Results

ABSTRACT: The main goal of this research was to investigate the risk of caprock failure due to the SAGDOX process, a hybrid steam and in-situ combustion recovery process for oil sands. A temperature dependency extension to the linear and non-linear constitutive models was developed and implemented in the GEOSIM software. The analysis has shown that there is no increased risk of caprock failure for SAGDOX process compared to SAGD. The study has shown that the overlying Wabiskaw formation experiences shear failure during both SAGD and SAGDOX due to its low initial cohesion, friction angle and proximity to pressure and temperature front, although the failure was mainly driven by pressure propagation. However, Clearwater shale above Wabiskaw can still provide proper zonal isolation to the steam/combustion chamber under SAGDOX operating conditions. Uncertainty in the analysis is due mainly to the sparse nature of geomechanical properties data for the oil sand reservoir and the caprock formations, especially at temperatures over 200 C.

1. INTRODUCTION

1.1 The SAGDOX process

Nexen Energy ULC (Nexen) has been evaluating SAGDOX - a post SAGD oxidation process (Kerr, 2012; Jonasson and Kerr 2013) - to improve the recovery and project economics of its Long Lake SAGD operation. SAGDOX is meant to be used after several years of SAGD operations when the bitumen between two SAGD well-pairs is mobile. In SAGDOX process (applied to a row of parallel well pairs) oxygen is coinjected with steam in every other SAGD injector well and starts an oxidation process by reacting with residual oil around the injection well. At this point the SAGD production well below the oxygen-steam injector is shut in and steam along with oxygen and combustion gasses fill the steam chamber voidage and push hot bitumen towards the neighbouring SAGD well-pair. The neighbour injection well is also shut-in and could be converted to a producer if need be. Various other well arrangements have been considered including those with vertical injection wells and infill horizontal production wells. Since oxygen is co-injected with steam, very high oxidation temperature of a pure combustion process are not generated as steam carries a large portion of the heat of combustion away from reaction front and temperatures are thereby moderated. Nonetheless, temperatures in the range of 400-600 deg C are expected in the oil sand zone. The high temperature combustion front where the oxidation reactions are active moves away from the oxygen injection wells as the residual oil left behind after steam displacement is consumed. The high temperature reaction zone has a tendency to move upward towards the cap rock under the influence of gravitational forces.

Summary

Producing from bitumen reservoirs overlain by gas caps can be a challenging task. The gas cap acts as a thief zone to the injected steam used during oil-recovery operations and hinders the effectiveness of processes such as steam-assisted gravity drainage (SAGD). Moreover, gas production from the gas cap can accentuate the problem even more by further depressurization of the gas zone.

Following a September 2003 ruling by the Alberta Energy Regulator (AER), the oil and gas industry in the province of Alberta, Canada, had approximately 130 million scf/D of sweet gas shut-in to maintain pressure in gas zones in communication with bitumen reservoirs. This decision led to the development of EnCAID (Cenovus' air-injection and -displacement process), a process in which air is injected into a gas-over-bitumen (GOB) zone, and combustion gases are used to displace the remaining formation gas while maintaining the required formation pressure.

An EnCAID pilot was started in June 2006, and preliminary results were reported in 2008. After 8 years of operations, the EnCAID project has not only proved to be effective at recovering natural gas and maintaining reservoir pressure, it has also shown it can heat up the bitumen zone and make the oil more mobile and amenable for production. This led to the development of the air-injection and -displacement for recovery with oil horizontal (AIDROH) process.

The AIDROH process is the second of two distinct stages. First, an air-injection well is drilled and perforated in the gas cap. The well is ignited and air injection is performed to sustain in-situ combustion in the gas zone. This phase is characterized by a radially expanding combustion front, accompanied by conduction heating into the bitumen below. The second stage begins when horizontal wells are drilled in the bitumen zone. The pressure sink caused by drawing down the wells alters the dynamics of the process and creates a pressure drive for the combustion front to push toward the producers in a top-down fashion, taking advantage of the combustion-front displacement and gravity drainage.

In light of the temperature increases observed in the bitumen overlain by the EnCAID project, a horizontal production well was drilled in late 2011 and commenced producing in early 2012. This paper provides an update of the EnCAID pilot results and presents a summary of the technical aspects of the AIDROH project, pilot results, and interpretation of the data gathered to date, such as observation-well temperatures, pre- and post-burn cores, and temperatures along the horizontal producer.

Results indicate that the AIDROH process has the potential to maximize oil production from GOB reservoirs, and efforts continue to be made to optimize its design and operation.

Oil sand characterization tests are essential for the selection and evaluation of sand control devices. Current approaches for screen selection and evaluation usually rely on Particle Size Distribution (PSD) and neglect the effect of important parameters such as porosity, grain shape and frictional properties. One aim of this study is to characterize oil sand's mechanical, geometrical and size characteristics that should be considered in the completion design. Another objective is to determine if natural mixture of oil sand could be reasonably replicated with commercial sands for large-scale sand control evaluation tests.

In this paper we present the results of a comprehensive image analysis and laser sieve analysis on oil sand samples from the McMurray Formation to quantify geometrical grain characteristics (sphericity, aspect ratio, convexity and angularity) of the sand grains and establish the PSD of the samples. Direct shear tests were performed to assess the frictional characteristics of different oil sands around the liner under variable stress conditions during the SAGD well lifecycle.

Image analysis, PSD, and direct shear tests showed that natural mixture of oil sand could be successfully simulated with commercial sands in terms of size and shape of grains and mechanical properties. This conclusion is significant to those performing large-scale sand control evaluation tests that usually require large quantities of sands that are not readily available and require significant preparation.

This paper provides the first comprehensive investigation of the granular, and geomechanical characteristics of oil sand from the McMurray Formation. The paper discusses the missing parameters in the design of sand control device, and evaluates test methods that measure those parameters. The proposed testing program could be used as a benchmark for oil sand characterization in relation to the design and evaluation of sand control device.

Connacher's first oil sands project, the Pod One facility at Great Divide, has been operational since 2007. The successful SAGD project has produced approximately 7 million barrels of bitumen. During the past three and a half years, the impacts of certain predicted reservoir challenges and opportunities have become apparent.

While the quality of the oil sands in this first phase of Pod One is generally good, Pad 101 South in particular has geological zones that affect SAGD operation. This includes a bitumen lean zone, and a gas cap overlying the main bitumen channel/s. Early field results matched with detailed simulations have shown positive results in maximizing well pair production. For the purposes of this paper a lean bitumen zone differs from an aquifer in two ways. The lean zone is not charged, and is limited in size. The operation is also complicated by the fact the gas bearing zone has been depleted through earlier production.

Connacher's operating practice at Great Divide attempts to achieve a pressure balance between the 3 zones (rich oil sands, lean zone, gas cap) to reduce steam loss and maximize production rates. Reducing the pressure encourages steam chamber development growth horizontally and ensures that steam contacts the highly saturated bitumen areas. How this is achieved with the highest positive impact on well productivity is illustrated with operational data and analysis including the results of simulations that recommended the optimum operating strategies.

Bitumen is too viscous to be produced by conventional recovery methods and significant amounts are too deep to be recovered by mining, necessitating enhanced in-situ oil recovery techniques. The majority of operating and planned in-situ bitumen projects employ thermal techniques to lower the bitumen's viscosity, allowing it to be produced. The viscosity characteristics of the bitumen consequently have a significant effect on production rates and recovery. Bitumen viscosity and chemical composition variation with depth within a single reservoir column has been reported for many heavy oil and oil sand reservoirs in the Western Canadian Sedimentary Basin and elsewhere in the world.

This study investigates, through reservoir simulation, the effects of viscosity variation with depth on the SAGD process and the resulting produced oil characteristics. Oil characteristics, including chemical component and viscosity profiles were built into a variety of reservoir simulation models. The simulation results indicate that the produced oil viscosity and component concentration vary as the steam chamber develops. The trend of the produced oil characteristics is related to the original in-situ profiles of and the reservoir flow barriers. In conjunction with oil rate, surface heave, or other available data, the produced oil characteristics may be used to suggest steam chamber development and the presence of barriers or baffles. The presented approach has potential to become a useful technique for SAGD steam chamber growth monitoring and production optimization.

1. Introduction
Oil viscosity and compositional gradients, both areal and vertical have been observed in various fields worldwide1. Differences in physical properties and chemical composition of oil are more significant in heavy oil and oil sands reservoirs2. Recently, more attentions has been paid to heavy oil and oil sands reservoirs in the Western Canadian Sedimentary Basin, where 172.7 billion barrels of bitumen and heavy oil are to be recovered3, mostly through thermal processes, such as CSS (cyclic steam stimulation) and SAGD (steam assisted gravity drainage). Erno et al. found that the viscosity increases towards the bottom of the reservoir for Clearwater B, McMurray, and Wabiskaw formations at Caribou Lake, and Waseca formation at Pikes Peak with up to an order of magnitude difference in the Clearwater B formation. It was suggested that the viscosity variation may affect the performance of CSS, the proposed recovery process for those reservoirs, and should be considered in reservoir characterization and modeling4. Chan et al.5 reported vertical variations of certain chemical compounds in a McMurray Formation corehole in the Athabasca area (re-plotted in Figure 1). They showed that the ratio of diasterane to regular sterane increases from the top to the bottom of the reservoir. Based on the observed baseline of chemical compound distribution, a field application was demonstrated that used the chemical compound concentration from the produced sample to diagnose the CSS performance5.

Traditional oil sands mining operations have used deterministic techniques to create a single resource model for mine planning. Stochastic modeling, commonly used for in situ oil sands evaluation, provides more realistic geology and allows for multiple realizations, which mining operations can use to assess the variability of recoverable bitumen volume estimates and develop mine plans accordingly. The existence of multiple realizations makes it possible to measure uncertainty, but eventually detailed mine planning will proceed based on a single realization. This paper discusses the processes of stochastic modeling and of determining the appropriate single realization for mine planning as applied to an oil sands mine currently in the planning stage (Fort Hills).

Geological models for mining operations have less uncertainty than models for in situ operations due to the much closer drill hole spacing and the better understood recovery process for mining, but the level of uncertainty is not zero. The same techniques that are currently being used to assess uncertainty for in situ oil sands leases can be applied to mining leases to quantify uncertainty for mine planning. In the case of Fort Hills, 100 realizations of ore grade were created using conditional simulation. Ranking solely by total bitumen in place was insufficient, so a new measure of heterogeneity related to vertical ore-waste changes was developed and is discussed in this paper. These two measures were combined to rank the realizations and to select mid, high, and low cases. The combined ranking resulted in ordering the realizations in a way that correlated with other measures of recoverable resource volumes, and lends support to the choice of the "mid?? model (centrally located in the ranking) for use in detailed mine planning.

The conditional simulation for Fort Hills marks the first time that stochastic modeling has been applied to full field modeling and then used for mine planning in an oil sands mine. The ranking method, including the methodology for assessing mining heterogeneity, is new and heretofore unpublished, and is the ultimate topic for discussion in this paper.

Oil sands geomechanics plays an important role in the oil sands recovery processes, such as surface mining, cyclic steam stimulation and SAGD, which are widely applied in the development of oil sands resources in Alberta, Canada. Coupled reservoir geomechanical simulation techniques have been developed and used for the design of in situ recovery processes, particularly for SAGD. Thus, a realistic geomechanical model of oil sands material is a critical component in these reservoir geomechanical simulations. This paper presents the development of an oil sands model based on the analysis of laboratory testing results provided by different researchers, including Oldakowski, Samieh and Wong, and Touhidi-Baghni. On the basis of this analysis, 25 numerical experiments were conducted to match these laboratory tests, including the stress paths, as those applied in the laboratory experiments. Consequently, a comprehensive geomechanical model of oil sands material was established based on these numerical experiments. The proposed strain softening model parameters, such as the modulus of elasticity, peak and post-peak friction angle, and dilation angle, can be applied in the coupled reservoir geomechanical simulations of thermal recovery processes, including the SAGD process.

INTRODUCTION

The development technology of oil sands reserves in Alberta, Canada, introduced a series of issues associated with the geomechanical properties of oil sands material. The surface mining technology and in situ recovery processes, such as cyclic steam stimulation (CSS) and steam assisted gravity drainage (SAGD), are both widely applied. Oil sands geomechanics has been studied in these areas, such as the improvement of surface mining efficiency, reservoir deformation regarding in situ thermal recovery, hydraulic fracturing during the cyclic steam stimulation process, and the prediction of in situ recovery performances.

The geomechanical properties of oil sands have been stud\ied extensively since 1970s(1) (2) (3) (4) (5) (6). With increasing experience in sampling and testing, good quality data can be obtained from lab testing. In this paper, the most recent laboratory testing results from Oldakowski(4), Samieh and Wong(5), and Touhidi-Baghini(6), are analyzed and simulated in order to obtain a representative geomechanical model of oil sands material.

Laboratory Testing on Oil Sands

Oldakowski's Lab Tests. Oldakowski(4) conducted a series of triaxial compression tests with different stress paths based on relatively undisturbed oil sands cores to characterize the stress-strain relationships of oil sands material. These oil sands cores were obtained from wells drilled at the AOSTRA Underground Facility Test Phase A site in 1987. In total, 23 oil sands samples were obtained from wells AT3 and AGI4 at two stratigraphic units, E and D, which consist of the richest oil sands at the UTF site.

These effects are easily reproduced with simple 1D column simulations. Solution gas has a material impact on SAGD production rates in general, which explains the need to use arbitrarily low permeabilities when history matching with no gas in the model.