Sendji Formation reservoir delineation based on 2-D and 3-D inversion, Yombo Fi
Seven fields with prolific Sendji
Formation production have been dis-
covered offshore Congo. These fields
have ultimate recoveries of 100 mil-
lion b/o or greater. Only the shallow
waters (less than 200 m) of the play
have been adequately explored. The
inability to predict the amount of eco-
nomically recoverable oil after initial
discovery and delineation is a major
production challenge. The Sendji
Reservoir is a complex sequence of
limestones, dolomites, and sand-
stones deposited as several distinct
facies. These facies vary both laterally
and vertically, and have vastly dif-
ferent reservoir characteristics.
Predicting the configuration of these
facies is the key to predicting reser-
voir performance and calculating the
amount of oil that is economically
recoverable. Predicting intra-Sendji
facies distribution is a significant
challenge to optimizing development
drilling.
The largest oil field in the Congo
is one example of this production
challenge. N’Kossa Field, discovered
by Elf in 1984, was estimated to have
900 million barrels of oil and con-
densate in place. Initial delineation
drilling and testing supported a
recovery factor assumption of 47%,
meaning more than 400 million bar-
rels could be economically produced.
Yet N’Kossa has significantly under-
performed and will probably pro-
duce only 300 million barrels, or 33%
recovery. This shortfall is due to reser-
voir heterogeneity revealed only after
producing several development
wells.
CMS Oil and Gas operates Yombo
Field within the Marine I Exploitation
block, which is in the Lower Congo
basin offshore the Republic of Congo
(Figure 1). Sendji reservoir hetero-
geneity is experienced in Yombo
Field. Amoco discovered Yombo in
1988. Five initial wells were drilled to
delineate the field. Commercial pay
was discovered in both the
Cretaceous Cenomanian-age Tchala
Formation and the Albian-age Sendji
Formation. Yombo currently pro-
duces about 16 000 b/d. Production
0000 THE LEADING EDGE APRIL 2001 APRIL 2001 THE LEADING EDGE 435
Sendji Formation reservoir delineation based
on 2-D and 3-D inversion, Yombo Field,
offshore Congo
JOHN VAN HORN, CMS Energy, Houston, Texas, U.S.
Figure 1. Yombo Field lies within the Marine I Exploitation block in the
Lower Congo basin, offshore the Republic of Congo. Commercial pay was
discovered in both the Cretaceous Cenomanian-age Tchala Formation and the
Albian-age Sendji Formation. Yombo currently produces 16 000 b/d on aver-
age.
Figure 2. The Upper Sendji 1 reservoir net oil isopach demonstrates a large
volume of oil in place in Yombo Field. A large portion of the field
produces at a rate well below 500 b/d. The region of economic production
is restricted to an elongate trend (green).
is almost evenly split between the
two formations.
The Upper Sendji 1 reservoir net
oil isopach map (Figure 2) illustrates
the large volume of oil in place for the
field. The five initial delineation wells
are highlighted on the isopach map.
A large portion of the field produces
at the low rate of less than 500 b/d.
The region of economic production—
more than 600 b/d—is restricted to an
elongate trend (green in Figure 2).
Conventional core description and
analysis show that this economic
trend is confined to a limited facies of
stacked tidal channels. Channels are
predominantly porous, permeable
quartz sandstone with varying
amounts of dolomite as grains and
cement. The uneconomic area to the
north is dominated by tidal flat and
lagoon facies. The area to the south is
dominated by open marine facies. The
tidal flat, lagoon, and open marine
facies have poor reservoir character-
istics, low permeability, and porosity.
Seismic inversion method to delin-
eate economically productive facies.
A seismic inversion method was rec-
ommended to accurately delineate
the extent and distribution of the eco-
nomically productive facies. If effec-
tive, seismic inversion was proposed
to evaluate other Sendji Formation
exploitation opportunities within the
lease block. A stepwise approach
allowed management to gain confi-
dence in the technique before com-
mitting significant resources where
the main expense is reprocessing the
3-D seismic volume. Given success,
this method is cost-effective because
inversion and interpretation can be
completed in-house with commer-
cially available software.
The stepwise approach employed
the following sequence: (1) attribute
cross-plotting and stratigraphic mod-
eling; (2) inversion of current 3-D
data; (3) analysis and review of
results; (4) 2-D test reprocessing of
selected 3-D sail lines and 2-D inver-
sion; (5) analysis and review of
results; (6) 3-D reprocessing of entire
volume and inversion of volume; and
(7) facies delineation using final
inversion volume.
Attribute crossplotting and strati-
graphic modeling. Amoco cored the
Upper Sendji Formation in the field’s
first development and delineation
wells. Core analysis has been vital to
understanding the facies present and
in developing a depositional model.
We now understand that different
facies have very different reservoir
properties.
Core analysis clearly shows that
permeability is facies dependent.
Figure 3 illustrates that the best reser-
voir permeability and best porosity
lie in the stacked tidal channel facies.
This relationship suggests that seis-
mic attributes would be appropriate
in mapping the tidal channel facies.
The technique of using acoustic
impedance inversion on a 3-D seis-
mic data set to delineate zones of rel-
atively high porosity is well
established. Inversion should theo-
retically be effective in detecting
high-porosity, high-permeability
tidal channel facies on seismic.
Porosity versus acoustic imped-
ance within the Upper Sendji was
crossplotted for the five delineation
wells (Figure 4). A strong linear rela-
tionship exists between increasing
porosity and decreasing acoustic
impedance. Core analysis demon-
strates that Yombo 1 and 4 wells pen-
etrated the stacked tidal channel
facies. Upper Sendji Formation in
Yombo 2, 3, and 5 is dominated by
less permeable and less porous facies.
Within the good tidal channel facies,
porosity varies from 18% to 35% at
the low acoustic impedance range
(6000-8000 gm/m3/s). Outside the
tidal channel facies, porosity is 15-
25% at a low acoustic impedance
range.
A well-log cross-section (Figure
5) using the five delineation wells
was flattened on the Top of Upper
436 THE LEADING EDGE APRIL 2001 APRIL 2001 THE LEADING EDGE 0000
Figure 3. The Upper Sendji 1 porosity-permeability crossplot illustrates
that the best reservoir permeability and best porosity lie in the stacked
tidal channel facies.
Figure 4. Porosity versus acoustic impedance within the Upper Sendji was
crossplotted for the five delineation wells. Note the strong linear relation-
ship between increasing porosity and decreasing acoustic impedance.
Sendji (red line). Porosity greater than
18% is black on the log curves. The
zone of interest is between the red
and yellow horizons. Yombo 1 and 4
have the highest average porosity
and high permeability because they
penetrated multiple, stacked tidal
channels.
The stratigraphic model, derived
from this cross-section, was convolved
with the wavelet extracted from the 3-
D volume. Figure 6 shows the result-
ing vertical incidence response. The
area of stacked tidal channels is char-
acterized by a strong trough/peak
response. The acoustic impedance dis-
play (Figure 7) exhibits a low imped-
ance response as predicted from the
attribute cross-plot (Figure 4). Note
that the acoustic impedance units for
the cross-plot are in gm/m3/s and the
units for the model are in gm/ft3/s.
Cross-plots indicate that the high
porosity zones of the Upper Sendji 1
have low acoustic impedance.
Stratigraphic modeling demonstrated
that this low acoustic response may
detect the zone of stacked tidal chan-
nel facies in the reservoir. The next step
was to extend the test to real seismic
data.
Inversion of 3-D data over Yombo
Field. Acoustic impedance inversion
was applied to a subset of the 3-D
seismic volume over Yombo Field. A
very stable wavelet was extracted
using the five delineation wells. The
resulting correlation coefficients for
the synthetic ties ranged from 0.55 to
0.76 (Figure 8).
An arbitrary line from the result-
0000 THE LEADING EDGE APRIL 2001 APRIL 2001 THE LEADING EDGE 437
Figure 5. A well-log cross section using the five delineation wells was flattened on the Top of Upper Sendji (red
line). Porosity greater than 18% is black on the log curves. The zone of interest is between the red and yellow hori-
zons. Yombo 1 and 4 have the highest average porosity and high permeability because they penetrated multiple,
stacked tidal channels.
Figure 6. This log model was convolved with the wavelet extracted from
the 3-D volume. The resulting vertical incidence response is shown. The
area of stacked tidal channels is characterized by a strong trough/peak
response.
Figure 7. The acoustic impedance display exhibits a low impedance
response as predicted from the attribute cross-plot (Figure 4). Cross-plots
indicate that the high porosity zones of the Upper Sendji 1 have low
acoustic impedance.
ing inversion 3-D volume was
extracted across the area of the well log
cross section and flattened on the Top
Upper Sendji, red horizon, as done in
the modeling. The resulting acoustic
impedance section (Figure 9) matches
the model. The thickest, most laterally
continuous zone of low acoustic
impedance lies within the zone of the
stacked tidal channel facies. The inver-
sion has detected the low acoustic
impedance zone, outlined on Figure 9,
which coincides with the economic
production trend area shown on the
net pay isopach map (Figure 2).
Comparing 2-D and 3-D data for
detailed inversion study. The result of
the inversion on the 3-D seismic vol-
ume showed that the productive facies
could be seismically delineated. The
next question was: Can we increase
the detail in the delineation by improv-
ing the seismic data and the resulting
inversion? To answer this question in
a cost-effective manner, we selected 12
sail lines for 2-D reprocessing. Each
sail line ties at least one well.
The current 3-D volume, used in
the inversion test presented above, is
a poststack reprocessed version of the
original 1990 processing. The original
data were whitened to 160 Hz over
the entire section. Data were acquired
and processed at 2 ms to avoid alias-
ing these high frequencies. Filter test-
ing showed that the signal/noise ratio
deteriorated rapidly over 120 Hz in
the zone of interest. The whitening,
therefore, boosted high frequency
noise. Upon completion of the post-
stack reprocessing, the whitening was
removed. The resulting data were far
less noisy, but the high frequency con-
tent of true signal was diminished.
The 2-D reprocessed data are less
noisy than the 3-D data. The high fre-
quency content is preserved. Figure 10
compares extracted wavelets and fre-
quency spectra of the 3-D and 2-D
data. The dominant frequency content
is 20-50 Hz in the 1998 poststack repro-
cessing. This has improved to 20-90
Hz in the 1999 2-D test reprocessing.
The processing steps contributing
most to the data improvement were
prestack radon demultiple; offset
DMO; prestack spectral balance (8-124
Hz); and one-pass Kirchhoff migra-
tion.
Most of these algorithms were not
available when the data were origi-
nally processed in 1990. Only one-pass
migration could be applied in the 1998
poststack reprocessing. The other algo-
rithms are prestack processes.
2-D inversion defines the target
facies. Inversion was applied to sev-
eral of the 12 reprocessed sail lines.
Figure 11 shows acoustic impedance
inversion of sail line YST-1183 with the
seismic data in wiggle trace overlay.
The zone of interest is between the
blue and light green horizons. The
strong trough/peak response in the
wiggle trace overlay is laterally con-
tinuous. This is the zone of lowest
acoustic impedance in the Upper
Sendji. The dashed line, enclosing the
low acoustic impedance area, delin-
eates the lateral extent of the stacked
tidal channel facies. The acoustic
impedance indicates that this facies is
thinning updip, and this is confirmed
438 THE LEADING EDGE APRIL 2001 APRIL 2001 THE LEADING EDGE 0000
Figure 8. A very stable
wavelet was extracted
using the five
delineation wells. The
resulting correlation
coefficients for the syn-
thetic ties ranged from
0.55 to 0.76.
by well data. Well Y-4 encountered
about 35 m of porous tidal channel
facies, but B-13 only encountered
about 7 m. The facies was absent in B-
6St.
The location of sail line YST-1183
is highlighted on Figure 2. The low
acoustic impedance outline matches
the extent of Yombo Field’s productive
trend along YST-1183. Cumulative pro-
duction from the wells is shown in
Table 1.
3-D inversion, reprocessing, and
reservoir facies delineation. The data
improvement from the 2-D test repro-
cessing and the successful delineation
from the 2-D and 3-D test inversions
gave CMS management confidence in
this technique and resulted in approval
to reprocess the entire 3-D volume.
Inversion was applied to the
reprocessed volume using the same
five delineation wells. Volume
0000 THE LEADING EDGE APRIL 2001 APRIL 2001 THE LEADING EDGE 439
Figure 9. An arbitrary line from the resulting inversion 3-D volume was extracted across the area of the well log
cross-section, and flattened on the Top Upper Sendji (red line) as done in the modeling. The resulting acoustic
impedance section matches the model. The thickest, most laterally continuous zone of low acoustic impedance lies
within the zone of the stacked tidal channel facies. The inversion has detected the low acoustic impedance zone
(outlined in black) which coincides with the economic production trend area on the net pay isopach map (Figure 2).
Figure 10. Comparison of extracted wavelets and frequency spectra of the 3-D and 2-D data. The dominant
frequency content is 20-50 Hz in the 1998 poststack reprocessing. This has improved to 20-90 Hz in the 1999 2-D test
reprocessing.
Table 1. Cumulative production from wells
Well
Y-4
B-1
B-13
B-6St
Facies
Thick stacked tidal channel
Thick stacked tidal channel
Thin tidal channel
Lagoon/tidal flat
Cumulative production
not produced/delineation well
1.92 MMBO
0.55 MMBO
0.015 MMBO
440 THE LEADING EDGE APRIL 2001 APRIL 2001 THE LEADING EDGE 0000
Figure 11. Inversion was applied to several of the 12 reprocessed sail lines. Acoustic impedance inversion of sail line
YST-1183 is presented with the seismic data presented in wiggle trace overlay. The zone of interest is between the
blue and light green horizons. The strong trough/peak response in the wiggle trace overlay is laterally continuous.
This is the zone of lowest acoustic impedance in the Upper Sendji. The dashed line, enclosing the low acoustic
impedance area, delineates the lateral extent of the stacked tidal channel facies.
Figure 12. Mean acoustic impedance was the volume attribute that best delineated the extent of the productive
facies. The contours of the net oil isopach map are overlain on the attribute map of mean acoustic impedance. The
facies extent to the north and east is clearly mapped and matches the limits of the productive area.
attribute analysis was performed on
the inversion data set. Mean acoustic
impedance was found to be the vol-
ume attribute that best delineated the
extent of the productive facies. The
contours of the net oil isopach map
are overlain on the attribute map of
mean acoustic impedance (Figure 12).
The facies extent to the north and east
is clearly mapped and matches the
limits of the productive area. A rea-
sonable interpretation of the attribute
map has the stacked tidal channel
facies extending southwest of well Y-
4, beyond the productive trend. Wells
have not been drilled in this area
because the oil-water contact has
begun to rise and new wells would
likely produce too much water.
Figure 13 shows two arbitrary lines
through the 3-D inversion volume.
Cumulative production for the Upper
Sendji 1 reservoir is next to each well.
Dashed lines indicate the interpreted
extent of the stacked tidal channel
facies. The inversion demonstrates that
this facies developed off the south-
west, the seaward-facing flank of the
structure. The northern portion of the
net oil isopach is attractive because it
is high above the water contact, with
potential undeveloped oil in place. The
facies transition from tidal channel, to
tidal flat and lagoon dramatically
decreases the permeability and lowers
the average porosity, which results in
low oil production rates. The inver-
sion volume is assisting in delineation
of the extent of the productive tidal
channel facies and optimizing the loca-
tion of future Upper Sendji develop-
ment wells.
Conclusions. A strong linear relation-
ship between increasing porosity and
decreasing acoustic impedance is
observed in Yombo Field, offshore
Congo. Crossplotting indicated that
the high-porosity zones of the Upper
Sendji 1 would have low acoustic
impedance. Stratigraphic modeling
demonstrated that this low acoustic
response detected the zone of the
stacked tidal channel facies in the for-
mation.
A seismic inversion method was
recommended to accurately delineate
the extent and distribution of the eco-
nomically productive facies. The thick-
est, most laterally continuous zone of
low acoustic impedance lies within the
zone of the stacked tidal channel facies.
The acoustic impedance indicates that
this facies is thinning updip, and this
is confirmed by well data. The inver-
sion detected the low acoustic imped-
ance zone that coincides with the
economic production trend area
shown on the net pay isopach map. LE
Acknowledgments: The author thanks CMS
Oil and Gas Company, Nuevo Energy
Company, Societe Nationale Des Petroles du
Congo. and the Ministry of Hydrocarbons,
Republic of Congo, for permission to publish
this paper. I thank Allen Brown, Barry Faulkner,
and Doug Jordan for key contributions. I thank
the management of CMS and Nuevo for sup-
porting the publication of this article.
Corresponding author: J. Van Horn,
jvanhorn@cmsenergy.com
John Van Horn received a bachelor’s degree in
geology from the State University of New York
at Binghamton and a master’s degree in geo-
physics from Michigan Technological
University. He joined Mobil in 1980 and
worked on a series of exploration and produc-
tion projects in a variety of domestic and inter-
national locations. In 1996, he joined Texaco
international exploration, working on the China
exploration team. In 1998, he joined CMS Oil
and Gas.
0000 THE LEADING EDGE APRIL 2001 APRIL 2001 THE LEADING EDGE 441
Figure 13. Two arbitrary lines through the 3-D inversion volume. Cumulative production for the Upper Sendji 1
reservoir is next to each well. Dashed lines indicate the interpreted extent of the stacked tidal channel facies. The
inversion demonstrates that this facies developed off the southwest, the seaward-facing flank, of the structure.