Fluid Flow and Sand Production in Heavy-Oil Reservoirs Under Solution-Gas Drive

Abstract

Summary. The production of heavy oil in Canada has led to a number of anomalous results, most of which have been excused as high-permeability channels resulting from sand production. The methods of soil mechanics predict gross formation failure resulting from high fluid compressibility, small cohesion, and high viscosity. Gross failure results in excellent productivity but reduced in-situ stress (and fracture stress). Solution-gas drive in these reservoirs involves simultaneous-mixture flow of gas as very tiny bubbles entrained in heavy oil. Stress, geometry, and permeability alteration resulting from matrix deformation combined with peculiar pressure-dependent multiphase-flow properties result in a new model of reservoir performance. A field observation of stress modification is discussed, as are the contributions of the four components discussed previously to the observed phenomena. Introduction A number of heavy-oil reservoirs are currently being produced under primary conditions around the world. Primary production in the Lloydminster, Alta., area has been under way for 40 years, and many anomalous behaviors have been noted. The oil gravities in the Lloydminster block are quite low, ranging from 1 to 0.93 g/cm [10 to 20 API]. The producing formations are compact, uncemented sands, and during early production considerable sand is produced by mechanisms modeled by soil-mechanics methods. The drive mechanism is generally believed to be solution-gas drive, and occasionally the suspected influx of natural water from below or from the flanks is credited with improved recovery. Because depleted pools show very low pressures and even watered-out wells generally have static pressures below the original aquifer pressure, a material-balance computation of the water and solution drive indices indicates that only solution-gas drive is of importance. Considerable water is produced during primary operations. As long as water production is low, quite high sand cuts can be tolerated by the production system. As water production increases, sand dropping out in the tubing and pump becomes a more severe problem and often leads to excessive operating costs and well suspension. Waterflooding has been carried out quite extensively with generally poor results, not unexpected from classic displacement calculations. Expected recoveries under primary depletion are in the 3 to 10% range, although much better recoveries have been achieved in some areas. Pressure-buildup analyses conducted throughout the Lloydminster area show in-situ kh/ values on the order of 100 mdm/mPas [300 md-ft/cp]. Measured viscosities of the produced stock-tank oils range from 200 to 35 000 mPas [200 to 35,000 cp]. Liquid-phase property (PVT) analyses show a decrease in viscosity as a result of the presence of dissolved methane, the reduction generally being around 50% at saturation conditions. In general, however, when substituted into kh/, the accepted values of u and h lead to k values on the order of tens of darcies, which is much higher than anything measured in the laboratory. The belief within the industry has been that sand production has caused an enhanced permeability or "negative" skin. Extensive sand production modifies the reservoir in the vicinity of the casing, but laboratory studies have shown that no amount of sand remolding or fines removal would increase the permeability to the extent apparently observed in pressure analysis. It is more likely that the laboratory-determined core permeabilities are overestimates of the undisturbed in-situ values, and it is most commonly observed that fines migration in the reservoir leads to reservoir plugging, not reservoir cleaning. Also, some engineers have believed for many years that the excellent primary performance of Lloydminster reservoirs, performance much in excess of the predictions of radial Darcy flow, was a result of the presence of horizontal holes, fractures, or other channels. Extensive sand production was correlated with better-than-average oil productivity, and the hypothesis of production-eroded "wormholes" was often quoted as the responsible mechanism. Many sudden failures in injection schemes (firefloods, waterfloods, and steamfloods) and in drilling and workover operations were also blamed on wormholes or channels. After all the anomalous observations were considered, it was seen that alternative hypotheses were required. Objectives. The objectives of this study included the following.1. To develop a model of primary production in the Lloydminster heavy-oil area and to be able to predict production, decline, recovery, pressure, and pressure-transient behavior.2. To explain the extensive sand production of the area and how and why sand production affects oil production.3. To explain sudden and rapid breakthroughs of injected fluids and chemical tracers and particulates in primary depleted areas. Hypotheses Flow of a Complex Fluid Within the Pore Space. The mobile fluid is a mixture of a true liquid, large molecular particles [asphaltenes as molecules with a molecular weight on the order of 1,000 (Fig. 1A) or micelles with a molecular weight on the order of 100,000), and gas bubbles of similar size (Fig. 1B). The fluid undergoes a phase transformation and has a high compressibility with time-dependent properties. Mechanical Failure of the Sand Matrix Around the Well Results in Sand Production. Failure begins with local plastic flow at perforations (Fig. 2A) and expands to general radial failure (Figs. 2B and 3a). The geometry of the failure is likely to be asymmetric relating to the native stress state (Fig. 3b). Bridging by the overburden is probable, and the sand likely fails selectively because cohesion is a function of grain size and water saturation. The loose-sand zone supports a stable plastic zone. Modification of Near-Well Geometry and Stress State. General loosening of porosity and increased permeability occur (Fig. 3a). Formation of loose-sand-filled regions (Fig. 3b) and of cavities, including possible piping failures (Figs. 2B and 3b through 3e), results in alterations of the near-well geometry. SPEPE P. 169^