COMPUTER SEDIMENTARY SIMULATION MODELS SEQUENCE STRATIGRAPHY

Christopher G. St. C. Kendall, Gregory L. Whittle, Robert Ehrlich, Philip D. Moore, Robert L. Cannon, Douglas R. Hellmann
University of South Carolina
Columbia, S.C.

A basic challenge in analyzing sedimentary basins is to determine three-dimensional stratigraphy from a limited set of observations. The StratMod Group at the University of South Carolina has developed a software package called Sedpak that uses an empirical approach for stratigraphic prediction via graphic simulation.

Sedpak has been used by oil companies to analyze the stratigraphic successions of basins in many regions of the world. It enables large amounts of data to be handled quickly and complex outcomes to be stored efficiently. Moreover, the assumptions driving the simulation can be systematically assessed in terms of validity and completeness.

Sedpak has a strong sequence-stratigraphic flavor, resembling in that sense the simulation procedures of Lawrence et al.1 Following the approach pioneered by Van Siclen,2 it recognizes the effect of eustasy upon facies progradation and reciprocal sedimentation.

Sedpak incorporates the rules of modern day sequence stratigraphy and its tie to high quality, multichanneled seismic data.3 This sequence-stratigraphic approach to the history of the fill of a basin emphasizes that basinal accumulations are responses to a complex interplay of processes.

This complexity is the reason for the use of large scale computer simulations. The output of such a simulation can be matched to an interpreted stratigraphic section by iteratively changing the input parameters, including the rate of sediment accumulation, crustal dynamics, and eustasy.

Sedpak starts with a hole to be filled; i.e., "accommodation."4 This space has an area and a shape and is related to the earlier history of the basin. It is filled by sediment whose constantly changing geometry reflects the sum of sedimentation and changes in water depth.

Sea level, sediment compaction, and crustal dynamics control the space to be filled. The predicted stratigraphy of the simulation can be checked against the control of the seismic and the measured stratigraphic sections.

SEDPAK SIMULATION

The Sedpak simulation is being used by petroleum exploration and production companies and by academia for stratigraphic interpretations of seismic sections.

Sedpak reproduces the interfingering geometry of clastic and carbonate sediments in hydrocarbon plays and oil fields and the basin fill in their vicinity. It identifies stratal elements with respect to time, tracking the evolution of sedimentary geometry (be it carbonate, sandstone, shale, or a mixture of the three) as a function of rate of sediment accumulation, subsidence behavior, and sea level.

Sediment geometries are plotted as they are computed, so the results can be viewed immediately (Fig. 1). Then, based upon these observations, parameters can be repeatedly changed and the program rerun until the resultant geometry is satisfactory.5

Simulations can provide a feeling for the geometric relationships of reservoir facies and aid in field development. In a wildcat setting (where control may consist of a few critical seismic lines and possibly one or two stratigraphic test wells), the simulation can be used to predict the sedimentary relationships on a seismic section while honoring the ages and lithofacies picks on this section.

On the basis of the simulation the magnitudes of the processes responsible for the geometry seen on the seismic are now known, being values of the parameters needed to recreate the geometries seen on the interpreted seismic sections.

The simulation can be used to identify potential migration pathways, reservoir, source rock, and seal geometries. Because it stores lithological and density data, Sedpak can be used to create synthetic seismic traces and to test lithofacies predictions, These same data can be used also as input to fluid flow and thermal models of the simulated section.

HYPOTHESIS TESTING

Simulations can be viewed as a medium for testing hypotheses that attempt to explain the geometries seen in stratigraphic sequences in terms of sea level variation, sediment accumulation, and tectonic movement.

Sedpak offers the researcher an opportunity to investigate the sensitivity of the evolution of a particular lithofacies model to varying sea level or to use different types of tectonic models and their effect upon simulation geometry. The user has the opportunity to evaluate whether one particular hypothesis is more reasonable than another. The sensitivity of a given play can be assessed. If a favorable exploration setting can only be produced from a very narrow set of assumptions then those assumptions must be very carefully evaluated. If on the other hand a significant stratigraphic geometry persists over a wider range of assumptions, this might be considered a safer prospect.

Sedpak models the effects of crustal rigidity or thermal subsidence, the latter determined from burial and crustal subsidence curves derived from seismic data or modeled sources. These values of vertical movement may be input directly to the simulation, just as a prescribed sea level curve is entered.

CONSISTENCY OF PARAMETERS

Another use of Sedpak is to test the consistency of parameters thought to be responsible for a particular geometry.

A common criticism of simulations is that similar synthetic stratigraphic records may be produced by sets of input parameters whose values are radically different. The accommodation filled by sediment is a product of eustatic movement, rates of sedimentation, and tectonic behavior.4 A variety of different values for these parameters might produce the same geometric response.

We cannot, however, expect tectonic movement or eustatic excursions to exceed certain rates or sizes or sedimentation to exceed certain rates and amounts. As a result, we can place upper bounds on these parameters and see whether we can produce realistic geometries using more than one set of parameter values for those bounds.

In fact, so many features commonly present in a seismic line produce constraints that it is extremely difficult (often impossible) to come up with more than one set of acceptable values (within limits) for parameters when basin stratigraphy is simulated at the scale of a formation.

EXAMPLE SIMULATION RUNS

Sedpak has been used for a variety of different depositional settings.

These include clastic to carbonate to mixed environments, including the Cretaceous/Tertiary of the South Carolina Coastal Plain, the Plio-Pleistocene of the Gulf of Mexico, the Neogene of the Canterbury basin of New Zealand, the Cretaceous of the Baltimore Canyon, and the Triassic Dolomite Alps of northern Italy.

S. CAROLINA COASTAL PLAIN

A well log cross section (E-E' of Colquhoun 6) stretching from the Fall Line to the Atlantic Ocean was modeled with Sedpak (Fig. 2). This illustrates how the simulation handles a predominantly clastic setting with occasional downdip carbonates, which match the geometries and thicknesses of each formation and the unconformities which occur between formations.

This simulation requires definition of accumulation modes. For instance, a "clastic damping exponent" reduces the relative carbonate accumulation as a function of detrital accumulation rate. The "lagoonal damping" factor varies from zero to unity and represents the reduction of carbonate accumulation expected to the lee of a reef. We define erosional events by setting the "basin surface" parameter to match the topography of the unconformity.

At the base, this geologic section consists of an updip lower delta plain facies and a downdip carbonate shelf facies. Two limestone units representing a deep carbonate shelf facies are found above them (Fig. 1).

The unconformity between the Piedmont crystalline basement and the Upper Cretaceous Middendorf formation was used as the initial basin surface. The updip and downdip thicknesses were measured from the well log cross section and used to estimate sedimentation rates for sand and shale.

The updip clastic and downdip carbonate facies were modeled with Sedpak by setting the sedimentation rate higher for shale than for sand and allowing the shale to be transported further into the basin than sand. The result is that the coarser-grained sand accumulates updip, while the finer-grained shale is deposited downdip, matching the facies observed in cross section. In order to allow only downdip accumulation of carbonate, lagoonal damping was set to zero, and the clastic damping exponent was turned on.

As clastic input increases, the effect of the damping factor increases, so the rate of carbonate accumulation decreases. Clastic damping only affects the area on the shelf considered to be lagoonal; i.e., it is the area behind a marginal buildup.

Sedpak can create multiple carbonate depth-rate curves, allowing the user to vary the carbonate accumulation rate through time. The carbonate accumulation rate curve is input as a rate in meters per kiloyears vs. depth. Thus, the user can control carbonate accumulation both laterally and vertically. In this way, carbonates and clastics can be balanced appropriately.

The Eocene Santee and Ocala limestones are composed totally of carbonate. To simulate their deposition, clastic supply was set to zero so that carbonate was the only lithology deposited at this time. The subsequent overlying formations consisted of an updip deep siliciclastic shelf facies and a downdip deep carbonate shelf facies similar to the Lumbee and Black Mingo Groups.

For deposition during this time interval, clastics were turned on again in the simulation with much more shale than sand being deposited. The shale represents the fine-grained material present on a deep siliciclastic shelf. The clastic damping exponent reduced carbonate accumulation just enough to allow the downdip deepwater carbonate accumulation.

Unconformities separate all of the formations in the section; these were simulated by defining the erosional planes at appropriate times throughout the simulation. For example, after a formation had been completely deposited (Fig. 3), the simulation was paused at the time step when the unconformity was to occur, and the erosional plane was estimated directly from the screen by recording location-depth pairs across the basin.

Sedpak linearly interpolates between these specified locations to create the erosional plane, so it was necessary to specify only a few vertices for the plane. The simulation was then rerun to view the results of the newly created erosional plane (Fig. 4).

Modifications were made iteratively until the erosional surface matched the stratigraphy. All sediment above the erosional plane was removed from the plane of section; an isostatic adjustment followed. Except for sediment loading, the sediment below the erosional surface is no longer accounted for by Sedpak.

In the time step immediately following creation of the unconformity, sedimentation resumes, and the next formation is deposited. This continues to the time of the next erosional unconformity. Geometries are easily mimicked when a unit is bounded by unconformities because the erosional surface can be created to reflect the topography of the upper boundary of the underlying unit.

Complicating this section is the Garner-Edisto Island Fault, which is located on the right side of section E-E'. Sedpak handles faulting by allowing different rates of vertical movement at adjacent locations to be prescribed.

Subsidence in the section was handled by specifying subsidence curves for the extreme left and extreme right of the section. The curve at the extreme right was set to subside at a slightly faster rate than that on the left to simulate the more rapid subsidence due to greater sediment load.

To model the fault in section E-E', two more subsidence curves were created: one on either side of the fault. Because the right side was downthrown, both the subsidence curve just to the right of the fault and the curve at the extreme right of the section were set to subside faster than the curve just to the left of the fault and the curve at the extreme left of the section during the time when the fault was active.

The cross section suggests the fault was active during the deposition of the Santee limestone in the Lutetian. Prior to and subsequent to the faulting period, subsidence rates for the curves on either side of the fault were identical to that of the extreme right side of the section. The result is a fault which affects all units up to and including the Santee limestone (Fig. 5).

LOUISIANA OFFSHORE

This section illustrates how salt diapirs may be handled using Sedpak." In this case, the salt intrusions seen on the seismic section (Fig. 6) are assumed to have remained at a constant depth, while the rest of the section subsided around them.

This example also shows Sedpak's ability to model facies changes, as sand was specified to be deposited during the lowstands of sea level, and shale was specified to be deposited during the highstands (Fig. 7).

The zoom capability of Sedpak enables viewing the crest of the salt dome (Fig. 8). This allows a better look at the actual geometries displayed in the simulation output.

CANTERBURY BASIN

Similar in appearance to the South Carolina cross section and the Louisiana offshore, this example (Fig. 9) illustrates how high rates of sedimentation can disguise an eustatic signal.9

10 It also demonstrates Sedpak's ability to recreate not only geometries but the facies changes as well (Fig. 10).

Here, progradation of the margin is occurring as the rate of sedimentation exceeds the accommodation created by tectonic subsidence and eustasy (Fig. 11). A low bypass angle causes some of the sediment to be transported downslope and deposited at the basin margin.

BALTIMORE CANYON

This example is from the Atlantic margin (Fig. 12)." It illustrates how a mixed carbonate (reef)-siliciclastic margin can be modeled.

A reef is considered by Sedpak to be the area of sea floor at which the carbonate depositional surface reaches sea level close to deep water. Behind this buildup, a lagoon may be created by damping deposition in the lagoon.

In this case, the lagoonal damping function of Sedpak allowed the Jurassic-Lower Cretaceous reef of this section to build up on the margin until a drowning event, caused by a rapid sea level rise, initiated the failure of the reef. The percentage of carbonate reef debris deposited in the lagoon and the distance of transport seaward for talus and turbidites to the sea were varied to produce the final geometry (Fig. 13).

The simulation output for this run is displayed in both basin/sea level mode and time-depth-elevation/sequence mode (Fig. 14). The time-depth-elevation plot shows two things: 1) the basin itself with the sequences colored different colors at the top, and 2) the modified burial history plot as a time-depth-elevation plot." The latter was produced for the pseudo well at column 70 (shown by the vertical line on the top plot). The differently colored sequences are defined by entering times for the sequence boundaries (Fig. 15).

PICCO DI VALLANDRO

This example illustrates how carbonate accumulation rate vs. depth curves can be varied to create progradational, aggradational, and backstepping carbonate margins.13

Like the Baltimore Canyon, Picco di Vallandro is a mixed margin (Fig. 16) in which carbonate interfingers with shale." This affects the resulting carbonate geometries (Fig. 17).

In the event of a relative sea level rise (i.e., increasing accommodation), one of three geometries may result: 1) progradation produced when a high carbonate accumulation rate exists throughout the water column so that carbonate accumulation exceeds accommodation (Fig. 18); 2) aggradation produced when the carbonate accumulation rate is high in shallow water and low in deep water and the margin keeps up with a relative sea level rise; and 3) a backstepping geometry produced when there is a low carbonate accumulation rate throughout the water column and the margin fails to keep up with a relative sea level rise.

SYSTEM DESIGN

Sedpak is a collection of integrated computer programs written in the "C" programming language, under the MIT X Window System and Motif, for a UNIX workstation environment.5 Versions for the Hewlett Packard, IBM RS6000, Sun Sparcstation, DECstation, and Silicon Graphics Iris have been installed in the field.

Porting Sedpak to a new environment requires few, if any, changes to the source code. The system design philosophy places importance on user interaction with the simulation. The user interface was designed for use by seismic interpreters, particularly persons with a sequence stratigraphic background. With this orientation, the user can easily test hypotheses and acquire rapid feedback comparable to that obtained from conventional seismic workstations.

At present, Sedpak is a two-dimensional simulation system. User suggestions have indicated that the next major stage in the evolution of Sedpak is the development of a pseudo three-dimensional model wherein a set of two-dimensional sections interact.

Ultimately, a true three-dimensional simulation would be desirable. This will require much larger amounts of empirical data, such as three dimensional seismic sections and, consequently, more powerful computing resources.

CONCLUSIONS

Sedpak is a graphical simulation program designed to aid in the analysis and classification of stratigraphic sequences derived from seismic and well data. It permits the extrapolation of facies and structure away from control, embodying the concepts and processes of sequence stratigraphy, for a wide variety of basins. It is designed to be a tool to increase the efficiency and extend the capabilities of professional sequence stratigraphers.

ACKNOWLEDGMENT

The authors wish to thank the University of South Carolina Stratigraphic Modeling Industrial Association, which has financially supported this research. Members include Japanese National Oil Corp., Texaco Inc., Mobil Corp., Den norske stats oljeselskap AS (Statoil), and Conoco Inc.

REFERENCES

1. Lawrence, D. T., Doyle, M,, and Aigner, T., "Stratigraphic Simulations of Sedimentary Basins: Concepts and Calibrations," AAPG Bull., Vol. 74, 1990, pp. 273-296.

2. Van Siclen, D. C., "Depositional topography-examples and theory," AAPG Bull., Vol. 42, 1958, pp. 1897-1913.

3. Van Wagoner, J. C., Mitchum, R.M., Posamentier, H.W., and Vail, P.R., 1987, "Key Definitions of Seismic Stratigraphy," AAPG Atlas of Seismic Stratigraphy, Vol. 1, 1987, pp. 11-14.

4. Jervey, M. T., "Quantitative Geological Modeling of Siliciclastic Rock Sequences and their Seismic Expression," in Wilgus, C. K., Hastings, B., Kendall, C. G. St. C., Posamentier, H., Ross, C., and Van Wagoner, J.C., eds., "Sea-Level changes - An integrated approach," SEPM Pub. 42, 1989, pp. 47-70.

5. Strobel, J., Cannon, R., Kendall, G. St. C,, Biswas, G., and Bezdek, J., "Interactive (SedPak) simulation of clastic and carbonate sediments in shelf to basin settings," Comput. Geosci., Vol. 15, 1989, pp. 1279-1290.

6. Colquhoun, D. J., 1990, "Surface and Subsurface Formations, Structure and Groundwater Aquifers of the South Carolina Coastal Plain," Dept. of Geology, University of South Carolina, 1990, 81 P.

7. Kendall, Christopher G. St. C., and Lowrie, Allen, "Simulation modeling of stratigraphic sequences along the Louisiana offshore," Transactions, Gulf Coast Association of Geological Societies, Vol. XL, 1990, pp. 355-362

8. Whittle, Gregory, Kendall, Christopher G. St. C., Fogerty, Michael A., and Lowrie, Allen, 1992, "Paleotectonic restoration and simulation applicable to the Louisiana offshore," Transactions, Gcags, Vol. XLII, 1992, pp. 429-444.

9. Fulthorpe, C. S., "Geological controls on seismic sequence resolution," Geology, Vol. 19, 1991, pp. 61-65.

10. Fulthorpe, C. S., and Carter, R. M., 1991, "Continental-shelf progradation by sediment-drift accretion," Geological Society of America Bull., Vol. 103, 1991, pp. 300-309.

11. Ehrlich, R. N., Maher, K. P., Hummel, G. A., Benson, D. C., Kastritis, G. J., Linder, H. D., Hoar, R. S., and Neeley, D. H, "Baltimore Canyon Trough, Mid-Atlantic OCS: Seismic Stratigraphy of Shell/Amoco/Sun Wells," Atlas of Seismic Stratigraphy, AAPG Studies in Geology, Bally, A. W. (ed.), Vol. 2,1988, pp. 51-65.

12. Lorenz, J. C., and Finley, S. J., "Fracturing of Mesaverde Reservoirs in the Piceance Basin, Colorado," AAPG Bull., Vol. 75, 1991, pp. 1738-1757.

13. Whittle, G., Moore, P., Kendall, C. G. St. C., Cannon, R., and Biddle, K., "Simulation of the Triassic Pico Vallandro Section of the Dolomite Alps," program of annual meeting of AAPG, AAPG Bull., Vol. 76, 1992, P. 777.

14. Biddle, K. T., Schlager, W., Rudolph, K. W., and Bush, T. L., "Seismic Model of a Progradational Carbonate Platform, Picco di Vallandro, the Dolomites, Northern Italy," AAPG Bull., Vol. 76, 1992, pp. 14-30.

Copyright 1993 Oil & Gas Journal. All Rights Reserved.