Shear-horizontal guided wave tomography
File(s)
Author(s)
Zimmermann, Andreas Armin Ernst
Type
Thesis or dissertation
Abstract
The petrochemical industry’s pipe-network is increasingly suffering from corrosion-induced
shutdowns and thus a quantitative measurement approach is vital for making adequate
service-life predictions and to guarantee safety. At pipe-supports, access is limited and
traditional point-inspection methods such as ultrasonic thickness (UT) testing are not
feasible. Guided wave tomography has been proposed as a solution to this problem. In
this technique ultrasonic guided waves are excited from an array around the defect and
received signals capture the interaction of the defect with these waves. From these signals,
velocity maps are reconstructed, and the dispersive nature of the waves utilised to invert
to thickness maps. Recently, it was found, that although reconstruction algorithms such
as the hybrid algorithm for robust breast ultrasound tomography (HARBUT) allow for
very high-resolution reconstructions, in practice, the mismatch of traditional guided wave
scattering of Lamb waves used in guided wave tomography with the acoustic scattering
models on which algorithms are based lead to poor resolution.
In this thesis, another type of guided wave mode is investigated instead, shear-horizontal
(SH) guided waves. Traditionally, these have been harder to excite but recent advances in
electromagnetic acoustic transducer (EMAT) excitation made them a practical alternative.
Since the fundamental SH0 mode is non-dispersive, the aim of this thesis is to assess
whether the first higher-order SH mode, SH1, is applicable to guided wave tomography.
SH1 exists at a higher frequency than those used in guided wave tomography with fundamental
Lamb wave modes, leading to a smaller wavelength and it offers no out-of-plane
displacement, which should avoid leakage into any pipe-loading. The thesis presents a
thorough numerical investigation of the SH1 mode, examining its scattering behaviour
and reconstruction of various defect types using three different SH excitation sources; (i)
pure SH1 point source (ii) SH surface point source and (iii) directional SH surface source.
These numerical results were validated experimentally, confirming the robustness of SH1
based guided wave tomography. In addition, the thesis investigates a technique to separate
out individual modes via the application of the nonuniform fast Fourier transform
(NUFFT) algorithm to non-uniformly sampled guided wave data.
The benefits of SH1 for guided wave tomography are shown throughout the thesis and
have been demonstrated by accurate and reliable thickness mapping of the defects. The
SH1 wave was able to achieve a 2.4 times better resolution than was previously achieved
using Lamb waves; yielding the best numerical and experimental guided wave tomography
reconstructions to date. The application of the NUFFT algorithm showed how filtering
in the frequency-wavenumber domain could be used to separate modal components even
without the application of time-based windowing approaches. This technique can also be
used to robustly determine bulk shear velocity even from directional data. Finally, this
thesis demonstrated proof of concept for SH1 guided wave tomography, suggesting that
this technique should be further investigated for employment in the accurate sizing and
quantification of corrosion in pipelines.
shutdowns and thus a quantitative measurement approach is vital for making adequate
service-life predictions and to guarantee safety. At pipe-supports, access is limited and
traditional point-inspection methods such as ultrasonic thickness (UT) testing are not
feasible. Guided wave tomography has been proposed as a solution to this problem. In
this technique ultrasonic guided waves are excited from an array around the defect and
received signals capture the interaction of the defect with these waves. From these signals,
velocity maps are reconstructed, and the dispersive nature of the waves utilised to invert
to thickness maps. Recently, it was found, that although reconstruction algorithms such
as the hybrid algorithm for robust breast ultrasound tomography (HARBUT) allow for
very high-resolution reconstructions, in practice, the mismatch of traditional guided wave
scattering of Lamb waves used in guided wave tomography with the acoustic scattering
models on which algorithms are based lead to poor resolution.
In this thesis, another type of guided wave mode is investigated instead, shear-horizontal
(SH) guided waves. Traditionally, these have been harder to excite but recent advances in
electromagnetic acoustic transducer (EMAT) excitation made them a practical alternative.
Since the fundamental SH0 mode is non-dispersive, the aim of this thesis is to assess
whether the first higher-order SH mode, SH1, is applicable to guided wave tomography.
SH1 exists at a higher frequency than those used in guided wave tomography with fundamental
Lamb wave modes, leading to a smaller wavelength and it offers no out-of-plane
displacement, which should avoid leakage into any pipe-loading. The thesis presents a
thorough numerical investigation of the SH1 mode, examining its scattering behaviour
and reconstruction of various defect types using three different SH excitation sources; (i)
pure SH1 point source (ii) SH surface point source and (iii) directional SH surface source.
These numerical results were validated experimentally, confirming the robustness of SH1
based guided wave tomography. In addition, the thesis investigates a technique to separate
out individual modes via the application of the nonuniform fast Fourier transform
(NUFFT) algorithm to non-uniformly sampled guided wave data.
The benefits of SH1 for guided wave tomography are shown throughout the thesis and
have been demonstrated by accurate and reliable thickness mapping of the defects. The
SH1 wave was able to achieve a 2.4 times better resolution than was previously achieved
using Lamb waves; yielding the best numerical and experimental guided wave tomography
reconstructions to date. The application of the NUFFT algorithm showed how filtering
in the frequency-wavenumber domain could be used to separate modal components even
without the application of time-based windowing approaches. This technique can also be
used to robustly determine bulk shear velocity even from directional data. Finally, this
thesis demonstrated proof of concept for SH1 guided wave tomography, suggesting that
this technique should be further investigated for employment in the accurate sizing and
quantification of corrosion in pipelines.
Version
Open Access
Date Issued
2021-09
Date Awarded
2022-03
Copyright Statement
Creative Commons Attribution NonCommercial Licence
Advisor
Huthwaite, Peter
Lowe, Michael
Sponsor
Engineering and Physical Sciences Research Council
Guided Ultrasonics Ltd (Firm)
Grant Number
EP/L015587/1
Publisher Department
Department of Mechanical Engineering
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Engineering Doctorate (EngD)