The Impact of Geologic Reservoir Characterization on the Flow Unit Modeling at the Kern River Field, California, USA
Elliott P. Ginger, William R. Almon, Susan A. Longacre, Cynthia A. Huggins, 1995. "The Impact of Geologic Reservoir Characterization on the Flow Unit Modeling at the Kern River Field, California, USA", Hydrocarbon Reservoir Characterization: Geologic Framework and Flow Unit Modeling, Emily L. Stoudt, Paul M. Harris
Download citation file:
A multidisciplinary Kern River Geology Team was charged with developing geologically-based reservoir models that improve understanding of the steam flood process and the encroachment of cool water in the Kern River field. The focus of the effort was on investigating reservoir layers, continuity and boundary conditions, as they relate to movement of steam, oil and water through the system. This paper presents the results of a pilot study of steam movement in a localized area in Kern River field known as Project D-159.
Sands of the Kern River Formation that serve as reservoir to hydrocarbons in the field were deposited in a set of braided river channels that repeatedly crossed the field area from late Miocene through Pleistocene time. Channel sand deposition within the study area was not continuous through time, but rather was episodic. At times the depositional system changed significantly, such that the channels shifted to another area and the only deposition in the study area was a succession of floodplain silts. Where these silt units are not breached by subsequent rejuvenated river channel systems, they are barriers between successive large sand reservoirs.
A spatial “description” of the sand and silt components in the reservoir system was developed using both conventional geologic methods and geostatistical methods. The conventional geologic methods include stratigraphic cross-sections, developing a layering pattern within the reservoir, and expressing that layering pattern on cross-sections and maps. The layering system was based on a set of “rules” that focused on the unconformable bases of the major sand episodes and that included the overlying silts with the sand body with which it was genetically linked.
Using GRIDSTAT - proprietary software developed at Texaco - a series of rapidly constructed geostatistical cross-sections were made that closely approximated the conventional cross-sections. A rigorous statistical comparison of the conventional and geostatistical cross-sections indicates that mean rates of differences in lithology assignment on the GRIDSTAT sections, relative to the conventional cross-sections, are less than 3 percent. The average difference in lithology assignment between GRIDSTAT and traditional techniques is approximately the same as would be present between cross-sections generated by different geologists. However, on GRIDSTAT cross-sections approximately 5 percent of the wells have lithology assignment error rates that are in excess of 10 percent. Such difference rates have the potential to produce significant errors in determining sand thickness and are unacceptably high; geological intervention is required.
Average total correlation difference between GRIDSTAT sections and traditional cross-sections is 5.5 percent. There is a tendency for GRIDSTAT to fail to find all geologic correlations; this tendency is significant at the 0.95 confidence level.
Stratamodel’s program SGM™ was used to model the 3-D inter-well distribution of reservoir properties in Project D-159. A special layering style was used in order to best represent the internal stratigraphic character of the reservoir. A cell thickness of five feet was chosen within each sequence in order to minimize the vertical averaging of well data while keeping the size of the model within reason. Two types of well data have been modeled at D-159 to date—short-normal resistivity (from fifty-four wells) and temperature measurements (monthly profiles from eleven temperature observation (TO) wells for January through November, 1992). Geobodies and model operations are used in evaluating and tracking the rate of heat growth or dissipation. If temperature geobodies are built in a time sequence, growth in the volume of successive geobodies based on specific temperatures would allow calculation of the heat added to the reservoir; while shrinkage would indicate heat dissipation.
Figures & Tables
This collection of papers presents documentation for (1) approaches to be taken in developing a geologic framework for explaining layering, heterogeneity, and compartmentalization of a reservoir; (2) the value of outcrop data in improving understanding of reservoir performance; (3) methods for integrating, analyzing, and displaying geologic, petrophysical rock property, and engineering data to be used during field evaluation, management, and simulation; (4) geostatistical approaches that are being used to characterize the spatial distribution of reservoir properties and augment geologic descriptions, and (5) methods of displaying quantitative models of reservoir properties and reservoir simulation in three dimensions.