Plastic pipes offshore - a look below the surface

TWI Bulletin, January/February 1995

Ian Munns graduated from the University of Hertfordshire in 1992 and joined TWI later that year. Since then he has been widely involved with the NDT of plastics, composites, ceramics and adhesively bonded joints. Ian is currently working in the Numerical Modelling Section in the Structural Integrity Department.

Glass-reinforced plastic pipes are increasingly being adopted for use on offshore installations. How should we be non-destructively testing them? Ian Munns reports ...

At present the majority of these pipes are used to transport non-hazardous fluids at relatively low pressure, usually below 25bar. In such cases, a minor leakage can be tolerated, but a total failure can lead to a halt in production and an associated loss in revenue. Therefore, an ability to inspect the quality of these low pressure pipe systems is desirable. With the onset of glass-reinforced plastic (GRP) piping in higher, more critical pressure applications, the ability to ensure quality is not merely desirable, it is essential.

It has been estimated that the flow line pressures needed to transport hydrocarbons in the Norwegian sector of the North Sea, would be 5-10 times higher than the operating pressures of existing GRP pipe systems. ^[1] If GRP is to be used in critical applications, such as this, then it is important that potential failure mechanisms are studied and understood. Much work has been carried out in this area and guidelines specifying the design, fabrication and testing of GRP pipelines for use onshore ^[2,3] and offshore ^[4] have been produced. Experience gained from GRP is generally good. The presence of documentation to aid the design and installation of GRP pipe systems has, without doubt, helped to reduce the number of failures. However, failures do still occur, and those that do are mostly due to poor installation, especially with regard to the adhesively bonded join.

Defect types in GRP pipelines

The majority of defects within GRP pipe are caused by poor handling of the pipe during installation. Impact damage is a particular problem. An isolated impact on the exterior of a GRP pipe can leave little or no visible trace. However, beneath the surface a combination of interlaminar cracking and delamination may penetrate a significant proportion of the pipewall thickness.

Other potential defects occurring in the GRP itself can often be attributed to errors in the fabrication of the pipe. For example, in filament wound pipe, the winding angle may be incorrect, a contaminate (such as oil or water) may have been introduced at an intermediate fibre layer during the winding process, or the resin curing cycle may have been incorrect.

Chemical degradation of the resin system is another type of defect that can potentially occur, although failures due to this are rare. ^[6] In a ten-year study of GRP piping by the Exxon Corporation, ^[7] ultraviolet degradation on above ground piping has been observed. No failures have been attributed to this problem and the significance of the observation is unknown. In summary, the information available at the present time suggests that the more critical defect types which may occur in GRP pipe itself, are delamination and resin cracking.

The most popular means of joining individual sections of GRP pipe is by adhesive bonding. Commonly, an epoxy type of adhesive is used, consisting of a resin and a hardener. Some typical adhesively bonded joint configurations are shown in Fig.l. Previous studies ^[5] have shown that the bonding procedure used governs the quality of the adhesively bonded joint and that a poor bonding procedure can ultimately lead to joint failure. Adhesively bonded joints are a potential source of failure and their integrity must be assessed.

Fig. 1. Typical adhesively bonded pipe connections:

a) Sleeve;

b) Straight/tapered;

c) Taper/taper bell and spigot

Adams and Cawley ^[8] categorised defects within adhesively bonded joints according to their location:

• Those within the adhesive layer itself;
• Those within the proximity of the adherend-adhesive interface.

These are illustrated in Fig.2. Of these defect types, NDT techniques exist for the detection of gross defects, such as disbonds, voids and porosity. However, there is no widely accepted technique for detecting poor cohesion of the adhesive, and, mostimportantly, poor adhesion at the adhesive-adherend interface. ^[9]

Fig. 2. Typical defects within adhesively bonded joints

Methods of NDT

This article concentrates on ultrasonic methods of inspection for composite materials. Other methods of NDT do exist for the inspection of these materials, for example - low energy radiography, acoustic emission, thermography, shearography and resonance techniques. Often though they are less versatile than ultrasonics, and some may require specialised safety procedures. In an offshore application it is envisaged that a composite pipeline would be inspected periodically once it has entered service. Unlike acoustic emission and some shearographic techniques, an ultrasonic technique would not require such a pipeline to be loaded above normal operating pressure in order to examine its integrity. Also, the use of an ultrasonic technique alleviates the need for specialised safety procedures which have to be enforced during a radiographic inspection. For these reasons the other methods of NDT will not be discussed here, this article will only concern itself with ultrasonic methods of inspection.

The conventional method of ultrasonic inspection is a normal incidence technique. A piezo-electric crystal is excited by an electrical impulse to generate longitudinal (compression) waves of known frequency, incident at an angle perpendicular to the composite surface. Any discontinuity present within the thickness of the composite material, such as a variation in resin/fibre content, a delamination or fibre breakage will cause a proportion of the incident ultrasonic energy to be reflected. In this way, an analysis of the reflected waveform can be used to detect possible flaws in a composite material.

However, the fibrous, anisotropic nature of many composite materials can increase the complexity of this ultrasonic inspection technique. The strands of fibre contained within the matrix material are not conducive to ultrasonic wave propagation. Ultrasound is scattered by the fibres as it propagates through the thickness of the composite material and the energy of the ultrasonic wave decays rapidly, i.e. attenuation is high. Attenuation of the shorter ultrasonic wavelengths (higher frequencies) is especially rapid and, as a result, it is often necessary to inspect composite materials at a frequency lower than that which would typically be applied to a similar thickness of metal. Inspection at a lower frequency may help to overcome the attenuation problem, but the use of larger ultrasonic wavelengths sets a limit to the detectability of small flaws.

Material anisotropy is another potential problem. For example, the fibre content of chopped strand mat GRP may vary from one position to another. Associated with this local change in fibre content is a local change in the through-thickness speed of ultrasound propagation. Consequently, the use of a conventional ultrasonic thickness gauge may lead to incorrect values for material thickness being obtained. ^[10]

Coupling ultrasonic energy into the material may also be problematic. Certain GRP interconnections, such as elbows and T joints, are fabricated in a fairly rudimentary manner and their surface finish is often uneven. In these instances, inspection of the adhesively bonded interconnection using a normal incidence ultrasonic technique may be complicated by a difficulty in achieving a standard contact between the ultrasonic probes and the composite surface.

Having discussed some of the difficulties in practically applying ultrasonic techniques to composite materials, a number of developments have been made with a view to overcoming these obstacles. Transducers with a specially controlled frequency bandwidth have been used to improve the ultrasonic inspection of GRP. ^[11] With this system, the signal to noise ratio is enhanced by effectively discarding the higher frequencies of the driving pulse. In addition, other systems have attempted to alleviate the problem of inconsistent coupling between probe and testpiece. Commercial water jet systems exist for the rapid scanning of composite components ^[12] and Shell Research have recently developed an Ultrasonic Through Transmission Testing technique capable of inspecting the adhesively bonded connections in liquid-filled GRP pipelines for lack of adhesive. ^[13] The t

As well as the significant improvements to composite inspection discussed above, TWI is developing a specialised ultrasonic technique with the potential to inspect adhesively bonded joints between composite components with increased sensitivity. Reference has already been made to ultrasonic techniques that are able to detect disbonding and a lack of adhesive in adhesively bonded joints. However, at present, there is no widely accepted technique capable of assessing the cohesive and adhesive properties of an adhesively bonded joint.

It has been shown that, at normal incidence, ultrasonic shear waves are more sensitive than compression waves for detecting small (submicron) liquid filled gaps in a layered structure. ^[14] Consequently, many researchers ^[15-19] are exploring the potential of ultrasonic waves with a transverse (shear) component for inspecting the adhesion qualities of adhesively bonded joints. TWI is considering a category of waves capable of introducing a non-destructive shear stress at the adhesively bonded interface. The waves considered are termed interface waves and are capable of propagating along the adhesively bonded interface itself.

Interface waves may be classified in two groups according to their energy flow characteristics. If, as the wave propagates, ultrasonic energy is leaked away from the interface region into the less dense material, then the wave is dispersive and termed a 'leaky' interface wave. If, however, the interface is capable of supporting the wave so that it may propagate non-dispersively along the interface region, then a true guided wave exists, termed a Stoneley wave.

Work at TWI ^[20] has concentrated on modelling the propagation of interface waves through an adhesively bonded lap joint. This work has demonstrated the versatility of an interfacial wave approach to bonded joint inspection. If the wavelength of an interface wave is selected to be larger than the thickness of the adhesive, then it is possible to assess the bulk properties of the adhesive layer using this technique. If, however, the wavelength of the interface wave is designed to be much smaller than the thickness of adhesive, then the technique is potentially sensitive to variations in the quality of the adherend-adhesive interface.

The TWI method of interface wave inspection relies on mode conversion from other wave types to generate waves at the interface. For example, on a bonded lap joint, one technique may be to propagate a wave, whose energy is confined to the surface, along the substrate towards the interface region. The instant this surface wave reaches the interface, its ultrasonic energy is reflected and transmitted in a number of directions. For certain material combinations it is possible for energy to be coupled to the interface and for interface waves to propagate. Figure 3 shows interface waves propagating along the two adhesive/adherend interfaces in an adhesively bonded joint. Despite offering, probably, the most sensitive means of detecting interfacial imperfections, true guided interface waves, i.e. Stoneley waves, are inherently difficult to generate and detect. For this reason, methods using leaky interface waves have been investigated. An experimental technique using leaky interface waves has succes

Fig. 3. Computer visualisation of an ultrasonic interface wave propagating through an adhesively bonded joint

Fig. 4. Experimental configuration for detection of leaky interface waves

The guided nature of an interface wave propagating along the adhesive layer between two metal adherends is shown in Fig.3. Due to the acoustic mismatch at each metal-adhesive boundary, the ultrasonic energy is effectively confined to the adhesive layer and very little energy is 'leaked' across each metal-adhesive boundary as the wave propagates. In the case of two composite adherends, however, the energy distribution is expected to be different. Composite adherends, with an epoxy-based matrix, will exhibit acoustic properties much similar to those of the adhesive layer. Consequently, as the ultrasonic wave propagates along the interface, more energy will be leaked from the interface into the composite adherends than for the metal adherends shown in Fig.3. In this way, a leaky interface wave can be detected by a surface mounted compression wave probe situated as in Fig.4. An analysis of this transducer response could feasibly be used to detect a weakness in adhesion and poor cohesio

In conclusion, reliable ultrasonic methods exist for the detection of gross defects, such as delaminations, inclusions and fibre agglomeration in composite materials.

However, there is no widely accepted technique for detecting poor cohesion and, most importantly, poor adhesion between adhesively bonded composite structures. Based on evidence from recently published research, TWI is investigating the use of ultrasonic interface waves as a potential solution to this adhesive inspection problem with a considerable level of success. This work is being carried out within a Group Sponsored Project.

References