From Bureau of Economic Geology, The University of Texas at Austin (
For more information, please contact the author.

E Exitep 2003, Encounter and Exhibition International of theOil Industry, Veracruz, Mexico, February 1619, 2003

3-D Geocellular Modeling of Gulf Coast Miocene Water-Drive Gas Reservoirs:
Example from StarFak Field, Louisiana

Mark H. Holtz, Larry W. Lake, Hongliu Zeng, Paul Knox, and Mike De Angelo


Because of a complex interplay between faulting and stratigraphy, it is difficult to determine original and remaining gas volumes in produced offshore Miocene Northern Gulf of Mexico gas reservoirs. Faults can be numerous, and their throws can often be too small to detect using conventional 3D seismic analysis. They can also be sealing, forming reservoir-flow compartments, or nonsealing faults that die-out within the reservoir. The fluvial-deltaic depositional environment, along with unconformable valley fills, multiplies this reservoir complexity because it produces both flow barriers and widely varying petrophysical properties. These complexities result in multiple gas-water contacts and a wide range of residual gas saturation to water influx, thus rendering gas volume calculations difficult. The hypothesis of this work is that this difficulty can be overcome by constructing reservoir models that integrate a seismically derived 3-D v-shale volume with an interrelated set of petrophysical property models. An example of such a model is given for the Starfak T1 reservoir.

The 3-D geocelluar reservoir model of the Starfak T1 reservoir is determined by integrating geology, geophysics, and engineering. We identify complex faulting by comparing seismic amplitude maps, stratal slices, and seismic lines with engineering fluid-contact and water-influx maps, and pressure-history plots. Near-subseismic faults are then detectable from the spatial alignment of subtle amplitude-map lineations and seismic-trace breaks along with variable fluid contacts and pressure depletion. At the same time we determine the compartmentalizing nature of depositional facies patterns by analyzing seismic stratal slices, wireline logs, geologic sandstone maps, and initial fluid levels. A 3-D seismic v-shale volume is combined with a wireline-derived, 3-D v-shale volume by first creating an accurate, interrelated set of petrophysical equations that include residual gas saturation. Secondly, a 3-D, seismically derived v-shale volume based on neural-network modeling of seismic was optimally averaged with a small-scale 3-D shale volume constructed from well data and geologic depositional facies patterns. The optimization was based on obtaining both the most geologically likely 3-D v-shale distribution and an initial and residual gas volume that resulted in reasonable recovery efficiency.

The resultant 3-D model of the Starfak T1 reservoir captures both the geologic and petrophysical character of this Miocene age reservoir. The resulting reservoir architecture contains separate reservoir compartments delineated by a sealing unconformity and sealing faults, and differing gas-water contacts. Petrophysical property equations interrelate v-shale, porosity, permeability, initial water saturation, capillary pressure, and residual gas saturation. The optimal model was obtained by a 70 %-30 % weighted average influence of log derived vs 3-D seismic derived v-shale. This inclusive technique, results in a model that both follows the geological depositional patterns interpreted in the seismic and well logs and produces volumes of the initial and remaining gas that give realistic recovery efficiency for a water drive gas reservoir.