SPWLA 47th Annual Logging Symposium, June 4-7, 2006

CASE HISTORY OF AUTOMATED EVALUATION OF MINERALOGY

AND POROSITY IN COMPLEX CARBONATES

N. Gomaa, ADMA OPCO

M. Herron, R. Ramamoorthy, P. Tilke, D. Allen, Schlumberger

Copyright 2006, held jointly by the Society of Petrophysicists and Well Log INTRODUCTION
Analysts (SPWLA) and the submitting authors.
th
This paper was prepared for presentation at the SPWLA 47 Annual Logging
Symposium held in Veracruz, Mexico, June 4-7, 2006. Middle East carbonates contain much of the world’s
petroleum reserves. They have been analyzed in the
past with conventional well logs as being essentially
ABSTRACT mixtures of calcite and dolomite with sporadic levels of
pure anhydrite. With additional mineralogical analysis
Significant oil and gas reserves in the U.A.E., Qatar, in the past decade, it is becoming clear that this is an
and elsewhere occur in carbonate formations containing oversimplification of the mineral composition. JJJ
anhydrite and quartz disseminated within calcite and Disseminated anhydrite and sporadic quartz/chert layers
dolomite reservoirs. Accurate evaluation of mineralogy have been found to be important in many Middle East
in these complex carbonates, while critical to wells. The underestimation of anhydrite by
computing porosity, hydrocarbon density and well-to- conventional logs has a significant effect in
well correlation, is challenging when a conventional underestimating reservoir quality.
logging suite is used. The number of unknowns in the
formation exceeds the number of available independent
measurements. MINERAL ASSEMBLAGES IN MIDDLE EAST
CARBONATES
Mineralogy evaluation of a complex carbonate in the
well studied was greatly improved when nuclear In the past several years we have analyzed hundreds of
spectroscopy logs were incorporated into the carbonate samples from the Middle East by Dual Range
evaluation. These logs measure calcium, sulfur and Fourier Transform Infrared spectroscopy (Herron et al.,
silicon which directly map to the key mineralogical 1997) for quantitative mineralogy. Each sample was
components – carbonates, sulfates (anhydrite) and also analyzed for major element chemical content by x-
quartz/chert. The resulting evaluation was far more ray fluorescence to provide an independent means of
accurate when compared to mineralogy measured on validating the mineralogy. In Table 1, we summarize
core samples from the same well. the mineralogy from 495 samples from 12 wells from
several oilfields in the Middle East.
Many such carbonate reservoirs have formation waters
with salinity in excess of 200,000 ppm. Drilling fluids Table 1. Average and maximum observed abundance
used to drill the well also have high salinity. In order of minerals in 495 samples from Middle East carbonate
to reduce the environmental effects on the neutron reservoirs.
porosity log, an epithermal neutron porosity tool was Mineral Average Maximum
run in the subject well. We demonstrate through Abundance (wt. Abundance (wt.
comparison to core data the improvement in the %) %)
accuracy of porosity evaluation through the use of Calcite 54 100
epithermal neutron data. Dolomite 36 100
Anhydrite 7 100
We show that the combination of nuclear spectroscopy Quartz 0.7 81
and epithermal neutron porosity improves both the Illite 0.4 36
accuracy and the precision of porosity and mineralogy Pyrite 0.3 24
evaluation. Detailed uncertainty analysis further
substantiates the accuracy and precision improvement
in lithology and porosity through the use of these The dominant minerals are calcite and dolomite, but
measurements. there is a surprisingly high concentration of anhydrite,
confirmed by chemical analysis, in these samples. We

1
SPWLA 47th Annual Logging Symposium, June 4-7, 2006
have found that the anhydrite is present in two forms, as
Figure 1. Anhydrite nodule in a Middle East core.
Nodules (Fig. 1) and as disseminated anhydrite(Fig. 2).
The nodules are relatively easy to identify visually by
the white patch. The disseminated anhydrite is much
more difficult to visually identify and this is probably
the reason it has been underestimated in the past.
Figure 2. Disseminated anhydrite in a Middle East
carbonate core. The light colored areas are the
anhydrite which is not in a nodular form.
The disseminated anhydrite is much more subtle in
visual presentation, and so has probably been under
2
represented in core descriptions. What is the impact of
misidentifying small concentrations of anhydrite?
Figure 3. Two computer processed interpretations of
the same depth interval. The processing on the left was
from conventional logs, and does not recognize the
disseminated anhydrite in the porous interval. The
interpretation on the right includes spectroscopy logs
with the direct measurement of sulfur, allowing
identification of the disseminated anhydrite. The
increase in porosity from inclusion of the anhydrite is
shown as the shaded porosity;
ELEMENTAL CAPTURE SPECTROSCOPY
LOGS
Accurate anhydrite estimation from logs requires the
use of a measure of the sulfur concentration of the
formation. This is provided by capture spectroscopy
logs which measure the relative yields of several
elements (Grau and Schweitzer, 1989) which also affect
the nuclear log response such as iron, silicon, calcium,
sulfur, titanium, gadolinium, barium, hydrogen and
chlorine, to name just a few of the elements. The
measure of barium permits the accurate correction for
the sulfur contribution of barite which is often present

in the mud systems. Hydrogen and chlorine helps us to
remove the effect of the borehole and pore fluids. The
algorithm used to convert the elemental yields to
elemental dry weights and dry mineral fractions is
known as SpectroLith (Herron and Herron, 1996)
Figure 3 shows two computer processed formation
evaluations, one with only conventional logs and one
including spectroscopy logs, which include a sulfur and
silicon analysis and an epithermal neutron log. The
inclusion of the sulfur and silicon logs enables an
accurate detection of the anhydrite and quartz/chert
concentrations. The quartz/chert zone is useful for
chronostratigraphic horizon mapping across the
reservoir as it likely represents a dust storm event . The
anhydrite zone is more important from a petrophysical
perspective. Anhydrite itself is rarely porous. Ite
presence increases the formation matrix density. The
increased matrix density due to the anhydrite means
that the original interpretation is too low in porosity by
approximately 10 relative percent. Thus reserves have
been underestimated by about 10 percent.
additional computed porosity is shaded orange. The
two advantages to the interpretation on the right of
Figure 3 are the inclusion of spectroscopy and the use
of the epithermal neutron log.
WHICH NEUTRON LOG?
There are two main classes of neutron log – thermal and
epithermal. Neutrons are emitted from these devices at
relatively high energy and are gradually slowed down,
primarily by collisions with hydrogen nuclei (Ellis,
1985). Eventually they are slowed to an epithermal
energy (0.2-10 eV) or to thermal energy (~ 0.025 eV).
Epithermal neutron tools are not affected by thermal
absorbers, elements with a sizable thermal neutron
capture cross section which remove thermal neutrons
and thus produce a too high porosity. Kamal et.al.
(2004) reported significant boron concentrations in
Khuff reservoirs. We have measured Middle East
carbonate samples containing as much as 100 ppm
boron which is equivalent to an overestimation of 10
porosity units by a thermal neutron log.
Epithermal neutron devices are also subject to much
less environmental correction due to borehole
conditions. Environmental corrections need to be
applied on thermal neutron measurements for bore hole
size, mud weight and type, borehole salinity, mud cake,
standoff, temperature and pressure and formation
capture cross-section. Epithermal corrections are only
required for hole size, mud weight, temperature and
pressure and standoff most of which are either known
SPWLA 47th Annual Logging Symposium, June 4-7, 2006
or measured downhole. Hence, reliable corrections are
available at the wellsite.
JJJ
Figure 4. Track 1 shows a computer processed
The interpretation. Track 2 shows raw and corrected
thermal neutron porosity. Track 3 shows the raw and
corrected epithermal neutron porosity. The correction
on the thermal is a factor of three higher than the
epithermal.
Another advantage of the epithermal neutron
measurement is close conformance of the depth of
investigation (DOI) to that of the density log. Both
sensors are consonant (Gossenberg et.al.; 1996). This
results in very robust and reliable elimination of light
hydrocarbon effects when these are present on the logs.
The thermal neutron, on the other hand, has a very deep
DOI – 12 inches to 17 inches – compared to the DOI of
the density log – 2 inches. This makes the hydrocarbon
correction unreliable in the presence of shallow
invasion affecting only the density log.
Figure 4 shows the computer processed interpretation
for an interval of a Middle East carbonate well. On the
right are the raw and environmentally corrected thermal
neutron porosity logs in blue and the raw and
environmentally corrected epithermal neutron logs in
green. The correction to the thermal log amounts to six
porosity units. The correction to the epithermal neutron
log is less than two porosity units. The reduced
environmental effects and freedom from influence of
thermal neutron absorbing elements make the
epithermal neutron log superior to thermal neutron
devices.
3
SPWLA 47th Annual Logging Symposium, June 4-7, 2006
AN AUTOMATED WORKFLOW
The existence of the spectroscopy-derived elemental
concentration logs, coupled with the enhanced accuracy
of the epithermal neutron log permit development of an
automated workflow to compute mineralogy and
accurate porosity in these Middle East carbonate
reservoirs. In this workflow, spectroscopy data is used
to define the silicon, calcium, iron, sulfur, gadolinium,
and titanium concentrations. These concentrations are
used to make an initial estimate of carbonate, clay,
anhydrite and quartz-feldspar-mica. Then, the
photoelectric effect and/or epithermal neutron and
density are introduced to split carbonate into calcite and
dolomite. This is a split that is based on forward
models of the calcite, dolomite, quartz and clay mineral
response from the Schlumberger Nuclear Parameter
(SNUPAR) modeling code (McKeon and Scott, 1998).
The computation of mineralogy then loops to address
all the log responses to make sure of complete
compatibility. Finally, from the final mineral
assemblage the matrix density is calculated. Thus an
extremely accurate porosity can be readily computed.
This automated workflow ensures that the answers have
minimal user dependence. The workflow also permits
us to model the uncertainty in the final answer based on
the uncertainties of each of the inputs as we show in the
next section. The flowchart of the automated workflow
is shown in figure 10 at the end of the paper.
UNCERTAINTY ANALYSIS – WHAT IS THE
VALUE OF ADDITIONAL LOGS?
To compute the value of the additional logs to the
interpretation of mineralogy and porosity, we employ
an uncertainty analysis. Each of the inputs to the
automated workflow interpretation in the previous
section has published uncertainties. Using those values
and Monte Carlo techniques, the measurement
uncertainties have been propagated through the
interpretation procedure to produce uncertainties in the
mineralogy and porosity. In this way, we can evaluate
how the addition of each new measurement influences
the uncertainty of the answers. This gives a means of
evaluating the sensitivity of each measurement as well
as the value reflected by a reduction in answer
4
uncertainty. In addition, since the true mineralogy and
porosity is known, the accuracy of a given
interpretation can be independently determined.
Published examples of uncertainty quantification in
petrophysical interpretation use either analytical
uncertainty propagation or a Monte Carlo method.
There are advantages to using Monte Carlo techniques
as opposed to an analytical approach for this analysis.
Analytical techniques are fast but approximate and are
suitable for normally distributed properties and linear
functions. Monte Carlo techniques are slower in
computation, but more accurate in assessing the answer
uncertainty. In addition, the technique is applicable for
any property distribution, e.g. log-normal, and is
suitable for non-linear as well as linear relationships.
In essence, a given set of inputs is constructed from log
inputs with known measurement uncertainties. The log
inputs were from a carbonate well in the Middle East
which had core porosity and mineralogy available.
This set of inputs is then computed through the
automated petrophysical workflow into a set of output
mineralogy, matrix density and porosity. Then the
Monte Carlo procedure resamples the parameters and
uncertainties for new sets of inputs to be converted to
mineralogy and porosity and in this study, the
resampling continued until more than 100 cases were
established. The net result is a set of mineralogy
answers and porosity computations. The process was
begun with just triple-combo logs of thermal neutron,
density, gamma ray, and then separate computations
were made after adding spectroscopy and replacing the
thermal neutron porosity with the epithermal neutron
porosity.
Figure 5 shows the results for the analysis with only
triple-combo logs; Figure 6 has the results when
spectroscopy and epithermal neutron are included. The
three tracks shown are porosity, limestone and
anhydrite. The red line shows the computed answers,
and the gray shading around the red line shows the
computed uncertainty. The core mineralogy is shown
as filled pink circles. We consider these results in
terms of accuracy and precision.
 SPWLA 47th Annual Logging Symposium, June 4-7, 2006
Porosity V Lim estone V Anhydrite
0
10
20
30
40
50
0.4 0.3 0.2 0.1 0 0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3 0.4
Figure 5. Triple-combo only results for porosity,
limestone and anhydrite in the Middle East carbonate
well. The red lines represent the mean answers and the
gray shading represents the uncertainty in the answer.
Core mineralogy is shows as the filled circles.
The lack of accuracy of the triple combo results for
estimating the formation mineralogy is seen from
comparing the red line with the core points. In the
upper zone, the core calcite is consistently 40% higher
than the log computation. Then in the middle of the
zone, the log computation averages about 30% calcite
while the core shows no calcite in sample after sample.
Finally at the bottom of the interval, the log-derived
answers are in better agreement with the core. For the
anhydrite, the computed anhydrite is ranging up to 35%
in the upper zone, which has little anhydrite according
to the core data. Then in the lower zone, which
contains about 15% anhydrite, the computed anhydrite
is zero. So there are significant errors in the accuracy
of the triple combo mineralogy.
Besides accuracy, the Monte Carlo procedure permits a
calculation of the uncertainty in the petrophysical and
mineralogical answers represented by the width of the
gray region surrounding the red logs in Figure 5. The
porosity uncertainty for the triple combo logging suite
is fairly constant at ± 5 porosity units (pu). The
limestone uncertainty is fairly constant at ± 20% calcite
over almost the entire depth interval. The anhydrite
uncertainty is as much as ± 20% over most of the
interval.
5
INCREASED ACCURACY AND REDUCED
UNCERTAINTY WITH SPECTROSCOPY AND
EPITHERMAL NEUTRON LOGS
By adding calcium and sulfur elemental data, the calcite
accuracy is significantly improved. In Figure 6 the
limestone log is significantly closer to the core
measurements than in the case of triple combo only
logs. The same is true for the anhydrite concentration,
which now is low when the core analysis is low and
moderately high in concentration when the core data
show significant anhydrite.
The improvement in petrophysical accuracy and
reduced uncertainty from adding spectroscopy and
epithermal neutron porosity logs is seen in Figure 6.
Porosity V Lim estone V Anhydrite
0 JJJ
10
20
30
40
50
0.4 0.3 0.2 0.1 0 0 0.2 0.4 0.6 0.8 1 0 0.1 0.2 0.3 0.4
Figure 6. Adding spectroscopy and epithermal neutron
porosity greatly increases the accuracy of the porosity,
limestone and decreases the uncertainty of the
estimates. Note the much better agreement between log
limestone and anhydrite with core compared to the
results in Figure 5.
The addition of spectroscopy and epithermal neutron
porosity data also greatly reduced the uncertainty of the
mineral analyses as shown in the much smaller gray
areas around the red mean concentration lines.
IMPACT OF SPECTROSCOPY AND
EPITHERMAL NEUTRON LOGS ON POROSITY
There are many more core porosity points than
mineralogy points and the agreement between log and
core can be assessed through a histogram of the
 SPWLA 47th Annual Logging Symposium, June 4-7, 2006

difference between log and core porosity. Figure 7 neutron are included and core porosity. There is no
shows the histogram of the difference in porosity for bias and the uncertainty of the estimation as estimated
the triple-combo set of logs. from the full width at half maximum for this
distribution is ± 2 pu.
80

ANALYSIS OF A CORED INTERVAL WITH

70
THE AUTOMATIC POROSITY AND
60 LITHOLOGY WORKFLOW

50 Figure 9 shows the summary petrophysical analysis and

Frequency

40
comparison to core data for a Middle East carbonate
well. The data set included triple-combo logs,
30 spectroscopy and epithermal neutron data. The log data
were processed through the automated workflow
20
described earlier. The first track shows the volumetric
10 solution of calcite, dolomite, anhydrite, quartz and
porosity. The last track shows matrix density computed
0 from this mineral assemblage as a continuous log. Core
-0.20 -0.10 0.00 0.10 0.20
Difference
plug data are shown in black and agree with the
computed matrix density. Also shown as a fairly
Figure 7. Histogram of the difference between the
continuous set of data are the matrix density
porosity computed from triple-combo only logs and
measurements made every foot on the whole core.
core porosity.
Aware of the level of heterogeneity in many Middle
East carbonate reservoirs, this is an excellent level of
The difference histogram shows that the porosity
agreement between the log and core data.
estimate is only slightly biased compared to the core
porosities and that the width of the distribution as
The third track in Figure 9 shows the calcite mineralogy
determined by the full width at half maximum is about
computed from the logs along with core plug data as
± 6 pu. red filled circles. In addition, foot long sections of the
whole core were homogenized to reduce heterogeneity
Figure 8 shows the porosity difference histogram effects and the calcite content of these is shown as
between the porosity computed from logs including green dots. The fourth track shows the same
spectroscopy and epithermal neutron compared to core comparison between computed log dolomite and core
porosity. There is a no bias in the log data and the full plug and core section dolomite. The comparison for
width at half maximum is reduced to about + 2 pu. these major minerals is very good.
This is a significant reduction in the porosity
uncertainty through the use of an optimized log The minor or trace minerals anhydrite and quartz are
acquisition and interpretation process. shown in the fifth and sixth tracks. These constituents
120 are more heterogeneously distributed as shown in
Figures 1 and 2 and so we do not expect as good a core-
100 log comparison as with the major minerals. Although
the individual core data are not in perfect agreement
80
with the log estimations, it is clear that the highest
anhydrite intervals in the log are the same as the highest
F req u en cy

core anhydrite intervals, and the same with the low

60
concentration intervals. As with the anhydrite, the
interval where there is a small sand signature is the
40
same interval where one core sample found 8% chert
and the core description, not shown here, also found
20
chert.

0
-0.20 -0.10 0.00 0.10 0.20
Difference
CONCLUSIONS AND RECOMMENDATIONS

Figure 8. Histogram of the difference between log The addition of capture spectroscopy log data and
derived porosity when spectroscopy and epithermal epithermal neutron porosity data are seen to
6
 SPWLA 47th Annual Logging Symposium, June 4-7, 2006
significantly improve the accuracy of the porosity and
mineralogy estimates in Middle East carbonate
reservoirs. In addition, the additional logs are seen to
reduce the uncertainty of the porosity and mineralogy
estimates. The root cause of these improvements are
that the spectroscopy data such as sulfur and silicon is
directly sensitive to such mineralogical components
such as anhydrite and chert.
The automated workflow used in this analysis (Figure
10) provides nearly instantaneous answers of mineral
fractions and total porosity with minimal user influence
while honoring the chemistry and physics. The original
elemental concentration logs from spectroscopy of
calcium (Ca), silicon (Si), iron (Fe), and sulfur (S) are
processed through the SpectroLith∗ analysis (Herron
and Herron, 1996) to produce total carbonate, clay,
quartz-feldspar-mica (sand) and anhydrite. Additional
information is provided by the photoelectric factor
(PEF) if considered reliable, and/or the combination of
epithermal neutron and density along with SNUPAR
modeling data to split the total carbonate into calcite
and dolomite fractions. The solution of dolomite
requires a slight modification in the elemental
concentration logs which are then reintroduced into the
loop until a convergent solution is found, usually only
two loops. At this point, the mineralogy is finalized
and a matrix density and total porosity is computed.
The mineralogy and total porosity are seen to be
validated by core data.
The uncertainty of the porosity estimate is seen to
decline from ± 5% to ± 2% with the additional data.
The accuracy and uncertainty of the mineralogical data,
and thus the matrix density leading to the porosity
answer, are significantly improved with with the
inclusion of the spectroscopy and epithermal data.
ACKNOWLEDGEMENTS
The authors would like to acknowledge Abu Dhabi
Marine Operating Company (ADMA OPCO) for their
support and permission to publish the data shown in
this paper. We would also like to thank Schlumberger
in Abu Dhabi, Schlumberger-Doll Research in USA
and Schlumberger Dhahran Carbonate Research for
their support for this study.
REFERENCES
Bowers, M.C. and Fitz, D.E., 2000, A probabilistic
approach to determine uncertainty in calculated water
saturation, Transactions of the 41st SPWLA Annual
Logging Symposium, Paper QQ.
∗
Mark of Schlumberger
7
Gossenberg, P., Galli, G., Andreani, M., and Klopf, W.,
1996, “A new petrophysical interpretation model for
clastic rocks based on NMR, epithermal neutron and
electromagnetic logs,” Transactions of the SPWLA 37th
Annual Logging Symposium, New Orleans, LA, USA,
June 16 – 19, Paper M.
Grau, J.A. and Schweitzer, J.S., 1989, “Elemental
concentrations from thermal neutron capture gamma-
ray spectra in geological formations,” Nuclear
Geophysics, Vol. 3, No. 1, pp 1 – 9.
Herron, M.M., Matteson, A. and Gustavson, G., 1997,
Dual-range FT-IR mineralogy and the analysis of
sedimentary formations, Paper SCA-9729 presented at
the 1997 International Symposium of the Society of
Core Analysts, Calgary, Sept. 7-10. JJJ
Herron, S.L. and Herron, M.M., 1996, Quantitative
lithology: An application for open and cased hole
spectroscopy, Transactions of the SPWLA 37th Annual
Logging Symposium, New Orleans, LA, June 16-19,
Paper E.
Kamal, Rami A., Al-Eid, Ghazi, Sellwood, B.W. and
Clerke, Edward A., 2004, “A comparative study of
visually estimated mineral composition, X-ray
diffraction, and trace element determination from cores
across the Khuff-C carbonate in Hawiyah, Ghawar
field, Saudi Arabia,” presented at the 6th Middle East
Geosciences Conference and Exhibition, 7th – 10th
March, 2004, Bahrain.
McKeon, D.C. and Scott, H. D., 1988, SNUPAR – A
nuclear parameter code for nuclear geophysics
applications, Nuclear Geophysics, Vol. 2, pp. 215-230.
Verga, F., Viberti, D and Gonfalini, M., 2002,
Uncertainty evaluation in well logging: Analytical or
numerical approach?, Transactions of the 43rd SPWLA
Annual Logging Symposium.
ABOUT THE AUTHORS
Nasser Gomaa graduated in 1983 with a BSc degree in
geology from Ain Shams University in Cairo. His 22
years of experience includes wireline operations, log
interpretation, reservoir monitoring, log data
management, routine and special core analysis. He
started in Khalda Petroleum Company (was a Conoco
joint venture company) in 1985 as a wellsite geologist
and later as an operational petrophysicist. He joined
Shell Egypt in 1991 as a petrophysicist providing
support to development, exploration and new business
SPWLA 47th Annual Logging Symposium, June 4-7, 2006
departments. In 1997 he was transferred to Shell
International Exploration and Development in The
Hague to work as a petrophysicist in the Integrated
subsurface field study department, and was involved in
many integrated field studies in different countries.
Back in Egypt, he worked with Shell Egypt, Rashid
Petroleum Company ( a BP joint venture company). He
currently works with ADMA-OPCO in Abu Dhabi as a
petrophysicist. He is a member of SPE, SPWLA, SCA
and ESG.
Michael Herron is a scientific advisor in the Geology
and Rock Physics Department at Schlumberger-Doll
Research in Ridgefield, CT. He joined Schlumberger in
1982. He received a B.A. degree in chemistry from the
University of California at San Diego and a Ph.D. in
geology from the State University of New York at
Buffalo.
Raghu Ramamoorthy joined Schlumberger Wireline
Services in 1982 and served as a logging engineer, and
later as a log analyst in the Middle East. Raghu
received his Master of Science in Petroleum
Engineering degree at the University of Texas, Austin
in May 1994. He then spent an extended stint as
Research Scientist (Petrophysicist) at Schlumberger-
Doll Research in Ridgefield, Conn., where he worked
on Reservoir Characterization and Carbonate
Petrophysics. In 1997 he returned to the field and
served as Principal Petrophysicist in Australasia, East
Asia and Saudi Arabia. Raghu is currently
8
Petrophysics Advisor in the Middle East based in Abu
Dhabi.
Peter Tilke earned a Ph.D. degree in geology from
Massachusetts Institute of Technology in 1986.
Following a post-doc at Harvard University and
research and exploration positions with Shell in
Houston and New Orleans, he has held various research
and engineering positions with Schlumberger since
1990. Currently Peter is a Principal Research Scientist
with Schlumberger-Doll Research in Boston. His
principal research interests are in characterizing and
modeling uncertainty in borehole measurements and
petrophysical interpretation, and characterizing and
modeling the heterogeneity of carbonates.
David Allen is a Petrophysics Advisor in the
Schlumberger Sugar Land Product Center. From 1999
to 2005, he lead the effort on carbonate case studies at
Schlumberger-Doll Research. David received a B.S.
degree in physics and a B.A. degree in economics from
Beloit College in 1978 and joined Schlumberger as a
field engineer in 1979. From 1995 to 1997, David was
the chief petrophysicist for Schlumberger Wireline and
Testing. He received Best Paper awards from the
SPWLA for a 1987 paper on invasion and a 1997 paper
on resistivity anisotropy.
 SPWLA 47th Annual Logging Symposium, June 4-7, 2006

CarbECS Grain
Volumes Calcite Dolomite Anhydrite Quartz Density
0 1 0 1 0 10 1 0 1 2.5 3

Anhydrite
visible in core

JJJ

Figure 9. Results of the addition of spectroscopy and epithermal neutron porosity data to standard triple combo data
in the automatic computation of mineralogy and porosity in a Middle East carbonate well. The first track is the
complete volumetric solution. The remaining tracks are the mineralogy and matrix density. In each case, core plug
data are shown as red filled circles and continuous homogenized lengths of core are shown as black connected dots.
For anhydrite and quartz there are also core description data showing when the minerals were observed.

9
SPWLA 47th Annual Logging Symposium, June 4-7, 2006

Fe, Si, Ca, S

SpectroLith*
Iterate

PEF Total
Carbonate
Clay
QFM Fe
N-D
Anhydrite Si
Ca
S
SNUPAR* Modeling

Calcite, Dolomite,
Clay, QFM,
Anhydrite, Siderite,
Results Porosity, Matrix
Density

Figure 10. The automated workflow used in processing these data. The inputs are Fe, Si, Ca, and S elemental
concentration logs from the spectroscopy logs along with photoelectric factor (PEF) and/or epithermal neutron
combined with density. The elemental concentration logs are processed through SpectroLith (Herron and Herron,
1996) to identify total carbonate, clay, quartz-feldspar-mica (sand) and anhydrite. The PEF and/or epithermal
neutron-density logs are chosen and used to split the total carbonate into calcite and dolomite using SNUPAR
(McKeon and Scott, 1998) forward model data. The dolomite is used to refine the elemental concentration logs
which are then input into the computational loop until convergence of the answers is obtained. The mineralogy is
used to compute a matrix density which is then combined with the bulk density to produce an accurate total porosity.

10