Download as pdf or txt
Download as pdf or txt
You are on page 1of 23

IPA03-G-161

PROCEEDINGS, INDONESIAN PETROLEUM ASSOCIATION


Twenty-Ninth Annual Convention & Exhibition, October 2003
INTEGRATED SEISMIC MULTI-ATTRIBUTE ANALYSIS FOR COMPLEX FLUVIO-DELTAIC
RESERVOIR PROPERTIES MAPPING, MINAS FIELD, CENTRAL SUMATRA
Sigit Sukmono*
Dardji Noeradi*
Faizil Fitris**
Tafsillison**
William C. Richmond**
Seffibudianti**
Pujiyono**

ABSTRACT
Minas field, operated by PT Caltex Pacific Indonesia
in Central Sumatra, is characterized by a complex
fluvio-deltaic depositional system where lateral and
vertical distribution of sand bodies is very
heterogeneous.
The existing high-resolution
sequence stratigraphic framework based on
conventional cores and well logs is inadequate for
accurately mapping reservoir-scale flow units. To
overcome this problem, seismic multi-attribute
analysis utilizing high-resolution 3D seismic data has
been investigated.
The prime objective is to
differentiate sands from shales using the seismic data,
especially for inter-well correlations (injector to
producer). The integrated attributes include: 1)
internal seismic attributes (amplitude, frequency, etc),
and 2) log property attributes (porosity, gamma ray,
etc) derived from cross-plots between the internal
seismic attributes and well data.
Cross-plot analyses were generated for sand/shale
intervals that roughly correspond to five main oilproducing reservoirs. These five intervals are defined
and bracketed by key field-wide sequence
stratigraphic marker horizons: 1) M1 - SB1, 2) SB1
SB2, 3) SB2 M6, 4) M6 SB4, and 5) SB2
SB2.1. For Interval 1, the acoustic impedance (AI)
can distinguish sands from shales, thus seismic
inversion is recommended for this interval. Intervals
2 to 5 have overlapping AI and it is very difficult to
differentiate sand from shales using the normal
amplitude sections.
*
**

Institute Technology Bandung


P.T. Caltex Pacific Indonesia

Therefore, the most applicable internal seismic


attributes are the derivatives of frequency. These are
used to generate seismic pseudo-gamma ray, pseudoporosity and pseudo-density sections, which are
derived from the multi-attributes analysis. Each of
these has proven to be effective in differentiating sand
and shale in different portions of Intervals 2 to 5.
Consequently, the system-tracts can also be
delineated more easily, thus providing an effective
tool for guiding well correlations.
The results of combining various derived attributes
provide additional support for the detailed correlation
and mapping of the location, geometry and quality of
individual sand bodies within Minas field. This will
lead to improved waterflood optimization and infill
well selection. Newly acquired well data have
confirmed the results obtained in this study.
INTRODUCTION
Minas field, located in the Riau Province of Central
Sumatra, Indonesia (Figure 1), is one of the largest oil
fields in Southeast Asia. The field was discovered in
1942 and production began in 1952. In 1970,
peripheral water injection was initiated owing to field
pressure decline. This was followed by secondary
infill development on 71-acre well spacing in the late
1970s. During 1970 and 1980, the peripheral
waterflood was expanded, and in the late 1980, a
successful pattern waterflood pilot development
project was undertaken. As a result, a full crestal
pattern waterflood (PWF) on 72 acre inverted 7-spot
patterns began as a staged development in 1993. The
pattern waterflood has been expanded to cover the

entire crestal portion of the Main Segment (Figure 1).


At this stage, the field has reached a very mature
stage of production and various tertiary recovery
methods have been explored over the past few years.
The objective of this study is to apply seismic multiattribute analysis on the Minas high-resolution 3-D
seismic data as a tool to aid in the lateral correlation
of sand bodies. The main Minas reservoirs are
believed to be a complex fluvio-deltaic depositional
system. The lateral and vertical distribution of sand
bodies is very heterogeneous. Even though well
spacing in Minas Field is already very dense (300 m
on crestal portion of field), classic correlation
methods, using wire line log data, are unable to solve
the ambiguity in sand body correlation between wells.
The study area in Minas field is located in the
southern portion of the Main Segment (Figure 1), and
contains 11 vertical and three directional wells as
control (true sonic and density logs, and check shot
surveys).
This paper discusses the application of seismic multiattribute analysis on the Minas high-resolution 3D
seismic data in order to generate pseudo-log cubes
(pseudo-gamma ray, pseudo-porosity, pseudo-density,
etc). The integrated analyses on the pseudo-log cubes
are then utilized to map the depositional facies, sand
bodies and their properties (porosity, etc), and for
guiding well correlations. This will in turn provide a
critical tool for reducing the risk in targeting new
infill wells, and for making decisions related to the
optimization of the current waterflood operations.

be termed deterministic reservoir parameter


prediction. Although relationships have been inferred
between these attributes and reservoir parameters, the
physical basis is not always clear, and we may want
to derive statistical, rather than deterministic,
relationships.
In the most general case, a relationship function is
used to convert m different attributes into the desired
property. This relationship can be written as:
P (x,y,z) = F[A1 (x,y,z),,A m(x,y,z)]

(1)

where:
P = the property as a function of coordinates x,y,z,
F = the functional relationship; and
Ai , I = 1, , m = the m attributes.
The simplest possible case would be a linearly
weighted sum:
P = w 0 + w1 A1 +wm Am

(2)

where:
wi , i = 0, , m = the m + 1 weights.

METHODOLOGY
Theoretical Background
Theoretical background on seismic multi-attribute
analysis is provided by Russell et al. (1997). Seismic
multi-attribute analysis is a broad term that
encompasses all geostatistical methods that utilize
more than one attribute to predict some physical
property of the earth. The idea of using multiple
seismic attributes to predict log properties was first
proposed by Schultz et al. (1994). They pointed out
that the traditional approach to deriving reservoir
parameters from seismic data has been to look for a
physical relationship between the parameter to be
mapped and some attribute of the seismic data, and
then use that relationship over a 2D line or 3D
volume to predict the reservoir parameter. This could

The number and types of attributes are determined by


using cross plots between the attributes and the
parameter to be measured. The associated correlation
coefficient is used to determine the quality of the
relationship between the attribute and parameter in
question.

Workflow of the Study


To achieve the above-mentioned objectives, the
following steps were performed (Figure 2):
1. Construct stratigraphic framework and define
system tracts, using conventional cores, wire-line
logs and 3-D seismic data

2.
3.
4.
5.
6.

Cross plot well log properties


Tie wells to 3-D seismic volume
Generate pseudo log cubes
Validate log property cubes
Integrated analysis

The stratigraphic framework provides the geological


constraints required by the multi-attribute analysis,
such as basin configuration, depositional/stratigraphic
sequences, system-tracts and sedimentary filling
pattern. The log properties cross plots are then used
to identify seismic properties which can be used for
differentiating sand from shales within sequences, and
for mapping sand bodies and their properties,
especially the porosity. Good well-to-seismic ties are
very crucial for obtaining good results in the
construction of pseudo log cubes from the seismic
attributes. In this step, convolving the reflectivity
coefficients with the best wavelet that has a frequency
response and bandwidth similar to the seismic
sections produces the synthetic seismogram. The
reflectivity coefficients are calculated from the edited
sonic and density logs. The constructed pseudo log
cubes are then utilized for the inter well mapping of
sand bodies and their properties such as by using the
pseudo gamma ray cube, pseudo porosity cube, etc.
To test the validity of the results obtained from multiattributes analysis, the predicted values derived from
the analysis are compared to the actual log values
from the wells not used in training. Also, the derived
values are compared with the actual gamma ray logs
from recent infill wells that were drilled during the
final stages of this study. After a strong confidence of
the multi-attrib ute results is obtained, the results of
the multi-attributes are then applied to get better
understanding on the geological and stratigraphic
conditions of the study area.
RESULTS
Stratigraphic Framework Analysis
The Minas reservoirs belong to the Sihapas Group of
Early Miocene age. The Sihapas was deposited
during the late stage of basin rifting (postrift).
Starting from this period, the basin setting was
changed from continental rift to back arc setting.
This changing of tectonic style implies a dynamic
component of basin evolution, which controls the
accommodation space, provenance and sedimentary
filling. The overall depositional environment of the

Sihapas Group in the Central Sumatra Basin includes


braided fluvial systems, braided delta plain, mouth
bars and prodelta/shelf mudstones.
A detailed
description of the regional geologic setting can be
found in Heidrick and Aulia (1996).
The Sihapas Group in the Minas field consists of
intercalated sandstone and shale deposited in a fluviodeltaic environment. Previous studies by Toha et al.
(1999), Richmond et al. (2002) and Institut Teknologi
Bandung (unpublished), found that the reservoir in
the Minas field can be divided into six depositional
facies:
Fluvial Channel, Distributary Channel,
Isolated Channel/Bar, Distributary Mouth Bar,
Prograding Bar and Tidal Flat/Estuarine. This model
implies that the variety of sand facies distribution is
difficult to delineate. This sequence stratigraphic
model, based on core calibrated log data, gave
ambiguous results in certain levels of the reservoir.
Moreover, the same log character sometimes did not
fall on the same reflector.
The stratigraphy of Minas field constitutes in a largescale regressive-transgressive cycle, punctuated by
several smaller-scale parasequences that are used to
define reservoir units (gross flow units). The
correlation chart in Figure 3 shows the relationship
between parasequences (SBn), reservoir units and
depositional intervals defined for this study. A reevaluation of the sequence stratigraphy in the study
area found that the reservoirs could be divided into
five parasequences (SBn) that are separated by
transgressive shelf and prodelta mudstones (Mn).
From bottom to top these parasequences are: 1) Top
basement to lower SB4, 2) Lower SB4 to SB3, 3)
SB4 to SB2.1, 4) SB2.1 to SB2 and 5) SB2 to SB1.
During the early stage of sedimentation (basement to
Lower SB4) the basin configuration controlled the
paleo-sedimentation (Figure 4). The sedimentation in
this period took place in a relative small basin that
was elongated in the northwest-southeast direction.
Basement highs in the northeast and southwest
restricted and funneled the mostly fluvial sediments
into the local basin from the north-northeast.
However, once this local basin begin to fill, a change
in sedimentary filling pattern occurred (Figure 5). In
an early stage, the sedimentation was from north to
south (Top basement-SB2.1), and in later stages from
the east and southeast (SB2 to SB1).
The sediments between SB4 to SB2.1 can be divided
into two packages (Figure 6): LST (Lowstand System

Tract) and TST (Transgressive System Tract). The


LST package has at least two sedimentary patterns
that are from north to south and from east to west.
The thickness of those packages tends to thin toward
the southwest, marking the prograding direction
(Figure 7).
The TST interval follows the LST package and
thinning toward northeast indicating the retrograding
direction (Figure 6). Gamma ray logs for the LST
generally show a blocky stacking pattern that indicate
stacked braided channels, while the TST generally
shows a lesser energy depositional system,
characterized by the thickness of shale interval.
Pseudo density and pseudo gamma ray cubes cannot
be used to interpret the geological condition due to
inconsistency between well data and the cube.
Therefore, the reservoir distribution can only be
interpreted qualitatively using the LST map
(Figure 7).
These results imply that vertically and laterally, the
reservoirs in the studied area cannot be considered as
single reservoir packages, but discrete packages that
are bounded vertically and laterally by stratigraphic
discontinuities.

cross-plots within the above listed interval, the


following conclusions can be drawn:
a. Interval 1 (M1 - SB1):
Reservoir sand can be differentiated from shale
using P-wave Acoustic Impedance, P-wave
velocity, and gamma-ray. Thus P-wave velocity
and gamma-ray are the target logs for this interval
which can be generated from multi-attributes
analysis. P-wave Acoustic Impedance can be
generated using post stack seismic inversion
b. Intervals 2-5 (SB1-SB2, SB2-SB2.1, SB2-M6,
M6-SB4)
Sand and shale can be differentiated using
gamma-ray, density and porosity, Therefore,
those three logs are the target logs for these
intervals. However, P-wave Acoustic Impedance
cannot be used.
Within the study area, seven wells that have sonic and
density logs were used. The best well-to-seismic ties
were obtained when the wavelet extracted from
seismic traces around the training wells were used.
Minor bulk-shifts and stretch-squeeze were also
applied. The average correlation coefficient between
the seismic and the synthetics is 0.74.

Log Properties Cross-Plots & Well-Seismic Tie


The objective of cross plotting is to obtain log
properties (called target logs) that can be used to
differentiate reservoir sands from shale within each
appropriate stratigraphic interval. The intervals are
defined as the ones that give the best results for
separating the sand and shale, as determined through
trial and error. The best results were obtained for the
following depositional intervals :
1.
2.
3.
4.
5.

Interval M1 SB1
Interval SB1-SB2
Interval SB2- M6
Interval M6 - SB4
Interval SB2 SB2.1

The five selected intervals are shown on Figure 3.


Examples of different log properties cross-plot for
interval M1-SB1 (P-wave Acoustic Impedance,
Gamma Ray, and Density), interval SB1-SB2 (Pwave Acoustic Impedance, Gamma-Ray, and
Porosity), and interval M6-SB4 (P-wave Acoustic
Impedance, Gamma Ray , and Density) are shown in
Figure 8, 9 and 10 respectively. Based on numerous

Pseudo-Log Cube Generation and Result


Validation
Seismic multi-attribute analysis is aimed to generate
pseudo log cubes from seismic cubes as illustrated in
mathematical equation (2) above. Initially, 25 seismic
attributes were generated, these are called internal
attributes. External attributes can also be added such
as acoustic impedance, etc.
Cross-plot analysis between the target log within the
selected stratigraphic interval and single seismic
attributes was performed. The best single attribute
was selected based on the best correlation coefficient.
This selected attribute was then used for cross-plot
analysis to find the second attribute from remaining
24 attributes. This process is continually repeated to
find the next best attributes. As more attributes are
used in this cross plot, the correlation tends to
improve, i.e., training error becomes smaller.
A cross validation is done by excluding one well from
the cross-plot analysis and predicting the log values
from the other 6 wells used in this project. A

validation error is computed by comparing the


predicted and the original log values within the
interval of the analysis. The process is repeated,
excluding in turn each of the other six wells.
To find the most optimum attribute, the errors of
validation and application are plotted against the
number of attributes. For instance, Figure 11 shows
that attribute number 7, i.e. the derivative, gives the
smallest validation error when it is used to generate
the pseudo gamma-ray in Interval SB1-SB2. This
implies that in the training wells, the pseudo gammaray log generated using the derivative attribute gives
the best match with the real gamma-ray log.

Figures 12-14 give further illustration of this


technique. Figure 12 shows a comparison between
pseudo gamma-ray logs generated by the multiattribute analysis and actual gamma-ray logs from
seven training wells for Interval SB1-SB2: The
predicted pseudo-logs are a good match to the actual
logs. Figure 13 gives a visual illustration of a pseudogamma ray section derived from multi-attribute
analysis compared to actual gamma-ray log data for
Interval SB1-SB2. The inset picture shows the crossplot between the predicted gamma-ray logs and the
actual gamma-ray logs for the seven training wells
used in the analysis. The correlation coefficient is
0.81 and the average error is 18.3 gamma-ray units
from data interval of 50 to 150 gamma-ray units.
Figure 14 gives a comparison between a normal
amplitude section (top panel) and a pseudo-gamma
ray section (bottom panel). Sand and shale cannot be
differentiated in the amplitude section as their AIs
overlap but they can be differentiated quite easily in
the pseudo gamma-ray section and thus provide a
good guidance for correlation of sandbodies between
wells.
Using the same technique, multi-attribute analysis
was conducted for all five intervals described in the
previous section. It was found that the dominant
attributes in these five intervals are mostly the
derivatives of frequency attributes, which are then
used to generate seismic pseudo gamma-ray, pseudo
density and pseudo porosity cubes.
For each interval, the best correlation between the
pseudo-logs and actual logs are obtained when the
following pseudo-log cubes are generated:

1. Interval M1-SB1: Pseudo gamma-ray and pseudo


Vp.
2. Interval SB1-SB2: Pseudo porosity (the best),
and pseudo gamma-ray.
3. Interval SB2-M6: Pseudo density (the best), and
pseudo gamma-ray.
4. Interval SB2-SB2.1: Pseudo density (the best),
and pseudo gamma-ray.
5. Interval M6-SB4: Pseudo density
Similar to Interval SB1-SB2, the sands and shales in
Intervals SB2-M6, SB2-SB2.1, and M6-SB4 also
have overlapping AI; therefore, it is very difficult to
differentiate the sands from shales using the normal
amplitude sections. However, using the pseudo log
sections (pseudo gamma-ray, pseudo density and
pseudo porosity) derived from multi-attributes
analysis, the sands and shales can be differentiated
more easily. Consequently, the system tracts can also
be delineated more easily and with more detail, thus
providing an effective tool to assist in well
correlations.
In order to determine the validity of the results, the
pseudo-gamma ray cubes derived from the multiattributes analysis are compared with the actual
gamma-ray logs from independent wells, not used in
training. Also, real wire-line log data from recent
infill wells acquired during the completion of this
study have been used to validate results.
The validation results using independent well 8D-88
is considered good to very good where actual trend of
log porosity, gamma-ray and Vp can be well predicted
by the multi-attributes pseudo-logs. The average
errors of predicted porosity, gamma-ray and Vp are
also quite small: 3.1%, 10% and 4%, respectively.
Figures 15 and 16 give examples on the comparison
of predicted inter well sand body correlations from
multi-attribute analysis (Intervals SB2-SB2.1 and
SB1-SB2) with newly acquired infill well data. The
infill data show that sand thickness in Interval SB2SB2.1 varies abruptly over a relatively short distance,
but the variation still can be well modeled by the
pseudo gamma-ray section derived from multiattribute analysis. The same condition is also
obtained for other intervals; the predicted sand

distribution from multi-attribute analyses matches


well with the actual sand distribution identified in
infill wells.
Integrated Analysis
The results of the multi-attributes analysis are utilized
to get a better understanding of the geological
conditions of the study area. The following results
are discussed in terms of each depositional interval.
1. It is found that for the Interval M6 to SB4, the
sedimentation most probably occurred in the post
rift period. This occurred in a relatively small
basin that was elongated in the northwestsoutheast direction and was bordered by two local
highs in the northeast and southwest portions of
the area. There is a possibility that a minor
tectonic pulse had induced a local unconformity
that created an erosional surface below SB4
(Lower SB4) and pseudo downlap in the
southwest basement high.
2. In the Interval SB4 to SB2.l, the sediments can be
divided into two packages: LST (Lowstand
System Tract) and TST (Transgressive System
Tract). The LST package at least has two
sedimentary patterns that are from north to the
south and from east to the west. The thickness of
those packages tends to thin towards the
southwest, which also marks the direction of
prograding system. Deposition of the TST
interval follows the LST package, and thinning
occurs toward the northeast indicating the
retrograding direction.
A north-south fault
system also played a significant role in
controlling the sediment distribution pattern.
3. A local high reappeared during the deposition of
the Interval SB 2.1 to SB 2 in the southwestern
part of the study area. This was most probably
the result of tectonic tilting as indicated by the
reversal of sedimentary filling direction from
generally north-south to become south-north.
The sedimentation tends to thicken to the north,
east, and southeast. The sand distribution map
shows
northwest - southeast
trend of
sedimentation.
4. In Interval SB2 to SB1, the local basement high
born in the previous period was flooded by
sediments indicating that the regional subsidence

was the dominant control of sedimentation.


Sedimentary packages between SB2 to SB1 can
also be divided into the LST and TST. The LST
package progrades northwardly, and the TST
retrograde southwardly. The LST and TST
configurations indicate a change on the general
sedimentation direction that was formerly from
north to the south (basement to SB2.1) becoming
from south to the north. Gamma-ray logs indicate
that the sedimentation in this interval was started
by a fluvial system followed by a bar system.
The sequence can be interpreted as deltaic
sedimentation.
The integrated analysis on the pseudo-log cubes can
assist in detailed correlation and mapping of the
location, geometry and quality of individual sand
bodies within the Minas field. This will lead to
improved pattern waterflood optimization and infill
well placement. For instance, in Figure 17 the pseudo
gamma-ray and pseudo-porosity sections (left panels)
are integrated with the pseudo gamma-ray and
pseudo-porosity horizon slices (right panels) to map
the distribution of porous LST sand in interval SB1SB2.
CONCLUSIONS
The difficulties in correlating the sand bodies in
Minas field are related to two main factors: 1) very
complex lateral and vertical sand distribution owing
to the nature of the fluvio-deltaic depositional
environment, and 2) overlapping AIp values between
sands and shales. This study finds that the seismic
multi-attribute analysis can be very effective to solve
this problem. It is because the technique does not rely
only on AI, but utilizes statistical techniques to
identify the most applicable seismic attributes for
differentiating sand from shales. To succeed, these
attributes should be applied in certain lateral and
vertical intervals constrained by stratigraphic
framework of the area.
In this regards, the study area was divided into five
depositional intervals: 1) M1-SB1, 2) SB1-SB2, 3)
SB2-M6, 4) SB2-SB2.1 and 5) M6-SB4. The
applicable attributes in these five intervals are mostly
the derivative of frequency attributes, which may be
related to the nature of the channeling in the fluviodeltaic depositional environment. The best results are
achieved when the attributes are converted to the
pseudo gamma-ray for Interval M1-SB1, pseudo

porosity for Interval SB1-SB2, and pseudo density for


Intervals SB2-M6, SB2-SB2.1 and M6-SB4. The
integrated analysis on the pseudo-log cubes can assist
in detailed correlation and mapping of the location,
geometry and quality of individual sand bodies,
which in turn will lead to improved pattern
waterflood optimization and infill well placement.

Annual Convention Of Indonesian Association of


Geologists (IAGI), Jakarta.

REFERENCES

Schultz, PS., Ronen, S., Hattort, M. and Corbett, C.,


1994. Seismic guided estimation of log properties:
Part 1, 2 and 3, The Leading Edge, Society of
Exploration Geophysicists.

Heidrick, T. L. and Aulia, K., 1996. Regional


Structural Geology, Chapter II. Petroleum Geology
of the Central Sumatra Basin, BPPKA/Pertamina,
Jakarta Indonesia, p. 13-156.
Richmond, W. C., Dwidjojuwono, H., Tastari, A., and
Toha, B., 2002. Reservoir Compartmentalization: An
Integrated Evaluation of Supermature Minas Oil
Field, Central Sumatra. Proceedings of the 28th

Russell, B., Hampson, D., Schuelke, J. and Quirein,


J., 1997. Multi-attribute Seismic Analysis. The
Leading Edge, Society of Exploration Geophysicists.

Toha, B., Aulia, K., and Primadi, H., 1999. High


Resolution Sequence Stratigraphy of the Minas Oil
Field: A key Reference for Reservoir Management
and EOR Oil Field Development. Proceedings of the
27th Annual Convention Of Indonesian Association of
Geologists (IAGI), Jakarta.

Sumatra

STUDY AREA

Figure 1 - Location of Minas Field and schematic picture showing the study area and location control
(training) wells.

3D Seismic Data

Well Log Data

Sequence Stratigraphic
Framework model

Log Properties
Cross-plots

Seismic key wells


Aligning & matching

Seismic multi attribute analysis and


log property cube estimation

Validation using nontraining


wells & new infill
data

Reservoir Bodies and their


Properties maps

Figure 2 - Workflow of Multi Attributes Study.

Minas Field Correlation Chart


Stratigraphic Markers

Interval

Reservoirs

M1
M2

X Sand

M3
SB1

M4
M4.3

SB2
SB2.1

A1 Sand

M4.5

M5
M5.3

M6
M6.3

SB4

Upper
B1 Sand
Lower
Upper
B2 Sand
Lower

M5.5
SB3

A2 Sand

D Sand

M7
S Sand

Figure 3 - Correlation Chart of Minas Field showing local stratigraphic marker horizons, reservoirs and
grouped depositional intervals defined for study. SB = Sequence Boundary and M = Flooding
Surface.

Basement
High

Basement
High

Isochron: Basement to SB4

Figure 4 - Isochron map of Basement to SB4 interval indicating a small basin elongated in the northeast-southwest direction and bordered by two local highs
in the northeast and southwest parts of the study area. Green/Yellow indicates thick and blue indicates thin.

SB1

SB2
SB2.1

Basement
High

Basement

Basement
High

No
sedimentatio
n area

Isochron: SB2 to SB2.1

Figure 5 - Sedimentary distribution within Interval SB2-SB2.1 illustrating the wedging-out of interval
owing to the presence of local basement high in southern study area.

Figure 6 - Illustration of Lowstand System Tract (LST) and Transgressive System Tract (TST) within
SB4-SB2.1 interval.

LST: SB4 SB2.1

Figure 7 - The Lowstand System Tract (LST) map within Interval SB4 SB2.1, indicating southward
thinning and progradation.

Sand

Shale

Sand

Shale

B
Figure 8 - A) Example of cross-plot AIp vs. GR for Interval M1-SB1 at well 8C74. Sands AI is greater
than shales AI. B) Example of cross-plot of AIp vs. GR vs. density for Interva l M1-SB1 at well
8C38. Sands density is higher than shales density.

8C-66_SB1-

Shale

Sand

Shale

Sand

Figure 9 - Example of cross-plot AIp vs. GR vs. Porosity for Interval SB1-SB2 at wells 8C38 and 8C66.
Sands AIp is equal to shales AIp . Sands porosity is higher than shales porosity.

Sand

Shale

Sand

Shale

Figure 10 - Example of cross-plots AIp vs. GR vs. Density for Interval M6-SB4 at wells 8C66 and 7C78.
Sands AIp is equal to shales AIp . Sands density is lower than shales density.

Attributes Used

Figure 11 - Plot of application and validation errors vs. attributes used. These plots are used to determine
number of attributes used for predicting gamma -ray -- Interval SB1-SB2.

Figure 12 - Plots of application errors: modeled GR log versus actual GR log in seven training wells.
Predicted logs are quite well matched to actual log -- Interval SB1-SB2.

Good Match

Figure 13 -

Illustration of pseudo-gamma ray section resulting from multi-attribute analysis compared to actual gamma-ray log data for Interval SB1-SB2.
Inset picture shows the cross-plot between the predicted gamma-ray log and the actual gamma-ray log for the seven training wells used in the
analysis.

No unique relationship between amplitude and sand bodies

Figure 14 - Illustration of normal amplitude section (top panel) compared to pseudo-gamma ray section (bottom panel) for Interval SB1-SB2. Sand and shale
cannot be differentiated in the amplitude section, as their AIps overlap; but they can be differentiated quite easily in pseudo gamma-ray section thus
providing effective guidance for inter well correlation.

8D-68

8D-77 -EP1

8D-78

M4

8D-79

SB1
SB2
SB2.1

M5.5
Figure 15 - Comparison of infill data and multi-attribute analysis results in predicting sand body distribution in Interval SB2 -- SB2.1.

8D-68

8D-77-EP1

8D-78

8D-78-EP1

8D-79

SB1

SB2
Figure 16 - Comparison of infill data and multi- attribute analysis results in predicting sand body distribution in Interval SB1 -- SB2.

Pseudo porosity-section

Pseudo porosity-slice

TST

Porous LST sand

Pseudo gamma-ray section

Porous LST sand

Pseudo gamma ray-slice

TST
LS

sb1

TS

sb2
LST Sand

Figure 17 - The pseudo gamma-ray and pseudo-porosity sections (left panels) are integrated with the pseudo gamma-ray and pseudo-porosity horizon slices
(right pan els) to map the distribution of porous LST sand.

You might also like