Simulation of Cold Production in Heavy-Oil Reservoirs: Wormhole Dynamics

Summary

The laboratory investigation described arose from a need to observe more directly how sand production can lead to enhanced oil recovery during the cold-production process. The production of oil and sand into a perforation was simulated by use of a sandpack with an orifice at one end. The pack was imaged throughout the experiment with a computed-tomography (CT) X-ray scanner. The experiment showed that (1) stable wormholes can develop in unconsolidated oil sands, (2) they likely develop in the weaker sands (which normally contain the most oil) within a formation, (3) a critical flow rate is required for the wormholes to grow, (4) the high sand cuts observed in the field at the start of cold production indicate that the wormholes are growing, and (5) the decrease in sand production rate observed in the field indicates that the wormholes have stopped developing and are being emptied out by scouring.

Introduction

Cold production is a nonthermal recovery process used in heavy-oil reservoirs whereby sand and oil are produced together under primary conditions. Solution-gas drive is generally believed to be the main drive mechanism.^1-7 The reservoir conditions of the fields in which cold production was applied in Alberta and Saskatchewan, Canada are listed in Table 1. A necessary condition for cold production is that the sand formation be unconsolidated. The viscosity range given in Table 1 is approximate. It is not certain at the moment if cold production is economical outside this range.

Economical recovery of these high-viscosity oils was possible only when large quantities of sand were produced with the heavy oils. The total volume of sand produced per well can range between 500 to 1000 m³ for the wells which produce typically at 10 to 20 m³/d. Examples of the average quantities of sand produced from the economical wells are 700 m³ (Elkins et al.⁸), 580 m³ (McCaffrey and Bowman⁹), 500 m³ (Yeung¹⁰) and 560 m³ (Metwally and Solanki¹¹). All operators report high sand cuts at the start of cold production: 10 to 50 vol% during 1 month (Elkins et al .⁸), 30 to 40 vol% during 2 weeks (McCaffrey and Bowman⁹ ), 20 to 30 vol% (Yeung¹⁰). These sand cuts then declined to low levels: between 0.1 and 2 vol% after a few months (Elkins et al.⁸ ), between 0.5 and 1 vol% after 1 year (McCaffrey and Bowman⁹), and less than 5 vol% after 1 year (Yeung¹⁰).

To optimize well spacing, several field methods were used to determine the effect on the formation of producing so much sand. The interpretation of the tests, however, differed widely. Most operators performed tracer tests to try to answer this question.^1-3,8,11 In several tests, the injected tracer took much less time to travel from an injector well to a producing well than expected from radial drainage between wells.^1-3,8 The concentration of the dye was almost the same as when it was injected. Most investigators found that the dye would show up only at certain wells that could be connected together for long distances. In one case, 12 wells were connected in succession over a distance of 2 km.³ Not all tracer tests showed communication between wells.¹¹ In one test, a radioactive tracer was injected.¹¹ Post-production compensated-neutron and gamma logs of the well after cold production were compared by Metwally and Solanki¹¹ with openhole logs. A few zones of much higher porosity were observed.

Some operators noticed lost circulation in drilling or cementing wells.^9,10 Injectivity tests were performed in addition to tracer tests by measuring the pressure during the injection of a fluid at a known flow rate. The pressure signal resembled that observed during hydraulic fracturing. The interpretation of these tracer and injectivity tests differed significantly. Most investigators agree that some form of communication channel must exist between the injector and producers. McCaffrey and Bowman⁹ and Loughead and Saltuklaroglu⁶ believed that these channels were fractures, whereas Squires³ and Yeung¹⁰ concluded that the channels were circular "wormholes." Smith¹ gives several explanations for the rapid communication between wells: (1) massive failure around the well, (2) wormholes, and (3) fracturing during injectivity tests. Some authors have also suggested the possibility of cavities growing around perforations as well as wormholes to explain the increase in oil production when sand is produced.

The formation of high-permeability channels by erosion has been observed previously in water-saturated sandpacks.^12-14 These channels are called pipes in the geotechnical literature. These studies were aimed at determining the conditions under which piping occurs in soil embankments or highways. A steady wormhole (pipe) was observed in silty clay¹³ but not in water-saturated sand¹⁴ because the cohesive strength of clay is significantly larger than for water-saturated sand.

The experiment was separated into two parts. In the first part, oil flowed through the pack at a rate of 0.6 cm³/min. The pressure gradient at the orifice end of the sandpack simulated the pressure gradient at a perforation (19 mm diameter) in a well with 150 perforations producing at a rate of 0.033 m³/d. In this calculation the formation permeability was assumed to be 3 darcies, and the viscosity was assumed to be that of a heavy oil (21.5 Pa·s). In the second part of the experiment, the flow rate was increased to 9.6 cm³/min once the wormhole had broken through the inlet face.

Experimental Apparatus and Procedure

Materials.

The Clearwater sand used in the experiment was obtained from the production tanks at Suncor's Burnt Lake project.² The sand was cleaned with toluene to remove bitumen, water, and other chemicals used to wash the tanks. The average particle-size distribution (PSD) of the sand in the pack is shown in Fig. 1 along with the PSD of two core samples at two different wells. The Clearwater sand was wetted with formation water obtained from the wellhead. A geochemical analysis of the formation water as well as an elemental analysis of the sand and water are given elsewhere.¹⁵

The oil used in the experiments was also obtained from the production wells at Suncor's Burnt Lake project.² This oil was then diluted with toluene and centrifuged several times to remove the fines. The toluene was then removed by heating the oil. The viscosity of the oil at reservoir temperature (18.5°C) was 27.0 Pa·s. The experiment was performed at the same temperature.

Materials.

The Clearwater sand used in the experiment was obtained from the production tanks at Suncor's Burnt Lake project.² The sand was cleaned with toluene to remove bitumen, water, and other chemicals used to wash the tanks. The average particle-size distribution (PSD) of the sand in the pack is shown in Fig. 1 along with the PSD of two core samples at two different wells. The Clearwater sand was wetted with formation water obtained from the wellhead. A geochemical analysis of the formation water as well as an elemental analysis of the sand and water are given elsewhere.¹⁵

The oil used in the experiments was also obtained from the production wells at Suncor's Burnt Lake project.² This oil was then diluted with toluene and centrifuged several times to remove the fines. The toluene was then removed by heating the oil. The viscosity of the oil at reservoir temperature (18.5°C) was 27.0 Pa·s. The experiment was performed at the same temperature.