Subscribe to our newsletter to receive the latest news and events from TWI:

Subscribe >
Skip to content

Long range inspection of engineering assets using guided ultrasonic waves (September 2008)

   
Chiraz Ennaceur, Peter Mudge, Tat Hean Gan

Paper presented at BINDT annual conference 2008, 15-18 September 2008, Macclesfield, Cheshire, UK.

Abstract

Long Range Ultrasonic Testing (LRUT) is a technology that has been developed over the past decade and is finding new potential applications throughout the energy, process plant, transports and engineering industries. LRUT is a non-destructive technique that has the potential to inspect engineering structures covering hundreds of metres from a single test location. This technique provides 100% screening coverage for in-service degradation and has the ability to inspect inaccessible areas. It also enables the inspection without removing the insulation or the coating, except at the location of transducers thus reducing costs of gaining access.

This paper will look into different types of application for guided ultrasonic wave technology to monitor and inspect various types of engineering assets with different geometry e.g. rails and offshore sheet pile. The functional capability of guided waves in terms of test range, defect detection capability, defect positioning and flaw sizing will be illustrated in this paper.

1. Introduction

Long Range Ultrasonic Testing (LRUT) is a technology that has been developed in Europe over the past decade and is finding new potential applications throughout the energy, process plant, transports and engineering industries.

The aim of this work is to develop new technologies for the maintenance and inspection of the following European engineering assets (rails and sheet piles) which are now ageing, thus posing a considerable risk of structural failure due to degradation mechanisms such as corrosion or various forms of environmentally assisted cracking (fatigue, corrosion fatigue, stress corrosion, etc.). This paper will present the results of three engineering assets which are rail, sheet pile and heat exchanger tubes. In Europe there is 512,000km of rails and there are thousands of kilometres of Sheet-Piled River Wall and Sea Defences. The annual European maintenance and inspection budget for the above engineering assets is approximately €225 billion. Conventional non-destructive testing (NDT) methods e.g. manual ultrasonic testing (UT) and eddy current for inspecting the above engineering assets have been in use for over 50 years. These techniques however possessed several drawbacks. For example, at a given position, only a very small area can be inspected as their maximum range is measured in tens of millimetres. Thus they require many man months of effort to inspect large structures e.g. pipelines and pipework, bridge cables, long sections of rail etc. These techniques require direct access to the entire structure, i.e. all insulation and coatings must be removed and buried components must be exposed by excavation. Access cost typically exceeds inspection cost by a factor of 5-10 times and overall cost is prohibitively high. In addition to the fact that they are so time consuming and costly that the amount of inspection actually carried out is far less than is required to ensure long-term structural integrity.

This paper looks into different types of LRUT applications:

  1. Asymmetrical structure (e.g. rail) and
  2. Large plate

2. Principle of Long range ultrasonic testing

Up to date, LRUT has been applied commercially to one-dimensional components (to a limited extent to rods, cables and rails) but almost exclusively to pipes, mainly in the oil, gas and chemical industries. The principles of LRUT for the inspection of pipes are contrasted with conventional NDT methods in Fig.1. All conventional methods (ultrasonic thickness gauging, eddy current, digital radiography, ACFM, etc.) inspect a volume of the component under the foot-print of the search device. (Fig.1(a)). In LRUT, a symmetrical circular wave is transmitted from a ring of piezoelectric or EMAT transducers clamped around the pipe.

spcesept08f1.gif

Fig.1a) Current inspection technologies can only inspect a small area (few cm2) underneath the transducer

Fig.1b) Proposed system using guided waves will revolutionise inspection tasks since it will be able to inspect hundreds of meters of engineering assets from one location even when these are buried underground or under insulation and coatings


The guided ultrasonic wave signals generated from the LRUT tool may be longitudinal or torsional. The signal frequency is low and can vary from 20kHz-100kHz. Features in the pipe (such as corrosion and erosion) will reflect transmitted ultrasound back to the transducer ring. The time-of-flight from the ring to the feature and back to the ring enables the position of the feature along the pipe to be established with acceptable accuracy (better than ±100mm). Symmetrical features such as welds reflect symmetrical waves back to the transducer array. However, asymmetrical features such as corrosion cause wave-mode conversion and this enables them to be distinguished from welds. The amplitude of the reflected signal is a function of the total change in cross section caused by the feature as a proportion of the total cross sectional area of the component. Flaws as small as 3% of the cross sectional area can usually be seen at the limit of detectability. However, the level of reliable reporting is limited to ~10% of the cross sectional area. (The industry would welcome improved sensitivity to 1% of the cross section). As the range to the feature increases, the signal-to-noise level reduces.

The maximum usable range is that at which signals arising from reportable features cease to be distinguishable from the noise floor with sufficient confidence. This range is determined by many factors associated with the specific pipe (viscosity of pipe contents, condition of pipe, geometrical complexity, and particularly the presence or otherwise of viscoelastic wrapping). Fig.2 shows a numerical model of guided ultrasonic wave propagation in a pipe with a branch. It can be seen that part of the energy has been diverted into the branch thus reducing the concentration of energy in the main pipe. This will also shorten the distance of wave propagation in the main pipe.

spcesept08f2.gif

Fig.2. Visualisation of ultrasound propagation in a pipe with a branch


3. Long range inspection of rail structure

3.1 Experimental setup for rail inspection

In this experiment, 3 types of investigations on rail structure were carried out, namely rail head, web and foot. First experiment was carried out on rail head. In order to achieve this, a 2x2 matrix of piezo-elements has been used. The sensor array was placed on the top of the rail head with the distance between the elements and the end of the rail is equivalent to the wavelength, λ and also the horizontal distance between the elements is equal to λ. The second experiment was carried out on the rail web. The investigation is focusing on exciting the rail web with a T2 wave mode at 70 kHz frequency. In order to generate T2 wave, the same matrix which was placed on the rail head was placed on the web. The third experiment was to investigate guided wave propagation in the rail foot. The objective of this experiment is to generate the F2 mode using the array of piezo-elements which have been used for the first two experiments. The wave propagation were modelled numerically.

3.2 Results and discussion

It has been known that the propagation of guided waves in complex geometries such as rail structure is extremely difficult. Rail track has irregular shape and this makes guided waves difficult to propagate. In order to achieve a good signal to noise ratio (SNR) for LRUT inspection, the theoretical study has been carried out to study the characteristic of the guided wave in the rail structure i.e. head, foot and web. Fig 3 shows the three types of wave modes that propagate on the 3 location of the rail structures. These wave modes were identified and selected for guided wave application because of their sensitivity and efficiency.

F3, T2 and F2 were selected for the head, web and foot of the rail to be excited accordingly. Furthermore, it has been found that for the rail, the selected wave modes for each section can be generated by a unique set of transducer arrays. These modes have been chosen for different reasons:

  • Ease of generation, by choosing the appropriate transducer array set-up
  • The selected modes have a similar group velocity, which reduces the complexity and defect position
  • The modes give 100% coverage for the cross-section.
spcesept08f3.gif

Fig.3. Modelling of different wave modes that present in different section of the rail structure

a) F3 wave mode in the head
b) T2 wave mode in the web and
c) F2 wave mode in the foot

3.3 Inspection of rail head

The F3 mode was successfully generated on the head as shown in Fig.4(b). In this experiment, a 1mm saw cut defect has been introduced to the rail head. This defect is located at 4.46m from the position of the piezo-elements arrays, as shown below in Fig.4(a). According to the theory, the saw cut defect should be represented on the A scan by a peak at 2877m/s. The experimental data in Fig.4(b) shows that the signal from the defect arrives at 2834m/s. This clearly shows good correlation. 

spcesept08f4a.jpg

Fig.4a) 2mm depth defect on the railhead

spcesept08f4b.gif

Fig.4b) Results obtained with a 2mm saw cut defect on the railhead

3.4 Inspection of rail web

The objective of these experiments is to generate the T2 mode in the web and verify the modelling results shown in Fig.3(b). Two defects were introduced in the web as shown in Fig.5(a) (Defect 1: at 6 m from the position of the piezo-element array and Defect2: a hole at 6.22m from the position of the piezo-element array). The theoretical arrival times for the Defects 1 and 2 are 4000µs and 4146µs respectively. In Fig.5(b), the experimental waveform shows that the signals from the two defects arrived at 3890µs and 4132µs respectively. Theses peaks thus arrive at the right time, and the defects are clearly represented on this A scan.

spcesept08f5a.jpg

Fig.5a) Defects 1 and 2 in the rail web

spcesept08f5b.gif

Fig.5b) Zoom on the wave reflected by the rails end and the 2 defects

3.5 Inspection of rail foot

The final experiment is to illustrate that F2 mode is able to detect defects in the foot. A 1mm saw cut was introduced at the edge of the foot as shown in Fig.6(a). The defect is located at 4m from the position of the piezo-element arrays and the collected results for the 4 actuators are shown in Fig.6(b). According to the theory, the saw cut defect should be represented on the A scan with a peak at 2643m/s and this correlates well with the experiments.

spcesept08f6a.jpg

Fig.6a) Defect in the foot

spcesept08f6b.gif

Fig.6b) Results obtained with a 5mm saw cut defect on the foot of the rail

4. Long range inspection of large plate structure

4.1 Experimental setup for large plate inspection

The plate used for these experiments is an aluminium plate with 158cm length, 125cm width and 1cm thickness. In this experiment, the guided waves have been generated by shear piezo-elements. A frequency sweep has been done and it brought out an optimal frequency at 70kHz. The dispersion curves were drawn for a 1cm thickness aluminium plate. The curves show that at 70 kHz, 2 modes could be generated: A0 and S0. The characteristics of these 2 modes are presented in Table 1.

Table 1 Characteristic of the modes A0 and S0 at 70 kHz on the aluminium plate with 10mm thick

A0
Frequency kHz Phase velocity (m/s) Group Velocity (m/s) Wave length λ (cm)
70 2110 3065 3.01
S0
Frequency kHz Phase velocity (m/s) Group Velocity (m/s) Wave length λ (cm)
70 5397 5310 7.71
The theory shows that A0 is very attenuative comparing S0 and for the inspection of the large sheet piles or sea defence plates. Therefore it is preferred to use S0 mode in the experiments. During the LRUT inspection the cancellation of the A0 mode could be achieved by using 2 arrays of transducers and by optimising the spacing between the transducers. Therefore, in order to cancel the A0 mode, these two rings were separated by the A0 wavelength at 100kHz (frequency used for the experiment). The excitation signal of one was also inverted with respect to the other. The outgoing A0 signal in each direction was thus zero since the A0 mode from each ring was cancelled.

Besides, inside the rings, the transducers spacing was optimised to maximise the S0 signal. This setting aims to reduce the scattering of waves to the lateral edges of the plate. A 4x2 array as shown in Fig.7, was used to run the experiment.

spcesept08f7.gif

Fig.7. Setting of the experiment to cancel the A0 mode

4.2 Results and discussion

It can be seen in Fig.8 that with an inverted array setting, the S0 mode has enhanced with a peak-peak amplitude of 50mV. The S0 signal has an arrival time of 600µs. With this setting the A0 peak-to-peak signal has drop to 10mV (signal arriving at 1000µs). This shows that the inverted array is effective in cancelling out the unwanted mode). Fig.7 also shows that without the inverted matrix, the A0 mode has been enhanced.

spcesept08f8.gif

Fig.8. Matrix cancelling A0 mode (one array inverted with respect to the other)


A second experiment was carried out to determine whether a defect in the plate can be detected using the above configuration. The defects were little trenches in the plate. Two sizes of defect have been studied, a 30x12x6mm defect and a 60x12x6mm defect.

Figs.9(a) and 9(b) show the results at 100kHz with the different defects where Fig.8(b) is the zoom in view of Fig.8(a). It can be seen from the figures that with the present of defect, additional signal has arrived earlier that the original S0 pulse (from the end of the plate).

spcesept08f9a.gif

Fig.9a) Pulse echo signal at 100kHz of a 4x2 matrix cancelling A0 mode (one array inverted with respect to the other) with different sizes of defect

spcesept08f9b.gif

Fig.9b) Zoom in the pulse echo signal at 100kHz of a 4x2 matrix cancelling A0 mode

5. Conclusions

This paper has illustrated the application of LRUT technology for complex and large geometries applications. First application was conducted on a rail structure. In this experiment, it was found that:

  • The use of a piezo-element with in-plane motion is better than a compression element, because it gives more directionality of the acoustic energy in the rail.
  • The use of a transducer array greatly reduces the coherent noise in the generated mode.
  • The most appropriate mode for inspection of the web of the rail is T2. The experiments have shown that T2 is able to detect defects in the web of the rail.
  • The most appropriate mode for inspection of the head of the rail is the mode F3. The results have confirmed that this mode is sensitive to defects in the head. During the experiments a saw cut of 2mm depth has been detected.
  • For the foot of the rail, the mode F2 is used for the long range ultrasonic testing and it has been shown that is sensitive to small defects in the foot (5mm deep)

As for the experiments on large plate, it can be concluded that:

  • There are two factors that complicate interpretation of the waveforms of received signals: almost in all cases by arrays are generated both A0 and S0 modes ofLamb waves and also Lamb waves in lateral directions.
  • The 'horizontal' transducer arrays in all cases generate both A0 and S0 modes of Lamb waves. Selective generation of only A0or S0 mode waves can be implemented only by application of 'vertical' or matrix transducer arrays.
  • Due to the simultaneous excitation technique the directivity patterns of all types of transducer arrays are symmetric with respect to both axis (horizontal and vertical). As a consequence, the waves in front and back directions aregenerated of the same amplitude. The waves generated in the backward direction are reflected by the edge of the plate and complicates interpretation of the results. Additionally in the case of the positioning of the transducer arrayclose to the edge of the plate the amplitude front signal can be affected by interference of direct and reflected by the edge signals.
  • The application of the delay time excitation technique enables essentially reduce the amplitude of the signal generated in the backward direction.

Acknowledgement

The Project is co-ordinated and managed by TWI Ltd and is partly funded by the EC under the Collective SME programme ref: Number Coll-CT-2005-516405.

This work was carried out under the European Commission funded LRUCM project: www.lrucm.eu.com

For more information please email:


contactus@twi.co.uk