Induction, Resistivity, and MWD Tools in Horizontal Wells

S. Gianzero: Halliburton Logging Services, Austin, Texas R. Chemali: Halliburton Logging Services, Austin, Texas S.-M. Su: Halliburton Logging Services, Austin, Texas

Abstract: Conventional induction and focused resistivity tools are designed to measure resistivity from a vertical borehole surrounded by a cylindrically invaded zone while minimizing the signal contribution from adjacent horizontal beds. In recent years, our understanding of these devices was extended to include beds exhibiting a large dip relative to the borehole, as in the case of a highly deviated well. We investigate the applicability of induction and resistivity devices to horizontal wells, where the borehole runs parallel to the bed boundaries. The presence of the borehole may be simply ignored for induction sondes, with the tool response being computed via an analytic solution. Because of the relative simplicity of the induction solution, the log response is computed for entire trajectories for the more common radii of curvature used in the drilling process. On the other hand, for focused resistivity devices such as the dual laterolog or the MWD toroid sonde, the borehole is an essential part of the problem. The tool response is evaluated using a numerical solution to simulate accurately the complex physical situation. The modeling results for the resistivity devices indicate that the measurement is more sensitive to conductive than to resistive shoulder beds. Typically, for the MWD sonde, 50% of the resistivity signal comes from the adjacent conductive bed when it is 0.5 ft away from the approaching borehole wall. A similar sensitivity to a resistive adjacent bed is not attained until the borehole has actually penetrated the bed. The reverse physical situation is in evidence with induction devices-namely, resistive adjacent beds are more readily detected than are conductive adjacent beds.

evaluation methods. Because all existing instruments were primarily designed for the traditional geometry of a borehole normal to the bedding, their response to bed boundaries that are more nearly parallel to the borehole axis is of essential importance (see Figure 1). This is particularly true for resistivity sondes such as induction, laterolog, and measurement while drilling because of their large ranges of investigation into the formation. In many situations, the measurement while drilling sonde or the open hole sondes are strongly influenced by the resistivity of a neighboring shoulder bed. This is a desirable feature if the sonde is used precisely to signal an approaching upper boundary, but is an unwanted source of error if the sonde is run to measure the resistivity ofthe formation traversed by the borehole. In any case, mathematical modeling is applied below to predict the response of standard resistivity devices in the more commonplace situations. Two configurations are modeled: first, the case of an approaching single bed boundary; then, the situation where the sonde is in the center of a bed surrounded by an upper and a lower shoulder of like resistivities. For completeness, the case of highly deviated wells is addressed using recently published modeling techniques. SIMULATED LOGS IN A HORIZONTAL WELL NEAR A BED BOUNDARY When drilling horizontally in a reservoir, it is generally desirable to stay near the top of the oil or gas zone, immediately below the impervious cap rock. The boundaries between the various media tend to be parallel to each other and parallel to the borehole. If one of the boundaries is close to the wellbore, its effect is likely to be dominant. Therefore, a natural first step toward understanding the response of induction, laterolog, and the measurement while drilling toroid sondes in horizontal wells is to simulate their behavior near a single horizontal bed boundary. Dual Induction The borehole has been omitted in the physical model shown in Figure 2. Indeed, past experience has shown that the response of an induction tool calculated in such complex geometries does not depend significantly on the inclusion of the borehole in the model. Analytic solutions developed for dipping bed situations (Anderson, 1986;
May- June, 1990

INTRODUCTION Horizontal wells for exploration and development of oil fields are rapidly gaining acceptance in many types of reservoirs. Principally because of the enhanced recovery, increased penetration of vertical fractures, and controlled coning, production is generally more favorable from a single horizontal well than from several vertical wells in a given area (Mahony, 1988). Horizontal wells even help gather data more effectively to study the geology and the petrophysics in complex fields (De Montigny et al., 1988). Furthermore, horizontal drilling is less troublesome than previously imagined. This new technology naturally calls for new formation
1

On@y

presented at 1989 SPWLA Symposium, Denver, paper N.

158

The Log Analyst

Horizontal Wells

Dual Induction Run with a Tool Pusher in a Horizontal Well

Ruppar

Induction

Shen and Hardman, 1986; Gianzero and Su, in press) are then directly applicable to this simplified configuration. The results of the modeling (Gianzero and Su, 1989) are plotted in Figure 2 as a function of the distance separating the axis of the borehole from the horizontal bed boundary. When the dual induction is in the lower formation, far below the boundary, it naturally reads RIo,,,. As the sonde approaches the upper resistive formation, both the medium and the deep measurements sense this medium long before it is reached. After several feet above the interface, both induction devices approach the value
Rupper

Measurement While Drilling in a Horizontal Well

Figure 1: Logging horizontal wells with wireline and MWD resistivity tools.

The distance at which the sonde senses the resistive formation cannot be quantified exactly because the progressive transition between the two resistivity values also includes horn-like features. There are two conflicting requirements illustrated in Figure 2 that present themselves to the log analyst. If the principal concern is the error in the resistivity reading introduced by the adjacent bed, then the error in the ILd is in excess of 10% when the borehole is within 5 ft of the resistive shoulder and 25% when the nearby shoulder is conductive. Obviously, this latter situation is obtained by reversing the perspective in the figure. These effects clearly are less severe for the ILm. On the other hand, if the principal concern is to detect an approaching bed, only a significant change in the measured resistivity is considered a positive indication of the presence of a neighboring bed. For that specific application, an approximate distance of investigation may be selected visually from the computed logs. From the conductive formation it appears that the deep induction sees the resistive shoulder from a distance of 1 ft. The medium induction with its shallower investigation identifies the

................

I1

10

100

Resistivity, ohm-m
Figure 2
May-June, 1990

Computed response of the dual induction i n a horizontal well near a bed boundary.
The Log Analyst 159

Gianzero et al.

Rupper bower

'

L10 1
' s l ' ' ' 'I
'

Formation Bounda

P
I

'

'

" " "

10

100

Resistivity, ohm-m
Figure 3: Computed response of the dual laterolog in a horizontal well near a bed boundary.

approaching resistive shoulder at only 0.5 ft away. In the reverse situation when the sonde is in the resistive bed approaching a conductive shoulder, the log trend toward lower resistivity values is not clear for the deep measurement until the sonde is l ft inside the conductive formation. Similarly, the medium induction needs to be 0.5 ft inside the conductive bed to be significantly affected by it. These trends are further complicated by the presence of a horn on the curve precisely at the bed boundary.

Dual Laterolog Because the borehole usually has a significant impact on the response of the dual laterolog, it has been included in the model (Chemali et al., 1983).

The physical model of a dual laterolog in a horizontal well approaching a bed boundary is depicted in Figure 3. Because no simple analytical solution is available for this complex configuration, the simulation of the tool response is carried out via a three-dimensional finite element program. Consequently, the amount of computation required is several orders of magnitude larger than for the induction model. It is apparent from the results in Figure 3 that the behavior of the dual laterolog is complementary to that of the dual induction. More specifically,the dual laterolog is more sensitive to a nearby conductive bed than to a resistive bed. An approaching conductive bed is positively identified at approximately 0.5 ft by the shallow

................ ..........................
1

F sistivity ohm -m

10

100

10

100

Resistivity] ohm-m

Figure 4: Comparison of induction and laterolog in a horizontal well.

160

The Log Analyst

May- June, 1990

Horizontal Wells

................
*
G
G w-

RWWr

2-

5 n
n
Mandrel Dia. = 7" dh = 9"
7
I
, 1 1 1 1 ,

I
-4I
I

dh

10

100

Resistivity, ohm -m
Figure 5: Computed response of the MWD resistivity tool in a horizontal well near a bed boundary.

laterolog and at 1 ft by the deep laterolog, whereas a resistive bed is not seen by the dual laterolog until the sonde has actually penetrated the bed. As in the case of the deep induction, the perturbing effect of a shoulder bed-on the response of the deep laterolog is measurable from several feet away. For example, in the configuration of Figures 2 and 3 the errors observed at 5 ft away are nearly identical for the dual induction and the dual laterolog. Overlaying the deep induction and deep laterolog in one chart, and the medium induction and shallow laterolog on another (Figure 4), helps to visualize their relative behavior.

MWD Toroid Sonde The same study was carried out for the toroid dual resistivity sonde, which is implemented on a drill collar for a measurement while drilling application. This tool consists of two measurements, one called the Lateral and the other called the Bit. The Lateral measurement is similar to the shallow laterolog, and the Bit measurement measures the formation resistivity near the drill bit assembly. A detailed description of this tool is given in various publications (e.g., Gianzero, Chemali et al., 1985). The computed response of the tool near a horizontal bed boundary is shown in Figure 5 . As expected, the Lateral

10 100 1000 Resistivity, ohm-m Figure 6: Computed response of the MWD resistivity tool in a pay zone.
1 May-June, 1990 The Log Analyst

161

Gianzero et al.

has nearly the same characteristics as the shallow laterolog.

MWD Sonde in a Thin Bed A somewhat more realistic case is illustrated in Figure 6, wherein the MWD resistivity tool approaches the cap rock within the pay zone. For purposes of clarity, imagine that the logs are generated for a small angle of attack away from the horizontal direction. Observe that the readings range from approximately 1 to 10 S2-m as we move throughout the bed.

Shoulder bed

Rs

Rt
I

M B l

dual induction sonde

Shoulder bed Rs CORRECTION FOR SHOULDER BED EFFECT IN HORIZONTAL WELLS In the case of vertical wells, the shoulder bed effect of adjacent formations has been extensively modeled both for induction (Schlumberger, 1972) and laterolog tools (Chemali et al., 1983). The study recently has been extended to the case of dipping beds or deviated wells (Shen and Hardman, 1986; Chemali et al., 1988). The question Shoulder bed Rs naturally arises as to the magnitude of that effect in the case of horizontal wells parallel to the upper and lower bed bundaries. The evaluation of the shoulder bed effect is conducted for each of the models shown in Figure 7. The sonde is I in a formation of resistivity R,parallel to the upper and 4 dual laterolog sonde lower boundaries with shoulder beds. A borehole filled with a fluid of resistivity R, is also included in the model of the dual laterolog. The calculations are limited to the Shoulder bed Rs situation where upper and lower shoulder beds have the same resistivity R, and the sonde is exactly at the center r I of the formation. We have omitted the computation for : Physical configuration of a dual induction or a dual the measurement while drilling lateral sonde because its Figure 7 laterolog in a thin bed. response is so similar to the shallow laterolog.

Dual Induction A set of correction charts for the shoulder bed effect is reproduced in Figure 8 for the I M and in Figure 9 for the ILm. As in the case of wells perpendicular to the bedding, the variable parameter from one correction chart to the next is the value of the shoulder bed resistivity R,. A cursory examination of these charts indicates that the correction for shoulder bed in horizontal wells is a monotonic function of bed thickness as long as the latter exceeds 4 ft. Dual Laterolog Correction charts for the shoulder bed effect are shown in Figure 10 for the L M and in Figure 1 1 for the LLs. In this case, the variable parameter from one chart to the next is the resistivity ratio between the mud and the shoulder bed. From the limited number of cases evaluated in this study it appears that the resistivity of the borehole fluid is less critical in horizontal wells than in vertical wells (Chemali et al., 1983).
162

As observed with the dual induction, for a given R,/ R,, the correction for shoulder bed effect exhibits a monotonic variation with bed thickness. Furthermore, the shallow laterolog has very little shoulder effect when the formation thickness is 10 ft or more. As mentioned previously, the shoulder bed effect of the Lateral measurement while drilling sonde is so similar to that of the shallow laterolog that the results are omitted.

SIMULATED LOGS IN HIGH RELATIVE DIP The common practice for drilling horizontal wells consists of starting out with a nearly vertical borehole and deviating progressively along a specified curved trajectory until reaching the desired depth and direction. Along such trajectories the resistivity sondes encounter the various informations at progressively increasing relative dip angles. Typical trajectories for reaching a horizontal direction are shown in the diagram of Figure 12a. They are genMay- June, 1990

The Log Analyst

Horizontal Wells

200

t 8 60
40
20

. . c
0

'8 6
4

0 . 8
0.6 0.4

0 . 2

0 . 1
0
4

12

1 6 20 24 28

12

1 6 20

24 28

bed thickness, ft

bed thickness, ft

bed thickness, f t

12 IS 20

24

ia

1 2

16

20

24

28

bed thickness, f t

Figure 8: Shoulder bed correction for the deep induction in a horizontal well.

May-June, 1990

The Log Analyst

163

Gianzero et al.

E
I

12

16

20

24

28

12

16

20

24

28

bed thickness, f t

bed thickness, f t

E
I
L 0

12

16

20

24

28

12

16

20

24

28

bed thickness, f t

bed thickness, f t

Figure 9: Shoulder bed correction for the medium induction in a horizontal well.

164

The Log Analyst

May- June, 1990

Horizontal Wells

2.0 1.8 1.6 1.4 1.2

1 .o

0.8

0.6
0.4

0.2

0.0
1

Bed Thickness, f t

10

20

40

2.0 1.8 1.6 1.4

3 1.2
2

OL 1.0 \

0.8 0.6
0.4

0.2
0.0
1

Bed Thickness, f t

8 10

20

40

Figure 10: Shoulder bed correction for the deep laterolog in a horizontal well.

erated by turning the drill bit assembly at rates of 6"/100', 20/100', or 191"/100'. Along such trajectoriesthe relative dip angle varies continuously from 0" to 90". By entering the relative dip angle at each point into analytic response equations for induction sondes, simulated logs are computed for the given trajectories. These logs are plotted in
May-June, 1990

Figures 12b and 12c for the ILd and ILm, respectively. Observe that the simulated logs are plotted versus true depth, and are measured normally to the bedding. For reference, the resistivity profile and the ILd and ILm logs computed for a vertical well are also shown on the plots. In the sections of the hole where the relative dip is high,
165

The Log Analyst

Gianzero et al.

'

'

" " I

.
8

10

20

40

Bed Thickness, ft

2.0

1.8
1.6 1.4

1.2
1 .o

0.8
0.6 0.4

0.2

0.0
1

6 8 10 Bed Thickness, f t

20

40

Figure 11: Shoulder bed correction for the shallow laterolog in a horizontal well.

the deep induction exhibits horns and increased shoulder bed effect as indicated in previous publications. By comparkon, because of the inherently smaller shoulder effect, the ILm is not as severely affected as the ILd. Computation times for the same profile for the dual laterolog are prohibitive and were omitted. However, the
166

simulation of the dual laterolog in a bed with a high relative dip is shown for the specific case of a 5 ft resistive bed with a relative dip of 75' (Figure 13). The computed log is plotted versus apparent depth along the wellbore and versus depth measured normally to the bedding. As pointed out in a previous publication (Chemali et al.,
May- June, 1990

The Log Analyst

Horizontal Wells
% 4

.
0

E" al

.............. Curvature 6 degJ1OOft ---- 20 degJl OOft

-.191 degJl OOft

.0
II)

U a -100

-e, 0

-200

N .L

o -300
xi00 xi50

Depth normal to bedding, ft

Deep Induction

..............
----.-

6 degJlOOft
20 degJl OOft

x200

x250
Well .............. Vertical 6 degJlOOft

Medium Induction
I

20 degJlOOft 191 deg./l OOft

l !

x200

x250

Depth normal to bedding, ft

Figure 12: Computed dual induction logs for typical well trajectories.

1988), the response of the dual laterolog in this case does not exhibit any unexpected features, except for a generally increased sensitivity to the shoulder bed effect. Finally, for the sake of completeness we have included
May-June, 1990

simulations of the DFL (Digitally Focussed Log) (Strickland et al., 1987), which is a shallow resistivity device for a thin bed with various dip angles (i.e., 0"-75"). The zero dip case was included for reference (see Figure 14).
167

The Log Analyst

Gianzero et al.

0.1

1 " " 1 " " I " " I " " ~ ' " ' ~ " " ~ " "

Dip Angle: 7 5 '

v)

CY

Q)

Deep

0.1 ! . -120

.---.Shallow
. . . I
1 -

Rm= 0.1 dh= 8" Bed Thickness= 5 f t

, ' I

" ~ " " , ' " ' " . " , " ' ' ~ " "

-90

-60 -30 0 30 60 90 Depth normal t o bedding, inches

1 0

Figure 13: Computed dual laterolog response in a highly deviated well.

Note also that there are certain computed points indicated restores the resolution of the DFL as indicated previously in the figure from this case that illustrate agreement be- for the dual laterolog in Figure 13. tween two distinct methods of computation, namely the CONCLUSION hybrid method (Gianzero, Chemali et al., 1985) and the Induction, resistivity, and MWD tools were modeled Finite Element Method (FEM) (Coggon, 1971). Once again, observe that plotting points on true vertical depth mathematically in horizontal wells. The response char168
The Log Analyst May- June, 1990

Horizontal Wells

E 10

................

E -c

.>
-w
v)

..v )
C Y
Q)

>; -w

--1

-.-

-----

Bed Thickness= 2 ft

0.1

Depth measured along well axis, inches

Figure 14: Computed DFL response in a highly deviated well.

acteristics to a single bed boundary revealed that the induction and resistivity tools reverse their roles from the normal vertical wells situation. Specifically, induction tools are more sensitive to resistive shoulder beds, whereas resistivity tools are more sensitive to conductive beds. A series of shoulder bed correction charts for either a centered induction or resistivity device are provided for
May-June, 1990

use to the log analyst. The results for the resistivity device indicated a reduced sensitivity to borehole conditions for horizontal wells than for vertical wells. Complete trajectories for the more commmon drilling conditions have been provided for the standard dual induction. Additionally, a log profile of the dual laterolog response is provided for the extreme case of 75" deviation
169

The Log Analyst

Gianzero et al. angle for purposes of comparison. Finally, log profiles were included for the DFL traversing a thin bed at various dip angles. paper A, in 26th Annual Logging Symposium Transactions: Society of Professional Well Log Analysts, p. A 1-22. Gianzero, S., Lin, Y., and Su, S.-M., 1985, A new high speed hybrid technique for simulation and inversion of resistivity tools, SPE-14 189, presented at 60th Annual Technical Conference and Exhibition: Society of Petroleum Engineers. Gianzero, S., and Su, S.-M., in press, The response of a n induction dipmeter and standard induction tools to dipping beds, submitted to Geophysics. Mahony, B. J., 1988, Horizontal drilling use on the rise; why and how: World Oil, V. 86, no. 2 October 88, p. 45-54. De Montigny, O., Sorriaux, P., and Louis, A. J., 1988, Horii l& zontal well drilling data enhance reservoir appraisal: O Gas Journal, V. 86, no. 27, July 4, p, 40-47. Schlumberger, 1972, Log interpretation principles: Schlumberger Limited, New York, p. 19-20. Shen, L., and Hardman, R., 1986, Effect of formation dip or hole deviation on induction logs, paper I, in 27th Annual Symposium Transactions: Society of Professional Well Log Analysts, p. 11-14. Strickland, R., Sinclair, P., Harber, J., and DeBrecht, J., 1987, Introduction to the high resolution induction tool, paper E, in 28th Annual Logging Symposium Transactions: Society of Professional Well Log Analysts, p. El-1 7.

REFERENCES
Anderson, B., 1986, The analysis of some unsolved induction interpretation problems using computer modeling, paper 11, in 27th Annual Logging Symposium Transactions: Society of Professional Well Log Analysts, p. 111-18. Chemali, R., Gianzero, S., Strickland, R., and Tijani, S. M., 1983, The shoulder bed effect on the Dual Laterolog and its variation with the resistivity of the borehole fluid, paper UU, in 24th Annual Logging Symposium Transactions: Society of Professional Well Log Analysts, p. U U 1-25. Chemali, R., Gianzero, S., and Su, S.-M., 1988, The Dual Laterolog in common complex situations, paper N, in 29th Annual Logging Symposium Transactions: Society of Professional Well Log Analysts, p. N1-25. Coggon, J. H., 197 1, Electromagnetic and electrical modeling by the finite element method: Geophysics, v. 36, n. 1, p. 132155. Gianzero, S., Chemali, R., Lin, Y., Su, S.-M., and Foster, M., 1985, A new resistivity tool for measurement while drilling,

ABOUT THE AUTHORS

Stan Gianzero is Manager of the Austin Research Center. He received his B.S. in physics from Fordham University in 1959 and his M.S. in physics in 1963 and Ph.D. in astronautics in Roland Chemali is a Research Group Manager with Halli1965, both from Polytechnic Institute of Brooklyn. He was previously employed with Schlumberger Doll Research Center, burton Logging Services in Austin. He graduated from Ecole Ridgefield, Connecticut, where he worked in Electrical Logging. Polytechnique in Paris in 1966 and from the French Petroleum In 1981 he joined Halliburton Logging Services (formerly Gear- Institute in 1967. He was employed by Schlumberger in 1969 hart Industries, Inc.). He has authored or co-authored many key and held various positions in Research and Operations. In 198 1 publications, holds patents in electromagnetic logging, and re- he joined Halliburton Logging Services (formerly Gearhart Inceived the Best Symposium Paper Award in 198 1 and Dis- dustries, Inc.). He has co-authored several papers and has pattinguished T e c h n i c a l Achievement award fiom SPWLA in 1989. ents in sonic and electromagnetic logging.
170
The Log Analyst

May-June, 1990

Horizontal Wells

Shey-Min Su is a Research Engineer with Halliburton Logging Services in Austin. He received his B.S.E.E. from Tamkang University in 1977 and an M.S.E.E. from the University of Houston in 1982. He then joined Gearhart Industries Research Center in Austin, now part of Halliburton Logging Services. His research interests are in mathematical modeling of induction and resistivity logging tools, and the theory of electromagnetic logging. He has co-authored several papers in electromagnetic logging.

May- June, 1990

The Lag Analyst

171