uAbstract
uFigures
1-5
uMethodology
uFigures
6-14
uAttributes
in porosity prediction
uFigures
15-17
uConclusions
uReferences
uAcknowledgments
uAbstract
uFigures
1-5
uMethodology
uFigures
6-14
uAttributes
in porosity prediction
uFigures
15-17
uConclusions
uReferences
uAcknowledgments
uAbstract
uFigures
1-5
uMethodology
uFigures
6-14
uAttributes
in porosity prediction
uFigures
15-17
uConclusions
uReferences
uAcknowledgments
uAbstract
uFigures
1-5
uMethodology
uFigures
6-14
uAttributes
in porosity prediction
uFigures
15-17
uConclusions
uReferences
uAcknowledgments
uAbstract
uFigures
1-5
uMethodology
uFigures
6-14
uAttributes
in porosity prediction
uFigures
15-17
uConclusions
uReferences
uAcknowledgments
uAbstract
uFigures
1-5
uMethodology
uFigures
6-14
uAttributes
in porosity prediction
uFigures
15-17
uConclusions
uReferences
uAcknowledgments
|
Return to top.
The following steps were taken in the
completion of this study: First, the horizons were picked in the well
logs; then synthetic seismograms were created to tie the wells to the
seismic data. The horizons were then picked in the seismic data, and
faults were mapped using a combination of coherency and conventional
amplitude seismic displays; the resulting structural features were
analyzed. We sought to create a porosity volume for the Trenton-Black
River interval using the total average porosity (PHIA), which was
generated from the average of the density-porosity and neutron-porosity
logs. We used the methodology of Hampson et al. (2001) to identify the
best combination of attributes for predicting PHIA. We then trained a
neural network to convert the seismic amplitude data to a porosity
volume. The porosity volume was combined with the fault mapping in order
to examine relationships between porosity and structural features.
Figure Captions
(6-14)
|
Figure 6. A schematic diagram of the
workflow followed during this study (adapted from Pearson and
Hart, 2004) |
|
Figure 7. a). Stratal slice through
the coherency volume 18ms above the Top Basement horizon, with
highly coherent reflections shown in white and less coherent in
black. The sinuous nature of the basement fault can be seen, along
with the step-over faults to the north and south. b). Inline 93 is
a transect through one of the positive flower structures that was
mapped. |
|
Figure 8. a). A simplified block
diagram of helicoidal deformation above a single basement fault
(after Mandl, 1988). b). The main Riedel shear shown with the Top
Basement structure map; its helicoidal nature is apparent. |
|
Figure 9. The strain ellipse for
Saybrook, using the general strike-slip criteria that the primary
stress is oriented at 45 degrees to the direction of movement. |
|
Figure 10. Looking down the fault
system at Saybrook, at the Trempealeau level and above
(Trempealeau time structure is shown). The main synthetic Riedel
shear is light green, and the less developed Riedel shears are in
darker greens. The en echelon nature of these faults is apparent
from this angle, which appears to be consistent with the
generalized shear model presented in Figure 13.
The inset to the
right shows the correct orientation of the entire fault zone in
3-D. |
|
Figure 11. Maximum curvature
extracted from the prestack Trenton horizon, draped over the 3D
surface. Negative curvature is seen in red and orange, while
positive curvature is in blue and green.
The main fault ridge is
clearly a positive anomaly (convex-up), while well locations
(black dots) are located in small negative curvature anomalies
(concave-up) along the ridge. |
|
Figure 12. Paleogeographic
reconstruction from the Mid to Late Ordovician of southeast
Laurentia, showing the location of Taconic activity on the
northeast edge of the New York Promontory (after Ettensohn et al.,
2002). |
|
Figure 13. A generalized left-lateral
Riedel shear model (rotated to have the same orientation as
Saybrook), in two stages of development: a) initially with
only
synthetic Riedels (R) and b) later with antithetic (R') (adapted
from Mandl, 1988 and Ahlgren, 2001). The regularly spaced
extension along sub-seismic antithetic faults combined with minor
dip-slip movement may have helped to develop fluid migration
pathways. |
|
Figure 14. Graph showing the
prediction error and validation error for the multiattribute
analysis. While the validation error decreases slightly for
predictions using 7 and 8 attributes, 6 is determined to be the
optimal number in order to avoid overtraining the data. |
The attributes used in the porosity
prediction were:
RMS amplitude
Perigram
Reflection Strength
Derivative of Instantaneous Amplitude
Integrated Trace
Cosine of Instantaneous Phase
Figure Captions
(15-17)
|
Figure 15.
a). The application of the neural network to the training data;
the average error was 0.96% and the correlation was 89%.
The prediction closely matches the target
log (PHIA), except at the bottom of some of the wells where it
under-predicted the values. b).
Crossplot of the predicted versus the actual values of porosity.
The high number of data points is indicative of a volume-based
approach, which gives more statistically significant results
(Hampson et al., 2001). |
|
Figure 16. a),b). By exporting the
created porosity volume into a program that allowed the faults and
seismic to be viewed together in 3-D, it was possible better to
visualize the relationship between the faults and the predicted
porosity. Values below about 5% porosity were made transparent so
that the high-porosity (producing) areas could be highlighted.
While some noise was predicted by the analysis, c) especially on
the edges of the survey, in general the porosity is closely
associated with the mapped fault network. The highest porosity
values are also concentrated in the areas of intense faulting,
especially where the flower structures are located.
It was also
apparent that although the main fault trend continues at about 122o
–302 o, the porosity development did not continue with
the trend. Instead the porosity is better developed along the NW
left-lateral step-over fault (Figure 7)
that trends 105 o
-285 o, where there is also a small flower structure. |
|
Figure 17. a). Transects through the
porosity volume; high porosity is shown in dark reds, while low
porosity is in dark blue and black. The volume was only generated
for the interval from the top of the Trenton to the Trempealeau
horizon. Inline 93 through the porosity volume with Strong UN #1
shown and Inline 33 with the productive Downes #3 on the left and
the non-productive Downes #2 on the right. The porosity
development is greatest in the areas between the limbs of the
flower structures. b). For comparison, the same inlines are shown
in the reflection seismic volume. |
The Saybrook fault system is consistent
with a left lateral strike-slip model, with the main fault movement
accommodated by synthetic Riedel shears. Fluid migration may have been
aided by the development of antithetic Riedel shears that formed between
the overlapping synthetic Riedel shears (flower structures). This
hypothesis is supported by the porosity prediction using seismic
attributes that illustrated a clear relationship between high porosity
values and areas where there are flower structures in the fault zone.
Through the combined use of seismic
attributes and fault mapping in 3-D, it is apparent that faulting is one
of the key controls on dolomitization, and hence porosity development at
the Saybrook Field. For plays similar to Saybrook in which the reservoir
development is related to a strike-slip fault environment, detailed
fault mapping should help to illuminate the impact these structures have
had on reservoir development.
Ahlgren, S.G., 2001, The
nucleation and evolution of Riedel shear zones as deformation bands in
porous sandstone: Journal of Structural Geology, v. 23, p. 1203-1214.
Ettensohn, F.R., J.C. Hohman,
M.A. Kulp, and N. Rast, 2002, Evidence and implications of possible
far-field responses to Taconian Orogeny: Middle-Late Ordovician
Lexington Platform and Sebree Trough, east-central United States:
Southeastern Geology, v. 41, p. 1- 36.
Hampson, D.P., J.S. Schuelke,
and J.A. Quirein, 2001, Use of multi- attribute transforms to predict
log properties from seismic data: Geophysics, v. 66, p. 220-236.
Larsen, G.E., 2000 (Hull,
D.N., 1990, chief compiler), Generalized column of bedrock units in
Ohio: http://www.ohiodnr.com/geosurvey/pdf/stratcol.pdf.
Mandl, G., 1988, Mechanics
of tectonic faulting: Models and basic concepts: Elsevier: Amsterdam,
Netherlands, 407p.
Middleton, K., M. Coniglio,
R. Sherlock, and S. Frape, 1993, Dolomitization of Middle Ordovician
carbonate reservoirs, southwestern Ontario: Bulletin of Canadian
Petroleum Geology, v. 41, p. 150-163.
Pearson, R.A., and B.S.
Hart, 2004 (in press), 3-D seismic attributes help define controls on
reservoir development: Case study from the Red River Formation,
Williston Basin, in G.P. Eberli, J.L. Masaferro, and J.F. Sarg,
eds., Seismic imaging of carbonate reservoirs and systems: American
Association of Petroleum Geologists Memoir 81.
We thank Pete MacKenzie, formerly with CGAS
Inc. for supplying the data used in this project. Funding was provided
by an NSERC Discovery Grant to Hart. Software was furnished by Landmark
Graphics Corp. and Hampson-Russell Software Services.
Return to top.
|