To offset production decline caused by normal pressure decline and reservoir drainage, secondary recovery using water injection has been administered in three Saskatchewan tight oil plays. This study will be focused on the Bakken formation, the area with the most-complete historical data set available to the authors.
The original completion and well-spacing plan has resulted in occasional direct water channeling from injection to production wells. This is believed to be caused by hydraulic fracturing. In some cases, this communication has limited waterflood sweep efficiency and is referred to as short-circuiting of injection fluid. Chemical and mechanical diversion techniques have been implemented to address short-circuiting with varying degrees of success.
Field Development
The simplest means to waterflood a multistage fractured well is to reverse the direction of flow and to bullhead water from the surface without diverting or compartmentalizing fluid flow. Because the water follows the path of least resistance, nonuniform fluid injection along the wellbore and into the formation may occur. In some instances, short-circuiting of producing wells through fracture paths may occur.
Using distributed temperature sensing (DTS), the operator was able to investigate wells that were suspected of short-circuiting or channeling. Many attempts were made to redirect flow from the paths of least resistance by use of various viscous fluids and solid diverters. The effectiveness of these treatments was inconsistent and, in some cases, required multiple applications. The operator began investigating the application of mechanical diversion and isolation of sections of the wellbore to enable physical control of flow rates into different areas along the lateral wellbore.
Compartmentalization With ICDs
The nozzle diameter of ICDs can be adjusted to compensate for variations in reservoir and fracture injectivity resulting from variable permeability effects and frictional pressure losses along the wellbore, known as the heel-to-toe effect. The installations can be run in openhole configurations and within cemented or noncemented tubulars.
In this study, tubing strings with several active ICDs with two-position sleeves were trialed. Hydraulic packers were used along with a circulation valve at the toe of the string that allowed fluid circulation during installation of the assemblies. The hydraulic packers were set using a common ball-drop system that closes the circulation valve. Once the packers were simultaneously set, a slim-diameter, expandable shifting tool was run on coiled tubing (CT) to open the sleeves. At that point, the well was ready to start or resume injection.
A key component in the systems used by the operator are the active ICDs (Fig. 1), which can be exercised (opened or closed) multiple times with a CT-deployed hydraulic shifting tool. In the open position, flow control is achieved using tungsten carbide nozzles. Each ICD can accommodate up to four nozzles, enabling flexibility in backpressure potential. When the flow area of a single nozzle cannot accommodate the desired flow rate adequately, a combination of two or more nozzles is used. For this study, a single nozzle size at each ICD point was sufficient to achieve the designated flow rates and diversion.
Blind-Modeling Approach. At the initial stage of the project, because of the limited data availability and time constraints, the near-wellbore reservoir region was the area of focus.
Reservoir matrix permeability was considered negligible in contrast with the conductivity of the hydraulic fractures present. The hydraulic fractures were modeled as zones with very high relative permeabilities. These permeabilities, along with other near-wellbore properties, were then increased throughout the boundaries of the reservoir simulation.
The gridblock size for the subject well was defined at 3×3×6 m, placed close to the wellbore to capture a high resolution for well-to-reservoir fluid movement. The farther away from the wellbore the gridblocks were positioned, the larger their size was in order to reduce simulation time to a reasonable level. The base-case (no ICD) simulation showed a nonuniform water-injection profile. This was believed to be the result of a combination of a higher drawdown and high-permeability zones (short-circuit zones).
The work flow provided valuable metrics required to design, rank, and optimize ICD configuration with limited data. Use of a uniform nozzle size for each compartment, and use of similar spacing between compartments, is recommended. This is referred to as the blind-modeling approach. This method will provide improved flow equalization through a pressure-drop effect across the wellbore. However, it may lead to cases where higher-permeability zones are not choked as optimally as they are in a variable-nozzle-size approach. Simulation outputs demonstrate that smaller-diameter ICD nozzles result in better injection distribution and will generally result in higher injection pressure.
Production-Log-Tool-Based Approach: Calibrating the Model. To understand injection profiles before and after ICD installations better, DTS analysis was incorporated into the work flow. DTS-interpretation software used for this purpose incorporates a similar analytical approach for flow modeling. The injection rates measured were matched with the reservoir-simulation model by performing various sensitivity optimizations with the reservoir permeability.
Fiber-optic-distributed temperature measurement uses an industrial laser to launch 10-ns bursts of light down an optical fiber line. During the passage of each packet of light, a small amount is backscattered from molecules in the fiber. This backscattered light can be analyzed to measure the temperature along the fiber. Because the speed of light is constant, a spectrum of the backscattered light can be generated for each meter of the fiber by using time sampling. A continuous log of spectra along the fiber is provided. A temperature log can be calculated every meter along the entire length of the fiber (and wellbore).
To collect the data, a single-ended wireline-deployed DTS fiber was run inside the tubing along the horizontal section of the subject well. An axisymmetric finite-element thermal model for the well and near-wellbore region was built using thermal-analysis software. Once a match is achieved, the distribution of the variable in the reservoir zones is defined uniquely and the model provides measured individual zone flow rates.
After a number of sensitivities on nozzle size had been established, an ICD configuration (Table 5 of the complete paper) was determined that provided the optimal injection distribution. The largest volume of injection was located in the stages closest to the heel. Because these stages had near-uniform distribution, they were grouped together in one compartment and nozzle size was designed to restrict flow within this zone and divert to the rest of the compartments.
Modeling injection profiles calibrated with DTS field-measurement inputs provide an accurate baseline injection profile. This allows for customization of ICD configurations to optimize individual wellbores on the basis of their specific parameters.
Field Results
Since late 2016, the contributing operator has successfully installed more than 50 ICD systems in southeastern Saskatchewan. The Viewfield Bakken area accounts for 26 of the successful ICD system installations as of Q1 2018. Improvements in injection rates of 200% on average have been achieved without direct channeling effects. Oil-production rates have increased 25% on average, with some wells reaching close to 100% improvement.
石油圈
