Role of Small-Scale Variations in Water Saturation in Optimization of Steamflood Heavy-Oil Recovery in the Midway-Sunset Field, California

摘要

In recovery of heavy oil by Steamflood, efficiencies can be realized by limiting the placement of steam to the portions of the reservoir with highest oil saturations, thus reducing the disproportionate loss of heat to connate water. This optimization strategy requires knowledge of water-saturation (S_w) distributions within the heavy-oil reservoir at the scale of operations. Application of this strategy has contributed to the successful reactivation of the shut-in Pru Fee property in the Midway-Sunset field 1 mile west of Taft, a U.S. Dept. of Energy (DOE) Class 3 oil-technology demonstration project. The 40 new wells drilled and logged within the 40-acre Pru Fee property, together with a single continuous core, have permitted 3D mapping of S_w within the 250- to 350-ft-thick pay zone in the Monarch sand reservoir. Water saturations are observed to vary at three different scales: 1. A systematic vertical reduction in S_w through a 100- to 150-ft interval above the oil/water contact (OWC) caused by dominant capillary influence where the buoyancy effects are diminished by the low-density contrast of the 13° API oil and formation water. 2. Lateral variations in S_w on the scale of 10 to 100 ft caused principally by prior oil production from the reservoir, but modified by its internal stratigraphic architecture. 3. Bed-to-bed variations in S_w on the order of a few feet or less constrained by grain-size-controlled differences in porosity/ permeability in these crudely graded sands. Overall production efficiency in the Steamflood has been improved by limiting steam injection to the upper one-half to two-thirds of the pay zone, where S_w is lowest. Knowledge of the lateral variations in S_w has permitted more accurate appraisal of the effectiveness of individual producers and nine-spot injector/ producer arrays. The recognition of the bed-to-bed variations has permitted a better petrophysical model for calibrating S_w calculated from logs.