Reservoir Simulation of Cyclic Steam Stimulation in the Cold Lake Oil Sands

Abstract

Summary Steam injectivity during cyclic steam stimulation (CSS) at Cold Lake can beachieved only by injecting at pressures high enough to fail the formationmechanically. The reservoir also exhibits water/oil relative permeabilityhysteresis. This paper describes enhancements made to a permeabilityhysteresis. This paper describes enhancements made to a thermal reservoirsimulator to incorporate these Cold Lake physics. The geomechanical modelallows both localized and global reservoir deformation (dilation andhistory-dependent recompaction). This representation allows the simulator tomatch injection and production pressures that are otherwise difficult toreproduce. The history-dependent relative permeability hysteresis modelcalculates gridblock relative permeabilities permeability hysteresis modelcalculates gridblock relative permeabilities that always lie on or betweeninput imbibition and drainage bounding curves. The model makes it possible touse laboratory-derived relative permeabilities in a simulation and still matchfield WOR's. permeabilities in a simulation and still match field WOR's. Introduction The Cold Lake oil-sand deposit in northern Alberta contains over 40 × 10(9)m3 of bitumen in place. The extremely high oil viscosity and low native watermobility in this reservoir result in negligible initial injectivity orproductivity. Steam injection at commercial rates requires injection pressureshigh enough to cause both localized fracturing and widespread PV increases inthe formation. The resulting complex geomechanical behavior determines theinitial steam flow paths and provides significant drive energy during CSSoperations. The reservoir also exhibits relative permeability hysteresis. Thesephenomena must be represented adequately in a reservoir simulator phenomenamust be represented adequately in a reservoir simulator to properly model Cold Lake reservoir performance during CSS. This paper describes the deformation, fracturing, and relative permeability hysteresis behavior in the Cold Lake Clearwater reservoir permeability hysteresis behavior in the Cold Lake Clearwater reservoir and the methods used by Esso Resources Canada Ltd. (ERCL)to include these phenomena in our simulators. The physics and modeling aspectsof each of these phenomena are described, and methods to determine values forthe additional input data required by the simulator are discussed. Examplesillustrate how the model enhancements improve the simulator matches of fieldobservations. The enhancements described have been incorporated into acommercially available thermal simulator and a thermal version of Exxon Production Research Co.'s MARS simulator. Although different in formulation, both are fully implicit models and give virtually identical results when usedto simulate the same problem. Reservoir and Operations Description ERCL's Cold Lake operations target the Clearwater formation, a near-shoredeltaic sand of Cretaceous Age at a depth of about 450 m. The sands are thick, often greater than 40 m, with a high net/gross thickness ratio. Porosity rangesfrom 30 to 35%, with oil saturations that average 70 % PV. At the initialreservoir temperature of 13 degrees C, the oil viscosity is about 100 Pas; oilviscosity decreases to 0.002 Pas at 250 degrees C. ERCL uses CSS at Cold Lake. Steam is injected at 225 m3/d for 30 to 60 days, followed by a productionperiod of 120 to 400 days, depending on cycle number. Average producing day oilrates decline from greater than 20 m3/d in the first cycle to about 7 m3/d inthe eighth cycle; WOR's increase from 1 to 2 to more than 4 during the sameperiod. Total daily production from almost 1,800 pilot and commercial wellsaveraged 14 000 m3/d in 1989. More detailed descriptions of pilot andcommercial operations are available. Deformation Modeling High-pressure injection into a tar-sand reservoir causes fracturing (whichis discussed later) and widespread PV changes (deformation) in the formation. Deformation involves both dilation and subsequent recompaction. Details of themodeling of these phenomena are discussed below. Dilation. Two observations lead to the conclusion that significant dilationof the reservoir occurs during steam injection at Cold Lake. First, surfacebenchmark arrays at Cold Lake have been used to measure surface uplifts asgreat as 45 cm during injection, far larger than can be attributed to thermalexpansion or tensile fracturing of the formation. Second, steam injectivity at Cold Lake is greater than might be expected from native reservoir properties. Acommon problem when simulating steam injection into tar sands is matchingrelatively high observed injectivities when reasonable fracture lengths androck compressibilities are used. To match observed injectivities, mostpublished simulations of Cold Lake CSS have used a rock compressibility one ortwo orders of magnitude larger than the actual value of about 1.0 GPa (-1) atreservoir conditions. Coats et al., who used this high-compressibility methodfor a California heavy-oil reservoir, called it the "spongy-rock"approach. The only published Cold Lake simulation that did not use anenhanced-compressibility approach required extremely long fractures to matchthe observed injectivity. The major problem with the spongy-rock approach isthat it predicts a steady increase in injection pressure with time, whereaspredicts a steady increase in injection pressure with time, whereas the fieldpressures increase rapidly initially and then level off for most of the cycle. This is illustrated in Fig. 1, which shows typical field and simulation resultsfor first-cycle injection. The spongy-rock simulation used a rockcompressibility of 35 GPa(-1). Surface heave and the high observed injectivitycan be explained by two mechanisms that increase porosity. First, oil sandsdemonstrate nonlinear compressibility behavior. Fig. 2 shows how thecompressibility increases dramatically as the effective stress nears zero. Inaddition, shear failure can be induced in the formation at sufficiently loweffective stresses in the presence of anisotropic stresses. Shear failurecaused by increasing pore pressure results in dilation of the pore system.pressure results in dilation of the pore system. As steam is injected into arelatively incompressible reservoir, pore pressure increases and the effectivestress decreases. At pore pore pressure increases and the effective stressdecreases. At pore pressures corresponding to low effective stresses, theformation pressures corresponding to low effective stresses, the formationcompressibility increases by about two orders of magnitude, and shear failuremay also occur. Both these phenomena result in dilation, with rapid increasesin porosity and permeability. This greatly increases the steam injectivity, andinjection pressures increase slowly thereafter. The dilation is reflected byuplift of the ground surface above the injection location. We model thisphenomenon by specifying a dilation pressure, PD, below which behavior iselastic and a low original PD, below which behavior is elastic and a loworiginal compressibility value is used. Above the dilation pressure, a higherdilation compressibility is used, enabling the porosity to increase rapidlywith pressure, as shown by the line labeled "dilation" in Fig. 3. Amaximum porosity is also specified, above which further dilation is notpermitted and the compressibility reverts to a low value. Fig. 1 shows theimproved match of injection pressure that results from use of this dilationmodel.