Don W. Allen
VIV Solutions Inc.
Shell Exploration and Production
Vibration in subsea risers caused by vortex shedding is a challenge to operations in deepwater. Proper management of vortex induced vibration (VIV) requires a balance between acceptable movement and the cost of suppression. To address the questions involved in planning configurations for marine risers, and to investigate the impacts of typical conditions affecting VIV, two programs were conducted consisting of tests along a tubular (just under 100 ft long) that was towed under a rotating arm at high speeds to simulate conditions similar to those a deepwater riser may encounter.
The test results reveal a number of phenomena regarding the performance of helical strakes. Among the most important are:
- While strakes provide significant additional damping, they offer the most benefit when placed at locations corresponding to the maximum velocities encountered by the tubular
- Strakes placed at low velocity locations, at the expense of high velocity locations, still contribute to VIV suppression but mostly by adding damping
- It is not critical to cover 100% of each joint or region (i.e. 100% density) though suppression performance improves with increased coverage density
- While strake alignment is desirable, it may often be unnecessary since the degradation in strake performance from misalignment can be fairly small
- Relatively low to moderate amounts of uniform marine growth cause very little degradation in strake performance of an isolated tubular (this is not necessarily true of tandem tubulars or arrays)
- For triple start strakes with a fin height of 0.25D, pitch lengths of both 12D and 17.5D were quite effective with only small differences in their performance, thereby indicating that strake performance is relatively insensitive to pitch lengths in the range of about 12D to 18D
- Strakes increase drag above that of a bare cylinder.
All of the experiments were conducted in the Rotating Arm Basin at the Naval Surface Warfare Center in Carderock, Maryland. The basin is a circular indoor basin approximately 260 ft (79 m) in diameter and 20 ft (6 m) deep. Models are towed in circular paths through still water by a rotating arm. A 2.5-in. (6.3-cm) diameter test pipe was placed under the arm, supported by a truss assembly, for these tests.
The test pipes at each end were supported by an assembly that allowed the ends to rotate while allowing the test pipes to be pre-tensioned the same value. However, the variation of drag on the various models tested produced different tensions during testing. The tensions were measured at each end and the accelerations were measured at four locations along the pipe axis.
|Rotating arm setup.|
Aluminum connectors were made to fit inside the main joints (as an inner sleeve) and bolted to the fiberglass to connect adjacent joints. Helical strakes were modeled by gluing strips of Buna-n rubber onto the pipe surface. The strake cross-sections were rectangular, ranging from a base width to height ratio of 0.85 to1.0 (square) with most of the cross-sections square.
For these experiments, the rotation of the arm ensures that the test cylinder experiences higher velocity at the outer end than at the inner end. This means that the vortex shedding frequencies are higher at the outer end than at the inner end since vortex shedding frequencies are proportional to velocity. Since the fundamental natural frequency in bending for a long tubular is relatively low, there is a range of bending modes that may be experienced by the test cylinder (these modes are typically categorized as those that are in-line with the flow and those that are transverse to the flow; however, the primary focus here is on the transverse bending modes since those modes usually cause the largest fatigue damage and respond with larger displacements than the in-line modes). The relative flow at the outer end of the test cylinder excites higher modes than the relative flow at the inner end of the cylinder.
Deepwater ocean tubulars can experience a similar phenomenon due to changes in ocean current magnitude with depth.
It is important to understand this concept since a vortex suppression device will only suppress vortex shedding excitation locally. The response modes and frequencies are further complicated because helical strakes not only change the cylinder vibration amplitude but they can, and usually do, also change the cylinder vibration response frequency (frv). An effective helical strake will reduce frv, which can thereby reduce the vibration mode and the bending stress on the cylinder. Thus, effective helical strakes can improve the fatigue life of the cylinder in two ways: By reducing the response amplitude; and by reducing the response frequency.
Bare cylinder results
To properly assess the performance of a helical strake configuration, the bare cylinder (the cylinder without any helical strakes present) must first be examined.
The results illustrate several things. For one, the values for adjacent accelerometers are very close; the displacement values vary with Reynolds number (i.e. test velocity) and most likely mode shape too; the accelerations increase with increased Reynolds number/test velocity which is not surprising since the mode numbers and response frequencies are increasing too; the accelerations at each end of the pipe vary, but are relatively close in magnitude indicating that very little attenuation is occurring for the bare cylinder; and the drag coefficients decrease slightly with increased Reynolds number, most likely due to an increasing portion of the cylinder length entering into the transition region for turbulent boundary layers (i.e., experiencing the well-known "drag crisis").
These bare cylinder values are useful for comparison with the straked cylinder values. The accelerations are particularly important since accelerations are directly proportional to the bending stresses that cause VIV fatigue.
One difficulty of designing deepwater tubulars is determining how much of the tubular to cover with VIV suppression devices. Since most deepwater areas have ocean currents that are highest in magnitude near the surface and decay in magnitude with increasing depth, this determination often focuses on how deep to cover the tubular with suppression devices, in this case helical strakes.
Cylinders were covered with 17 ft, 9 in. and 35 ft, 6 in. of helical strakes on the outer end. (Unless otherwise designated, most of the helical strakes tested herein have a 0.25D fin height and a pitch per start of 17.5D; most have three starts but some data from the second test program is for strakes with four starts, called "quad-start" helical strakes). For a 17 ft 9 in. coverage length, a reduction in displacement of about 50% was observed. However, the corresponding reduction in acceleration was far more than 50% for this same coverage level. Further, this experiment indicated a greater reduction in both displacement and acceleration at the outer (high velocity) end of the cylinder than at the inner (low velocity) end. This is due to the strong local effectiveness and damping of the helical strakes.
At the outer end, the strakes reduce the excitation tremendously and thus the vibration is quite small. However, there is still a large amount of excitation present in the bare regions of the cylinder that excite VIV. The excitation in the bare regions occurs at a lower frequency (since the excitation frequency is proportional to the flow velocity) and thus excites lower modes. Since these lower modes are exciting a system with more damping present than what was present for the bare cylinder (due to the presence of the strakes), the displacements are lower and the displacements at the inner end are lower than those on the outer end. This indicates that the helical strakes are also providing substantial local damping.
Increasing the strake coverage length from 17 ft, 9 in. to 35 ft, 6 in. had the effect of reducing the displacements at both ends of the cylinder, at the expense of a small increase in the overall drag (it is well known that helical strakes have higher drag than most non-vibrating bare tubulars, and thus it is not surprising that additional strake coverage increases the overall drag). The displacements were also lowered by the additional strake coverage, again due to increased damping introduced by the additional strakes.
In short, covering only 18.4% (17 ft, 9 in.) of the outer end of the tubular reduced the accelerations at the inner end by a factor of well over two and reduced the accelerations at the outer end by a factor of about four to six (the exact amount of reduction is dependent upon the actual test velocity/Reynolds number). A suppression efficiency of over 80% was generated at the outer end simply by covering 18.4% of the tubular with helical strakes. Covering just 36.8% (35 ft, 6 in.) of the tubular reduced accelerations at the inner end by a factor of three to eight, and at the outer end by a factor of well over 10. Presumably the reductions might be even greater if currents were not experienced by the tubular on the bare region.
In the second test program using the same setup, quad-start helical strakes were tested with a wider range of coverage lengths than what was tested in the first test program. Starting at the outer end, one, two, three, and five joints were covered with helical strakes and the results compared with the bare cylinder results (from that same test program). To provide an overall view of the results, each of the displacement, acceleration, and frequency values were averaged for all of the tests and for all of the accelerometers.
As the strake coverage length was increased, accelerations and displacements decrease. Since the strakes decrease both displacement and frequency, the acceleration values were lower than the displacement values. At a coverage density just under 80%, the strakes reduced the average accelerations by over an order of magnitude.
|Figures a and b present the displacement and acceleration results for a bare cylinder with a relatively low level of roughness. These results were from the first of the two test programs. Note that the Reynolds number was chosen for the x-axis, but the test speed could just as well have been chosen. The Reynolds number was chosen since, in some instances, the fact that a portion of the test cylinder length experienced critical Reynolds numbers (transition to turbulent boundary layers on the cylinder surface) influences the results.|
Coverage density and location
While it may be attractive to simply strake a specific coverage length to obtain the desired suppression efficiency, this may not be possible due to the presence of connectors, anodes, etc. or it may be uneconomical. In these cases it is important to understand the impact of a lower density of coverage. Additional tests were conducted with 12 ft, 9 in. of triple start helical strakes on the outer end of each of the two outer joints (so that the total coverage length was 25 ft, 6 in.).
If the outer two joints had been fully covered, the coverage length would have been 35 ft, 6 in. While covering only the outer 12 ft, 9 in. of each 17 ft, 9 in. joint (thus, the outer end of the outer joints were covered) is not as effective as covering the entire joint length, the effectiveness with partial coverage was still high. Accelerations at the inner end of the test pipe were reduced by a factor of about four to five, while accelerations at the outer end of the test pipe were reduced by a factor of eight to 10. It is not critical to fully cover each joint with helical strakes to achieve a relatively high efficiency.
Even if the strakes are not located in the most desirable area (in this case, after the outer joint is straked, the inner most joint would be considered the worst possible joint to cover and the next to outer joint would be considered the most desirable joint to cover), they can still provide a significant amount of VIV suppression. The addition of strakes on the inner most joint reduced the accelerations by at least one-third.
Just because strakes are present locally does not imply that stronger acceleration reduction will accompany that presence. The decrease in acceleration by adding strakes to the inner most joint is not much larger at the inner end than it is for the outer end. Interestingly, with strakes on the inner end, the accelerations are still a little larger at the inner end than at the outer end. This also indicates that the addition of strakes on the inner most joint has mostly contributed only damping to the entire system. Most of the vortex shedding excitation is located closer to the outer end and not the inner end, thus, while the inner joint strakes suppress vortex shedding, they are not suppressing vortex shedding at the frequencies that are most influencing the cylinder response.
In practice, most helical strakes are made in short segments 5 ft to 6 ft long. Usually it is simple and quick to visually line up adjacent strake sections; however, there are some instances when lining up adjacent sections may be problematic (such as when retrofitting with an underwater vehicle). To examine the importance of strake alignment, an experiment was performed with strakes glued to the outer two joints of the test cylinder in three different arrangements: 1) The strakes attached in a continuous helix 2) The strakes attached in 15-in. sections and staggered 60° apart so that the next section started exactly halfway in between the previous section (the most severe misalignment possible) and 3) The strakes attached in 30-in. sections and staggered in similar fashion to the 15-in. section length strakes (note that since the pipe OD was 2.5 in., the 15-in. section is 6D long, while the 30-in. section is 12D long).
Results from this experiment showed that while the intentional strake misalignment produced a small degradation in the strake performance, all three alignment configurations were effective at suppressing VIV.
The majority of VIV analyses for deepwater tubulars use computational models built upon smooth cylinder data for both the bare tubular as well as for the suppression devices.
Overall, surfaces simulating soft marine growth (i.e., foam and carpet) had very little effect on the helical strake effectiveness, indicating that the strakes tolerate a modest amount of soft marine growth well.
While helical strakes can effectively suppress VIV using a range of coverage lengths, densities, and groupings, they also usually increase the mean drag forces on the tubular. This larger drag can sometimes be problematic (since it increases deflection of the tubular) and the choice of a VIV suppression device with lower drag such as a fairing may be worth considering.
Based on a paper presented at the ASME 33rd International Conference on Ocean, Offshore and Arctic Engineering, held in San Francisco, California, June 8-13, 2014.