Effects of Steam Quality and Injection Rate on Steamflood Performance
- K.C. Hong (Chevron U.S.A. Production Co.)
- Document ID
- Society of Petroleum Engineers
- SPE Reservoir Engineering
- Publication Date
- November 1994
- Document Type
- Journal Paper
- 290 - 296
- 1994. Society of Petroleum Engineers
- 5.1.1 Exploration, Development, Structural Geology, 5.5 Reservoir Simulation, 4.1.5 Processing Equipment, 5.1.5 Geologic Modeling, 1.2.3 Rock properties, 5.3.4 Reduction of Residual Oil Saturation, 5.2.1 Phase Behavior and PVT Measurements, 5.2.2 Fluid Modeling, Equations of State, 5.4.6 Thermal Methods, 2.4.3 Sand/Solids Control, 4.1.2 Separation and Treating, 5.1 Reservoir Characterisation
- 1 in the last 30 days
- 931 since 2007
- Show more detail
- View rights & permissions
|SPE Member Price:||USD 12.00|
|SPE Non-Member Price:||USD 35.00|
A comprehensive numerical simulation study showed that for a typicalheavy-oil reservoir, steamflood performance improves monotonically withincreases in steam quality and injection rate. Hence, for this reservoir, theoptimum quality and injection rate are the highest values that wereinvestigated. In general, however, optimum steam conditions depend on reservoirtype and operating mode (e.g., line drive or pattern flood). For mostsituations, published "rule-of-thumb" steam conditions are not optimum.Therefore, we recommend that optimum steam conditions for a specific reservoirbe determined through economic comparison of predicted oil recoveries forranges of steam conditions, as demonstrated in this study.
Steamflooding involves injection of heat into a reservoir by means oftwo-phase steam. For the process to be effective, steam of sufficient qualitymust be injected at sufficient rates. However, the cost for generating steam ishigh, accounting for about one-half of all steamflood operating costs. Becauseof the high steam cost and the difficulty in obtaining generator permits,optimization of the use of injected steam is necessary.
Previous attempts to optimize steamflood performance involved the use ofnumerical or physical simulators or observed field data to determine optimumsteam quality and injection rate. This was done mainly for homogeneousreservoirs with simple geometries to avoid masking of the quality and injectionrate effects by effects of other factors in complex reservoirs. The optimumconditions so determined were applied as a rule of thumb to all situationsregardless of reservoir type or mode of operation. Currently, the most widelyaccepted rule-of-thumb values are an injection rate of 1.5 B/D cold-waterequivalent (CWE) per acre-foot of reservoir volume and 40% steam quality at thesandface.
Because reservoirs respond differently to steam injection, no single valueof steam injection rate or quality can be optimum for all reservoirs and allsteamflood situations. Thus, using a set of steamflood conditions that isoptimum for one reservoir may produce different, less-than-optimum results whenapplied to another reservoir. An understanding of how the optimum conditionsdiffer for different reservoirs clearly is needed.
A comprehensive numerical simulation study was conducted to investigate theeffects of steam quality and injection rate on steamflood performance for avariety of steamflood situations. This paper presents the simulation resultsand demonstrates how they are used to determine optimum steam quality andinjection rate for different steamflood situations.
Laboratory radial-flow model and analytical model studies showed thatsteamflood performance improves monotonically with increasing injection rate.On the basis of a laboratory simulation of the steamdrive process, however, Maet al. concluded that an optimum steam injection rate exists. This wascorroborated by Bursell and Pittman, who found that a 1.5-B/(D-acre-ft)injection rate yielded the lowest steam/oil ratio for Kern River steamfloods. Areview of field projects by Farouq Ali and Meldau also confirmed an optimuminjection rate, showing that all successful projects were operated withinjection rates between 1.0 and 2.0 B/(D-acre-ft). The average, 1.5B/(D-acre-ft), thus became the rule of thumb.
Laboratory steamfloods by Doscher and Huang and Shen showed that oilrecovery increased with increasing steam quality. Gomaa et al. made a similarobservation through a numerical simulation study of a thick Monarch sand of theMidway-Sunset field. These studies indicated that steam quality should be ashigh as possible to maximize oil recovery. This thinking prevailed until 1980,when Gomaa presented a new simulation study that attempted to show the presenceof an optimum steam quality for a typical heavy-oil reservoir in the Kern Riverfield. When oil recovery is plotted vs. net heat injected, an intermediatesteam quality ( 40%) yielded the best production. While this finding is limitedto the particular reservoir investigated, this value became widely accepted andused for lack of other studies challenging it.
Chu observed that high-quality steam is favored for thin reservoirs whereaslow-quality steam is more desirable for thick reservoirs. Separately,Moughamian et al. studied the effect of steam quality as part of a linedrivesimulation study of a steeply dipping reservoir. They found that oil recoveryvs. time increased with increasing steam quality, as others had observed.However, when plotted vs. heat injected, oil recovery decreased with increasingsteam quality, which was contrary to the results obtained by other studies forno-dip reservoirs. The different results were attributed to the difference inreservoir orientation.
No single value of steam injection rate or quality can be optimum for allreservoirs and all steamflood situations because reservoirs respond differentlyto steam injection; therefore, the same set of values should not be used in allreservoirs.
Reservoir and Fluid Models
Reservoir models constructed and used for this study can be divided into twogroups: pattern flood and line drive. The first is used for studying steamfloodin no-dip reservoirs and the second for studying steamflood in dippingreservoirs.
Pattern Flood. The reservoir model was a 3D, 7x4x5 grid system representingone-eighth of a five-spot pattern (Fig. 1). The pattern area is 2.5 acres, andthe distance between the injector and producer is 233 ft. We also studied alarger 5-acre pattern where the distance between the injector and producer was330 ft. The reservoir has 75-ft gross thickness, which was divided into fiveequal communicating layers, each 15 ft thick. Steam was injected into the twobottom layers, and the producer was completed in all layers.
Table 1 shows reservoir parameters used in the simulation study. For thehomogeneous model, the reservoir was assumed to have uniform properties. Forstudying the effect of permeability variation, the horizontal permeability waseither decreased upward as in a fining-upward sequence of a channel sand ordecreased downward as in a coarsening-upward sequence of a bar sand. Table 1compares the resulting permeability distributions with those of the homogeneousmodel. Vertical permeability in each layer was assumed to be one-half thehorizontal permeability. Fig. 2 shows the water/oil and gas/liquid relativepermeability curves used for the simulation. Table 1 also givestemperature-dependent irreducible saturation and endpoint relative permeabilitydata.
The oil was assumed to be composed of two components: methane and a 14API-gravity dead oil with a molar mass of 400 kg/mol. A small amount ofmethane, 1.5 mol% in the oil phase, was used to create the initial gas cap inthe model. Ref. 13 gives other details of the reservoir and fluid models.
|File Size||615 KB||Number of Pages||7|