METHODS AND SYSTEMS FOR DETERMINING A THICKNESS OF AN ELONGATE OR EXTENDED STRUCTURE

Abstract

A method of determining a thickness of an elongate or extended structure (2; FIG. 8) using elastic waves is disclosed. The method comprises receiving at least one time-domain signal from a transducer (12), generating a frequency-domain signal in dependence upon the at least one time-domain signal, reducing noise in the frequency-domain signal to provide a de-noised frequency-domain signal, comparing the de-noised frequency-domain signal with at least one reference signal, each reference signal corresponding to a respective thickness; and determining the thickness of the elongate or extended structure in dependence comparing the de-noised frequency-domain signal with the at least one reference signal.

Claims

1. A method of determining a thickness of an elongate or extended structure using elastic waves, the method comprising: receiving at least one time-domain signal from a transducer; generating a frequency-domain signal in dependence upon the at least one time-domain signal; reducing noise in the frequency-domain signal to provide a de-noised frequency-domain signal; comparing the de-noised frequency-domain signal with at least two reference signals to determine how closely each reference signal matches with the de-noised frequency-domain signal, each reference signal corresponding to a respective thickness; and determining the thickness of the elongate or extended structure based on which reference signal most closely matches the de-noised frequency-domain signal.

2. The method of claim 1, comprising: receiving at least two time-domain signals from the transducer; wherein generating the frequency-domain signal comprises: converting the at least two time-domain signals into at least two frequency-domain signals; and combining the at least two frequency-domain signals into the frequency-domain signal.

3. The method of claim 1, comprising: receiving at least two time-domain signals from the transducer; wherein generating the frequency-domain signal comprises: combining the at least two time-domain signals into a single, combined time-domain signal; and converting the single, combined time-domain signal into the frequency-domain signal.

4. The method of claim 2, wherein the at least two time-domain signals comprise three time-domain signals.

5. The method of claim 4, wherein the three time-domain signals comprise first, second and third time-domain signals corresponding to measurements of first, second and third excitations at first, second and third frequencies respectively.

6. The method of claim 1, wherein reducing the noise in the frequency-domain signal comprises: reducing or removing coherent noise from the frequency-domain signal.

7. The method of claim 1, wherein reducing the noise in the frequency-domain signal comprises: reducing or removing incoherent noise from the frequency-domain signal.

8. The method of claim 1, wherein reducing the noise in the frequency-domain signal comprises: performing Welch's method on the frequency-domain signal.

9. The method of claim 1, further comprising: windowing each of the at least two time-domain signals prior to generating the frequency-domain signal.

10. The method of claim 1, wherein comparing the de-noised frequency-domain signal with at least two reference signals comprises: performing a convolution of the de-noised frequency-domain signal with each of the at least two reference signals.

11. The method claim 1, wherein comparing the de-noised frequency-domain signal with at least two reference signals comprises: performing a cross-correlation of the de-noised frequency-domain signal with each of the at least two reference signals.

12. The method claim 1, wherein comparing the de-noised frequency-domain signal with the at least two reference signals comprises: multiplying a matrix comprising a set of masks by a first vector containing measured signals values for different frequencies to obtain a second vector, wherein each mask contains a series of values extending along a first direction corresponding to values at different frequencies, the masks are arranged along a second, orthogonal dimension and the contains a series of measured signal values extending along the second direction.

13. A method comprising: performing a guided wave ranging measurement of a reference feature in an elongate or extended structure using at least one transducer in a first set of transducers to determine a value of a distance to the reference feature; calculating a multiplication factor for a velocity of a guided wave mode using the value of the distance; and performing the method of claim 1 using a velocity adjusted using the guided wave ranging measurement.

14. A method of claim 14, wherein performing the guided wave ranging measurement comprises using a T(0,1) mode or SH0 mode.

15. A method comprising: receiving a nominal value of a thickness of an elongate or extended structure; providing an excitation signal to at least one transducer in a first set of transducers, the excitation signal having a frequency range which contains a cut-off frequency for a dispersive guided wave mode in the elongate or extended structure having the nominal value of the thickness; and receiving guided wave(s) using at least one transducer in a second, different set of transducers.

16. A method comprising: receiving a signal from at least one transducer in a set of transducers in an inspection ring, the signal having a frequency range lying between 100 kHz to 800 kHz; measuring a frequency of a characteristic feature contained in the signal; and determining a thickness of an elongate or extended structure using the frequency of the characteristic feature.

17. A method comprising: providing at least one excitation signal to at least one transducer in a first set of transducers, the at least one excitation signals covering a sufficiently broad frequency range to contain cut-off frequencies for at least two dispersive guided wave modes in an elongate or extended structure; and receiving guided waves using at least one transducer in a second, different set of transducers.

18. The method of claim 17, wherein the sufficiently broad frequency range is contained within a range between 50 kHz to 800 kHz and at least a portion of the sufficiently broad frequency range extends above 100 kHz.

19. (canceled)

20. A computer program product comprising a computer-readable medium storing a computer program which, when executed by at least one processor, causes the at least one processor to perform the method of claim 1.

21. Apparatus for determining a thickness of an elongate or extended structure using elastic waves, the apparatus comprising: at least one processor; and memory; wherein the at least one processor is configured to perform the method claim 1.

22. The apparatus of claim 21, which is a computer; wherein the computer further comprises: a network interface; wherein the computer is configured: to receive the at least one time-domain signal from a guided wave inspection system; and to determine the thickness of the structure in dependence upon the at least one time-domain signal.

23.-37. (canceled)

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Certain embodiments of the present invention will now be described, by way of example, with reference to the accompanying drawings, in which:

[0040] FIG. 1 shows dispersion curves for a set of guided waves in a plate;

[0041] FIG. 2 schematically illustrates a broadband excitation having a bandwidth Δf.sub.w which can be used to excite a set of shear horizontal guided wave modes in a pipe of a first wall thickness including a first cut-off mode;

[0042] FIG. 3 schematically illustrates a broadband excitation having a bandwidth Δf.sub.w which can be used to excite a set of shear horizontal guided wave modes in a pipe of a second wall thickness excluding a first cut-off mode;

[0043] FIG. 4 schematically illustrates synthesis of a broadband excitation having a bandwidth Δf.sub.w using three, different narrowband excitations;

[0044] FIG. 5 illustrates a frequency range used for broadband excitation with a frequency range typically used for guided wave inspection;

[0045] FIG. 6 illustrates an example of a de-noised frequency-domain signal;

[0046] FIG. 7 illustrates an example of a frequency-domain mask which can be used to search for guided wave modes in a de-noised frequency-domain signal and so find a thickness of a wall of a pipe;

[0047] FIG. 8 is schematic diagram of a pipe and a guided wave testing system which includes a transducer assembly, guided wave instrumentation and a computer system;

[0048] FIG. 9 is a transverse cross section through the pipe and transducer assembly shown in FIG. 8;

[0049] FIG. 10 is a schematic block diagram of the computer system shown in FIG. 8;

[0050] FIG. 11 illustrates propagation of a guided wave for a thickness measurement in a wall of a pipe;

[0051] FIG. 12 schematically illustrates determining a thickness of a wall of a pipe by comparing a frequency-domain response for a pipe of unknown wall thickness to reference frequency-domain responses for pipes of known wall thicknesses;

[0052] FIG. 13 illustrates comparing a frequency-domain response for a pipe of unknown wall thickness to a frequency-domain response (or “mask”) for a pipe of known wall thicknesses;

[0053] FIG. 14 shows plots of a frequency-domain response for a pipe of unknown wall thickness with a frequency-domain response for a pipe of known wall thicknesses;

[0054] FIG. 15 is a process flow diagram of a method of determining a thickness of a wall of a pipe;

[0055] FIG. 16 is a process flow diagram of a method of pre-processing a set of time-domain signals into a single, de-noised frequency-domain signal for comparison with other frequency-domain signals;

[0056] FIGS. 17a, 17b and 17c show first, second and third time-domain response signals;

[0057] FIGS. 18a, 18b and 18c show first, second and third windowed, time-domain response signals;

[0058] FIG. 19 shows a frequency-domain signal following combination and transformation of time-domain signals into the frequency domain;

[0059] FIG. 20 shows a de-noised frequency-domain signal following processing using Welch's method; and

[0060] FIG. 21 is a process flow diagram of a method of wave speed calibration.

DETAILED DESCRIPTION OF CERTAIN EMBODIMENTS

Introduction

[0061] FIG. 1 illustrates dispersion curves of a set of guided wave modes which can be excited in a plate (not shown). The dispersion curves may be used as a good approximation for dispersion curves for the guided wave modes propagating on a curved plate of the same thickness if the curvature is large compared to the wall thickness (for example, a 168 mm diameter pipe with a wall thickness of 7 mm).

[0062] The guided wave modes propagating between the transducers of the guided wave system for pipe inspection described herein are one such example. Therefore, even though in the following pipes are discussed, reference is made herein to shear horizontal, symmetric and antisymmetric modes, which are plate guided wave modes.

[0063] The structure need not be a pipe, but can take other forms of elongate or extended structure, such as plates, rails, beams, pillars and the like.

[0064] In FIG. 1, the guided wave modes (shown in continuous, bold lines) are symmetric and antisymmetric shear horizontal modes. However, the following discussion is not limited to shear horizontal modes. Other dispersive guided wave modes may be used, such as symmetric or antisymmetric Lamb wave modes (shown in dashed lines).

[0065] Referring to FIG. 1, each guided wave mode shown (shown in continuous bold lines) has a respective cut-off (through-thickness resonance) frequency starting from a first mode with a cut-off frequency, f, followed by a series of cut-off frequencies for higher-order modes spaced apart by the frequency, f. The frequency depends on the thickness of a wall of the pipe. In particular, the frequency decreases as the pipe wall thickness increases.

[0066] Some of the methods of determining a thickness of a wall of an elongate or extended structure herein described involve, in a pitch-catch configuration, exciting at least two guided wave modes regardless of the thickness of the pipe, measuring a response which includes contributions from multiple guided wave modes and using the response to determine the thickness of the pipe wall. Thickness determination can use the fact the spacing of features attributable to two different modes depends on the thickness of the pipe wall and/or employ a set of thickness-dependent references to find a matching reference and, thus, determine the thickness of the pipe wall (using a process which is herein referred to as “masking”).

[0067] Excitation of at least two guided wave modes is achieved by choosing an excitation that has a sufficiently wide bandwidth for a range of expected pipe wall thicknesses (for example, 5 to 25 mm). Thus, the same broadband excitation can be used for different pipes.

[0068] Referring to FIGS. 2 and 3, provided the bandwidth, Δf.sub.w, of the broadband excitation is greater than twice the first cut-off frequency, f, then at least two guided wave modes should be excited. In some cases, the first cut-off mode is excited (as illustrated, for example, in FIG. 2). In other cases, the first cut-off mode falls outside of the excitation range (as illustrated, for example, in FIG. 3). Nevertheless, even when the first cut-off mode falls outside of the excitation range, thickness determination is still possible.

[0069] Referring in particular to FIG. 3, for at least two modes to fall within an excitation range, Δf.sub.w, the following criterion should be satisfied, Δf.sub.w>2f, where f is the frequency of the first cut off frequency, f.sub.b. This criterion defines a minimum thickness, but not a maximum thickness that can be measured with this method. In practice, however, other factors, such as frequency bin size, can place a limit on maximum thickness.

[0070] The wall or plate thickness, t, is related to shear speed, C.sub.s, (which depends on material and temperature) and the first cut-off frequency, f:

f=C.sub.s/(2×t)  (1)

[0071] Thus, the sufficiently broad frequency range can be determined using:

[00002]$\begin{array}{cc}\Delta .Math.f_w>\frac{C_s}{\mathrm{wall}.Math..Math.\mathrm{thickness}}& \left(2\right)\end{array}$

[0072] For example, a 10 mm steel wall, assuming C.sub.s=3250 m/s, Δf.sub.w needs to be at least 325 kHz to guarantee capturing at least two modes.

[0073] If processing is extended to include all modes with a cut off rather than only shear horizontal modes, then the criterion becomes

[00003]$\Delta .Math.f_w>\frac{C_l}{C_s}.Math.f_b,$

where C.sub.l is the longitudinal speed. In this case the sufficiently broad frequency range can be determined using:

[00004]$\begin{array}{cc}\Delta .Math.f_w>\frac{C_l}{2*\mathrm{wall}.Math..Math.\mathrm{thickness}}& \left(3\right)\end{array}$

[0074] A broadband excitation can be achieved in a number of different ways. For example, a single broadband excitation signal, for instance, in the form of a chirp can be used.

[0075] Referring to FIG. 4, another way to obtain a broadband excitation is to use at least two, separate excitation signals, each excitation having a respective frequency range, where the frequency range of each excitation overlaps with that of a neighbouring excitation.

[0076] Referring to FIG. 5, an inspection ring having two rows of transducers or two inspection rings each having a row of transducers can be used to measure pipe wall thickness using methods herein described. The inspection ring(s) can be used for long range guided wave pipe inspection using torsional-mode guided waves having frequencies falling in a range between 1 kHz and 100 kHz. Other modes, such as longitudinal or flexural modes can be used. More than one different mode can be used. As will be explained in more detail hereinafter, thickness determination methods can use waves having frequencies falling in a range between 50 kHz and 800 kHz.

[0077] Masking

[0078] A masking process is used to “pick” the thickness of the wall employing multiple modes to measure pipe wall thickness. The process is not reliant on any particular mode and can have one or more advantages including increased robustness to noise, optimised resolution (regardless of wall thickness) and the potential for resolution to be finer than the frequency step of the size of the result signal.

[0079] Referring to FIGS. 6 and 7, the masking process takes a de-noised, broadband frequency-domain signal (for example one obtained by combining two or more frequency-domain signals into a single frequency-domain signal which is then filtered) and returns a thickness value.

[0080] The thickness of the pipe is determined by comparing the combined, de-noised frequency domain signal with a set of masks. A mask is an artificially-generated signal such as that shown in FIG. 7. A mask is built for a range of frequencies, f, (i.e., the frequency hereinbefore described) in the specified search range. Each mask is built to be the same length, i.e., NFFT, as the real signal. In other words, the mask is built in the time domain with the same time step (which is the inverse of sampling rate between samples), dt, as the real data. This ensures that frequency step is the same for the masks as it is in the real data. This can make operations, such as multiplication of the mask with the real data, simpler, easier and more meaningful. Expressed differently, the real data and the mask share a single frequency axis.

[0081] Each mask forms a row in a matrix, M, of all masks. By multiplying M with a column vector containing the real signal, P, a vector, C, is obtained which describes how closely each mask matches with the real signal and the length of C will be equal to the number of masks (rows) in M.

C=MP  (4)

[0082] The index of the maximum value in C can be cross-checked against the value of f used to create the corresponding mask. Frequency f can be used to calculate the thickness if a value for shear speed is known.

[0083] To make the process more efficient but maintain the desired resolution, initially the process runs with a coarse step size of f. The best matching value of f is then used as the median of a range of more finely-spaced masks. The process is repeated until the resolution meets a pre-specified threshold.

[0084] The exemplary mask shown in FIG. 7 is constructed to search only for shear horizontal modes, but other masks can be constructed to search for other guided wave modes.

[0085] The use of a masking method for picking a frequency allows the use of multiple modes to size the wall. This can have several benefits. First, there is no reliance on any single mode. For example, if the AH1 mode falls below the excitation range, then the wall can still be correctly sized using the higher-order modes. The more modes that can be used, the more robust to noise the measurement becomes.

[0086] Resolution can be defined as being the minimum change in wall thickness that can be detected between measurements. The bin size on the frequency axis is the most important factor determining resolution. Referring to FIG. 6, consider the situation that the frequency step size is 500 Hz and the wall thickness is 27 mm. If just the AH1 mode is used, then a peak would be expected at 60.2 kHz. A change is only seen when this value becomes 60.7 kHz, which corresponds to thickness of 26.77 mm and, thus, resulting in a resolution of 0.23 mm. However, by using multiple modes, it is possible to exploit the greater resolution available at higher frequencies. For the signal shown in FIG. 6, the process would be sensitive to changes on the highest frequency mode shown which would correspond to a resolution of 0.04 mm. This would be the resolution if using only a simple peak-picking method. However, another benefit of the masking method is that it can be sensitive to changes that are smaller than the frequency bin size.

[0087] To illustrate this, consider the situation that the bin size is 500 Hz and a signal that has a centre frequency falling somewhere between 100 kHz and 100.5 kHz. A number of masks can be built with centre frequencies ranging between 100 kHz and 100.5 kHz. Whilst both the mask and the result will have a peak at either 100 or 100.5 kHz, a much better estimate of centre frequency can be obtained from the best matching vector C.

[0088] Using multiple modes can help to reduce the amount of computation used to determine the thickness of the wall. If, however, a single mode is used, then zero padding can be added before transforming from time to frequency domain to improve resolution, although this can increase computational overheads.

[0089] Using multiple modes also takes advantage of the fact that the thickness of the pipe is encoded in the spacing between the modes. Using a single mode does not exploit this. Moreover, in an automated set up, it would be difficult to measure the thickness without prior knowledge of the approximate thickness of the wall or making assumptions.

[0090] If a measurement is reliant on a specific mode, then it is necessary to ensure that the mode falls within the excitation range. There may be two issues with this. First, unless using an extremely wideband excitation, it may be necessary to change the excitation settings from pipe to pipe which increases complexity. Secondly, the transduction system should have an acceptable frequency response. For example, external factors (such as transducer body resonances) might introduce unwanted resonances in the frequency range of interest.

[0091] In some embodiments, thickness determination can be carried out by a narrow range of excitation frequency in the range 100 kHz to 800 kHz, chosen based on a nominal thickness. Thus, a specific mode is excited and a resonance peak in the response in a range 100 kHz to 800 kHz can be used to determine the thickness of the structure using a narrower frequency range containing only one cut-off frequency.

[0092] Inspection System 1

[0093] Referring to FIGS. 8 to 10, a system 1 for inspecting a pipe 2 or other similar structure using guided waves is shown. The inspection system 1 includes a transducer assembly 3 (or “inspection ring”) which is permanently or removably attachable to the pipe 2, guided wave instrumentation 4, and a signal processing system 5.

[0094] The inspection ring 3 comprises a band 10 (or “collar”) which supports first and second arrays 11.sub.1, 11.sub.2 of transducers 12 for generating guided waves 13 in the pipe 2 and detecting waves 14 reflected from defects or features (not shown). As will be explained in more detail hereinafter, certain dispersive ultrasonic waves 15 (herein also referred to as “ultrasonic wave modes” or simply “modes”) can be generated by transducers 12 in the first array 11.sub.1 and detected by transducers 12 in the second array 11.sub.2 which can be used to measure the thickness, t, of the wall of the pipe 2. The transducers 12 preferably take the form piezoelectric transducers and an example of suitable transducers can be found in GB 2 479 744 A which is incorporated herein by reference. Each array 11.sub.1, 11.sub.2 may comprise, for example, 16 or 32 transducers 12, although there may be fewer than 16, between 16 and 32 or more than 32 transducers 12. The transducers 12 are grouped into sectors 16 (or “channels”), for example, eight sectors 16, each sector 16 consisting of between 2 to 9 or more transducers 12.

[0095] Each array 11.sub.1, 11.sub.2 are arranged such that, when the inspection ring 3 is installed, the transducers 12 are disposed around the periphery of the pipe 2. The first and second arrays 11.sub.1, 11.sub.2 are offset across the width of the band 10 such that, when the inspection ring 3 is installed, the two arrays 11.sub.1, 11.sub.2 are offset along a longitudinal axis 17 of the pipe 2. An example of a suitable inspection ring is the gPIMS® ring available from Guided Ultrasonics Ltd. (London, UK). Two separate rings 3, each having only a single array of transducers, can also be used.

[0096] The guided wave instrumentation 4 includes a signal generator (not shown) capable of generating rf signals 18 having a suitable frequency, which is usually of the order hundreds of kilohertz (kHz), and a suitable shape, such as, for example, a k-cycle suitably-windowed tone burst or a chirp signal, where k is a positive number equal to or greater than 1, preferably an integer or half integer, preferably taking a value in the range 3≤k≤10, and where a suitable windowing function can be a Gaussian function. The signal generator (not shown) feeds the rf signal 18 to a transmitter transducer 12 which converts the signal 18 into a guided wave in the pipe wall 2.

[0097] The receiver transducer 12 converts a received guided wave into an electrical signal 19. The receiver transducer 12 feeds the electrical signal 19 to a signal receiver (not shown). The signal receiver (not shown) may include an amplifier (not shown) and an analogue-to-digital converter (not shown) which generates a digitized signal of the electrical signal 19.

[0098] The guided wave instrumentation 4 excites transducers 12 in the first array 11.sub.1 and receives signals from transducers in the second arrays 11.sub.2 respectively in pitch-catch mode. The guided wave instrumentation 4 can excite transducers 12 in a sector 16 in the first array 11.sub.1 and receives signals from corresponding transducers 12 in the corresponding sector 16 in the second array 11.sub.2. The guided wave instrumentation 4 can excite all transducers 12 in the first array 11, and sample, in sequence, transducers 12 in the second array 11.sub.2. This can be employed with multiplexing and can be used to help simplify operation.

[0099] The guided wave instrumentation 4 and signal processing system 5 may be integrated into a single unit. The signal processing system 5 may take the form of a lap-top, tablet or other form of portable computer having one or more CPUs and, optionally, one or more GPUs. The signal processing system 5 may be remotely located, e.g., in a server farm, connected to the rest of the system via a communications network 6 which may include, for example, the Internet, or a local connection (e.g. USB). Examples of suitable guided wave instrumentation include G4 Mini (Full), Wavemaker G4, gPIMS Mini Collector and other instruments available from Guided Ultrasonics Ltd. (London, UK).

[0100] Referring also to FIG. 10, the signal processing system 5 is implemented by a computer system 20 which comprises at least one processor 21, memory 22 and an input/output module 23 interconnected by a bus system 24. The system 20 may include a graphics processing unit 25 and a display 26. The system 20 may include user input device(s) 27 such as keyboard (not shown) and pointing device (not shown), a network interface 28 and storage 29 for example in the form of hard-disk drive(s) and/or solid-state drive. The storage 29 stores thickness-measurement software 30, measurement data 31. If the guided wave instrumentation 4 and signal processing system 5 are co-located (e.g., the signal processing system 5 takes the form of a lap-top computer connected directly to the instrumentation 4) or integrated into a single unit, then the computer system 20 may be used for controlling guided wave instrumentation 4 and so the storage 20 may include guided wave testing software 33.

[0101] Thickness measurement may be implemented by the guided wave instrumentation 4.

[0102] Referring in particular to FIGS. 8 and 9, the system 1 may be used to inspect the pipe 2 to detect and/or to monitor development of cracks, corrosion and other defects (not shown) in the pipe 2 using long-range guided waves 13,14 in pulse-echo mode. The system 1 may also be used to measure a thickness, t, of the wall of a pipe 2 using through-thickness resonance using guides waves 15 in pitch-catch mode.

[0103] Measuring Pipe Wall Thickness—Introduction

[0104] Referring to FIGS. 8, 9, 11 and 12, transducer(s) 12 (“a transmit transducer(s)”) in the first array 11.sub.1 is (are) used to excite a sequence of guided waves 15, preferably shear horizontal wave modes 15, at different frequencies in the pipe wall 2, and transducer(s) 12 (“receive transducer(s)”) in the second array 11.sub.2 is (are) used to detect the ultrasonic waves 15. The transducer(s) 12 is excited by a series of excitation signals having centre frequencies, f.sub.1, f.sub.2, f.sub.3, where f.sub.3>f.sub.2>f.sub.1. f.sub.1 is 110 kHz, f.sub.2 is 220 kHz and f.sub.3 is 440 kHz. Other centre frequencies, however, can be used. The second and third centre frequencies f.sub.2, f.sub.3, need not be multiples of the first centre frequency f.sub.1.

[0105] Referring in particular to FIGS. 11 and 12, each time-domain response 19 generated by the receive transducer(s) 12 resulting from a respective excitation signal 18 at a given frequency (e.g., f.sub.1) is processed and, optionally, combined with other processed responses resulting from excitation signal 18 at one or more other frequencies (e.g., f.sub.2, f.sub.3) to provide a frequency-domain response 41.

[0106] The time-domain responses 19 are preferably transformed into respective frequency-domain responses before they are combined (or “merged”). Coherent and incoherent noise is reduced, e.g., removed, from the frequency-domain responses and/or the combined frequency-domain response to obtain a cleaner (i.e., less noisy) frequency-domain response. Coherent noise such as transient signals and incoherent noise can be reduced by performing Welch's method, by single-spectrum analysis or by another suitable method of reducing coherent and incoherent noise on a frequency-domain signal.

[0107] The frequency-domain response 41 can include one or more characteristic features 42.sub.1, 42.sub.2, 42.sub.3 (for example, a peak) corresponding to modes 15 generated in the pipe 2. In the example shown in FIG. 12, there are three characteristic features 42.sub.1, 42.sub.2, 42.sub.3 lying at first, second and third frequencies f.sub.α, f.sub.β, f.sub.χ respectively. In this case, the characteristic features 42.sub.1, 42.sub.2, 42.sub.3 are generally equally-spaced in frequency and correspond to three shear horizontal modes. Although three characteristic features 42.sub.1, 42.sub.2, 42.sub.3 are shown, the number of characteristic features 42.sub.1, 42.sub.2, 42.sub.3 do not need to be the same as the number of excitation signals. Fewer or more modes and, thus, characteristic features, may be generated by the excitation signals 17. The characteristic features can have different amplitudes and can have different relative positions is depending on, for example, the type of modes. In particular, if there are different modes, then the characteristic features 42.sub.1, 42.sub.2, 42.sub.3 need not be equally spaced.

[0108] Referring also to FIGS. 13 and 14, the response 41 is compared (e.g., by convolution or using equation (4) above) with a set of one or more mask signals 43.sub.i (or “masks” or “reference frequency-domain signals”) where i=1, 2, . . . , m, each mask 43.sub.i corresponding to a frequency-domain signal for a pipe wall 2 of a given thickness, to obtain a respective score 45.sub.i. The masks 43.sub.i may be generated (e.g., simulated) or measured. Each mask 43.sub.i includes one or more characteristic features 44.sub.i,j where j=1, 2, . . . , p. Each mask 43.sub.i preferably contains the same number of characteristic features as the frequency-domain response 41.

[0109] The mask signals 43.sub.i do not necessarily need to be replicas of the expected signals. In particular, mask amplitudes do not need to be the expected or measured amplitudes.

[0110] However, mask amplitudes are preferably used which can help to maximise probability of a correct measurement.

[0111] If the positions of the characteristic features for the response 41 and the mask 43.sub.i are closer, then the corresponding score 45.sub.i is higher. The thickness of pipe wall 2 is chosen to be the thickness of the mask 43.sub.i resulting in the highest score 45.sub.i.

[0112] Measuring Pipe Wall Thickness—Process

[0113] Referring to FIGS. 8 to 10, and FIGS. 15 to 20, a method of determining a thickness of a wall of a pipe using ultrasonic waves will now be described in more detail. The method is performed by the processor(s) 21 under the control of thickness-measurement software 30. For clarity, the process is described using only one processor 21. A measurement of local thickness is carried out for each sector of the pipe (i.e., each part of the circumference of the pipe) and so provide several measurements of thickness around the circumference of the pipe (as opposed to one, for example, averaged value obtained for the pipe wall at the location of the ring).

[0114] For each channel, starting with a first channel, i.e., a set of transducers 12 in a given sector 16 (step S1), the signal processing unit 5 receives a file which includes a set of time-domain responses 19 (or “signals”) (step S2). FIGS. 17a, 17b and 17c show examples of measured time-domain responses 19 resulting from excitation signals at 110 kHz, 220 kHz and 440 kHz respectively.

[0115] The processor 21 processes the signals 19 (step S3).

[0116] The processor 21 trims each time-domain response 19 by performing windowing, i.e., extracting a portion of the response 19 in a given time window 51 (step S3.1). FIGS. 17a, 17b and 17c show examples of the measured time-domain responses 19′ after windowing. The processor 21 transforms each time-domain response 19′ into a frequency-domain response (step S3.2). The processor 21 combines the frequency-domain responses into a single combined (or “merged”) frequency-domain response (step S3.3). FIG. 19 show an example of a frequency-domain response 40. The processor 21 then cleans (i.e., de-noises) the merged frequency-domain response (step S3.4). FIG. 20 show an example of a frequency-domain response 41.

[0117] The processor 21 then looks for the best-matching mask so as to determine the thickness of the pipe wall (step S6).

[0118] The processor 21 can use a seed value of thickness (herein referred to simply as a “seed”) to help reduce processing time and/or increase reliability (by reducing the probability of an incorrect selection).

[0119] The processor 21 searches for the best-matching thickness across a wide range (or “default range”) of thicknesses based on the excitation frequencies used. The default range may be 5 mm to 25 mm. The default range may be determined to be the range of thicknesses in which at least two target modes fall in the frequency range being excited in the pipe.

[0120] The search range may be reduced. For instance, taking a seed of 8 mm as an example, the processor 21 searches for a thickness between 7 mm and 9 mm.

[0121] By default, a previously-measured thickness for a specific channel (i.e., a specific transducer 12) of a given guided wave instrumentation 4 (FIG. 8) may be used as the seed for a subsequent measurement. Thus, a first thickness measurement for a specific channel for the guided wave instrumentation 4 (FIG. 8) is unseeded. A user-defined seed can be supplied either to override a default seed or supply a seed when one is unavailable. Conversely, a seed can be overridden to force a full range test even if seeds are available.

[0122] The processor 21 determines whether it has a set of masks in the search and if a mask for a given thickness in the range does not exist, then the processor 21 can create one (step S5).

[0123] The processor 21 then determines the best matching mask (step S6).

[0124] If a mask is found and the thickness of the pipe wall 2 is determined with sufficient resolution, e.g., to within 20 μm, then the process ends. If, however, the best match does not have sufficient resolution, then the range of the search is increased and the process of finding a best match continues (step S7).

[0125] Temperature Compensation/Velocity Calibration

[0126] The thickness measurement process hereinbefore described can take advantage of the fact that the apparatus which is used to measure wall thickness can also be used to measure a distance to a feature or a defect along the elongate structure and thus compensate for changes in shear velocity which depends on temperature.

[0127] Referring to FIG. 21, the signal processing system 5 (FIG. 8) can be used to generate an initial set of dispersion curves for a given plate-like structure (i.e., a suitable elongate or extended structure) using initial values for shear velocity and density (step S11). Dispersion curves 51 can be generated for structures of any thickness and can be used to determine cut-off frequencies and, thus, which masks to use. The guided wave instrumentation 4 (FIG. 8) is used to perform a ranging measurement preferably using, in the case of a pipe, torsional modes between the inspection ring 3 and a reference feature 52, such as a weld or support (step S12) to find a time of arrival between the inspection ring 3 and the reference feature 52. The guided wave instrumentation 4 receives, from the user, a value of the distance, L, between the inspection ring 3 and the reference feature 52 (step S13). Knowing the distance, L, and the time of arrival, t, the speed of the torsional mode, C.sub.T(0,1), in the pipe wall (which depends on temperature) can be extracted from a pulse-echo time-of-flight measurement using C.sub.T(0,1)=2×L/t. This speed, C.sub.T(0,1), is equal to the bulk shear speed, C.sub.S, which can then be used to generate a corrected set of dispersion curves.

[0128] During monitoring, variations in shear speed, C.sub.S, (due to, for example, changes in temperature) can be corrected. A reference time-of-flight measurement can be performed to find, T.sub.ref, and the reference value stored. A subsequent measurement of time-of-flight, T, and the using the reference value, T.sub.ref, are used to calculate a multiplication factor, α (step S14). The computer system 5 (FIG. 8) uses the multiplication factor, α, to correct the shear velocity, C.sub.S, and, thus, generate a new set of dispersion curves 51′ (step S15).

[0129] Although torsional modes are described, other modes (such as longitudinal or flexural modes) can be used, although the process of determining shear speed is more complex. For a plate, a shear horizontal mode may be used. For a bar, a torsional mode may be used.

[0130] Thickness Measurement Using Individual Peak(s)

[0131] As explained earlier, a masking process can be used to determine the thickness of the elongate or extended structure (e.g. the thickness of the wall of a pipe). This can make use of information provided not only by the position(s) of the peak(s) but also by the separation of the peaks.

[0132] Notwithstanding this, in some embodiments, the thickness of the elongate or extended structure can be determined using two rows of transducers in pitch-catch mode using one row of transducers and an excitation signal (or composite excitation signal) lying in a range between 100 and 800 kHz to generate guided waves in the pipe, another row of transducers receive the guided waves, and the signals are measured to find the peaks in the frequency-domain signal and calculate thickness using t=v/(2*f), where f is the measured frequency and v is velocity.

[0133] Modifications

[0134] It will be appreciated that various modifications may be made to the embodiments hereinbefore described. Such modifications may involve equivalent and other features which are already known in the design, manufacture and use of guided wave inspections systems and component parts thereof and which may be used instead of or in addition to features already described herein. Features of one embodiment may be replaced or supplemented by features of another embodiment.

[0135] Although claims have been formulated in this application to particular combinations of features, it should be understood that the scope of the disclosure of the present invention also includes any novel features or any novel combination of features disclosed herein either explicitly or implicitly or any generalization thereof, whether or not it relates to the same invention as presently claimed in any claim and whether or not it mitigates any or all of the same technical problems as does the present invention.

[0136] The applicants hereby give notice that new claims may be formulated to such features and/or combinations of such features during the prosecution of the present application or of any further application derived therefrom.