Designing steam-injection flow-control devices for SAGD oil recovery

The use of flow-control devices (FCDs) has been shown to improve steam distribution along lateral wellbores.⁸ By using FCDs on an injection string, steam chamber growth will be more balanced, and the risk of steam breakthrough can be reduced. The use of an FCD with a closing sleeve also provides the operator with the ability to cease steam injection, in a particular region of the lateral bore, if there is some indication that steam breakthrough might have occurred. Some experimental work also has shown that the use of inflow control devices or autonomous inflow control devices, on the producer, can prevent, or limit, the impact of steam breakthrough.⁷

Tubing-deployed FCDs, to evenly distribute steam along the lateral bore, are now very common in the SAGD market. Multiple FCDs are used in series to control the amount of steam flow along the lateral bore by controlling the amount of pressure decrease that occurs at each device. An FCD can be designed with or without a closing sleeve, based on the amount of flow control necessary and reservoir heterogeneity. The device considered in this project consists of a center nipple, top sub, bottom sub, closing sleeve and seals.

The purpose of this project was to design, build and test an FCD that would survive downhole SAGD conditions, and operate reliably, throughout the life of the well pair. The design process began by identifying the most significant requirements:

• Axial steam injection, to prevent direct impingement on the slotted liner
• A design that would have the mechanical integrity to survive the most severe conditions while running in hole
• A seal system that would allow adequate isolation of a particular steam injection zone, at SAGD temperature and pressure, should the sleeve need to be closed.

With axial steam injection in mind, the next consideration was how to best use the internal flow area to condition the steam flow, so that the FCD, itself, would not experience any negative effects from direct internal steam impingement. To determine the best way that the steam flow could be conditioned from the bore, before entering the nozzle throat, four closing sleeves were considered. The designs consisted of a different geometric pattern used for fluid communication between the sleeve ID and outside/outer diameter (OD); a fifth design did not include a sleeve to serve as a control.

The benchmark, to determine how well the flow was conditioned by each design, was to examine how close the peak flow stream velocity was to the bulk velocity and the percentage of total flow traveling through each nozzle (left and right). The ratio between the two serves as a representation of how effectively the orifice cross-section is being used. The greater the percentage of the orifice cross-section used, the lower the peak velocity considered when determining the risk from liquid droplet impingement erosion (LDIE). A CFD analysis was used to determine which design achieved the lowest peak-to-bulk velocity.

When FCDs are installed, they are forced through the dogleg, between the vertical and lateral portions of the wellbore, and should be able to handle combined tensile and bending loads. The requirement of axial steam injection lends itself to a design, which consists of greatly differing center nipple and end sub diameters. Having two different moments of inertia can lead to bending stress concentrations at the interface between the two parts. Special care was used to design an OD feature profile transition that limited stress concentrations and limited the potential for crack initiation. This was done in consideration for the metal-to-metal seals and threads on the center nipple and end sub connection components. FEA was performed on the FCD to demonstrate that the worst-case dogleg and peak tensile loading would not pose any risks to the FCD or completion.

The sealing system designed for the closing sleeve considered an operating envelope of 520°F and no less than 500 psid. ISO 14998 testing was performed to stress the seals, both mechanically and thermally, to simulate the worst-case operating life for both the seals and collet. The testing included pressure reversals and temperature cycling. However, the sleeve was never opened with a pressure differential, because the end of the tubing string is open for steam flow. The ISO test also served as the validation of the collet feature at temperature. Until now, there has not been a clear set of criteria that covers the most important design aspects and methodology to validate an FCD for use in SAGD steam injection.

LIQUID DROPLET IMPINGEMENT EROSION

During SAGD operations, wet steam can pose problems to downhole equipment and the liner. Steam flowing at high velocity, through a nozzle or orifice, can thin the wall directly downstream. When steam follows a preferential flow path to the producer, the high velocity flow with entrained sand can destroy slotted liners and screens, resulting in a loss of sand control. Whenever possible, it is important that the velocity limitations for a tool be clearly defined and the peak velocities managed. To help ensure that erosion is limited in SAGD operations, it is important to understand the primary mechanisms of erosion.

Erosion occurs when material is removed by physical means.¹⁰ Three primary mechanisms are the cause of this removal—cavitation, liquid particle impingement, and solid particle impingement. When cavitation occurs, the surface is repeatedly loaded and unloaded, causing fatigue and pitting. Solid particles—such as sand entrained in steam, oil or gas—can be carried in the fluid and abrade tubing, or other downhole tools made of ductile materials. In some cases, material loss can be a result of repeated removal and buildup of a passive surface layer caused by erosion/corrosion.¹⁰

API RP 14E has defined the fluid velocity limitations of oil and gas production two-phase flow systems, but some studies have shown the methodology to be overly conservative. The erosional velocity limitations imposed by API RP 14E are particularly conservative for sand-free two-phase systems, and situations in which liquid droplet impingement is the primary mechanism. The velocity limitations are defined by: 

in which V_e is the erosional velocity in ft/sec, p is the fluid density in lbm/ft³, and C is the constant between 100 and 125. In two-phase systems, which are sand-free, C=100 is typically used to determine the erosional velocity limit for continuous service, and C=125 is considered only for intermittent service. Based on a review of experimental data, and a comparison of the limits proposed by API RP 14E, Salama concluded that the API equation was overly conservative. Particularly in the case of sand-free systems, C=300 should be an acceptable empirical constant.¹⁰

The Erosion/Corrosion Research Center research group at the University of Tulsa developed an erosional model to predict velocity limitations, caused by liquid droplet impingement, and compared the values to those calculated using API RP 14E. The model showed that the API equation was much more conservative than those velocity limitations proposed by this new flow model. This work created a method for predicting erosional velocities caused by liquid droplet impingement, by considering droplet size, impact velocity, and the type of geometry in which the fluid stream interacts. Validation of the model was performed by comparing experimental results from an ASTM STP 474 test on 1018 carbon steel to the erosional velocity predicted by the correlation model. In cases where liquid droplet impingement was the primary mechanism of erosion, the paper showed that the API equation greatly underestimated the fluid velocity limitations.⁴

Fig. 1. The simulations were performed using a 1/4 cut of the cross-section.

Navas et al. proposed a new standard of C factors to be used for estimating erosional velocity limitations, using the API equation. Erosion testing conducted was evaluated to help determine which corresponding C factor was associated with the erosional velocity limit. A variety of geometries were considered in this study, to create a relevant standard to determine appropriate C factors (straight pipe, elbow and valve assembly). The study varied the sand particle size and concentration, in addition to the fluid in which the particles were entrained (water, wet gas and dry gas). The results showed that, in wet gas (>1,000 psi), C factors exceeding 300 might be acceptable for determining the erosional velocity limits in a straight pipe section. This paper demonstrates that using C factors between 100 and 200 for production or injection tubing and downhole tools was overly conservative.⁹

CFD SIMULATIONS

To determine which closing sleeve geometry best conditioned the flow entering the steam chamber within the FCD, two-phase CFD simulations were completed for each case. Phase-change was not considered. Constant concentration and fluid properties were considered for the steam and water droplets. ANSYS CFX Modeling software was used to perform CFD analyses.³ By conditioning the flow, the nozzle cross-section could be used more effectively, reducing the peak velocity in the nozzle and improving FCD reliability. In addition, the velocity of the flow exiting the diffuser needed to be determined. The steam flow velocity should be less than 100 ft/sec, exiting the diffuser, to help ensure there was no LDIE risk to the slotted liner.

Table 1. Design cases 1 and 2–velocity ratios.

Each case used the same center nipple, top sub and bottom sub, and the sleeve geometry represented the variable being tested. The simulations were performed by considering a ¼ cut of the cross-section, to reduce both mesh size and time necessary for simulation convergence, Fig. 1. The FCD in the analysis was the top device on the injection tubing string, meaning it would encounter the highest steam pressure and flowrate through the bore. Table 1 presents the boundary conditions considered for each simulation.

Fig. 2. Nozzle profile (Design 1).

This project considered five cases; the first had no closing sleeve to interact with steam flow, and the other four used various flow slot geometries, in the closing sleeve, to alter the steam flow. The criteria used to determine which design performed the best consisted of three elements:

• Lowest ratio between the nozzle bulk velocity and peak velocity in the flow stream 
• Closest to 50/50 balance of flow between the left and right nozzle considered 
• No diffuser exit velocities above 100 ft/sec.

Fig. 3. Streamlines (Design 1).

Based on the results from Design 1, it was determined that the recirculation that occurred in the steam chamber had a negative impact on the velocity ratio. The flow stream tends to be restricted in the steam chamber, because most of the internal area is consumed by a large flow recirculation. Figures 2 and 3 show the flow stream and nozzle flow profile for Design 1. The nozzle profile shows the peak velocity concentrated near the top half of the nozzle diameter, and the streamlines show how the recirculation consumes the entire steam chamber.

Next, a closing sleeve was included in the simulation, which was in the open position. The fluid communication path used in Design 2 consisted of 16 evenly spaced radial straight slots. The results from this design were similar to those in the first iteration, without the closing sleeve included. Figures 4 and 5 show the nozzle profile and streamlines for the simulation completed for Design 2. Design 2 had a velocity ratio higher than the control case, with no closing sleeve to interact with the flow. The radial flow slots, which span the entire length, allow for recirculation to consume the entire steam chamber. The flow slots, which span the chamber, allow the fluid stream through the FCD bore to interact directly with the chamber and induce recirculation. Velocity ratios and the flow balance between the left and right nozzle are included in Table 2.

Fig. 4. Nozzle profile (Design 2).

Fig. 5. Streamlines (Design 2).

Results from the first two design cases show that full chamber length flow paths in the closing sleeve might not be the most effective means to condition the flow stream. Three more designs were created to attempt to reduce the negative effects of recirculation on the nozzle velocity ratio. Designs 3 through 5 were created with a flow path that would diffuse the inertial energy in the flow stream that causes circulation. The alternate designs use offset slots and varied geometry to break the flow stream before entering the internal steam chamber.

Table 2. Design cases 1 and 2–velocity ratios.

Design cases 3 and 4 still exhibited some localized recirculation, while it was less pronounced in the steam chamber. Even though both design cases used offset slot patterns or varied slot sizes, particularly near the nozzles, multiple instances of recirculation were present in the flow stream. It appears that using a slot, even in the case of a slot that does not run the full length of the steam chamber, allows the main flow stream through the bore to influence the streamline behavior in the steam chamber. However, Design 5, which incorporated progressively smaller holes in place of slots as the fluid communication, showed much less recirculation in the steam chamber flow streamlines.

Table 3. Design cases 3 to 5–velocity ratios.

It is counterintuitive to consider progressively smaller, rather than progressively larger, diameter pathways. In most cases, the primary mechanism that governs fluid flow is fluid pressure, rather than the fluids inertia. By limiting the amount of continuous cross-sectional area exposed to the main flow stream, the amount of recirculation that occurs can be minimized. Velocity ratios for each of the remaining designs were obtained from the simulation, Table 3.

The simulations showed that the casing would not encounter a steam flow velocity >100 ft/sec, meaning that liquid droplet erosion would not be a concern. Results also demonstrated that, even without a closing sleeve to condition the flow steam, a near 50/50 balance between the left and right nozzles could be achieved. By reducing the velocity ratio between the peak and bulk velocities, more of the nozzle cross-section was used effectively. Reducing the velocity ratio also helps to ensure that no portion of the FCDs’ internal components will be exposed to a steam flow velocity that exceeds the recommended erosional velocity limit. Design 5 represents the optimal design case because of its near 50/50 flow balance, minimal velocity ratio and lack of significant recirculation present in the steam chamber flow streamlines.

MECHANICAL TESTING

Full-scale mechanical testing was performed, to help ensure that the design, which has a widely varied OD, would not have excessive stress concentrated at the interface between two ODs. The test was conducted in a horizontal load frame, rated for one million pounds, with the capability for applying tensile and bending loads simultaneously. Before being installed into the load frame, the ends subs were installed with 1,200 ft-lb of make-up torque. Following assembly, eight uniaxial strain gages were mounted onto the top and bottom end subs (four on each). Gages were placed at 0°, 90°, 180° and 270°. The strain gages were mounted on sections with the same ID, OD, and wall thickness, to help ensure that bending on any axis would be detected. Additionally, four tri-axial strain gages were positioned at critical locations along the FCD, to calculate the bending moment and strain.

Load testing consisted of applying a 126,000 lb/f tensile load then a 4,200 ft-lb bending moment through the assembly, which equates to 17°/100-ft well deviation. The load was held for 5 min., during which time the axial displacement was measured with a displacement transducer, to help ensure that the center nipple and end sub connections were not yielding. The bending moment through the assembly was also monitored, to help ensure that the expected deformation was occurring, and the strain values were not changing over time. No deformation or yield was observed.

ISO 14998 TESTING

Laboratory testing was performed at Southwest Research Institute (SwRI) in San Antonio, Texas. The testing plan, for the two seal designs, was based on Annex D of ISO 14998, which stipulates that the test article be opened and closed the maximum rated number of cycles, at the maximum rated temperature. It requires that, after cycling, a 15-min. pressure hold, at the maximum rated pressure, be performed on the bore, followed by one pressure reversal, at the maximum rated pressure, from the bore to the annulus. A pressure reversal involves a pressure hold on the bore and a pressure hold on the annulus, followed by another pressure hold on the bore. Leakage is measured during each pressure hold, and pressure holds should last at least 15 min.

For Grade V0 and Grade V1 of Annex D, at least one temperature cycle is necessary, in which the temperature of the test article is cooled down by at least the maximum rated temperature cycle range. Pressure holds are necessary at the low end of the temperature cycle range, and after heating back up to the maximum rated temperature. Grade V2 of Annex D requires no temperature cycle. The acceptance criterion for Grade V0 is zero bubbles of gas accumulated in a graduated cylinder over a hold period, and the acceptance criterion for Grade V1 and Grade V2 is no more than 20cc of gas accumulated in a graduated cylinder over the hold period.

A custom test fixture for shifting the test article shifting tool was designed and fabricated at SwRI. The test article shifting tool was coupled to a 31/4-in. bore hydraulic cylinder with a 1-in. diameter shaft and a 12-in. stroke length. The connection between the test article and the hydraulic cylinder included a tension and compression load cell for measuring the force necessary to shift the tool. The assembly was mounted to a 12-ft section of wide flange, with gussets welded in place to react the shifting force. The fixture was designed to accommodate a shifting force of up to 10,000 lb/f.

A port in one of the test article end caps was used to supply pressurized nitrogen to the bore of the test article, and to measure any leakage from the bore when the test article was pressurized from the annulus. Similarly, one of the ports in the annulus of the test article was used to supply pressurized nitrogen to the annulus of the test article, and to measure any leakage from the annulus when the bore of the test article was pressurized. A dynamic gland seal was incorporated into the test article end cap facing the hydraulic cylinder, to accommodate the reciprocation of the closing sleeve while pressurized with nitrogen. 

An arrangement of automated isolation valves was used to direct pressurized nitrogen to the test article or direct leakage from the test article to an aquarium tank containing an inverted graduated cylinder, submerged in water for quantifying leakage. A heat exchanger was necessary to cool the nitrogen leaving the test article before coming in contact with the automated isolation valves, which were rated at a temperature lower than the temperature at which the test article was heated for testing. The pressures of the bore and annulus were measured with pressure transmitters.

The test article was wrapped in heater bands to raise the fixture’s temperature to the maximum rated temperature of 520°F. A proportional integral derivative (PID) temperature controller was used to control the power to the heater bands, based on feedback from thermocouples. Three thermocouples, two of which were used as inputs to the PID controller and one of which was used for data collection, were secured to the outside of the test article.

A hydraulic pump and a manifold of automated isolation valves were used to control the hydraulic cylinder. The pressure supplied to the cylinder by the pump was measured to quantify the force necessary to shift the tool, in addition to the measurement made by the tension and compression load cell.

Table 4. Summary of test results.

Two seal package designs were tested within the same test article. For this test, intermediate pressure holds were performed before reaching the maximum rated number of cycles. The maximum rated number of cycles for the test article, inside of which the seals to be tested were installed, was 10. For seal Design 1, pressure holds on the bore were performed every two cycles, and two pressure reversals and a temperature cycle were performed after 10 cycles. For seal Design 2, pressure holds on the bore were performed after two cycles, six cycles, and 10 cycles. If the seal design had not failed already, because of more than 20cc of leakage during a pressure hold, one pressure reversal and a temperature cycle were performed after 10 cycles. Tests were conducted at the maximum rated temperature for the test article of 520°F. For temperature cycling, the temperature of the test article was reduced to below 400°F.

Table 4 presents a summary of the test results. Seal package designs 1 and 2 met the acceptance criterion for Grade V1. Note that seal package design 2 was tested at a maximum rated pressure of 750 psi, as opposed to seal package design 1, which was tested at a maximum rated pressure of 500 psi. The table also shows the ranges measured for the force necessary to open and close the test article with the load cell throughout cycling for each test.

CONCLUSIONS

• Under typical SAGD well conditions and flowrates, it’s possible to develop an FCD design that can provide steam axially in both directions, which prevents steam from impinging on the slotted liner. It also was shown that the threshold velocity for liquid droplet impingement erosion is much higher than the velocity of any flow with which the casing will interact.
• Using the geometry of the fluid communication flow paths in the closing sleeve, it’s possible to condition the flow, such that the bulk velocity and peak velocity are very similar, thereby limiting internal nozzle velocities to those below the LDIE threshold velocity.
• The FCD design presented should not encounter any issues downhole, because it was tested well beyond the tensile and bending values it should encounter normally.
• The seal package and collet design performed well under SAGD conditions and can help provide reliable functionality throughout the life of the well.
• A set of design criteria and benchmarks for validating future designs was presented, which will serve as a starting point for future developments of FCDs intended for use in SAGD injection.  

ACKNOWLEDGEMENTS

This article was adapted from SPE paper 174490-MS.

REFERENCES

1. Ajumogobia-Bestman, S., R. Jobling, E. Pederson, et al, “Quantifying the impact of flow control devices in SAGD applications in the Surmont area through numerical analysis,” SPE paper 170112-MS, presented at the SPE Heavy Oil Conference-Canada, Calgary, Alberta, June 10–12, 2014.
2. Ali, S.M., “Is there life after SAGD?,” Journal of Canadian Petroleum Technology, 36 (6), PETSOC-97-06-DAS, 1997.
3. ANSYS CFX, Release 15.0, ANSYS, Inc.
4. Arabnejad, K.H., S.A. Shirazi, B.S. McLaury, et al, “A guideline to calculate erosional velocity, due to liquid droplets for oil and gas industry,” SPE paper SPE-170951-MS, presented at the SPE Annual Technical Conference and Exhibition, Amsterdam, The Netherlands, Oct. 27–29, 2014.
5. Das, S., “Improving the performance of SAGD,” SPE paper SPE-97921- MS, presented at the SPE International Thermal Operations and Heavy Oil Symposium, Calgary, Alberta, Nov. 1-3, 2005.
6. Gotawala, D.R., and I.D. Gates, “SAGD subcool control with smart injection wells,” SPE paper SPE-122014-MS, presented at the EUROPEC/EAGE Conference and Exhibition, Amsterdam, The Netherlands, June 8-11, , 2009.
7. Least, B., S. Greci, R. Huffer, et al, “Steam flow tests for comparing performance of nozzle, tube, and fluidic diode autonomous ICDs in SAGD wells,” SPE paper 170083-MS, presented at the SPE Heavy Oil Conference-Canada, Calgary, Alberta, Canada, June 10–12, 2014.
8. Medina, M., “Design and field evaluation of tubing-deployed passive outflow control devices in SAGD injection wells,” SPE paper 165563-MS, presented at the SPE Heavy Oil Conference-Canada, Calgary, Alberta, June 11–13, 2013.
9. Navas, G., H. Nguyen, and K. Sun, “Choosing better API RP 14E C factors for practical oil field implementation,” NACE paper 11248, presented at the NACE International Corrosion Conference, Houston, Texas, March 13–17, 2011.
10. Salama, M.M., and E.S. Venkatesh, “Evaluation of API RP 14E erosional velocity limitations for offshore gas wells,” OTC paper 4485-MS, presented at the Offshore Technology Conference, Houston, Texas, May 2, 1983.