KENDRICK, J.W., Shell Deepwater Development, Inc., New Orleans, LA, USA

Abstract: Turbidite Reservoir Architecture in the Gulf Of Mexico- Insights from Field Development

Deepwater oil and gas exploration has been premised on the ability of turbidite reservoirs to deliver sufficient hydrocarbon volumes at high enough rates per well to justify the large capital investments required for field development. Early well performance at deepwater fields in the northern Gulf of Mexico have met or surpassed early expectations. Increasingly, efforts are being directed toward questions about fine-scale reservoir architecture that can be used to achieve efficient appraisal, optimal development-well placement, and improved sweep efficiency of turbidite discoveries.

The development and early production of several fields is making available new geological, fluid, and pressure data with which to validate, or modify, existing models of reservoir architecture. This paper summarizes the observations and learnings from several developed turbidite reservoirs. The reservoirs represent examples from each of several major reservoir architecture types, including layered/amalgamated sheet sands (Auger field); channel-levee deposits (Tahoe Field); and amalgamated channel complexes (Ram/Powell field).

The "S" sand is the major producing reservoir at the Auger field. Previous authors (McGee et al., 1993) have described the reservoir as a layered- and amalgamated- sheet sand deposited within a structurally confined salt minibasin. Production performance reveals that the pressure of the reservoir has declined more slowly than expected, indicating greater lateral continuity and aquifer support than originally expected and implying the existence of sands that extend across the entire minibasin. Sequential production logs of a downdip well indicate that the reservoir is not watering out in a uniform, bottom-to-top pattern, however (Figure 1). The occurrence of water-bearing zones above oil-bearing ones indicates the presence of effective internal seals within the reservoir that, like the sands, are laterally extensive.

Thin-bedded reservoirs, such as those that produce gas at prospect Tahoe, are interpreted as the overbank portion of channel-levee complexes. The thin-bed reservoirs comprise millimeter to centimeter thick sand beds separated by mudstones of similar thickness. Net sand percentage of such reservoirs ranges from 30-55 percent, and log shapes exhibit an upward-thinning trend in sand distribution. Early concern about the lateral extent of such thin-bedded sands led to pre-development production testing (Shew et al., 1993), which indicated that the beds were laterally continuous over areas exceeding several hundred acres.

Subsequent drilling and production provide additional, and somewhat surprising, insights into the reservoir architecture of the channel-levee system. Pressure profiles in newly drilled wells exhibit stratigraphically varying pressure depletion. Sands in the upper levee facies appear to be laterally continuous, even across the channel. Sands in the lower portion of the reservoir, however, are much less connected. The contact between channel and levee facies has been interpreted as a barrier to fluid flow because the levee facies on either side of the channel exhibit different hydrocarbon geochemistry and different fluid contacts. Evidence of some pressure depletion across the channel, however, indicates that early views may be too simplistic. Channel-levee architecture appears to evolve from localized to extensive overbank sedimentation. The thicker beds, which occur at the base of the levee facies, may not be the more continuous.

Most problematic, from a development perspective, are channel-fill reservoirs, such as the.Miocene "N" sand at Ram/Powell.The reservoir appears to comprise multiple, partially stacked, laterally accreting sand units deposited within a pre-existing scour depression; they exhibit large thickness differences, even over short distances. Paradoxically, all wells indicate some degree of pressure communication, but the occurrence of numerous perched water levels implies the existence of internal reservoir compartments. Presently, the "N" sand is interpreted to comprise two episodes of erosion and fill, with the later episode removing portions of the older deposit. Partial amalgamation of the two sand members allows fluid communication to occur, but the lateral heterogeneity in sand thickness creates structurally low closures in which water is trapped (Figure 2).

Our observations illustrate a spectrum of turbidite reservoir architectures and their influence on production performance. The occurrence and widespread lateral extent of turbidite sheets is confirmed by production performance of reservoirs like the one at Auger. Similarly, levee/overbank sands, despite their thin-bedded character, exhibit evidence, albeit less complete, of high lateral continuity. Issues remain concerning the evolution and internal architecture of these deposits, however. The ability to develop channel-fill complexes effectively and to achieve high per-well ultimates remains problematic and requires tools with the resolution to differentiate episodes of scour and fill.

The elements important to reservoir characterization are sometimes below the limits of seismic resolution. It is valuable therefore to integrate the diverse information that comes from development and production activities in order to develop the reservoir architecture models that will facilitate future planning and decision making.

AAPG Search and Discovery Article #90923@1999 International Conference and Exhibition, Birmingham, England