Discovery of large gas and gas/con densate reserves presents operators with the challenge of safe, expedient, and economic recovery of these resources. Most of the recent large gas discoveries are offshore, often far from existing production and transportation infrastructure. The economic benefits of many such discoveries can be realized only if gas production rates are very high and recovery costs are minimized. While producers always seek to minimize the number of wells drilled and to maximize the production rate from each well, these considerations are even more important in the case of remotely located reservoirs.
Fortunately, many recent gas discoveries have characteristics that permit extremely high production flow rates without damage to the geologic formation. With these higher flow rates, it becomes extremely important to be able to accurately predict the different types of aerodynamic forces on completion equipment and use those predictions to determine whether aerodynamic effects will complement or hinder operation of the equipment.
One of the key elements in the production string is the surface-con troll ed subsurface safety valve (SCSSV). The SCSSV is a fail-safe (fail-closed) device that is held in the open position by application of hydraulic pressure at the surface, which is transmitted to the SCSSV via capillary. The pressure places a force on a hydraulic piston, which moves downward, compresses a spring, and pushes a flapper (biased in closed position by a second spring) out of the flow path to permit well production. Upon loss of signal, the spring (and well pressure) force the piston upward and allow the flapper to swing into the flow path, thus closing the safety valve and preventing well production. It is imperative that the SCSSV, the most important - or often the sole - safety device in the tubing string, close upon loss of signal in all scenarios, regardless of fluid velocity or tubing pressure.
Achieving the desired flow rates from large gas reservoirs dictates the use of large-diameter completion tubulars. Production tubulars larger than 5.5-in. OD are becoming commonplace, and very large-bore tubulars (7-in. and larger) are used to maximize recovery rates in reservoirs with extremely high production flow rates. Currently, the size of the tubular employed is limited by the availability and performance of surface and subsurface completion equipment. If field-proven equipment is not available in the tubing size desired, a mixed-sized tubing string might be considered. With the advent of these mixed-sized strings comes the possibility of larger-than-normal aerodynamic and mechanical loads on a SCSSV.
The worst-case load condition for any SCSSV is the catastrophic case - complete loss of the wellhead, resulting in uncontrolled flow to the atmosphere. The maximum flow rate achieved will be the absolute open-hole flow (AOF) rate as defined by reservoir and wellbore characteristics. The SCSSV must be capable of securely shutting in the well under AOF conditions.
In recent years, considerable advances in machine technology, coupled with the use of high strength, precipitation-hardened round stock, have enabled continued optimization of safety valve diameters. As a result of these developments, many operators can now use homogeneous strings with full-size accessories that were previously impossible to manufacture or impractical to employ.
Currently, the most challenging completion design involves the use of tubing strings with very large external diameters (9 5/8-in. and larger). Subsurface accessory equipment designed for such large-diameter production tubing is just now becoming available. The obvious benefit of such a design is the very large production bore available - up to 8 1/2-in. through a tubing-retrievable safety valve (TRSV). This large through-bore results in lower pressure drop from reservoir to wellhead and lower velocities for a given production rate. Thus, an operator can flow each well at a much higher rate than previously possible. The challenge to the operator is handling and maintaining the equipment.
During a gas slam test performed in 1998, engineers noticed that a 7-in. TRSV performed inconsistently at very high flow velocities. The objective of the test was to slam the valve several times at 330 MMscfd and record resulting flow velocities in excess of 250 ft/sec. Two test programs were conducted, but with results that did not conclusively meet the test objectives. This was attributed to a new phenomenon discovered during the testing.
Preliminary analyses indicated that flow velocity, flow geometry upstream of the valve, and flapper (hinge) orientation with respect to the gravitational field may have created conditions that prevented the flapper from closing. To better understand the factors that contribute to the proper functioning of flapper closure mechanisms, laboratory tests, computational fluid dynamics (CFD) simulations, and full-scale tests were conducted.
Lab testing and results
Laboratory experiments using a smaller valve (4 1/2-in.) and air flowing at atmospheric pressure showed that the measured pressure distribution on the bottom surface of the flapper is similar to the pressure distribution in the gap between the flapper and the flapper housing.
The average pressure value that was calculated based on the measured pressure distribution on the bottom surface of the flapper was nearly the same as the average pressure value on the top side (in the gap between the flapper and the housing). The differences in the average pressure values were only within the accuracy of the incline tube manometers that were used to measure the pressure distributions. The important observations from the laboratory testing on the 4 1/2-in. valve were:
- Pressure distribution around the flapper was independent of the flow velocity from 50 ft/sec to 200 ft/sec (corresponding to Reynolds numbers ranging from about 100,000 to 400,000).
- Average pressure value on the bottom surface of the flapper (beneath the open flapper) was nearly the same as the average pressure value behind the flapper (in the gap between the flapper and the housing) for all of the test cases.
- For all experiments, the average pressure in the gap (between the top surface of the flapper and the housing) was slightly higher than the average pressure on the bottom surface (beneath the open flapper). This difference would favor closing the flapper.
- For velocities as high as 200 ft/sec, measured average pressure differences across the flapper were very small.
Measurements of pressure distribution were conducted at atmospheric pressure (low density) and did not show appreciable pressure difference across the flapper surface. Therefore, computational fluid dynamics (CFD) simulations were undertaken to increase the understanding of the aerodynamic effects on the flapper closing mechanism.
Preliminary CFD simulations inside the 7-in. TRSV flapper housing were carried out using a simplified axisymmetric model. The fluid was assumed to be gas (methane), and a k-e turbulence model was used for the simulations. The simulations were carried out at Reynolds numbers of approximately 400,000 and 40,000,000.
The results indicate that the axial pressure distributions on the surface of the flapper, when normalized by the dynamic pressure (0.5 * density*velocity2), are nearly identical for the two cases considered. This important observation suggests that the trend of pressure distribution (not values) over the surface of the flapper is nearly independent of the Reynolds number for the cases considered (when the Reynolds number is sufficiently high).
Furthermore, the axisymmetric CFD investigations indicated that the axial orientation of the flapper (degree of tilting from heel to toe) has an important effect on the distribution of pressure on the surface of the flapper. To determine the effects of design changes on the original design of the 7-in. SCSSV, three-dimensional CFD simulations were carried out. As of this writing, three different geometries have been simulated. The first was the original configuration of the valve prior to any design changes.
CFD simulations (3D) were carried out for inlet flow velocities of 303 ft/sec and 562 ft/sec. A finite-element mesh with approximately 700,000 elements and 140,000 nodes was used. As we found in lab testing of the 4 1/2-in. TRSV, the calculated values of the pressure distribution on the flapper, indicate that the distributions of pressure on the bottom and top surfaces of the flapper are similar. However, small differences exist.
Another important result is that the patterns (not values) of pressure on the top and bottom surfaces of the flapper are identical for the two velocities that were considered. This can be easily observed by comparing inlet velocities 303 ft/sec and 562 ft/sec. The main differences are that the values (magnitude) of the pressure for the flow velocity of 562 ft/sec are about four times larger than the pressure values for the flow velocity of 303 ft/sec. It should be noted that similar behavior was observed in the measured pressure distributions in the laboratory for flow velocities of 50, 100, and 200 ft/sec.
Using the calculated pressure distribution on the bottom and top surfaces of the flapper, the net forces that act on the flapper can be calculated. The average pressure acting on the flapper (or the net differential pressure across the flapper) was calculated using the following equation: - where P is the local surface pressure. The integral of PdA is determined by the CFD software, where As is the flapper surface area (also determined by the software).
The results indicate that the small pressure differences across the flapper bottom and the top surfaces cause a net force and moment to develop. These act in opposition to the safety valve's torsion spring force.
Because a 7-in. (or larger) TRSV flapper has a very large surface area, even a small pressure differential, from one side of the flapper to the other, produces a relatively large force and moment on the flapper. When the net force and moment acting on the flapper are higher than the torsion spring force, the flapper can remain in the open position.
A close examination of the results indicates that the pressure differences as well as the force and moment that develop are all proportional to the square of flow velocities. Thus, an important conclusion from the axisymmetric and three-dimensional CFD simulations is that the aerodynamic forces that develop on the flapper are proportional to the dynamic pressure of the flowing fluid (0.5*fluid density*flow velocity2).
Although the trend of the CFD simulations appears to agree with the laboratory experimental results, the accuracy of the calculated pressure differences across the flapper is unknown. The calculated average pressure difference across the flapper for a flow velocity of 303 ft/sec is only a fraction of a psi. This pressure difference is much smaller than the normal operating pressures of downhole safety valves (5,000 psi in this case), and such a small pressure difference would be very difficult to measure in full-scale test facilities.
Thus, the pressure difference across the flapper at high pressure and high velocities must be validated. The trend of the CFD results, however, agrees with the laboratory tests, and therefore the predicted trends can be a powerful tool for evaluating aerodynamic forces that may develop on the flapper.
Three-dimensional CFD simulations of a modified TRSV (interior) geometry were also conducted. The trend of the results is similar to that of the original geometry, except the magnitude of the ratio of pressure differences is decreased. The results of the CFD calculations on the (initial) modified geometry clearly indicate that the aerodynamic forces that are acting on the flapper can be reduced. Full-scale test results, discussed below, also confirmed that simple modifications to components inside the flapper housing can reduce the aerodynamic forces that develop on the flapper.
Full-scale test results
Full-scale gas flow testing was conducted on 7-in. and 9 5/8-in. TRSVs at a third-party test facility late in 1999. Natural gas from the supply grid for northeast England was used to provide high-volume, high-velocity flow - up to 20 MMscf/hr (480 MMscf/d).
Electronic data acquisition systems were used to collect flowing gas characteristics, TRSV flow tube position (an indication of valve functionality), and control fluid pressure data. In an attempt to simulate hydraulic conditions in a typical well, 2,000 ft of 1/4-in. control line were installed between the TRSV and the control panel. A common water-based control fluid was used.
At the time of preparations for the first round of full-scale tests, the CFD analysis was in the early stages and designers were not sure which flow parameters were most important for simulating actual gas well conditions. Test facility gas pressure was predicted to be in the range 600-800 psi, while flowing wellhead pressures were estimated to be approximately 1,200-2,400 psi. Deciding to err on the side of the most comprehensive test, initial gas flow rates were determined based on both Reynolds numbers and dynamic pressures for normal production and AOF conditions for several completion scenarios.
In an effort to improve understanding of the dynamic pressure inside the TRSV, a DP transducer attached to ports drilled in the side of the TRSV flapper housing, behind and opposite the location of the open flapper was used. As mentioned above, capturing very small pressure differences in a system of relatively high pressure is problematic.
Electronic flow tube position sensors were used to determine the location of the end of the flow tube. At no time did the flow tube interfere with the function of the safety valve. This information eliminated any question regarding the origin of the abnormal closure phenomenon mentioned below.
The 9 5/8-in. TRSV design was one of the first valves tested. It closed on command every time and continued to function nearly perfectly throughout the test. The original 7-in. TRSV design experienced difficulty closing when the through-bore velocity exceeded a critical limit, around 250-300 ft/sec. The valve's reluctance to close on command confirmed the suspicious results first encountered at the test facility. This phenomenon is now attributed to aerodynamic forces on the flapper.
The 7-in. and 9 5/8-in. TRSV designs were different in several ways. These differences led to modifications to the 7-in. TRSV design that resulted in aerodynamic performance improvement. Closure performance improvement was significant, but mechanical difficulties prevented completion of the test.
The results of the slam test series confirm that the anticipated aerodynamic forces are present and large enough to prevent the flapper from moving to the closed position. These tests also confirm that the flow tube does not play a role in the phenomenon.
Dynamic pressure differences are the most important consideration when predicting the magnitude of aerodynamic forces acting on SCSSV components.
The design enhancements mentioned above were based on this hypothesis, and preliminary CFD model results. The success of intermediate and final design support the belief that the hypothesis is correct. Successful, repeated slam closures conducted on both 7-in. and 9 5/8-in. TRSVs support the conclusion that large-bore, high-flow-rate gas wells can be safely controlled using existing SCSSV designs. However, designers and operators of large-bore, high-flow-rate gas completions should make themselves aware of the unique aerodynamic and mechanical loading conditions present in the event of an emergency shutdown.;
This article is a summary of paper OTC11881, to be presented at the 2000 Offshore Technology Conference in Houston.