Está en la página 1de 25

SPWLA Twenty-Eighth Annual Logging Symposium, June 29-July 2, 1987

ENHANCED VERTICAL RESOLUTION PROCESSING


-
OF
DUAL DETECTOR GAMMA-GAMMA DENSITY LOGS

C. Flaum, J. E. Galford

Schlurnberger Well Services, Houston, Tezas

A. Hastings

Schlumberger-Doll Research, Ridgejield, Connecticut

ABSTRACT

In today’s difficult economic environment for hydrocarbon exploration it is becoming


increasingly important to extract the maximum useful information from the data ob-
tained by wireline logs. One of the essential characteristics of these logs is their vertical
resolution.
One of the most important petrophysical characteristics of reservoir rock is its porosity,
and the most widely used porosity log is the gamma-gamma density log. In the modern
M
application of density logging, a dual detector compensation is implemented.
,---
The response of the compensated density log is dominated by the far detector, and
thus the resulting vertical resolution is determined by the far detector spacing, which
in most existing tools is between 1 and 1.5 ft. (In contrast, the near detector is usually
spaced somewhat closer than 0.5 ft from the source.) In addition, statistical averaging,
standard practice in the industry, further degrades the resolution. This means that
beds thkner than about 2 ft cannot be properly evaluated.
In this paper, we introduce a technique that utilizes a combination of the vertical in-
formation contained in the near-detector response with the compensated response of
the standard “spineand-ribs” density. The resulting density log exhibits a vertical
resolution of better than 6 in., while retaining the essential advantages of dual de
tector compensation. This technique is very different from deconvolution in that no
assumption is made as to the expected bedding features, and no serious degradation in
statistical uncertainty is introduced. Optimum results are obtained with data acquired
at high depth sampling rate (0.1 ft).
The technique has been successfully applied to a number of examples and demon-
strate the potential of significant improvement in evaluating beds of thickness be-
tween 6 in. and 2 ft. Comparisons with other high resolution measurements, such as
the EPT* electromagnetic propagation log and the Formation MlcroScanner* images,
can demonstrate the validity of this technique.

* Mark of Schlumberger -1-


SPWLA Twenty-Eighth Annual Logging Symposium, June 29-July 2, 1987

INTRODUCTION
Vertical resolution is a quantity describing the ability of a logging instrument to read
or resolve responses in sediments of limited vertical extent. We shall not delve into the
details of the absolute definition and quantification of the term. We can say, however,
that for most logging measurements, the vertical resolution is very strongly dependent
on the size of the “active” part of the tool, or on source-to-detector (transmitter-to-
receiver) spacing. In this paper, we shall refer to vertical resolution only in a relative
sense, meaning that tools with longer spacings will exhibit poorer vertical resolution.
To simplify the discussion we will sometimes refer to the “magnitude” of the vertical
resolution as being the same as the corresponding source-to-detector spacing.

The scale of the vertical resolution of existing tools varies from many feet for deep
resistivity or induction measurements, through 1 to 2 ft for most nuclear meaam-
ments, down to between a fraction of an inch and a few inches for microresistivity and
electromagnetic propagation (dielectric) measurements.

In the past, many attempts have been made to enhance the vertical response of log-
ging measurements using deconvolution techniques. l-S Unfortunately, when applied to
typical nuclear measurements, these either suffer from amplification of statistical un-
certainties or rely heavily on assumptions about the expected frequency and shape of
the layers.

One of the ultimate limitations of logging measurements is the correspondence between


the magnitude of the vertical resolution and the magnitude of the depth of investigation.
In general, ‘good” vertical resolution also implies “shallow” depth of investigation,
which leads to stronger susceptibility to borehole and mudcake effects.

The discussion of vertical resolution is further complicated when dealing with mul-
tisensorydevices. Here, we combine measurements having different vertical response
functions as well as different depths of investigation, usually in order to compensate
for borehole or mudcake effects. This can have some unexpected implications on the
vertical response of the combination, ss will be discussed in the following section.

In this paper, we wish to demonstrate a technique to opttilze the way to combine


two (or more) sensors responding to a common source, in order to obtain the best
vertical resolution, and at the same time, maintain the best borehole compensation.
Specifically, we will apply this technique to the Litho-Density* tool (LDT).
A very similar technique has been previously applied to the CNL* compensated neutron
log.4 The application to the density measurement described in this paper is simpler and
has the potential of more dramatic improvement in the results.

,.

* Mark of Schlumberger -2-


SPWLA 13venty-EighthAnnual Logging Symposium, June 29-July 2, 1987

OPTIMUM COMBINATION OF TWO SENSORS


----
As mentionedin the introduction,it is a nontrivialtask to find the best way of com-
biningtwo common sourcemeasurementswith differingvolumesof investigation.The
issue becomes crucial if msximum vertical resolution is desired. In particular, since the
effective measure points of the two sensors (peaks in the individual vertical response
functions) may not be at the same depth, it is important to consider the appropriate
depth correction required for an optimum match.
We shall start by schematically describing the present state of the art.
Figs. la-c demonstrate the different ways to combine the measurements in terms of
the geometry of the volumes of investigation. On the left are shown the volumes of
investigation of the two sensors. The difference between these two can be taken to be
the effective volume of investigation of the result. The right side shows the typical
vertical response function of the combination.

The first and simplest method is to take the two sensor readings simultaneously and
combine them without any depth correction (i.e., assuming that both readings come
simultaneously from the same depth) or averaging. Any further processing is performed
on the combination only. This yields a combination of two sensor readings from the
same source (transmitter) position, as illustrated in Fig. la. A borehole compensation
is made by “subtracting” the effect of the volume of investigation of the near sensor M
... from the reading of the far sensor. As can be seen from the figure, the volume of
investigation of the result, and thus the vertical resolution, is dominated by the zone
between the near and far sensors. For this configuration, the measure point (effective
reading position) is somewhere between the two detectors. Unfortunately, if the depths
of investigation of the two sensors are significantly different, the volume of investigateion
of the combination can extend all the way back to the source. Depending on the relative
weighting of the two signals, this can yield a complicated vertical response function, as
shown on the right of the figure. A main lobe, whose width is similar in magnitude to the
near-to-far detector spacing, and an undershoot below are observed. This asymmetric
response is commonly seen on nuclear, especially neutron, logs. It appears in the form
of unusual fluctuations near bed boundaries, which are repeatable (nonstatistical), but
which are not a true representation of the formation.
For measurements whose depth of investigation does not change dramatically with
spacing, the vertical response function would be much more symmetric. This is true
for acoustic and dielectric measurements, for which this scheme is the optimum.
Another drawback of this method of combination is that the net volume of investigation
contains a large section of the zone near the borehole wall. This makes the result very
susceptible to borehole anomalies psasing between the two detectors, and somewhat
susceptible (in the opposite sense) to anomalies between the near detector and the
,.. source.

-3-
SPWLA Twenty-Eighth Annual Logging Symposium, June 29-July 2, 1987

Another way to combine the two sensor readings is demonstrated in Fig. lb. Here, a m
depth shift is performed on one of the readings to put the two sensors “on depth” before
a combination is made. The resulting “M” shape of the vertical response function is a
characteristic feature resulting from the resolution mismatch of the two sensors. Note
that the total width, or vertical resolution, is now dominated by the distance between
the far detector and source. This is probably the least desirable way to combine the
two readings; nevertheless, it is often used. Not only is the vertical resolution degraded
over the previous case, but the sensitivity to borehole anomalies is enhanced in the
vicinity of the far detector and near the source. Because of the unusual shape of the
vertical response function, nonstatistical fluctuations will also be introduced into the
reading. This is the method currently applied to standard LDT logs.
A third method is similar to the second, but with a crucial additional step. The
response of the near detector is “smoothed,” or averaged, so that the resulting vertical
resolution of the near sensor is the same as that of the far sensor. The vertical response
function in Fig. lC is well behaved and symmetric, and very little of the zone near
the borehole wall is present in the resulting volume of investigation. This is the best
method of combining the two readings in terms of the systematic compensation for
borehole effects. Unlike the two previous cases, this scheme will not introduce unusual
fluctuations in the log. The vertical resolution, however, is apparently worse than in
the tist method, since it is fully controlled by the far detector-to-source spacing.
From the above chcussion it can be concluded that there is an “incompatibility between
optimizing systematic effects and vertical resolution.

DESCRIPTION OF THE METHOD


In the previous section it has been shown that one method of combiniig two sen-
sors has better systematic behavior, while another exhibits better vertical resolution.
This apparent paradox can be resolved if certain requirements are met and if certain
assumptions can be accepted as being approximately valid.
We have neglected to mention in the previous section that the near detector by itself
has a well behaved and symmetric vertical response function. It exhibits a vertical
resolution that is dominated by its distance from the source. For many devices, this
could be even smaller than the near-to-far spacing that dominates the resolution of the
‘instantaneous” method. This is true for most dual detector gamma-gamma density
tools. However, the near detector suffers dramatically from borehole effects. On the one
hand, the “depth and resolution matched” method hss the best borehole compensation
systematic; on the other hand, the near detector by itself has the best vertical response
function. A combination of the two could yield a result that has the beat features of
both.
Figs. 2~d demonstrate a very straightforward method for generating such a combi-
nation. Fig. 2a shows the true density of a hypothetical %hin~ bed compared to the ..

-4-
SPWLA Twenty-Eighth Annual Logging Symposium, June 29-July 2, 1987

response of a standard compensated density (as obtained, for example, using the “depth
.!%
and resolution matched” method). In Fig. 2b the standard density is replaced by that
computed from the near detector. It has a much sharper response to the bed bound-
aries compounded by a systematic offset; presumably because of environmental effects
on the uncompensated single-detector response. The first step of this technique is to
filter, or average, the near-detector response to match the vertical resolution of the
compensated density. Fig. 2C shows the comparison between the unaveraged and the
averaged near-detector density. In addition, the actual compensated density is repeated
here (dashed line). Schematically, the shaded area between the averaged and unaver-
aged near-detector response contains the high frequency information that is lost in the
compensated density. Thus, application of this difference directly to the compensated
reading results in a curve whose shape reproduces that of the near-detector response,
but whose average value is identically equal to the compensated density. This is shown
in Fig. 2d. The difference between the averaged near-detector response and the com-
pensated response is a measure of the systematic perturbation of the near detector.
The above description can be summarized in one simple equation:

where pen~is the enhanced vertical resolution result, pb is the standard “spine-and-ribs”
result, ~nearis the unaveraged near-detector density, and z is its value averaged to M
-..
match the vertical resolution of ~b. The term in parentheses will average to zero,
ensUriI’kgthat, on average, systematic effects on ~b from pnearwill cancel out.

This scheme is very easy to implement, but like most simplistic solutions, it has some
drawbacks. In order for the technique to be successful, a very important criterion has
to be met; namely, that the vertical behavior of the near detector truly represents the
vertical variation in the density of the downhole formation. Two conditions have to be
satisfied.
First, one must require that the density sensitivity and statistical uncertainty of the
near detector are adequate to obtain the precision needed in the final result. The near
detector of the Lith&Density tool achieves a precision of 0.017 g/cm9 for a l-second
measurement at 2.6 g/ems (this corresponds to precision of .01 g/cm9 for standard
vertical averaging). This is the same precision as that achieved by the compensated
“spine-and-rib~ result. This means that imposing the vertical response of the near
detector on the final result will not degrade its statistical uncertainty.
Second, one must assume that the systematic perturbations on the near-detector re-
sponse (mostly a result of environmental effects), are slowly varying with depth when
compared to the actual density variations. This is to ensure that the variations ob-
served by the near detector are reflections of actual density changes and not caused
.-- by sudden changes in systematic deviations. For the density tool this assumption is

-5-
SPWLA Twenty-Eighth Annual Logging Symposium, June 29-July 2, 1987

valid under most conditions when the borehole is in ‘re~onablyn good shape. The ,...
assumption obviously falls apart in rugose, washed-out hole and in cases of localized
massive buildup of mudcake (especially barite mudcake in fractures).

In spite of these limitations, excellent results have been obtained in a variety of situa-
tions. For the LDT tool, with standard sampling and three-level depth averaging, the
expected vertical resolution of the normal compensated density is about 2 ft. With this
technique, resolution slightly better than 6 in. can be achieved, particularly if higher
sampling rates and slower logging speeds are employed. In the examples presented,
typical depth sampling used was 1.2 in. Logging speed of 900 ft/hr is preferred to
maintain reasonable per sample statistics, but in porous reservoirs (low densities) a
logging speed of 1800 ft/hr is adequate.

FIRST EXAMPLE - SYNTHETIC LOG

In order to illustrate the effect of this technique and to support the conclusions of
the discussion on different ways of combining two sensor readings, a synthetic log,
exhibiting beds of varying thickness with sharp bed boundaries, was generated. The
hypothetical tool used in this reconstruction had far detector-to-source spacing twice as
large as that for the near detector. The top bed in the example waa the same thickness
as the near-to-source spacing.

The synthetic formation’s ideal response was inverted to yield expected near- and
far-detector readings with realistic individual vertical response functions. The two
detector responses were then combined in the various ways, described previously, to
yield simulated log responses. These were then compared to the initial ideal response.
In addition, the proposed technique was applied to the data and the resulting “enhanced
vertical resolution” response was compared to the ideal response. The results are shown
in Fig. 3. The near- and far-detector responses are plotted on the left and presented
“on depth.” The result of using the instantaneous, or “off depth,” combination is shown
immediately to the right of the count rate curves. This is followed, in order from left
to right, by the depth-matched result, the resolution-matched result, and, finally, the
result of the proposed technique. All the results are overlaid with the initial formation
model.

The results confirm the behavior predicted in Figs. la-c and demonstrate that the
proposed method yields the best vertical resolution and a symmetric response.

SECOND EXAMPLE - LAYERED TEST FORMATIONS

The tec~lque was &st applied to LDT logs acquired in a layered test well in Houston.
Fig. 4 shows the results of standard processing, using both the normal 6-in. sampling
rate and a high sampling rate. No averaging was performed on the data. The known
layer densities are also shown in the figure. It is apparent that the vertical resolution

-6-
SPWLA Twenty-Eighth Annual Logging Symposium, June 29-July 2, 1987

&-- in both cases is somewhat worse than 1 ft (examine the triplet of l-ft beds near the
top) and that higher sampling provides only minimal improvement. Fig. 5 shows a
comparison of the near-detector counting rates (raw and averaged to match far-detector
resolution) to the far-detector counting rate. As expected, the character of the averaged
near-detector response is very similar to that of the far detector, whereas the raw
near-detector response exhibits much more detail and much sharper boundaries. It is
the difference between the unaveraged and averaged near responses that contains the
required high frequency information.

Fig. 6 compares the result of the proposed method compared to the standard processing,
using high sampling rate. A dramatic improvement can be seen. The vertical resolution
hss been improved by at least a factor of two. The rise time of the response at bed
boundaries is now somewhat better than 6 in.

THIRD EXAMPLE

The next example comes from a commercial well. The interval contains beds exhibiting
large density contrasts. The data were acquired with a sampling rate of 1.2 in.
Fig. 7 shows the counting rate comparison, similar to that from the previous example
(Fig. 5). It is again apparent that the averaged near-detector response closely resembles
the shape of the far-detector response, whereas the raw response exhibits considerably
M
,.“.. greater detail. Fig. 8 shows the density comparison. The improved bedding definition
indicated by the enhanced resolution curve has subsequently been confirmed by whole
core examination. Note, also, the difference between the resolution-matched correct ion
and the original one. This is an indication of the nonstatistical fluctuations falsely
introduced into the standard processing result. The quality curve in the first track will
be discussed later.

FOURTH EXAMPLE
The next example comes from a carbonate sequence exhibiting thin, porous beds. The
comparison between the standard density and the enhanced resolution density in one
such bed is shown in Fig. 9. EPT and Formation MicroScanner (FMS) data were also
available and are shown on the figure. The Formation MicroScanner log provides a
detailed microresistivity image of the borehole wall.

The density curves are also presented in a “variable density (VDL)” format in order
to make a direct visual comparison with the FMS image. Dark areas in the density
displays indicate low density, or high porosity, while on the FMS image they indicate
high conductivity; i.e., high porosity. The bed is about 4 in. thick. The standard density
curve exhibits the characteristic “Mn shape around the bed, which is also mirrored in
the VDL display as two darker areas above and below the bed. The enhanced resolution
. result shows a clear bed and the VDL display haa a remarkable resemblance to the

-7-
SPWLA Twenty-Eighth Annual Logging Symposium, June 29-July 2, 1987

Formation MicroScanner image. The EPT curves further confirm the presence of the .
bed.

FIFTH EXAMPLE
The last example comes from a sand-shale sequence. The shales have an unusual
feature; they are laminated with distinct sideriterich layers, varying in thickness from
2 to 10 in. These siderite streaks are very correlatable between wells as far m 800 ft
apart. A considerable amount of core data has been collected for other wells in the field,
although none was available in this well. LDT and EPT logs and acoustic borehole
image data were acquired.

Fig. 10 shows a comparison of the standard bulk density, enhanced resolution density,
photoelectric factor (P~), and electromagnetic propagation time (tPl). The LDT data
were recorded with 1.2-in. sampling. Again, in this example, the new density curve
brings considerably more detail, which is well correlated with the tpl curve.

It is well known that the vertical resolution of the Pe measurement is only a few
inches because of the poor penetrating power of low-energy gamma rays. Formation
evaluation based on combining the Pe data with density data would suffer greatly if
standard density processing were used, as the resolution mismatch between the two
curves could cause erroneous results. On the basis of detailed core information from
this field, a reliable, quantitative correlation between Peand siderite content has been
derived. The correlation was used to determine the siderite content of the beds in this
example, which, in turn, was used to compute the expected density of the bed. Fig. 11
shows in detail a section of the interval in Fig 10. The borehole image is shown in
Track 2. Track 4 contains the comparison of the two density curves. In addition, the
core correlation-derived densities are indicated by the open squares. The agreement in
beds thicker than 4 in. has been found to be excellent.

QUALITY INDICATOR

As mentioned at the beginning, this technique has some drawbacks. In particular, it is


bound to fail in cases where the environmental effects vary aa quickly as formation char-
acteristics, as a function of depth. This often occurs in rugose boreholes. Fortunately,
by taking the difference between the density obtained from the averaged near-detector
response and the compensated “spine-and-rib” density, one obtains a direct measure
of the magnitude of the variation of the environmental effects. The difference (it is
closely analogous to the correction curve) can be used to judge whether the required
conditions are satisfied and, thus, whether the log quality is acceptable.

Fig. 12 shows a comparison of the enhanced resolution density to the standard result.
Track 1 contains, in addition to the gamma ray and caliper curves, the above mentioned
difference, renamed as the “quality factor.” Zone A shows a dramatic difference between -

-8-
SPWLA Twenty-Eighth Annual Logging Symposium, June 29-July 2, 1987

,- the two density curves. The standard curve exhibits the characteristic “M” shape,
which indicates the bed is thinner than 1 ft. Zone B contains similar differences. The
quality factor in Zone A, however, shows very little activity, in contrast to that in Zone
B. This indicates that the result in Zone A can be trusted, while that in Zone B is in
doubt (in places indicated by the shading on the quality curve). Referring to Fig. 8,
very little activity is seen on the quality factor in this interval.

CONCLUSION

A technique, by which the vertical resolution of the dual detector compensated density
log can be considerably improved, has been presented. The importance of understand-
ing the consequences of various schemes of combining two seqsor measurements has
been explained. The necessary conditions and limitations encountered have been de-
scribed. In addition, a method for judging the quality of the result was proposed.
All this can be achieved without any changes to the LDT hardware or to the existing
data acquisition capability. On the other hand, this knowledge can be useful to guide
development of new tools. As previously mentioned, this technique can be applied to
any multisensorytool, within the limitations and requirements described.
To summarize the impact of this technique, density neutron overlays acquired in the
layered test well are presented in Figs. 13 and 14. Fig. 13 shows the overlay resulting
from the standard processing method, while Fig. 14 shows the enhanced resolution M
.
results for both the density and the neutrord measurements. The l-ft marble and
dolomite layers in the top half of the figure are known to have near zero porosity. The
density-neutron crossplot porosities (from standard chartbook charts) seen in these
beds are reduced from about 7 pu, using standard processing, to about 1 pu, with the
proposed method.
With the implementation of this technique, a more accurate evaluation of layered reser-
voirs will be possible.

ACKNOWLEDGEMENTS
The authors thank the oil companies who have helped in this work, not only by giving
permission to use data from their wells but also by providing supporting informa-
tion, thereby increasing the confidence in our results. We also thank Don McCall and
Charles Towsley for their significant contributions. We are grateful to several people
at Schlumberger-Doll Research for preparing the supporting evidence for one of the
examples.

-9-
SPWLA Twenty-Eighth Annual Logging Symposium, June 29-July 2, 1987

REFERENCES
1. Lyle, W. D. and Williams, D. M.: “Deconvolution of Well Log Data, an Innovations
Approach,” paper O presented at the 1986 SPWLA Annual Logging Symposium,
Houston.
2. Czubek, J. A.: “Quantitative of Gamma-Ray Logs in the Presence
Interpretation
of Random Noise,” paper KKK presented at the 1986 SP WLA Annual Logging
Symposium, Houston.
3. Dyes, C. J.: “Inversion of Well Log Data by the Method of Maximum Entropy,”
paper H presented at the 1986 European Formation Evaluation Symposium, Ab-
erdeen.

There are mang more references on the above subject. The ones listed above are
$ome of the most recent.

4. Galford, J. E., Flaum, C., Gilchrist, W. A., and Duckett, S. W.: ‘Enhanced Res-
olution Processing of Compensated Neutron Logs,” paper SPE 15441 presented at
the 1986 SPE Annual Technical Conference and ExMbltion, New Orleans.

-1o-
ABOUT THE AUTHORS

C. Flaum J. E. Galford A. Hastings

C. FLAUM received a BSc degree from McGill University in Montreal and a PhD in
nuclear physics from the University of Rochester. He joined Schlumberger in 1977 as a
field engineer. Mr. Flaum has been engaged in research, development, and applications
of nuclear measurements at various locations including Schlumberger-Doll Research M
and the engineering center in Clamart, France. He is presently working on nuclear tool
interpretation in the Nuclear Department at Houston Headquarters.

J. E. GALFORD received a BS degree in physics from West Virginia University in 1974


and joined Schlumberger as a field engineer in 1975. Before transferring to his present
assignment in the Nuclear Department at Houston Headquarters, he was applications
development engineer for the Northeast Division in Columbus, Ohio. He was involved
in the development of the Dual Water Global* interpretation software. More recently,
he has been involved with the development and implementation of environmental cor-
rections for compensated neutron logs.

A. HASTINGS graduated in mechanical engineering from Lanchester Polytechnic,


Great Britain in 1971. Following graduation, he joined Schlumberger as a field en-
gineer in Spain and Italy. In 1974 he became the manager of the Engineer Training
Center in Livingstone, Scotland. He was manager of field operations in Qatar from
1976 to 1979. Before transferring to his current position as coordinator of interpre-
tation case studies at Schlumberger-Doll Research in 1985, Mr. Hastings held various
management positions in data processing and interpretation development in Paris and
Latin America.

* Mark of Schlumberger - ll-


SPWLA Xventy-Eighth Annual Logging Symposium, June 29-July 2, 1987

VOLUMES OF VERTICAL RESPONSE


INVESTIGATION FUNCTIONS

N lnstantaneous

S
Depth Matched

3
F

Depth Matched
N and
Resolution Matched

Fig. 1 - Methods of combining measurements from two sensors


with different vertical resolutions.

- 12 -
SPWLA Twenty-Eighth Annual Logging Symposium, June 29-July 2, 1987

True p
/

a)

P;ear
/

f
Effect of
b) Environment
*/

v
Pnear
/

c)

_4”

d)

Fig. 2- Pictorial representation of the enhanced resolution process for the Lithe-Density
tool.

-13-
SPWLA llventy-Eighth Annual Logging Symposium, June 29-July 2, 1987

,.,,

\\\
P L,/
I

4
5
,/’

-J--
Depth

II
Instantaneous RMy;~t~ Enhanced
Near Detector Combination Matched Resolution
CountingRate Response Response Reaponae Reaponae
. . . . . . . . . . . . . . . -------- -------- -------- ,-------
Far Detector
Counting Rate Ilwe Response

Fig. 3- Synthetic log responses as a result of applying different processing


techniques to signals from two detectors and a common source.

-14-
SPWLA Twenty-Eighth Annual Logging Symposium, June 29-July 2, 1987

.,.., :ement

Lime

Warble

Sand

‘olomite

Lime

Sand

Marble

Lime

Iolomite

rft

Caliper
L Density Correction
. ...<.., ... .. .......... . ...
1 in. 16
-0.4 glcms 0.1
Standard 6-in. Bulk Density
2.0 g/cm3 3.C
Standard l-in. BuIIKDensity
.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ,
2.0 g/cm3 3.C

------- ------Actual Bulk


------ Density------ ---
2.0 g/cm9 3.C

.. Fig. 4- Comparison of standard processing LDT bulk density for data sampled
at l-in. and 6-in. intervals in the layered test well.

-15-
SPWLA Twenty-Eighth Annual Logging Symposium, June 29-July 2, 1987

ement

Lime

Sand

)Iomitt

Lime

Sand

.\
Aarble

Lime

.
domit(

,,

r
ft

L Caliper Raw Near Detector Counts


) in. 16 log CPS
Averaged Near Detector Counts
.. ..... ..... .... ... ... .......... ... ... ... . ... ... ... .. ..
log CPS
. . . . . . . . . . . . .Far
. . .Detector
. . . . . . .Counts
..................
log CPS

Fig. 5- Comparison of the detector counting rates, sampled at l-in. intervals,


in the layered test well.

-16-
SPWLA ‘Rventy-Eighth Annual Logging Symposium, June 29-July 2, 1987

-;.
‘j

~
)

yement ~

,:

. ..
~. ,
,.-
+

/
~
Lime -1

$

ZziT .:
,.
,-
Sand . ,
.(
-(- —
olomite ,.,
‘ --.,
, .-
,‘
<
Lime
j
,..
$.

~,
L
Sand 4;

.:
:>
Uarble :.

Lime /

‘ -..
1,
.,-
./
..
M
,

>
...
,$
.. .
.. . -:.
d
;r-
1
. .’

r
1-
} ?

;
,
.. ...
.. ft ..
,.,
...

Caliper
..

L Density Correction
-i-
<’

;. -

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


in. It -0.4 g/cm3 o.
Quality Factor Enhanced Resolution Bulk Density
.. ... ..... .......... . ...
5 0.! 2.0 g/cm3 3.(
Normal Bulk Density
.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ----
2.0 g/cm3 - 3.1
Actual Bulk Density
------- ------ ------ ------ --,
2.0 g/cm$ 3.(

,... Fig, 6- Comparison of enhanced resolution LDT bulk density and standard
bulk density for data sampled at l-in. intervals in the layered test
well.
-17-
SFWL.A‘Rventy-Eighth Annual Logging Symposium, June 29-July 2, 1987

I
I
I

I
I
I
)

1-
1

I
1

..~-s-- ---

;
--- =-~.~.~-.-=_ -
1’
I
I
I

-..= .=-=.=-

ft

/ 1

Gamma Ray Raw Near Detector Counts


o API Units 200 log CPS
caliper Averaged Near Detector Counts
---- ---- ---- --! . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6 in. 16 log CPS
. . . . . . . . . . . . . .Far
. . .Detector
. . . . . . .Counts
..................
log CPS

Fig. 7- Comparison of the detector counting rates. sampled at 1.2-in. inter-


vals, from Example 3.

-18-
SPWLA Ttventy-Eighth Annual Logging Symposium, June 29-July 2, 1987

M
,.-,

r
ft

Resolution Matched
Gamma Ray Density Correction
-------- -----.
D API Units 20C -0.35 g/ems 0.15

-------
Caliper
------. I............................
Normal
Density Correction
6 in. Ie
I-0.35 g/cm3 0.15
Quality Factor Enhanced Resolution Bulk Densitv
.. . .. . . . . .. .. . . .. . . . . .. . .. .
.0.5 0.6 2.0 gfcmg 3.0
Normal Bulk Density
2.0 g/cmg 3.0

Fig. 8- Comparison of enhanced resolution LDT bulk density and standard


,,.- bulk density for data sampled at 1.2-in. intervals from Example 3.

-19-
I


✎✞
561

1
\
\ .-------.------

..........-
1’
S62
/’
\\
\\
\
;
#;#
“w’
,,”>

I
4‘ ~

Gamma Ray
Ii

Enhanced Resolution Bulk Density


L Jf&i---
) API Units 150 Lo g/cm9 3.0 Formation MicroScannerBorehole Image

)
caliper---
--------
in. 16

. . . .Attenuation
)
-------
!.0
Standard 1.2-in. Bulk Density
-------
g/cm3
-------

Electromagnetic Propagation Time


-
3.0

. .. . . . . . . .. . . . . . 0 .. .......... . ......... .......... .......... ..


dB/m 500 ~5 na/m 5
~1~ Enhanced Resolution Bulk Density Variable Density Display
Standard Bulk Density Variable Density Display

Fig. 9- Comparison of enhanced resolution LDT bulk density and standard bulk density with EPT log and Formation MicroScanner
borehole image from Example 4.
SPWLATwenty-Eighth
AnnualLoggingSymposium,June29-July2, 1987


I ‘—
I

I I
I Enhanced Electromagnetic !
Resolution
Bulk Densitv I ‘PW&$o’-” I
L%Z9:G!?W::

Fig. 10- Correlation of standard 1.2-in. bulk density and enhanced resolution
bulk density with electromagnetic propagation time in a sequence of
thin beds from Example 5.

-21-
SPWLA Twenty-Eighth Annual Logging Symposium, June 29-July 2, 1987

1
I
I
I
-3kG
w-
I

i
1
I —
I
I
I
I
I --
— —
I
I
I
I
I
I
I
I
I
I
I
1
I
1

i
i
I
I
I
I
I
I
I
I
I
/
i
I
I
I
I
I
I
I
1
I
I

Enhanced
Pho#=eJ~rtric Resolution
Gamma Ray Bulk Density
O API Unitsi 200 o 10 2.2 glcmg 3.2
Borehole Image
Standard 1.2-in.
Caliper Bulk Density
------ ---- ------- ---
6 in. 16 2.2 g/cm9 3.2

Fig. 11- Comparison of standard 1.2-in. bulk density and enhanced resolution
bulk density with borehole image and core correlation derived densities
from Example 5.

-22-
SPWLA Twenty-Eighth Annual Logging Symposium, June 29-July 2, 1987

...., —
560

570

B
b

580

Gamma Ray Density Correction


. .. .. ... ..... ..... ... .. ....
) API Units 150
-0.35 g/cm3 0.15
Caliper Enhanced Resolution Bulk Density
------- ------ .
1 in. 18 2.0 g/cm3 3.0
Quality Factor
. . ... ... ... . ... ... .... ... . Normal Bulk Density
.........................................
0.5 0.5 Z.o g/cm3 3.0

Fig. 12- Using the quality factor to judge the validity of the enhanced resolution
result.

,.

-23-
SPWLA Twenty-Eighth Annual Logging Symposium, June 29-July 2, 1987

——

Cement

——

5 ft

Lime

.
Marble

Sand
=
Dolomite

Lime

——
Sand

?
Marble ) .
.
.

Lime

Dolomite

— —
caliper Density Correction
A . . Photoelectric
. . . . . . . . . . .Factor
. . . . . . . ,.. . . . . . . . . . . . . . . . . . . . . . . . .
I in. 16 ) 10 -0.4 g/ems 0.1
Normal Density Porosity
)0 Limestone Porosity Units -lo

------- Normal Neutron


------ ------ Porosity
------ ---
10 Limestone Porosity Units -lo

Fig. 13- Normal neutron-density overlay.

-24-
SPWLA Twenty-Eighth Annual Logging Symposium, June 29-July 2, 1987


1 1
Cement

U
s it

Lime

Marble

Sand
I

= —
Dolomite

Lime

— —
Sand

~
Marble

Lime

Dolomite

— —

caliper
6 in. 16
.Photoelectric Factor
. . . . . ..- --------- I Density Correction
. .. . ... ... ... ........... ... .
10 I.04 g/cm3 0.1

Enhanced Resolution Density Porosity


Limestone Porosity Units -1(
Enhanced Resolution Neutron Porosity
------- ------ ------ ------ -.
Limestone Porosity Units -1(

Fig. 14- Enhanced resolution neutron-density overlay.

-25-

También podría gustarte