Detection of corrosion in offshore risers using guided ultrasonic waves
Graham R Edwards
Paper presented at 26th International Conference on Offshore Mechanics and Arctic Engineering, OMAE 2007, San Diego, California, 10-15 June 2007. Paper no. 29407.
Guided wave ultrasonics is now an established technology for detecting corrosion in pipes. The technique is principally applied to pipe that is insulated, buried or otherwise inaccessible to conventional NDT. TWI has experience in applying the technique to offshore risers, where there are specific problems. Access beneath the cut-off valves of the platform is restricted and the clamps used to hold the riser and the coatings used to protect it in the splash zone can affect test performance. This paper will describe these problems in detail and discuss ways of overcoming them. Case studies will be used to illustrate the current state-of-the-art and TWI's research and development programme in long range ultrasonics for inspecting a wide range of components from pipes to railway lines will highlight future developments.
The risers that connect the pipelines on the sea-bed with the processing pipe-work on the production platform present particular challenges for Non-Destructive Testing (NDT). They are regarded as a critical part of the oil/gas production system and are therefore subject to frequent inspections There is the possibility of not only internal corrosion and erosion caused by the pipe contents, but also external corrosion and erosion caused by the surrounding sea-water with attendant currents and wave action. Moreover, production platforms are busy places, where collisions with support vessels are a hazard. Even global warming has an effect by increasing the risk of damage from storms.
The first challenge is to gain good access for the NDT sensors. Risers are often difficult to reach ( Figure 1), may be encased in a caisson ( Figure 2) or covered in a thick protective coating.
Fig.2. Risers surrounded by a caisson
Thorough visual inspection is the primary method of detecting damage. This may be done under the platform from staging. Further down the need for scaffold staging can be avoided using rope access ( Figure 3). Below sea level, divers (Figure 4), or more commonly remote operated vehicles may be deployed ( Figure 5). The greatest access problem lies near and through the 'splash zone', where ROVs cannot operate at all.
Fig.3. Rope access (Courtesy of CAN)
Fig.5. ROV Deployment of NDT
Visual inspection will detect external corrosion, most likely associated with damage to the riser's protective coatings. But for detecting damage under the coatings and on the internal riser surface NDT techniques are needed.
With access from the outside of the riser, corrosion and erosion on the inside can be detected using pulse-echo ultrasonics. A simple portable instrument can be carried by a rope access technician down to the inspection site and measurements made of residual wall thickness. For survey purposes, the readings are taken over a grid of so-called TMLs or Thickness Measurement Locations. This will detect and monitor general corrosion, but will not detect isolated corrosion pits. Corrosion pitting is very difficult to detect by any NDT method. This is particularly true if it is caused by microbes and can occur almost at random. If corrosion pitting is suspected then the pulse echo ultrasonics must be carried out with mechanical probe scanners ( Figure 6) and B-scan and C-scan images used to identify the corrosion pits. Pits as small as 5mm diameter and 1mm deep can be detected and measured in this way. But the inspection process is slow. A raster scan of an area100mm wide around a 12" pipe may take half an hour. More rapid corrosion mapping ultrasonic mapping systems are available, but their resolution may not be good enough to detect isolated pits.
Fig.6. AUT for corrosion mapping
Ultrasonic pulse-echo methods cannot be applied through coatings. Coupling of the ultrasound probes to the pipe surface is always a problem. Marine growth and the hard calcareous deposits that fix it to the metal surface must be removed. In recent years this has led to the introduction of non-contact methods such as pulsed eddy-currents, where the sensor can stand-off from the pipe surface by several centimetres. Eddy-current techniques have been used extensively in the detection of corrosion in aircraft. On wing skins, the eddy-current field is generated beneath an excitation coil in a small hand-held probe. This field can penetrate the skin into the sub-structure and detect corrosion in the spars. The technique is sensitive to changes in the metal volume however, and does not measure the remaining wall thickness. This is a limitation of all electromagnetic techniques. Another is that the depth of penetration of the eddy-current field into the metal is only a few millimetres, even though the coil may stand-off from the metal surface by centimetres. This is a result of 'skin effect'. This also determines that the effective depth of penetration of the eddy-current field will decrease as its AC frequency increases. Test frequencies in the 100KHz range are the most common, rising into the MHz range for surface crack detection only. Pulsed eddy-currents techniques overcome this effect by using pulses of eddy-currents instead of continuous currents. Pulses of eddy-currents contain a spectrum of frequencies, the lower frequencies provide the penetration. Modern electronic filters are able to detect the decay in the eddy-current pulse in the pipe wall and give sensitivity to metal volume loss through several centimetres of steel. This is in addition to being capable to testing through thick, non-conductive coatings.
Another electro-magnetic technique that can be used to detect corrosion in the pipe wall, either on inside or outside surfaces uses a powerful magnet to saturate the pipe wall and sensors to detect flux leakage caused by the metal wall loss. Used extensively in the survey of storage tank floors, the technique has been adapted for use on pipes. Like eddy-currents, the magnetic flux leakage technique is sensitive to volume loss rather than residual wall thickness, but cannot be used through thick coatings.
The electromagnetic sensors may be deployed externally from scanners that are able to operate through the critical splash zone ( Figure 7). Alternatively, if access is possible to the inside of risers, then 'pigs' that carry the NDT sensors can be used. It is impracticable to use the type of 'pig' found in normal pipeline surveys, that move with the flow of the pipe contents, because of the vertical nature of the riser. Instead tethered pigs have been developed ( Figure 8). These are lowered into the riser to the test position.
Fig.7. NDT Deployment through the Splash Zone
Fig.8. Tethered 'Pig' ( Courtesy of RTD)
Guided Wave Ultrasonic methods
To meet the inspection needs of its industrial members in the oil, gas and petrochemical industries for an NDT technique that could detect corrosion under insulated pipe, TWI developed the long range Guided Wave Ultrasonic (GWU)technique during the early 1990s. By using the pipe as a wave guide, TWI had found that pulses of ultrasound could be propagated over distances of 100m. Extensive studies determined the essential parameters that influence the test performance and the first test system became available commercially through TWI's subsidiary, Plant Integrity Ltd in 1997.
Guided waves are complex in nature. Like plate or Lamb waves, they are caused by the interaction between the two parallel boundaries of the media through which the ultrasound is propagating. They are therefore formed only where the boundaries, the inside and outside surfaces of the pipe, are close enough to interact. As the frequency of the ultrasound decreases, so the wavelength increases and the wall thickness through which guided waves propagate can be expected to increase. However, effective GUW techniques are restricted to pipe walls less than 1½" thick to keep test frequencies above 20 KHz.
In their simplest forms the interaction between the inside and outside surfaces are asymmetric or asymmetric longitudinal waves ( Figure 9). The wave may also be a simple torsional wave, where the particle movement is transverse to the direction of propagation. Pipes are also able to flex giving rise to another family of wave modes; flexural waves ( Figure 10). As the interactions become more complex, so do the wave modes. To differentiate the fundamental longitudinal, torsional and flexural waves from each other, they are given the letters L, T and F respectively with a suffix containing 2 numbers; the first the wave order and the second its harmonic. The simplest longitudinal wave is therefore the L (0,1) , the next is L (0,2) . As well as the complexity of the waves, there is also the problem that they propagate at different velocities and are dispersive. Dispersion describes the property of waves that propagate at velocities that change with frequency. A pulse of waves will by its nature contain a spectrum of frequencies, the lower frequencies propagating more slowly than the higher ones. Therefore as the pulse propagates, it will lengthen or 'disperse'.
Fig.10. Wave modes in pipe
The changes in velocity with frequency have been calculated and are shown in ( Figure 11). The so-called 'dispersion curves' illustrate a number of interesting characteristics. Firstly the number of wave modes increases as the frequency increases. For each wave mode, there is therefore a 'cut-off' frequency below which it does not exist. For example the L (0,2) wave does not exist at frequencies below about 15KHz for this specific pipe diameter and wall thickness. Secondly the torsional waves are non-dispersive, which makes them an attractive option for inspection. Thirdly, each wave mode is very dispersive at low frequencies, that is to say a small change in frequency causes a relatively large change in velocity. As the frequency increases, so the level of dispersion decreases. In fact, all the curves can be seen to converge on a value that is in fact the velocity of surface or Rayleigh waves.
To avoid using dispersive ultrasound waves, frequency selection is therefore very important in GWU. The frequency of the wave mode is selected so that the velocity is on a flat part of the dispersion curve.
Dispersion curves can be calculated and plotted for any pipe-wall and wall thickness combination. An important first step in any GWU procedure is to find the pipe diameter and wall thickness. From these measurements the dispersion curves can be plotted and the velocity of the wave at any given frequency found. Fortunately, this is usually done within the software of the equipment.
Like conventional ultrasonics, GWU uses an A-scan with calibrated time base to measure the distance to reflectors ( Figure 12).
Fig.11. Dispersion curves for a 10", 8mm wall pipe
Fig.12. Schematic of GWU A-scan
An example of guided wave equipment is shown in ( Figure 13). The transducers are set into a ring that envelopes the pipe. Alternatively, the guided waves can be generated by the magneto-strictive effect caused by pulsing a current through a coil wrapped around the pipe. The propagated wave, usually an L (0,2) or T (0,1) must be symmetrical around the pipe and the transducers are therefore packed closely together, otherwise interfering flexural modes may be introduced. There must be at least two rings if the forward directing pulse is to be distinguished from the backward one. When using L waves, the transducers, which are shear wave propagators, are aligned parallel with the pipe axis. When using T-waves the transducers are aligned transverse to the pipe axis. With L waves a third ring is needed to remove unwanted L (0,1) waves by destructive interference. The dispersion curves show us that L (0,1) waves are always present with L (0,2) waves. L (0,1) waves are highly dispersive, travel at about half the velocity of L (0,2) waves and if not removed entirely, cause unwanted 'ghost' signals on the A-scan. L (0,2) waves are selected because they are almost non-dispersive above a certain frequency, shown by a flattening of the dispersion curve. In reception mode, the transducers work as a phased array and are able to distinguish flexural from symmetrical modes.
A liquid couplant is not necessary with GWU transducers. The resonating motion of the transducers is parallel with the test surface and not perpendicular to it and the coupling is by friction. The transducers need only bind tightly with the surface for good coupling. This can be achieved by wrapping an inflatable collar around the transducer rings. Inflating this collar to the pressure of a bicycle tyre is enough to exert sufficient pressure on the transducers for them to generate the shear motion in the surface of the pipe. A slight roughening of the surface can improve coupling even further.
In addition to the transducer rings, a pulser-receiver is needed close at hand to generate the pulses and receive the signals. An umbilical of up to 100m length conducts power and signals between the pulser-receiver and a lap-top computer which contains the software for driving the pulser-receiver and collecting, analysing and displaying signals.
A typical A-scan is shown in ( Figure 14). This one shows three A-scans superimposed upon one another. One for the signals from received symmetrical waves (either L-wave or T-wave) coloured black, another for received F-waves polarised in the vertical direction coloured blue and another for F-waves polarised in the horizontal direction coloured red. The vertical and horizontal motions are detected by having the rings divided into quadrants and the receiver sensing the relative motion in each. Reflected F-wave signals are normally associated with corrosion, reflected symmetrical waves with geometric features such as butt welds and pipe collars.
Fig.14. Typical GWU A-scan display
The A-scan shows symmetrical signals occurring at equal intervals. These are from welds in the pipe. They follow a distance amplitude correction or DAC curve. The DAC curve is a blue line. Remarkably, the DAC from pipeline welds is consistently 14dB less than (20%) of total reflection form the pipe end, shown in the top right-hand corner by a black line. For this reason the weld signals are used to calibrate the A-scan, rather like the side-drilled holes used in conventional ultrasonics. The recording threshold (green line) is taken as 18dB below this level and is equivalent to a loss of 9% in cross-sectional area of the pipe wall.
An interesting feature of GWU is that it does not rely on specular reflection of the ultrasound. This is because the reflections are caused by the acoustic impedance change of the pipe wall wherever there is a change in thickness. This may be due to an increase in wall thickness as well as a decrease in wall thickness. For this reason welds produce signals. The corrosion may be on the inside or the outside surfaces. If the corrosion is on one side of the pipe, then the reflection will have a mainly flexural component and the plane of the flexural wave will indicate the O'clock position of the corrosion. The intensity of the reflected ultrasound will be proportional to the loss in cross-sectional area. Therefore a wide shallow area of corrosion may give the same signal as a narrow deep area of corrosion. It is not a technique that can be relied upon to detect isolated corrosion pits.
GWU used on risers
The portability of the equipment makes it well suited for inspecting risers off-shore. The transducers are simply wrapped around the riser ( Figure 15), the cables connected to a pulser-receiver a short distance away and an umbilical ran away to the lap-top, which may be up on the platform. Often the transducer ring has to be placed just above a caisson that protects the risers through the splash zone ( Figure 16). The aim of the inspection is to propagate the guided waves so that they penetrate down through the splash zone as far as the J bend on the sea-bed. Corrosion will reflect echoes that appear as signals on the A-scant hat can can be compared with those from welds. An example is shown in ( Figure 17) with the cut section of the pipe with external corrosion.
Fig.16. Placement of transducer ring above a caisson
Fig.17. Corrosion detected next to a weld
The range of the guided waves is severely limited by coatings around the pipe. If they are 'sticky', they will adsorb the ultrasound. It is then not possible to reach the bottom of the riser. In some cases the sticky nature of the riser coating in the splash zone is enough to absorb all of the ultrasound. If at least two weld signals cannot be seen on the A-scan, then the DAC curve cannot be set and the test becomes invalid The riser clamps are another feature that can attenuate the energy of the ultrasound energy.
Overcoming the attenuation caused by 'sticky' coatings in the splash zone has become the main aim in GWU developments for riser inspection from the surface.
One approach is simply to increase the amplitude of the guided wave pulse. New types of transducer are being investigated, including ones that are layered to amplify the pulse, and others that are flexible to provide better coupling on the curved pipe surface. The transmitted pulse can also be amplified by increasing the number of rings and applying a delay to the transducer excitation pulse causing constructive interference and an increase in pulse energy. However, simply increasing the signal amplitude may not be an advantage if the noise increases proportionally.
The detection of signals in noisy A-scans has become the objective of new signal processing techniques such as 'time-reversal'.
Another approach is to use the transducer ring as a phased array and focus the transmitted pulse. Phased array ultrasonics is a recent innovation in NDT, although it has been used for many years in clinical ultrasonics. By using an array of small transducers instead of one monolithic transducer, the individual elements can be excited in such a way that the individual wave fronts can be converge at a focal point.
The individual transducers in the ring can now be used as the elements in the array. Again the pipe acts as a wave-guide to accentuate the effect ( Figure 18). The focal point can be orbited around the pipe as well as backwards and forwards along the pipe. By stacking all the A-scans together an image from a phased array 'sweep' can be produced of the pipe surface, which is similar to a C-scan image used in conventional ultrasonics to map corrosion ( Figure 19).
Fig.18. Focussed guided wave
Fig.19. Imaging with dynamic focused GWU
The good reproducibility of guided wave ultrasonic tests make it an ideal technique for monitoring corrosion in a pipe. The transducers can be set precisely at inspection points and the calibrations made identical on the same weldseach test. The next step is to permanently install the transducers on the pipe.
A current project sponsored by the European Commission is developing an LRUT system to monitor steel catenery risers at the touchdown point on the sea bed. Named Risertest, the project involves marinisation of the transducer rings to operate at depths of 2000m. The main hurdle is the presence of thick coatings. The transducers will have to be installed at the surface as the SCR is installed. Long term reliability is a key issue.
Guided wave ultrasonics is now an accepted tool for the NDT of risers. Currently it is used to screen risers for corrosion. It can locate the areas that require further evaluation with techniques such as ultrasonic corrosion mapping. Future developments may include techniques that allow GWU to map corrosion.