Devices”, Transactions of the SPWLA 28th Annual Logging
Symposium, London, June 29-July 2, 1987, Paper H.
17 B. Anderson & T. D. Barber, “Using Computer Modeling to Provide Mksing
Information for Interpreting Resistivity Logsn, Transactions of the SPWLA
29th Annual Logging Symposium, San Antonio, June 5-8, 1988, Paper H.

14
SPWLA 31st Annual Logging Sympsium, June 24.27, 19

-.

2-0 -J
Lr
m203D40E416i37il si2
Depth (ft) Depth (ft)

Figure I: Comparison of Dipping Bed Code and FEM resdts in a thin bed sequence

,,-

d--’-
4--’-
m;

470+-”-
“.

1
-48-26-24 -12 0 D 24 36
True Depth (in.)
,-.
Figure 2: Comptison of experimental test tank results and modeled results

15
~g
........—
......
SPWLA 31st Annual Lagging Symposium,June 24-27, 199+1

01

Rt

‘h
--”-”---” R P s

5 3 ““”’”””’””””””””-””-””-””
Figure 3: OkJahOmaII sequence
~= modeled for the CDR
> ~ , ,
“g 9 ;1 ~ tool at o“ dip
.-
U)
L?=
a

E
E
s
~
Figure 4: Oklahoma II sequence
=> mod.ded for the CDR
>
= tool at 45” dip
~
w
2

ii
~Q
g~
Figure 5: Oklahoma II sequence
h
.=
>q mod.ded for the CDR
.- S? too] d 60° dip
~
z
=2

o 20 40 60 80 100 liO 140 I&O 180


True Depth (ft)
-
.
1
!“’”””””””’’”’’’”’””””:””””’”

LaggingSymposium,
June24-27,1
k
...!
#“”~i..h-
..................;
...........
o (
o ~= (
o t
. ne ,
~~
SPWLA 31st Annual Lc.gging Symposium, June 24-27, 199+3

::
.,
H:.,
No MI)
+
40°
+
Dip 70° DiIJ

%
Rt X 10

+
x 1~

1*OOL
1000
1
,.
KK E H-

m“ .
RsX1O /
....
‘t $-’OO .......... ~ o- !
.,

(CDR) ]

1100
1000

P
P“
RS X 10

‘t 3’00 ““”””’’’’””””’””’’”’’””f
““’’’-!”””” s .-
(Induction)

“0° 0.02 O.* 0.o1 200


~~ ~~~ WLIDPH~ WL IDP!4
0.02 ml 0. .._ ~L - _,.._.+.~ *m
..... ... w~ ,W
$* . . . . . . . . . . . . . ~D,,
..—
. . . . . . . . ..*~ .-” ygp-::’”>.+i
~g.............
&----------- RAO---------~
0.2 ~~~
0; Rm —
moo
Rw ~
sow 0.2 0.2 Rwo
0.2 Rt
P
Rt — m—
I
Figure 7: Additional modehg results for the Gulf of Mexico example

18
SPWLA31st Annual logging Sympxium, June 24-27, 19

1000 —

1050

1100

!0
,: !!
,.- c% ‘0 ).2
LOCIRad 2a ii
. . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.2 2000
Log Rps
).02 200
Model Rad
.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
).02 200
Model Rps
).02 200
Model Rt

Figure 8: A second dipping bed log exsmpIe from the Gulf of Mtico with modding results

19
RPS RPS

L 4

F 1
RAD RAD

. ~o
. . . . . . MO
–“— ~~o
—.. no

— %
o.1~ + 4
-80-40 -S6-24 -12 0 12 24 M 42 0 40-40-20-24-12 0 w 24 80 4 -eO-40-2e -24-12 o 12 24 a 40 a -42-S0 -24-12 0 12 2436426 I
lhle Depth (Ilwhes) lkue Depth (Inches) ltue Depth (inches) hue Depth (Inchm)

Figure 9: The polarization horn modded for several resistivity levels and contrasts
SP~ 31stAmualLogging
Sympo.siurm June 24-27,19

1000-! ‘-; ‘“’’”: ““’”; ““-:

..............
(1%%) z i

1020- :
0.2
----------------- RAD-----------~
0.2
RPS
0.2 2000
Rt

Figure 10a: A North Sea field log showing the Figure 10b: Modeled reproduction of th
polarization-horn polm”zation-hom

1
0.00 m 100.00 0.20 ATTENUATION RESISllVllV 2000.00
--------------------------------
0.20 PliASE SHIFT RESISllWTY 2W0.00

Figure ha: A second North Sea fidd log showing Figure 11 b: Modeled reproduction of
the polarization-horn polarization-horn

.,-

21
SPWLA 31s[ Annual Imgging Synpsium, June 24-27, 1990
1
E
...,.
.-
E
c
o
..,!
-..-: -.-.....,
g
3
u ‘ -=+::...
9ro Zro Soo *OO 000 9ro 7$0 800 >00 000 ‘
asuodsaH l13a!J.laA
SPWLA 31s[ AnmualLagging Symposium.June24-27, 19

Attenuation Resistivities Phase Shift Resistivities

25 0h Sand & 75 0k S h ale

10

, 1

0.6
-200 -,60 .,,0 .60 ,00 tso 200
o.s.~oo
Depth ylnchei~ Depth (Imhes)

St)yo Sand & 5 00/’ Shale


,003 ,,,

~1111111111111111111111111111111111111111~
$
~ ,0. 10
=x ,..
>
.
=
:

*- ,

0.s. 04
-200 .,s0 -,00 -50 ,00 *SO 200 -200 -*O .,00 -60 ,00 ,s0 *OO
Depth iwheif Depth ~lnche~f

750/0 Sand & 25% Shale


100. 100

~111111111111111111111111111111111111111~
~
E
3 ,0, ,0
,,.
=- —@-
>
~
~
:

1, 1

0,s, 0.s
.200 .,80 -,00 -60 mo ,,0 200 -200 -1s0 -too -s0 ,00 ,,0 200
Depth &he8Y Depth ~lnohegf

Figure 13: Modeled lamination sequences

23
Attenuation Reslatlvlty CDR - ahhv = 2

E
‘“”m
Anl.c.tropy
Dk.1.o Iamlnstlena

=*
>0
~ 10
= 0:

&
~ --
: . ----- < 75% sand & 25% shale
m
: . ------- - < 50% sand & 50% shale :
. . . . ..--------”- . . .. . . . . . .. . . . . . ..
a 0
-.
..-
. ----- < 257. sand & 75°h shale

. . ---------
— RPS --------- RAD

o 10 20 30 40 50 60 70 80 90
Dip angla (dagreee)
Figure 15a: CDR response to a homogeneous ankotmpic medium
Phaae Shift Raalativlty

x

‘“”m
Anl.otrepy
I
I
~ ‘------ MWO Iamlnul.nw
cE .s0% sand & 50% shale
~
75% sand & 25% shale

25% aand & 75% shale


=*
>0

a
~
1~ — R-Signal -------- X-Signal
o 10 20 So 40 m @o 70 80 00
Dip angla (degrees)
o 10 20 30 40 50 60 70 8Q h
Figure 14: Compmison of CDR response in thinly laminated Dip angle (degraee)
and homogeneous miaotropic media Figure15b: Induction response to a homogeneous anisotropic
medimn
SPWLA 31st Annual Lagging Sympxium, June 24-27, 19
Anisotropic
... . . . ................
.........
3---,



> .

25

You might also like