Wei Li · Xin’an Yuan · Jianming Zhao ·

Xiaokang Yin · Xiao Li

Alternating Current
Field Measurement
Technique
for Detection and
Measurement
of Cracks in Structures
Alternating Current Field Measurement Technique
for Detection and Measurement of Cracks
in Structures
Wei Li · Xin’an Yuan · Jianming Zhao ·
Xiaokang Yin · Xiao Li

Alternating Current Field

Measurement Technique
for Detection
and Measurement of Cracks
in Structures
Wei Li Xin’an Yuan
Department of Mechanical and Electrical Department of Mechanical and Electrical
Engineering Engineering
China University of Petroleum (East China) China University of Petroleum (East China)
Qingdao, Shandong, China Qingdao, Shandong, China

Jianming Zhao Xiaokang Yin

Department of Mechanical and Electrical Department of Mechanical and Electrical
Engineering Engineering
China University of Petroleum (East China) China University of Petroleum (East China)
Qingdao, Shandong, China Qingdao, Shandong, China

Xiao Li
Department of Mechanical and Electrical
Engineering
China University of Petroleum (East China)
Qingdao, Shandong, China

ISBN 978-981-97-7254-4 ISBN 978-981-97-7255-1 (eBook)

https://doi.org/10.1007/978-981-97-7255-1

This work was supported by China University of Petroleum (East China).

© The Editor(s) (if applicable) and The Author(s) 2025. This book is an open access publication.

Open Access This book is licensed under the terms of the Creative Commons Attribution 4.0 International
License (http://creativecommons.org/licenses/by/4.0/), which permits use, sharing, adaptation, distribu-
tion and reproduction in any medium or format, as long as you give appropriate credit to the original
author(s) and the source, provide a link to the Creative Commons license and indicate if changes were
made.
The images or other third party material in this book are included in the book’s Creative Commons license,
unless indicated otherwise in a credit line to the material. If material is not included in the book’s Creative
Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted
use, you will need to obtain permission directly from the copyright holder.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, expressed or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd.
The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721,
Singapore

If disposing of this product, please recycle the paper.

Preface

The nondestructive testing (NDT) method is an effective means to find and evaluate
defect, which provides support for the safety pre-warning and the maintenance deci-
sion of structure in the industrial field. In the 1980s, there were urgent requirements
for an NDT technology for inspection and evaluation of fatigue cracks at welded inter-
sections in the offshore underwater structures in North Sea. The conventional NDT
methods were not practicable in this special underwater environment and quantitative
evaluation requirements. Thus, the alternating current field measurement (ACFM)
technology was developed originally to inspect the underwater structures by the
researchers in mechanical engineering department at University College London. In
the past 40 years, the theory model, inspection method and equipment of ACFM have
achieved rapid progress for the advantages of the noncontact capability, the reduc-
tion of lift-off effect and the size, which are used widely in the ocean engineering,
power industry, rail traffic and special equipment field. In the past years, colleagues
mainly focused on the study of basic theory, probe and system design and structure
inspection. This book is divided into three parts. In the first part, three chapters are
employed to introduce the basic theory of ACFM, which lays the theoretical foun-
dation for the following parts. In the second part, three chapters are introduced to
introduce the design and testing of ACFM probe, instrument and software, which
provide guidance and reference for the development of the ACFM instrument. In

v
vi Preface

the third part, three chapters are introduced to explain the visualization research in
ACFM, which provide guidance for technical engineering application.

Qingdao, China Wei Li

[email protected]
Xin’an Yuan
[email protected]
Jianming Zhao
[email protected]
Xiaokang Yin
[email protected]
Xiao Li
[email protected]
Contents

Research on Real-time and High-Precision Cracks Inversion

Algorithm for ACFM Based on GA-BP Neural Network . . . . . . . . . . . . . . . 1
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
2 The FEM Model of Electromagnetic Coupling ACFM Probe . . . . . . . . . . 2
3 Real-Time Determination Method for Defect Characteristic Signals . . . . 4
4 Real-Time Crack Inversion Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
4.1 Detection System Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
4.2 Defect Inversion Algorithm Based on GA-BP . . . . . . . . . . . . . . . . . . 8
5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Identification of Tiny Surface Cracks in a Rugged Weld by Signal
Gradient Algorithm Using the ACFM Technique . . . . . . . . . . . . . . . . . . . . . 15
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2 Insensitive Signal to Lift-Off Variations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.1 Simulation Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.2 System Set Up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.3 Lift-Off Variation Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3 Signal Gradient Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.1 Detection of Tiny Cracks in the Weld and the HAZ . . . . . . . . . . . . . . 22
3.2 Signal Gradient Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.3 Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
Visual Evaluation of Irregular Cracks in Steel by Double Gradient
Fusion Algorithm Using Composite ACFM-MFL Testing Method . . . . . . 31
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2 Finite Element Method Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3 Double Gradient Fusion Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

vii
viii Contents

4 Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
4.1 Testing System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
4.2 Testing Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.3 Verification of Double Gradient Fusion Algorithm . . . . . . . . . . . . . . . 40
5 Conclusion and Further Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
Design and Experiment Research of Oblique Crack Detection
System for Rail Tread Based on ACFM Technique . . . . . . . . . . . . . . . . . . . . 47
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
2 Finite Element Method Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
2.1 Simulation Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
2.2 The Influence of the Scanning Path on the Characteristic
Signals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
3 Testing System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
3.1 Probe Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
3.2 Detection System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4 Experimental Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.1 Different Depth Oblique Crack Detection Experiments . . . . . . . . . . . 57
4.2 Detection Experiment of Inclined Cracks of Different Lengths . . . . 59
5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
Design and Testing of High-Resolution Probe Arrays Using
Alternating Current Field Measurement Technique . . . . . . . . . . . . . . . . . . . 67
1 Finite Element Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
1.1 Model Building . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
1.2 Defect Disturbance Law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
2 Probe Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
2.1 Sensor Array Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
2.2 Excitation Module Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
2.3 Multiplex Module Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
2.4 Design of Probe Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
3 Experimental Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
3.1 Experimental System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
3.2 Experimental Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
3.3 Defect Inversion Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Design and Experimental Study of Inner Uniform Electromagnetic
Probe in Stainless Steel Pipe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
1 Finite Element Emulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
1.1 Distribution Law of Electric Field in Pipeline . . . . . . . . . . . . . . . . . . . 84
1.2 Crack Disturbance Law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
Contents ix

1.3 Effect of Crack Size on Characteristic Signal . . . . . . . . . . . . . . . . . . . 87

2 Detection System Construction and Experiment . . . . . . . . . . . . . . . . . . . . . 89
2.1 Detection Probe Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
2.2 Detection System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
2.3 Test Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
Research on the Detection of Surface Cracks on Drilling Riser
Using the Chain Alternating Current Field Measurement Probe
Array . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
2 Principle of Alternative Current Field Measurement . . . . . . . . . . . . . . . . . . 100
3 Finite Element Method Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
3.1 Modeling Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
3.2 Excitation Probe Spacing Simulation Analysis . . . . . . . . . . . . . . . . . . 104
3.3 Sensor Spacing Simulation Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 107
4 Experimental Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
4.1 Design of the Testing System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
4.2 Sensor Circuit Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
4.3 Chain Probe Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
4.4 Test Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
4.5 Analysis of Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
An Electromagnetic Helmholtz-Coil Probe for Arbitrary
Orientation Crack Detection on the Surface of Pipeline . . . . . . . . . . . . . . . 119
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
2 Simulation of the Electromagnetic Helmholtz-Coil Probe . . . . . . . . . . . . . 120
2.1 Finite Element Model of the Electromagnetic Helmholtz-Coil
Probe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
2.2 Result Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
3 Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
3.1 System Set Up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
3.2 Result Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
Circumferential Current Field Testing System with TMR Sensor
Array for Non-contact Detection and Estimation of Cracks
on Power Plant Piping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
2 Circumferential Current Field FEM Model . . . . . . . . . . . . . . . . . . . . . . . . . . 132
2.1 Model Set up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
2.2 Frequency Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
x Contents

3 Circumferential Uniform Current Field Testing System . . . . . . . . . . . . . . . 136

3.1 Testing System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
4 Experiments and Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
4.1 Crack Depth Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
4.2 Crack Length Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
5 Conclusion and Further Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
Research on Real-time
and High-Precision Cracks Inversion
Algorithm for ACFM Based on GA-BP
Neural Network

1 Introduction

Alternating current field measurement (ACFM) technology was initially applied to

weld inspection on offshore drilling platforms. Due to its characteristics of non-
contact measurement, no calibration required, low lift-off effect, and accurate math-
ematical model, it has been widely applied in fields such as petrochemicals, railway
transportation, nuclear energy, etc. [1–3]. The principle of ACFM technology is
shown in Fig. 1, where the excitation probe induces a uniform current on the surface
of the workpiece. When the current passes through the defect vertically, it will bypass
the ends and bottom of the defect, causing distortion in the surrounding magnetic
field. The X-direction magnetic flux density Bx produces a trough, and the depth of
the trough can reflect the depth of the defect. Similarly, the Z-direction magnetic flux
density Bz generates peaks and troughs, and the distance between them can indicate
the length of the defect [4, 5].
In the field of non-destructive testing, obtaining characteristic signals from defects
is referred to as the forward model, while obtaining defect shapes from character-
istic signals is referred to as the inverse model [6]. It is generally easier to obtain
characteristic signals from defects, while it is more complex to invert defect shapes
from characteristic signals. However, the information of defect shape can reflect
the degree of structural damage, which is a matter of great concern in structural
assessment [7]. The ACFM technique has an accurate mathematical model, where
the characteristic signal Bz reflects the crack length information and Bx contains
the crack depth information, providing good conditions for defect profile inversion
(length and depth). Traditional ACFM interpolation inversion algorithms are rela-
tively simple, but they have low accuracy and poor real-time performance [8, 9]. The
inversion algorithms that utilize characteristic signals have high requirements for the
sample database, and the mutual influence between characteristic signals makes the
inversion process complex. It is necessary to analyze the characteristic signals or

© The Author(s) 2025 1

W. Li et al., Alternating Current Field Measurement Technique for Detection and
Measurement of Cracks in Structures, https://doi.org/10.1007/978-981-97-7255-1_1
2 Research on Real-time and High-Precision Cracks Inversion Algorithm …

Fig. 1 The principle of

ACFM

determine the butterfly chart after completing the defect scanning, making real-time
determination and inversion difficult to achieve [10].
To solve the above problems, the author established the electromagnetic coupling
simulation model of ACFM probe, obtained the sample database of crack inversion
characteristic signals in real time through the energy spectrum and phase threshold
determination method, and introduced Back Propagation, BP neural network and
Genetic Algorithm (GA) are used to retrieve ACFM crack length and depth with
high precision. In this paper, the GA-BP neural network based ACFM real-time high-
precision inversion algorithm provides a new method for real-time crack inversion.

2 The FEM Model of Electromagnetic Coupling ACFM

Probe

Select the 3-D circuit coupling stranded coil CIRCU124 in ANSYS, set its options
as needed to make it a power supply, and generate a coupling unit with the K node
and the node at the position where the probe model needs to be combined with the E
command. Since the current in the coil and the potential drop through the coil terminal
are unique, the degrees of freedom of coil current per turn (CURR) and potential
drop (EMF) degrees of freedom through the coil terminal are separately coupled
together. A 3D ACFM probe electromagnetic coupling motion simulation model
was established, as shown in Fig. 2. In the model, the excitation probe is a U-type
current-carrying coil [11, 12]. Under the coil, X coil (to extract the induced voltage
value of the X direction magnetic field) and Z coil (to extract the induced voltage
value of the Z direction magnetic field) are provided. In the simulation model, the
crack length is 15 mm, the width is 0.8 mm, the depth is 5 mm, the loading excitation
voltage is 3 V, the frequency is 6 K Hz, and the lifting height of the detection coil is
2 mm.
2 The FEM Model of Electromagnetic Coupling ACFM Probe 3

Fig. 2 The FEM model of electromagnetic coupling ACFM probe

A harmonic solver was selected, a path was established along the top of the crack,
and the voltage Ex and Ez in the coils in the X and Z directions above the path
were extracted respectively, as shown in Fig. 3. Ex maintains a certain value in the
non-cracked region (the background magnetic field is in the X direction), and when
it enters the cracked region (14–29 mm), the magnetic flux density in the X direction
decreases, resulting in a trough in Ex. Because the extracted induced electromotive
forces are all positive, Ez is basically 0 in the region without entering the crack, and
two peaks appear in the aggregation area at both ends of the crack (14 and 29 mm).
The Ex and Ez rules are consistent with the ACFM principle, which shows that the
simulation model of the electromagnetic coupling ACFM probe established in this
paper is correct.

Fig. 3 The simulation results of crack characteristic signals. a Bx. b Bz

4 Research on Real-time and High-Precision Cracks Inversion Algorithm …

3 Real-Time Determination Method for Defect

Characteristic Signals

Traditional ACFM technology often adopts the feature recognition method, that is,
according to the features of Ex and Ez or the butterfly diagram composed of Ex and
Ez [13]. Because the identification method based on characteristic signal or butterfly
diagram needs to distinguish according to the characteristics of the obtained signal
after the probe has completely swept the crack, it is difficult to realize the real-time
data storage and processing of the defect characteristic signal. In this paper, the
energy spectrum and phase threshold discriminant methods can be used to save the
effective signal characteristic values of defects in real time according to the threshold,
which lays a foundation for real-time and high-precision defect inversion.
As shown in Fig. 3, since the distance between the peaks of Ex (13–29 mm) is
larger than the peak-to-peak distance of Ez (14–28 mm) and larger than the crack
length, Ex contains more defect information (depth and length, especially the safety
margin for length estimation). In this paper, the electromotive force Ex of the X
coil is converted into the energy spectrum by the square of the Fourier transform, as
shown in Fig. 4. Within the peak distance of Ex (dashed line range), the maximum
value of the energy spectrum is taken and a horizontal line is numerically drawn.
The intersection position of the horizontal line and the Ex energy spectrum (solid
line region) is obviously greater than the Ex peak value, then the energy spectrum
information located below contains all the length and depth information of the crack.
Therefore, the peak value of Ex energy spectrum is selected as the threshold for
real-time crack determination.
At the same time, because the direction of the magnetic flux density inside the
crack zone changes when the detection coil enters the crack zone, the phase of Ez
will change when the Z coil enters the defect. As shown in Fig. 5a, Ez phase has
a sudden change at the crack region (14–29 mm). Ez phase derivative is shown in
Fig. 5b, and it can be seen that Ez phase derivative has an obvious peak value in the
crack region. Therefore, the derivative of Ez phase is selected as the basis for real-
time crack determination. The crack characteristic signal can be obtained in real time
according to the Ex energy spectrum and Ez phase threshold real-time determination
method. The real-time determination method designed for crack characteristic signal
is shown in Fig. 6. The main steps are as follows:
(1) Perform phase-lock amplification and data processing on the collected Ex, Ez
and excitation signals to obtain the Ex energy spectrum and Ez phase.
(2) Determine the relationship with the threshold size. If the value is less than or
equal to the value, the data is saved. Take the derivative of Ez phase to get Dz.
(3) Determine whether Dz meets the requirements (there is a margin for phase
judgment in this paper, and the threshold of Ez phase derivative is set to 180).
If Dz is greater than or equal to 180, it is regarded as valid data, and the saved
data is regarded as the defect characteristic signal sample database. If Dz is less
3 Real-Time Determination Method for Defect Characteristic Signals 5

Fig. 4 The energy spectrum

curve of Ex. a Ex. b Ex
energy spectrum

Fig. 5 The phase of Ez. a Different position phase of Ez. b Pase reciprocal of Ez
6 Research on Real-time and High-Precision Cracks Inversion Algorithm …

Fig. 6 The defect real-time

determination method

than 180, increase the threshold, clear the saved defect data, and continue to
determine the threshold size. If the value is greater than the value, the data is
discarded.

By updating the threshold value in real time, the algorithm can effectively avoid
the misjudgment of the crack characteristic signal, and effectively save the crack
characteristic signal in time when the detection coil enters the crack zone, so as to
achieve the purpose of real-time crack determination without completely sweeping
the crack.
4 Real-Time Crack Inversion Experiment 7

4 Real-Time Crack Inversion Experiment

4.1 Detection System Design

According to the principle of ACFM technology, the crack real-time inversion exper-
imental system is designed, as shown in Fig. 7a. In this paper, the ACFM excitation
probe uses U-type current carrier coil, and the detection coil uses coil sensor. The
signal generator generates a sinusoidal signal of amplitude 1 V and frequency 6 kHz,
which is transmitted to the excitation coil through power amplification. The exciting
coil generates a uniform exciting current region on the specimen surface, and the
space magnetic field distortion is caused by the current passing through the defect.
The detection coil picks up the Bx and Bz information and converts it into elec-
trical signals Ex and Ez. Ex and Ez are transmitted from the acquisition system to
the computer through conditioning circuits (amplification and filtering). The defect
characteristic signal can be obtained in real time by using Ex energy spectrum and Ez
phase threshold method. The real crack real-time inversion detection system finally
built is shown in Fig. 7b.
The specimens in this paper are mild steel plates, on which rectangular artificial
cracks of different sizes are carved by EDM technology. A PLC controlled scanning
platform was used to drive the ACFM probe (excitation and detection coil) along one
of the cracks (length 40 mm and depth 6 mm) for uniform scanning. The measurement
and control software collected the position information of Bx, Bz and the probe, and
the defect characteristic signals obtained were shown in Fig. 8.
In order to realize real-time and high-precision inversion quantification of defects,
sensitivity is introduced as a characteristic quantity of defect size inversion [14]:

Mx Ex0 − Ex min Ez max

Sx = = , Sz =
Ex0 Ex0 Ex0

Fig. 7 The real-time and high- precision cracks inversion system for ACFM. a System schematic
diagram. b Physical inspection system
8 Research on Real-time and High-Precision Cracks Inversion Algorithm …

Fig. 8 The characteristic signals of crack

LZ : EZ : peak-to-peak interval of component waveform; EX 0 : component signal

amplitude away from the crack; EX min : minimum signal amplitude; EZ max : indicates
the maximum value of signal distortion.
By analyzing the crack characteristic signals obtained by software in real time,
the results show that Ex sensitivity Sx is 9.2%, Ez sensitivity Sz is 26.63%, and the
distance between Ez peaks is 38.46 mm. At the same time, cracks of different depths
and lengths were detected, and the signal characteristic quantities of different crack
sizes were obtained, as shown in Table 1.

4.2 Defect Inversion Algorithm Based on GA-BP

As shown in Fig. 9, this paper uses the newff function in MATLAB to create a 4-layer
BP neural network. By default, the default initnw is used to initialize weight and bias
[15]. Although the initnw method enables the active regions of neurons in each layer
to be roughly flat distributed in the input space, the network trained by this method
has the problem of network structure uncertainty, that is, the initial weights and bias
values are random. Therefore, BP algorithm has some characteristics, such as slow
convergence speed, unguaranteed convergence to the global minimum point, and
difficult to determine the network structure [16].
Genetic algorithm, based on biological evolution, has good convergence and
robustness, and the calculation time is less when the calculation accuracy is high
[17, 18]. In this paper, a step genetic algorithm is added to optimize the weight and
bias of the neural network (GA-BP neural network model) before the training of BP
4 Real-Time Crack Inversion Experiment 9

Table 1 The characteristic signals database of different crack size

S. No. E X sensitivity S X / E Z sensitivity S Z / E Z peak-to-peak Defect Defect
% % interval L Z /mm depth D/ length L/
mm mm
1 39.31 41.44 8.29 8 10
2 31.65 38.69 18.85 8 20
3 25.52 35.64 2861 8 30
4 10.97 33.87 38.82 8 40
5 31.68 39.31 8.33 6 10
6 24.62 35.74 18.69 6 20
7 17.84 31.58 28.53 6 30
8 9.20 26.63 38.46 6 40
9 23.42 33.24 8.45 4 10
10 17.38 26.76 18.62 4 20
11 10.92 23.11 28.72 4 30
12 7.58 18.87 38.69 4 40
13 14.61 12.47 8.46 2 10
14 10.30 14.23 18.52 2 20
15 7.10 12.59 28.36 2 30
16 6.28 9.91 38.75 2 40
17 9.87 23.12 38.73 5 43
18 14.23 24.85 23.76 4 25
19 27.95 37.43 13.14 6 15

Fig. 9 The model of layer

BP neural network
10 Research on Real-time and High-Precision Cracks Inversion Algorithm …

network, so as to determine an optimal set of initial weights and bias, so as to optimize

the structure of BP artificial neural network [19–21]. The main program instruction
of BP neural network based on genetic algorithm initialization is as follows:

[net, IW, IB, W1, W2, B1, B2]

= fgaBP(XX, YY, eranum, popsize, pCross, pMutation, pInversion, options).

In the GA-BP neural network for defect inversion in this paper, XX input vector
includes three characteristic quantities of sensitivity and sensitivity as well as pek-
valley distance of component waveform, while YY output vector is two characteristic
quantities of defect length and depth. The neuron matrix of the four-layer BP network
model established by this method is [2, 5, 6]. Groups 1–16 in Table 3 are taken as
training samples and groups 17–19 as test samples. Since the value range of defect
size is also quite different, in order to facilitate the convergence of BP-GA network
and improve the prediction accuracy, the sample data in Table 3 should be normalized
within the interval [− 1, 1] before training. The weight and bias process of BP neural
network solved by genetic algorithm is shown in Fig. 10a, and the training process
of BP neural network is shown in Fig. 10b.
The GA-BP neural network established above was learned, and the cracks of
group 17 (length 43 mm and depth 5 mm) in the sample database were predicted.
The crack prediction results were shown in Fig. 11. The inversion results show that
the crack size of group 17 is 40.70 in length and 4.72 in depth, with length error of
5.35% and depth error of 5.60%.
Similarly, GA-BP neural network was used to invert the cracks in 18 groups and
19 groups of samples, and the inversion results were shown in Table 2. The inversion
error of crack length and depth in group 18 is 6.12% and 7.00% respectively. The
inversion error of crack length and depth of group 19 samples is 9.40% and 6.51%
respectively. It can be seen that GA-BP neural network has better prediction ability
for deeper cracks and longer cracks. Compared with the actual crack size, the relative
error of crack size prediction based on GA-BP neural network is less than 10%, which
meets the requirements of engineering practice.

5 Conclusion

In this paper, the electromagnetic coupling ACFM simulation model was established
with the help of ANSYS software, the characteristic crack signal was obtained in
real time by using the energy spectrum and phase threshold determination methods,
the ACFM crack inversion experimental system was built and the crack detection
experiment was carried out, and the GA-BP neural network was used to achieve
real-time and high-precision real-time inversion of crack size. The main conclusions
were as follows:
5 Conclusion 11

Fig. 10 The training process of GA-BP. a Genetic algorithm optimization process. b BP Neural
network training process

(1) The electromagnetic coupling ACFM simulation model established in this paper
can accurately obtain the voltage Ex and Ez inside the X coil and Z coil above
the crack.
(2) The threshold determination method based on Ex energy spectrum and Ez phase
derivative can obtain the crack characteristic signal in real time.
(3) The crack inversion algorithm based on GA-BP neural network can effectively
predict the length and depth of defects, and the prediction error is less than 10%,
which meets the requirements of engineering practice.
12 Research on Real-time and High-Precision Cracks Inversion Algorithm …

Fig. 11 The test results of GA-BP in 19th data

Table 2 The forecasting result of test sample based on GA-BP neural network
S. No. Depth inverse Relative Length inverse Length inverse
normalization/mm error/% normalization/mm normalization/%
17 4.72 5.60 40.70 5.35
18 3.72 7.00 23.47 6.12
19 6.51 8.50 13.59 9.40

References

1. Lugg MC (2012) The first 20 years of the A.C. field measurement technique. In: 18th world
conference on non-destructive testing (WCNDT), South Africa, pp 16–20
2. Li W, Yuan XA, Chen GM et al (2015) Research on in-service detection for axial cracks on drill
pipe using the feed-through alternating current field measurement. J Mech Eng 51(12):8–15
3. Nicholson GL, Davis CL (2012) Modelling of the response of an ACFM sensor to rail and rail
wheel RCF cracks. NDT&E Int 46:107–114
4. Li W, Yuan XA, Chen GM et al (2014) A feed-through ACFM probe with sensor array for pipe
string cracks inspection. NDT&E Int 67:17–23
5. Qi YL, Chen GM, Zhang YT (2004) Numerical simulation on alternating current field
measurement and sensitivity analysis of detected signal. J Univ Pet China (Edn Nat Sci)
28(3):65–68
6. Ramuhalli P, Udpa L, Udpa SS (2002) Electromagnetic NDE signal inversion by function-
approximation neural networks. IEEE Trans Magn 38(6):3633–3642
7. Maazi M, Benzaim O, Glay D et al (2008) Detection and characterization of buried macroscopic
cracks inside dielectric materials by microwave techniques and artificial neural networks. IEEE
Trans Instrum Meas 57(12):2816–2819
References 13

8. Li W (2007) Research on ACFM based defect intelligent recognition and visualization

technique. China University of Petrol, Dongying
9. Li W, Chen GM (2009) Defect Visualization for alternating current field measurement based
on the double U-shape inducer array. J Mech Eng 45(9):233–237
10. Noroozi A, Hasanzadeh RPR, Ravan M (2013) A fuzzy learning approach for identification of
arbitrary crack profiles using ACFM technique. IEEE Trans Magn 49(9):5016–5027
11. Li W, Chen GM, Yin XK et al (2013) Analysis of the lift-off effect of a U-shaped ACFM
system. NDT&E Int 53:31–35
12. Li W, Chen GM, Li WY et al (2011) Analysis of the inducing frequency of a U-shaped ACFM
system. NDT&E Int 44:324–328
13. Lugg M, Topp D (2006) Recent developments and applications of the ACFM inspection method
and ACSM stress measurement method. In: Proceedings of ECNDT, Berlin, Germany
14. Nicholson GL, Kostryzhev AG, Hao XJ et al (2011) Modelling and experimental measurements
of idealised and light-moderate RCF cracks in rails using an ACFM sensor. NDT&E Int 44:427–
437
15. Li P, Zeng LK, Shui AZ et al (2008) Design of forecast system of back propagation neural
networks based on MATLAB. Comput Appl Softw 25(4):149–150
16. Li W, Chen GM, Zheng XB (2007) Crack sizing for alternating current field measurement
based on GRNN. J Univ Pet China (Edn Nat Sci) 31(2):105–109
17. Cui T, Sun YZ, Xu J et al (2011) Reactive power optimization of power system based on
improved niche genetic algorithm. Proc CSEE 31(19):43–50
18. Yang GJ, Cui PY, Li LL (2001) Applying and realizing of genetic algorithm in neural networks
control. J Syst Simul 13(5):567–570
19. Li WC, Song DM, Chen B (2006) Artificial neural network based on genetic algorithm. Comput
Eng Des 27(2):316–318
20. Li S, Liu LJ, Xie YL (2011) Chaotic prediction for short-term traffic flow of optimized BP
neural network based on genetic algorithm. Control and Dec 26(10):1581–1585
21. Li S, Luo Y, Zhang MR (2011) Prediction method for chaotic time series of optimized BP
neural network based on genetic algorithm. Comput Eng Appl 47(29):52–55

Open Access This chapter is licensed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits use, sharing,
adaptation, distribution and reproduction in any medium or format, as long as you give appropriate
credit to the original author(s) and the source, provide a link to the Creative Commons license and
indicate if changes were made.
The images or other third party material in this chapter are included in the chapter’s Creative
Commons license, unless indicated otherwise in a credit line to the material. If material is not
included in the chapter’s Creative Commons license and your intended use is not permitted by
statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder.
Identification of Tiny Surface Cracks
in a Rugged Weld by Signal Gradient
Algorithm Using the ACFM Technique

1 Introduction

The welding procedure is widely used in the manufacturing industry. As a critical

connecting portion, it easily introduces cracks in the weld and the HAZ due to the
discontinuous material, corrosive environment, varying temperatures, and complex
stress. Welds should be inspected routinely using a nondestructive testing (NDT)
method in in-service time or before use [1–4].
Magnetic particle testing (MT) and penetrant testing (PT) are effective methods
used to detect surface cracks in a weld. However, the surface of the structure should
be cleaned thoroughly, including the coating, the attachment, and the greasy dirt
[5, 6]. Additionally, the testing result depends on the experience of the operator.
Ultrasonic testing (UT) is usually used for the detection of internal defects [7]. Eddy
current testing (ET) is easily confused by the lift-off variations, making it hard to
identify a tiny crack in a rugged weld [8].
The ACFM technique was developed for the detection of cracks in underwater
structures. Due to the advantages of non-contact testing, less cleaning, and quanti-
tative detection, the ACFM technique has been widely used on land and underwater
[9–11]. The principle of ACFM is shown in Fig. 1. The excitation coil induces the
local uniform current field in the specimen. When a crack is present, the current field
is disturbed. The disturbed current field makes the space magnetic field distorted.
The magnetic field in the X direction, Bx (parallel to the crack), shows a trough in the
center of the crack which contains depth information of the crack. The magnetic field
in the Z direction, Bz (perpendicular to the specimen), shows two opposite peaks at
the tips of the crack, which reflects the length of the crack. Normally, a so-called
butterfly plot (the Bx against the Bz) is presented to decide whether a crack is present
or not. In practice, the cracks are identified by the number of the loops in the butterfly
plot after detection [12].
However, when the ACFM is used to detect tiny cracks in a weld, the ripples and
reinforcements can be regarded as the distance variations between the probe and the

© The Author(s) 2025 15

W. Li et al., Alternating Current Field Measurement Technique for Detection and
Measurement of Cracks in Structures, https://doi.org/10.1007/978-981-97-7255-1_2
16 Identification of Tiny Surface Cracks in a Rugged Weld by Signal …

Fig. 1 Principle of ACFM

specimen [13] which causes the lift-off of the probe variations, as shown in Fig. 2.
The lift-off variations bring many interference signals, making the signal to noise
ratio (SNR) poor [14, 15]. The response signals of a probe are harsh, even on a crack.
The Bx and Bz are confused by the noise caused by the lift-off variations. As a result,
the conventional butterfly plot cannot effectively identify tiny cracks. Of course, the
size of the crack cannot be evaluated by the Bx and Bz at this stage. As the first step,
how to find a tiny crack in a weld is critical work in the ACFM field.
The probability of detection (POD) using the ACFM technique decreases dras-
tically for tiny cracks whose lengths are less than 10 mm [16]. Smith et al. applied
the ACFM technique to inspect the welds in stainless steel nuclear storage tanks
[17]. The results showed that most of the tiny cracks (length less than 10 mm, depth
less than 1 mm) could not be identified effectively. Mostafavi et al. presented the

Fig. 2 Weld ripples and reinforcement cause lift-off variations

2 Insensitive Signal to Lift-Off Variations 17

mathematical method for the theoretical prediction of the magnetic field distribution
around a short crack (length 30 mm) [18]. Yuan et al. presented the two-step interpo-
lation algorithm for the measurement of long and short cracks in the pipe string using
the uniform alternating current field [19]. Leng et al. proposed the combined metal
magnetic memory (MMM) and the ACFM technique for the detection and evalu-
ation of critical weld joints in a jack-up offshore platform [20]. Rowshandel et al.
presented the artificial neural network (ANN) for sizing the important subsurface
section of the multiple cracks using ACFM [21]. These methods obtained good char-
acteristic signals from the cracks without lift-off variations [22, 23]. However, the
characteristic signal and butterfly plot are cluttered when ACFM is used to identify
tiny surface cracks in the rugged weld.
The aim of this paper is to find out tiny surface cracks in rugged welds. This paper
is organized as follows: in Sect. 2, the insensitive signal to the lift-off variation is
pointed out by simulations and experiments; in Sect. 3, the tiny crack is detected in
the rugged weld and the signal gradient algorithm is presented to identify the tiny
crack; and in Sect. 4, the conclusion and further work are given.

2 Insensitive Signal to Lift-Off Variations

2.1 Simulation Model

In terms of the theory, the Bx is produced by the decrease of the current density in
the depth direction of the crack. Thus, the Bx keeps an individual value (background
signal) when a crack is not present. The Bz is produced by the deflection of the
induced current field, thus the Bz is zero when a crack is not present. When the lift-
off varies, the background signal changes—but the current field is still uniform (the
current density value changes, but still no deflection). Thus, the Bx is easily affected
by the lift-off variations and the Bz is insensitive to lift-off variations.
The 2D simulation model of the ACFM probe was set up as shown in Fig. 3a.
In the model, the frequency is 2 kHz and the current amplitude is 50 mA [24]. The
material of the yoke is Mn–Zn ferrite and the specimen is a steel plate. The relative
permeability of the yoke is 2000. The conductivity and relative permeability of the
specimen are 5 × 106 S/m and 200, respectively. Because steel permeability is much
greater than air permeability, the magnetic line of force propagates from the U-shaped
yoke to the steel. Meanwhile, due to the skin effect, most of the magnetic line of
force gathers in the thin surface of the steel. The magnetic field produces a uniform
current field in the center of the specimen [25, 26].
When the lift-off increases, more magnetic field leaks into the air. Thus, the
background signal of the Bx increases when the lift-off goes up. While there is no
magnetic field in the Z direction, the Bz keeps zero because of the uniform current
field, as shown in Fig. 3b. We can make a conclusion that the Bx is easily affected
by the lift-off variations and the Bz is insensitive to the lift-off variations.
18 Identification of Tiny Surface Cracks in a Rugged Weld by Signal …

Fig. 3 A 2D simulation model of ACFM probe. a Distribution of magnetic field; b background

signal with different lift-off

2.2 System Set Up

The ACFM testing system is shown in Fig. 4a. The system consists of three main
parts: a probe, a signal processing box, and a 3-axis scanner. The probe includes an
excitation coil, a MnZn-ferrite yoke, a detection sensor, and a primary processing
circuit, as shown in Fig. 4b. The excitation coil is wound on the MnZn-ferrite yoke
with 500 turn enameled wires whose diameter is 0.15 mm. The detection sensor is
a 2-axis tunnel magneto resistance (TMR) sensor that is set at the bottom center of
the yoke [27, 28]. The 2-axis TMR sensor (Type: TMR2303, Capacity: ± 80 Oe,
Sensitivity: 3 mV/Oe, made by MULTI DIMENSION, China), is used to measure
the Bx and Bz at the same time. The Bx and Bz are amplified 50 times and 100
times, respectively, by the primary processing circuit. The thickness of the probe
shell bottom is 1 mm to keep the lift-off of the sensor 1 mm. The probe is installed
on a 3-axis scanner which is controlled by the programmable logic controller (PLC).
The signal processing box includes a power module, an excitation module, a
filtration module, and an acquisition module, as shown in Fig. 4b. The power module
provides the power source to keep the modules running for more than six hours.
The excitation module produces a sinusoidal signal, whose frequency is 2 kHz and
amplitude is 10 V. The sinusoidal signal is loaded on the excitation coil in the probe.
The Bx and Bz are filtered by the filtration module using a low-pass filter whose cut-
off frequency is 20 kHz. The Bx and Bz are gathered by the acquisition module and
2 Insensitive Signal to Lift-Off Variations 19

Fig. 4 ACFM system. a Photo of the system; b probe and signal processing module

transmitted to the personal computer (PC) for the signal processing and analyzing
software.

2.3 Lift-Off Variation Experiments

To verify the insensitive signal to lift-off variations, the probe is driven up and down
above the specimen by the scanner. The first specimen is a mild steel plate with a
semi-elliptical artificial crack (length: 50 mm, width: 0.5 mm and depth: 5 mm).
There is no weld on the plate and the surface is flat. In the lift-off upward testing
experiments, the probe is driven by the scanner to scan along the artificial crack with
a lift-off of 1 mm at a speed of 2 mm/s. The probe is raised up 1 mm by the scanner
and then dropped after 2 s, as shown in Fig. 5a.
As shown in Fig. 5b, the Bx shows a trough in the center of the crack, while
the Bz shows opposite peaks at the tips of the crack. The characteristic signals are
20 Identification of Tiny Surface Cracks in a Rugged Weld by Signal …

Fig. 5 Testing results when lift-off upward. a Probe scan path; b Bx, Bz, and butterfly plot

consistent with ACFM theory. When the lift-off is upward, the Bx shows a maximum
peak and the Bz remains the same. This is because the background signal increases
when the lift-off goes up. Meanwhile, the induced current field is still uniform and
the Bz almost remains the same. There are obvious loops in the butterfly plot for the
identification of a crack in the flat specimen. The Bx distorted value caused by the
1 mm lift-off variation is much larger than that caused by the crack, which produces
an interference signal in the butterfly plot.
In the lift-off downward testing experiments, the probe was lowered down 1 mm
by the scanner and then raised after 2 s, as shown in Fig. 6a. When the lift-off is
downward, the Bx shows a minimum trough and the Bz almost remains the same, as
shown in Fig. 6b. This is because the background signal decreases when the lift-off
goes down. In the same way, the crack can be identified by the butterfly plot. The
lift-off downward also produces an obvious interference signal in the butterfly plot.
To simulate the uneven surface of a weld, plastic humps were set on the surface
of the specimen in front of the crack, as shown in Fig. 7a. When the probe scans
the humps, the lift-off varies seriously from 1 to 4 mm. The characteristic signals
are shown in Fig. 7b. The Bx shows hash signals, while the Bz varies slightly in the
hump area. The results show that the Bx is affected seriously by the lift-off variations
and the Bz is insensitive to the lift-off variations. The Bz can be set as the insensitive
signal to identify the tiny surface crack in the weld.
2 Insensitive Signal to Lift-Off Variations 21

Fig. 6 Testing results when probe downward. a Probe scan path; b Bx, Bz, and butterfly plot

Fig. 7 Testing results when lift-off varies seriously. a Probe scan path; b Bx, Bz, and butterfly plot
22 Identification of Tiny Surface Cracks in a Rugged Weld by Signal …

3 Signal Gradient Algorithm

3.1 Detection of Tiny Cracks in the Weld and the HAZ

The second specimen is a plate with one weld, as shown in Fig. 8. The material of the
specimen is the same as the first specimen. The thickness of the plate is 4 mm. The
width of the weld is 10 mm and the height of the weld reinforcement is 2 mm. (The
distance between the weld and the HAZ is 2 mm.) The weld ripples are rugged and
the thickness of the ripples range from 0.1 to 1 mm. Thus, when the ACFM probe
scans along the weld, the lift-off of the probe varies from 0.1 to 1 mm.
There are three tiny surface cracks in the weld with the same length (4 mm) and
different depths (3.5, 3.0 and 2.5 mm). There are other 3 cracks with identical length
of 4 mm and the same depths (W1, W2, and W3) in the HAZ. The depths of the
cracks are given in Table 1.
The probe is driven by the scanner to scan the weld from W1 to H3 at a speed
of 5 mm/s. The probe can shake up and down with the ripples. The testing results
are shown in Fig. 9a. The Bx is chaotic because there are many interference signals
caused by the lift-off variations. Thus, the cracks cannot be identified clearly by the
Bx signal. There are six characteristic signals in the Bz when a crack is present. The
SNR of the Bz is much better than that of the Bx, because the Bz is insensitive to the
lift-off variations. As shown in Fig. 9b, the butterfly plot is irregular, which cannot
be used to identify the 6 tiny cracks.
The third specimen is shown in Fig. 10. There are two different length cracks in
the HAZ and three different length cracks in the weld. The cracks are at the same
depth (3.0 mm). The lengths of the cracks are shown in Table 2.
The weld was scanned at the same speed, and the testing results are shown in
Fig. 11. The Bx shows two troughs, and the Bz shows two troughs and peaks at

Fig. 8 The second specimen with different depth cracks

Table 1 The depth of the cracks

No. W1 W2 W3 H1 H2 H3
Depth/mm 3.5 3.0 2.5 3.5 3.0 2.5
3 Signal Gradient Algorithm 23

Fig. 9 Characteristic signals

of different depth cracks in
the weld and the HAZ. a Bx
and Bz; b butterfly plot

Fig. 10 The third specimen with different length cracks

Table 2 The length of the cracks

No. H4 H5 W4 W5 W6
Length/mm 1.0 2.0 4.0 6.0 8.0
24 Identification of Tiny Surface Cracks in a Rugged Weld by Signal …

Fig. 11 Characteristic
signals of different length
cracks in the weld and the
HAZ. a Bx and Bz;
b butterfly plot

the W5 and W6 cracks. The characteristics of the Bx and Bz are incomplete at the
W4 crack. For the H4 and H5 cracks, there is one response in the Bx, while the Bz
shows slight perturbations because of the very short length of the crack. As shown
in Fig. 11b, there are two obvious loops in the butterfly plot, which can be used to
identify the last two longer cracks in the weld. Other cracks cannot be identified
effectively by the irregular tracks in the butterfly plot.

3.2 Signal Gradient Algorithm

The gradient method can find the amplitude change rate of the signal. The Bz changes
are small when lift-off varies. Thus, the gradient of the Bz will be zero when a crack is
not present. What’s more, there are opposite peaks at the two ends of the crack in the
Bz. As a result, a slight perturbation in the Bz caused by the crack is amplified many
times by the gradient method. Thus, the gradient of the Bz will show an outstanding
3 Signal Gradient Algorithm 25

Fig. 12 Signal gradient algorithm for the processes of the Bz

peak in the center of the crack. The SNR of the crack identification signal will be
improved greatly in the near zero background signal.
The signal gradient algorithm is presented to process the Bz by the following steps,
as shown in Fig. 12. First, the gradient of the Bz is calculated in real time. If a crack
is not present, the value is near zero. When a crack is present, the gradient of the Bz
will show some negative values and a major positive value. Then the negative values
are decreased and the positive value is magnified. To make the curve smooth, the
signal is processed by a 6-th order Butterworth low-pass filter. Thus, the identification
signal is presented with a great positive peak and near zero value. The threshold is
set beforehand to compare the identification signal with the threshold automatically.
If the identification signal is greater than the threshold, the crack can be identified
effectively because of the high SNR.
The Bz signals of different depth cracks (from Fig. 9a) and different length cracks
(from Fig. 11a) are processed by the signal gradient algorithm. The identification
signals of the tiny surface cracks with lift-off variations in the weld and the HAZ are
shown in Fig. 13. As shown in Fig. 13a, there are six obvious peaks in the identification
signals caused by the different length cracks. Meanwhile, there are three obvious
peaks caused by the three longer cracks in the weld and two slight peaks caused by
the two shorter cracks in the HAZ, as shown in Fig. 13b. To remove the background
signal and identify the tiny cracks effectively in real time, the threshold can be set in
advance using 1.5 times the maximum noise amplitude in the identification signal.
In these experiments, the maximum noise amplitude was 53 and the threshold was
set as 80. Thus, all six different depth cracks and five different length cracks can be
identified effectively, as shown in Fig. 13c, d, respectively.

3.3 Discussions

The SNR of the Bz signal is better than that of the Bx in the crack testing results, as
shown in Fig. 9a. This is because the Bx signals are greatly disturbed by the lift-off
variations caused by the rugged weld ripples and reinforcements. The Bz is relatively
stable because it is insensitive to the lift-off variations—thus, the Bz is used as the
characteristic signal to identify the tiny crack with lift-off variations. As shown in
26 Identification of Tiny Surface Cracks in a Rugged Weld by Signal …

Fig. 13 Identification of tiny cracks by signal gradient algorithm. a Identification signal of different
depth cracks; b identification signal of different length cracks; c identification of different depth
tiny cracks; d identification of different length tiny cracks
References 27

Fig. 11a, where the length of the crack is shorter than 4 mm, the distortion of the
Bx and Bz decays seriously. This is because the disturbance of the induced current
field declines sharply around the short cracks. For the 1 and 2 mm length cracks, the
induced current field turns slightly and the magnetic field distorts weakly.
As shown in Fig. 13a, the last three peaks are less than the first three peaks. This is
because the last three cracks are in the HAZ and the distance between the probe and
the HAZ is greater than the distance between the probe and the weld. Because the
first three cracks are in the rugged weld, the background signal in the identification
signal has a bit more clutter than that of the last three cracks. As shown in Fig. 13b,
the peaks of the three longer cracks are obvious and the peaks of first two shorter
cracks are weaker. Because the length of the tiny cracks has a significant impact on
the Bx and Bz signals, it is difficult to identify the shorter cracks (less than 2 mm)
in the weld and the HAZ—the shorter cracks are easily covered by the noise signal.
Overall, in these experiments the SNR of the identification signals must be relatively
high to identify all the tiny cracks in the weld and the HAZ by threshold compared
to the conventional butterfly plot.

4 Conclusions

In this paper, a signal gradient algorithm was presented to identify tiny cracks in
welds using the ACFM technique. The insensitive signal to the lift-off variations
was pointed out by the simulations and experiments. The ACFM probe with a 2-axis
TMR sensor and testing system were developed. The different depth and different
length tiny surface cracks were detected in the rugged weld and the HAZ using
an ACFM testing system. The results show that the Bz signal was the insensitive
signal to the lift-off variations. The SNR of the crack response signal was greatly
improved by the signal gradient algorithm. Thus, all the tiny surface cracks can be
effectively identified in the weld and the HAZ by the signal gradient algorithm using
the ACFM technique. Further work will focus on the identification of other type
defects, evaluation of tiny defects in the weld, and the HAZ.

References

1. Taheri H, Kilpatrick M, Norvalls M, Harper WJ (2019) Investigation of nondestructive testing

methods for friction stir welding. Metals 9:624
2. Wang R, Liu ZH, Wu JF, Jiang BY, Li B (2019) Research on phased array ultrasonic testing
on CFETR vacuum vessel welding. Fusion Eng Des 139:124–127
3. Gao XD, Du LL, Ma N, Zhou XH, Wang CY, Gao PP (2019) Magneto-optical imaging char-
acteristics of weld defects under alternating and rotating magnetic field excitation. Opt Laser
Technol 112:188–197
28 Identification of Tiny Surface Cracks in a Rugged Weld by Signal …

4. Andreaus U, Baragatti P, Casini P, Iacoviello D (2016) Experimental damage evaluation of

open and fatigue cracks of multi-cracked beams by using wavelet transform of static response
via image analysis. Struct Control Health Monit 105:227–244
5. Shelikhov GS, Glazkov YA (2011) On the improvement of examination questions during the
nondestructive testing of magnetic powder. Russ J Nondestr Test 47:112–117
6. Pashagin AI, Shcherbinin VE (2012) Indication of magnetic fields with the use of galvanic
currents in magnetic-powder nondestructive testing. Russ J Nondestr Test 48:528–531
7. Takiy AE, Kitano C, Higuti RT, Granja SCG, Prado VT, Elvira L, Martinez-Graullera Ó (2017)
Ultrasound imaging of immersed plates using high-order lamb modes at their low attenuation
frequency bands. Mech Syst Sign Pr 96:321–332
8. Ricci M, Silipigni G, Ferrigno L, Laracca M, Adewale I, Tian GY (2017) Evaluation of the
lift-off robustness of eddy current imaging techniques. NDT&E Int 85:43–52
9. Katoozian D, Hasanzadeh RPR (2017) A fuzzy error characterization approach for crack depth
profile estimation in metallic structures through ACFM data. IEEE T Magn 53:6202100
10. Lewis AM, Michael DH, Lugg MC, Collins R (1988) Thin-skin electromagnetic fields around
surface-breaking cracks in metals. J Appl Phys 64:3777–3784
11. Raine A (2002) Cost benefit applications using the ACFM technique. Insight 44:180–181
12. Akbari-Khezri A, Sadeghi SHH, Moini R, Sharifi M (2016) An efficient modeling technique
for analysis of AC field measurement probe output signals to improve crack detection and
sizing in cylindrical metallic structures. J Nondestr Eval 35:9
13. Deng ZY, Sun YH, Yang Y, Kang YH (2017) Effects of surface roughness on magnetic flux
leakage testing of micro-cracks. Meas Sci Technol 28:045003
14. Lu MY, Xu HY, Zhu WQ, Yin LY, Qian Z, Peyton AJ, Yin WL (2018) Conductivity lift-off
invariance and measurement of permeability for ferrite metallic plates. NDT&E Int 95:36–44
15. Andreaus U, Casini P, Vestroni F (2005) Nonlinear features in the dynamic response of a cracked
beam under harmonic forcing. In: DETC’05, proceeding of the ASME 2005 international design
engineering technical conferences & computers and information in engineering conference,
Long Beach, CA, USA, 24–28 Sept 2005, pp 2083–2089
16. Shen JL, Zhou L, Rowshandel H, Nicholson GL et al (2015) Determining the propagation
angle for non-vertical surface-breaking cracks and its effect on crack sizing using an ACFM
sensor. Meas Sci Technol 26:115604
17. Smith M, Laenen C (2007) Inspection of nuclear storage tanks using remotely deployed
ACFMT. Insight 49:17–20
18. Mostafavi RF, Mirshekar-Syahkal D (1999) AC fields around short cracks in metals induced
by rectangular coils. IEEE T Magn 35:2001–2006
19. Yuan XA, Li W, Chen GM, Yin XK, Yang WC, Ge JH (2018) Two-step interpolation algorithm
for measurement of longitudinal cracks on pipe strings using circumferential current field
testing system. IEEE T Ind Inform 14:394–402
20. Leng JC, Tian HX, Zhou GQ, Wu ZM (2017) Joint detection of MMM and ACFM on critical
parts of jack-up offshore platform. Ocean Eng 35:34–38 (in Chinese)
21. Rowshandel H, Nicholson GL, Shen JL, Davis CL (2018) Characterization of clustered cracks
using an ACFM sensor and application of an artificial neural network. NDT&E Int 98:80–88
22. Yuan XA, Li W, Chen GM, Yin XK, Jiang WY, Zhao JM, Ge JH (2019) Inspection of both
inner and outer cracks in aluminum tubes using double frequency circumferential current field
testing method. Mech Syst Sign Pr 127:16–34
23. Tian GY, Sophian A (2005) Reduction of lift-off effects for pulsed eddy current. NDT&E Int
38:319–324
24. Yuan XA, Li W, Chen GM, Yin XK, Ge JH (2017) Frequency optimisation of circumferential
current field testing system for highly-sensitive detection of Iongitudinal cracks on a pipe string.
Insight 59:378–382
25. Li Y, Yan B, Li D, Li YL, Zhou DQ (2016) Gradient-field pulsed eddy current probes for
imaging of hidden corrosion in conductive structures. Sens Actuators A Phys 38:251–265
26. Li Y, Ren ST, Yan B, Abidin IMZ, Wang Y (2017) Imaging of subsurface corrosion using
gradient-field pulsed eddy current probes with uniform field excitation. Sensors 17:1747
References 29

27. Atzlesberger J, Zagar BG, Cihal R, Brummayer M, Reisinger P (2013) Sub-surface defect
detection in a steel sheet. Meas Sci Technol 24:084003
28. AbdAlla AN, Faraj MA, Samsuri F, Rifai D, Ali K, Al-Douri Yarub (2019) Challenges in
improving the performance of eddy current testing. Rev Meas Control 52:46–64

Open Access This chapter is licensed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits use, sharing,
adaptation, distribution and reproduction in any medium or format, as long as you give appropriate
credit to the original author(s) and the source, provide a link to the Creative Commons license and
indicate if changes were made.
The images or other third party material in this chapter are included in the chapter’s Creative
Commons license, unless indicated otherwise in a credit line to the material. If material is not
included in the chapter’s Creative Commons license and your intended use is not permitted by
statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder.
Visual Evaluation of Irregular Cracks
in Steel by Double Gradient Fusion
Algorithm Using Composite ACFM-MFL
Testing Method

1 Introduction

The steel is the most common material in the industrial and civil facilities. Because of
the corrosion, stress and fatigue, the cracks are easily introduced in the surface of the
steel. Generally, the incipient crack grows and tends to different directions resulting
in the irregular crack due to the complex stress and special structural shape [1–3]. In
order to prevent the corrosion, there are various coatings on the surface of the steel.
The irregular crack is covered by the coating, which leaves hidden trouble in the
facility [4]. Thus it is of prime significance to propose an effective nondestructive
testing (NDT) method to achieve visual detection and evaluation of the irregular
crack under the coating [5, 6].
Usually, several NDT methods are proposed to detect and evaluate the crack in the
steel. The visual testing (VT), magnetic particle testing (MT) and penetrant testing
(PT) are common methods to inspect the surface crack [7]. The surface morphology
of the crack can be shown visually by the gathered magnetic particle, penetrant or
optical instrument [8, 9]. However, the attachments, contaminations and coatings
should be cleaned thoroughly. After inspection, the coating should be painted again,
which is time-consuming and high-cost. What’s more, these NDT methods cannot be
applied in the service time in most case. The ultrasonic testing (UT) is mainly used to
test the inner defects in the steel, which needs coupling medium during application
[10]. Because of the radioactivity, the radiographic testing (RT) is forbidden in many
industrial field.
The electromagnetic nondestructive testing (ENDT) techniques are excellent
methods to test the surface crack due to the advantages of high sensitivity, non-
contact, low cost and no coupling required [11–14]. Many scholars have made broad
scale researches on the detection and evaluation of cracks using the ENDT technique.
Rowshandel, et al. proposed the artificial neural network (ANN) to learn the inverse
relationship between the crack pocket length and the ACFM signal for a given cluster
of rolling contact fatigue (RCF) cracks in the railway rails [15, 16].

© The Author(s) 2025 31

W. Li et al., Alternating Current Field Measurement Technique for Detection and
Measurement of Cracks in Structures, https://doi.org/10.1007/978-981-97-7255-1_3
32 Visual Evaluation of Irregular Cracks in Steel by Double Gradient …

Many scholars have made broad scale researches on the inspection of cracks using
the ACFM technique. Dover, Lewis, et al. developed the classical theoretical model
of the ACFM technique [17, 18]. The model is developed on the assumption that
the injected current field flows around one crack vertically. Pasadas, et al. presented
the Tikhonov regularization inversion algorithm to obtain the geometrical profile of
defects in the 2D surface view [19]. Feng, et al. proposed an optimized ACFM probe
for the detection of inner cracks inside the pipeline [20].
These works have obtained good results of surface cracks when the induced current
field is perpendicular to the crack. In other words, the probe should scan along the
direction of the crack. However, the direction of the crack is unclear under the coating
before the inspection in practical work. Thus, some scholars did research on the
inspection of different angle cracks. Rowshandel, et al. proposed the artificial neural
network (ANN) to learn the inverse relationship between the crack pocket length
and the ACFM signal for a given cluster of rolling contact fatigue (RCF) cracks in
the railway rail [21]. Shen, et al. studied the relationship between ACFM signals
and propagation angles of surface-breaking cracks [22]. In our previous work, the
rotating alternating current field measurement (RACFM) technique was presented to
overcome the limitations of the classical theoretical model of the ACFM technique.
The rotating current field is induced by two orthogonal excitation coils [23]. Thus, the
induced current field is always perpendicular to arbitrary-angle cracks. These works
all focus on the inspection of one crack with some specific angles [24, 25]. However,
the irregular crack usually grows and tends to different directions, which results in
some partial cracks. The development of the irregular crack should be monitored by
the periodic detection. On one hand, the rotating current field cannot pass through the
irregular vertically always because the rotating current field is cut off by the discon-
tinuous area. On the other hand, the RACFM inspects cracks by the conventional
amplitude of the magnetic signal, which cannot show the surface morphology of the
crack visually [26]. What’s more, two excitation coils and excitation signals should
be used in the RACFM system making the RACFM probe huge and complicated.
In this paper, the double gradient fusion algorithm is presented to achieve visual
evaluation of the irregular crack under the coating by the composite ACFM-MFL
testing method. On one hand, the excitation coil can excite AC primary magnetic field
to produce the leaked magnetic field caused by the transverse crack. The primary
magnetic field induces current field in the steel to produce the distorted magnetic
field caused by the longitudinal crack. Thus the magnetic field can be measured by
the same probe with AC excitation, which is beneficial to achieve visual evaluation
of the irregular crack. On the other hand, the composite ACFM-MFL testing method
is sensitive to the orthogonal directional cracks. As a result, any part of the irregular
crack can be detected and evaluated effectively.
The rest of the paper is organized as follows. In Sect. 2, the FEM model of the
ACFM-MFL with an irregular crack is presented to analyze the leaked and distorted
magnetic field. In Sect. 3, the double gradient fusion algorithm is developed to
reconstruct the morphology of the irregular crack using the simulation results. In
Sect. 4, the irregular crack testing experiments are carried out and the irregular crack
2 Finite Element Method Model 33

is evaluated visually by the double gradient fusion algorithm. Finally, the conclusion
and future work are drawn in Sect. 5.

2 Finite Element Method Model

The FEM model of the ACFM-MFL with the irregular crack is set up, as shown in
Fig. 1. The model includes an excitation coil, a U-shaped yoke, a steel specimen and
an irregular crack. The excitation coil is wound on the U-shaped yoke with 500 turns.
The lift-off of the U-shaped yoke is 3 mm to the specimen (1 mm is for the shell
of the probe and the lift-off of the sensor is 2 mm). The irregular crack is consists
of 4 cracks whose angle is 0°, 30°, 60°, 90°. The depth and length of the 4 cracks
are 6 mm and 30 mm respectively. The material parameters of the model is given in
Table 1.
The excitation coil is loaded by a sine signal whose frequency is 1000 Hz and
current magnitude is 50 mA. According to the Lenz’s law, the primary magnetic field
is excited in the X direction, while the induced uniform current field flows in the

Fig. 1 FEM model of ACFM-MFL with irregular crack. a Model FEM. b Principle of distorted
magnetic field around irregular crack. c Vector current field around irregular crack. d Current density
around irregular crack

Table 1 Parameters of EFM model

Model Air Excitation coil Steel specimen
Resistivity/Ω m 0.5 × 104 1.7 × 10–8 9.78 × 10–8
Relative permeability/μr 1.0 1.0 1000
34 Visual Evaluation of Irregular Cracks in Steel by Double Gradient …

Y direction, as shown in Fig. 1b. When the irregular crack is present, the primary
magnetic field pass through the 90° crack, 60° crack and 30° crack. As a result, the
primary magnetic field leaks in the air and produces the magnetic leakage signal.
Meanwhile, the induced uniform current field is disturbed by the irregular crack, as
shown in Fig. 1c, d. Especially, the current is interrupted by the 0° crack, 30° crack
and 60° crack, which makes the space magnetic field distorted. For the 30° and 60°
cracks, both the magnetic leakage effect and current field perturbation effect exist at
the same time [27].
The space magnetic field is extracted above the irregular crack with the lift-off
of 2 mm, as shown in Fig. 2. The magnetic field in the X direction (Bx) shows the
maximum peak at the 90° crack due to the strongest magnetic leakage effect. Because
the magnetic leakage effect is much greater than the current field perturbation effect,
the trough of Bx caused by the decreased current density (Secondary magnetic field)
is very tiny relatively at the 0° crack, as shown in Fig. 2a. The magnetic field in the
Z direction (Bz) shows peak and trough at each side of the 90° crack, as shown in
Fig. 2b. This is due to the opposite direction of the leaked magnetic field in the Z
direction [28]. Meanwhile, the surface current field turns around the irregular crack
greatly and gathers at the tips of the irregular crack. Thus, the disturbed current field
makes the space magnetic field distorted along the 0° crack, 30° crack and 60° crack,
as shown in Fig. 2a.
On one hand, because the Bz is mainly caused by the surface disturbed current
around the irregular crack and the leaked magnetic field besides the crack, the outline
of the Bz image is more similar to the morphology of the irregular crack (Comparing
Fig. 2b with Fig. 2a). On the other hand, the Bz is insensitive to the lift-off effect
as the zero background signal, which has a higher signal to noise ratio (SNR). Thus
the Bz is more beneficial to detect and evaluate the irregular crack under the coating.
The distorted magnetic field is measured by the magnetic sensor no matter what’s
the magnetic leakage effect or current field perturbation effect. Thus the Bz signal

Fig. 2 Distorted magnetic field around irregular crack. a Magnetic field in X direction. b Magnetic
field in Z direction
3 Double Gradient Fusion Algorithm 35

can be used to reconstruct the morphology of the irregular crack using the composite
ACFM-MFL testing method.

3 Double Gradient Fusion Algorithm

As the Bz signal changed greatly from the trough to the peak at each side of the
irregular crack, the gradient field can be used to find the maximum variation direction
and value of the scalar field, which is defined in Eq. (1) [29–32].
( )
∂u ∂u ∂u
grad u(x, y, z)= , , (1)
∂x ∂y ∂z

where the u(x, y, z) is the scalar field, the grad u(x, y, z) is the gradient field of the
scalar field.
Because there are 4 cracks from 0° to 90°, the peaks and troughs change greatly
in the X direction and Y direction respectively. Thus the X direction gradient field
(GXBz ) and Y direction gradient field (GXBz ) of the Bz scalar field (Bz value at each
location from Fig. 2b) should be calculated using the Eq. (1), as shown in Fig. 3. The
results show that the 90° crack, 60° crack and 30° crack can be presented obviously
in the X direction gradient of the Bz scalar field, as shown in Fig. 3a. Meanwhile the
0° crack, 30° crack and 60° crack can be presented in the Y direction gradient of the
Bz scalar field, as shown in Fig. 3b.
To get the whole morphology of the irregular crack, the back ground signals of
the GXBz and GYBz are removed to form the GXRB and GYRB respectively. Then
GXRB and GYRB are normalized to 0–1 to form the GX0-1 and GY0-1 respectively.
The maximum distorted value is set as 1 and the minimum distorted value is set as
0, as shown in Fig. 4.

Fig. 3 Double direction gradient field of Bz. a X direction gradient field. b Y direction gradient
field
36 Visual Evaluation of Irregular Cracks in Steel by Double Gradient …

Fig. 4 Normalized gradient field. a Removed background signal of X direction gradient field.
b Normalized X direction gradient field. c Removed background signal of X direction gradient
field. d Normalized Y direction gradient field

Then the GX0-1 and GY0-1 are added up, as shown in Fig. 5a. The outline of the
added gradient is similar to the morphology of the irregular crack. In the end, the
added gradient image is converted to the greyscale map. Thus the irregular crack can
be shown visually and obviously, as shown in Fig. 5b. We can make a conclusion
that the irregular crack can be reconstructed visually by the double gradient fusion
algorithm.
Based on the analysis above, the double gradient fusion algorithm is presented in
Fig. 6. The double gradient fusion algorithm includes 6 steps:
⎡ ⎤
a11 a12 · · · a1n
⎢ ⎥
⎢ a21 a22 · · · a2n ⎥
⎢ ⎥
• Step 1. Get the scalar field Bz = ⎢ ⎥
⎢ a31 a32 · · · a3n ⎥. amn is the value of the Bz
⎢ ⎥
⎣· · · · · · · · · ⎦
am1 am2 · · · amn
at each location above the irregular crack.
3 Double Gradient Fusion Algorithm 37

Fig. 5 Visual results of irregular crack. a Added gradient field. b Greyscale map of irregular crack

Fig. 6 Double gradient fusion algorithm

38 Visual Evaluation of Irregular Cracks in Steel by Double Gradient …

• Step 2. Get the double gradient field of the Bz. Calculate the X direction gradient
of each row of⎡ the Bz (∂ain /∂x, i = 1, 2,⎤3…m.), forming the X direction gradient
∂xa11 ∂xa12 · · · ∂xa1n
⎢ ⎥
⎢ ∂xa21 ∂xa22 · · · ∂xa2n ⎥
⎢ ⎥
field GXBz = ⎢ ⎥
⎢ ∂xa31 ∂xa32 · · · ∂xa3n ⎥.
⎢ ⎥
⎣· · · · · · · · · ⎦
∂xam1 ∂xam2 · · · ∂xamn
At the/same time, Calculate the Y direction gradient of each column of the
⎡ (∂amj ∂y, j = 1, 2, 3…n.),
Bz ⎤ forming the Y direction gradient field GYBz =
∂ya11 ∂ya12 · · · ∂ya1n
⎢ ⎥
⎢ ∂ya21 ∂ya22 · · · ∂ya2n ⎥
⎢ ⎥
⎢ ∂ya31 ∂ya32 · · · ∂ya3n ⎥.
⎢ ⎥
⎢ ⎥
⎣ · · · · · · · · · ⎦
∂yam1 ∂yam2 · · · ∂yamn
• Step 3. Remove the background signal of the gradient fields. Find the extreme
value of the GXBz (PGXBz ) and the extreme value of GYBz (PGXBz ). If the
extreme value is greater than 0, the negative value in the direction gradient field
is removed. If the extreme value is less than 0, the elements in the gradient field
multiply − 1 and then the negative value is removed. The X direction and Y
direction removed background are called GXRB and GXRB respectively.
• Step 4. Get the normalized gradient field. The GXRB is normalized to 0–1 forming
the GX0-1 . The maximum value of GX0-1 is 1 and the minimum value of GX0-1
is 0. The GYRB is processed in the same way to get the GY0-1.
• Step 5. Get the fusion gradient field. The fusion gradient field is calculated by the
equation: GXY = GX0-1 + GY0-1 . The GXY is plotted with position coordinates
forming the fusion gradient field image.
• Step 6. Get the greyscale map. The fusion gradient field image is converted to the
greyscale map to achieve visual detection and evaluation of the irregular crack.

4 Experiments

4.1 Testing System

The diagram of the ACFM-MFL testing system is shown in Fig. 7a. The enameled
wires are wound on the U-shaped yoke with 500 turns as the excitation coil. The signal
generator produces a sine signal with the frequency 1000 Hz and the voltage ampli-
tude 10 V, which is loaded on the excitation coil. The excitation coil excites magnetic
field and induces uniform current field in the steel specimen. The tunnel magneto
resistance (TMR) chips (Type: TMR2303, Sensitivity 3 mV/Oe, made in MULTI
DIMENSION, China.) is located in the probe bottom shell (Thickness 1.0 mm) for
4 Experiments 39

Fig. 7 ACFM-MFL testing system. a Diagram of system. b Internal structure of probe

the measurement of the distorted magnetic field, as shown in Fig. 7b. The measured
magnetic signal is amplified 100 times by the AD620 chip in the primary circuits.
The amplified signal is smoothed by the filtering circuits and then captured by the
acquisition card. In the end, the signal is processed by the software in the computer.

4.2 Testing Experiments

The specimen is a Q235 steel plate with an irregular crack, as shown in Fig. 8a. The
irregular crack consists of 4 cracks, whose angle is 0°, 30°, 60° and 90°. The length of
the crack is 20 mm and the depth is 2 mm and the width is 0.8 mm. The polyethylene
(PE) coating (Thickness 2 mm) is fixed above the irregular crack. The probe is driven
by a 3-axial scanner to scan the coating area (90 × 90 mm2 ) to measure the distorted
magnetic field, as shown in Fig. 8b. The scan path is a rectangular grid and the step
size is 0.5 mm in the X direction and the Y direction.
The Bx and Bz signals are picked up at each location to form the scalar field, as
shown in Fig. 9. In the Bx scalar field, there are peaks at the 90° crack and 60° crack.

Fig. 8 a Photo of specimen with irregular crack (partial view). b Photo of testing system
40 Visual Evaluation of Irregular Cracks in Steel by Double Gradient …

Fig. 9 Testing results of irregular crack. a Bx scalar field. b Bz scalar field

In the Bz scalar field, there are troughs and peaks at each side of the irregular crack.
The gathered current field produces peak in the Bz at the tips of 0° crack. The leaked
magnetic field produces obviously peak and trough in the Bz at each side of the 90°
crack. The Bz scalar field is more similar to the morphology of the irregular crack,
which is in accord with the simulation results. Thus the Bz scalar field is set as the
characteristic signal to achieve visual evaluation of the irregular crack.

4.3 Verification of Double Gradient Fusion Algorithm

The Bz scalar field is processed by the double gradient fusion algorithm step by
step. The X direction and Y direction gradient field of the Bz are shown in Fig. 10a,
b respectively. Then the background signals of the GXBz and GYBZ are removed
to get the GXRB and GXRB , as shown in Fig. 10c, d respectively. Next, the GXRB
and GYRB are normalized to GX0-1 and GY0-1 , as shown in Fig. 10e, f respec-
tively. The 30° crack, 60° crack and 90° crack can be seen in Fig. 10e because of
the magnetic leakage effect. Meanwhile, the 0° crack, 30° crack and 60° crack can
be seen in Fig. 10f due to the current perturbation effect. The experimental results
are consistent with the theory and simulation results.
In the end, the GX0-1 adds the GY0-1 to form the GXY , as shown in Fig. 11a.
Figure 11a is converted to the greyscale map, as shown in Fig. 11b. The morphology
of the irregular crack is shown obviously, which accords with the true shape of the
irregular crack in the steel. To achieve visual evaluation of the irregular crack, the
tips of the irregular crack are picked up in the greyscale map. The irregular crack
is reconstructed and plotted by the location of the tips, as shown in Fig. 11c. The
reconstructed image is in accord with the full morphology of the irregular crack.
4 Experiments 41

Fig. 10 Double gradient fusion algorithm for experimental Bz scalar field. a X direction gradient
field GXBz . b Y direction gradient field GYBz . c Removed background signal GXRB . d Removed
background signal GYRB . e Normalized GX0-1 . f Normalized GY0-1
42 Visual Evaluation of Irregular Cracks in Steel by Double Gradient …

Fig. 11 Visual evaluation results of irregular crack. a Fusion gradient field GXY . b Greyscale map.
b Location of crack tips

Table 2 Evaluated results of irregular crack

Name 0° crack 30° crack 60° crack 90° crack
Length/mm 20.1 20.6 21.1 20.0
Angle 5.7° 29.1° 58.6° 90.0°

The length and angle of the irregular crack can be evaluated by the location of
the tips, as given in Eqs. (2) and (3). The evaluated results of the irregular crack is
given in Table 2. The maximum angle error is 5.7° and the maximum length error
is 1.1 mm whose maximum relative error is 5.5%. Because the magnetic leakage
effect is stronger than the current perturbation effect, the evaluated angle and length
of the 90° crack is more precise than that of the 0° crack. The results show that the
irregular crack can be evaluated visually and accurately by the location of the tips.
We can make a conclusion that the irregular crack in the steel under the coating
can be reconstructed and evaluated visually by the double gradient fusion algorithm
using the composite ACFM-MFL testing method.
/
Length = (X 1 − X 2 )2 +(Y 1 − Y 2 )2 (2)

Angle = artan(|Y1 − Y2|/|X1 − X2|) × 180/π (3)

5 Conclusion and Further Work

The double gradient fusion algorithm is presented to achieve visual evaluation of the
irregular crack using the composite ACFM-MFL testing method. The leaked primary
magnetic field and distorted secondary magnetic field are analyzed by the FEM
model. The Bz is defined as the characteristic signal to reconstruct the morphology
References 43

of the irregular crack. The double gradient fusion algorithm is presented to process
the Bz scalar field for the visual evaluation of the irregular crack. The ACFM-MFL
testing system is developed to detect the irregular crack in the steel under the coatings.
The efficiency of the double gradient fusion algorithm is verified by the experimental
Bz scalar field. The results show that the composite ACFM-MFL testing method is
sensitive to any part of the irregular crack due to the combination of the magnetic
leakage effect and current field perturbation effect. The whole morphology of the
irregular is reconstructed accurately by the double gradient fusion algorithm. As a
results, the irregular crack is evaluated visually by the greyscale map of the crack
morphology. The maximum evaluated angle error is 5.7° and the maximum evaluated
length error is 1.1 mm in the experiment. The double gradient fusion algorithm and
composite ACFM-MFL testing method can be used in the detection and evaluation
of any direction cracks in the steel. The morphology of various cracks can be shown
visually without removing the coatings. Further work will focus on the evaluation of
the profile of irregular cracks, the development of high resolution sensor array for
online visual evaluation of crack under the coatings.

References

1. Corcoran J, Hooper P, Davies C, Nagy PB, Cawley P (2016) Creep strain measurement using
a potential drop technique. Int J Mech Sci 110:190–200
2. Cao W et al (2019) Microstructural material characterization of hypervelocity-impact-induced
pitting damage. Int J Mech Sci 163:105097
3. Lofrano E, Paolone A, Ruta G (2020) Dynamic damage identification using complex mode
shapes. Struct Control Health Monit Sci 27:e2632
4. Gao X, Wang Y, Zhang Z, Tan J, Cheng L, Yang W (2019) Conductive nano-carbon coating
on silica by pyrolysis of polyethylene. Mater Lett 255:126567
5. Earls CJ (2013) Bayesian inference of hidden corrosion in steel bridge connections: non-contact
and sparse contact approaches. Mech Syst Signal Process 41(1–2):420–432
6. Chen Q, Huang Y, Weng X, Liu W (2021) Curve-based crack detection using crack information
gain. Struct Control Health 4(8):1–27
7. Khan A, Ali SSA, Meriaudeau F, Malik AS, Soon LS, Seng TN (2017) Visual feedback-based
heading control of autonomous underwater vehicle for pipeline corrosion inspection. Int J Adv
Robot Syst 3:177–189
8. Andrushia D, Anand N, Arulraj P (2020) A novel approach for thermal crack detection and
quantification in structural concrete using ripplet transform. Struct Control Health 11:1–20
9. Lee JS, Hwang SH, Choi IY, Choi Y (2020) Estimation of crack width based on shape-sensitive
kernels and semantic segmentation. Struct Control Health Monitor 27(4):1–21
10. Zhao J, Li W, Ge J (2020) Coiled tubing wall thickness evaluation system using pulsed
alternating current field measurement technique. IEEE Sensors J 18:10495–10501
11. Xu Y, Yang Y, Wu Y (2020) Eddy current testing of metal cracks using spin Hall
magnetoresistance sensor and machine learning. IEEE Sensors J 20(18):10502–10510
12. Xie SA et al (2020) Features extraction and discussion in a novel frequencyband-selecting
pulsed eddy current testing method for the detection of a certain depth range of defects. NDT
E Int 111:102211
13. Wang C, Fan M, Cao B, Ye B, Li W (2018) Novel noncontact eddy current measurement of
electrical conductivity. IEEE Sensors J 18(22):9352–9359
44 Visual Evaluation of Irregular Cracks in Steel by Double Gradient …

14. Li X, Gao B, Woo WL, Tian GY, Qiu X, Gu L (2017) Quantitative surface crack evaluation
based on eddy current pulsed thermography. IEEE Sensors J 17(2):412–421
15. Rowshandel H, Nicholson GL, Shen JL, Davis CL (2018) Characterisation of clustered cracks
using an ACFM sensor and application of an artificial neural network. NDT E Int 98:80–88
16. Raine A (2002) Cost benefit applications using the alternating current field measurement testing
technique. Mater Eval 1:49–52
17. Dover WD, Collins R, Michael DH, Thompson RB (1986) The use of AC-field measurements
for crack detection and sizing in air and underwater. Phil Trans Roy Soc Lond Ser A Math
Phys Sci 1554:271–283
18. Lewis AM, Michael DH, Lugg MC, Collins R (1988) Thin-skin electromagnetic fields around
surface-breaking cracks in metals. J Appl Phys 8:3777–3784
19. Pasadas DJ, Ribeiro AL, Rocha T, Ramos HG (2016) 2D surface defect images applying
Tikhonov regularized inversion and ECT. NDT E Int 80:48–57
20. Feng Y, Zhang L, Zheng W (2018) Simulation analysis and experimental study of an alternating
current field measurement probe for pipeline inner inspection. NDT E Int: 123–129
21. Rowshandel H, Nicholson GL, Shen JL, Davis CL (2018) Characterisation of clustered cracks
using an ACFM sensor and application of an artificial neural network. NDT E Int 98:80–88
22. Shen JL, Zhou L, Rowshandel H, Nicholson GL, Davis CL (2015) Determining the propagation
angle for non-vertical surface-breaking cracks and its effect on crack sizing using an ACFM
sensor. Meas Sci Technol 26(11):115604
23. Li W, Yuan X, Chen G, Ge J, Yin X, Li K (2016) High sensitivity rotating alternating current
field measurement for arbitrary-angle underwater cracks. NDT E Int 79:123–131
24. Zhang N, Ye C, Peng L, Tao Y (2019) Novel array eddy current sensor with three-phase
excitation. IEEE Sensors J 19(18):7896–7905
25. Wang Y, Ye C, Wang M (2019) Synthetic magnetic field imaging with triangle excitation coil
for inspection of any orientation defect. IEEE Trans Instrum Meas 69(2):533–541
26. Zhang N, Ye C, Peng L, Tao Y (2020) Eddy current probe with three-phase excitation and
integrated array tunnel magnetoresistance sensors. IEEE Trans Ind Electron 68(6):5325–5336
27. Ge J et al (2017) Analysis of signals for inclined crack detection through alternating current field
measurement with a U-shaped probe. Insight Non-Destr Test Condition Monitor 59(3):121–128
28. Wu D, Liu Z, Wang X, Su L (2017) Composite magnetic flux leakage detection method for
pipelines using alternating magnetic field excitation. NDT E Int 91(1):148–155
29. Gong Q, Zhu L, Wang Y, Yu Z (2021) Automatic subway tunnel crack detection system based
on line scan camera. Struct Control Health 8:1–22
30. Gao Y, Su Y, Li Q, Li H, Li J (2020) Single image dehazing via self-constructing image fusion.
Signal Process 167:107284
31. Wu L, Liu Y, Liu N, Zhang C (2016) High-resolution images based on directional fusion of
gradient. Comput Vis Media 4(1):31–43
32. Luo L, Feng MQ, Wu J, Bi L (2021) A vision-based surface displacement/strain measurement
technique based on robust edge-enhanced transform and algorithms for high spatial resolution.
Struct Control Health Monit 28(9):1–31
References 45

Open Access This chapter is licensed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits use, sharing,
adaptation, distribution and reproduction in any medium or format, as long as you give appropriate
credit to the original author(s) and the source, provide a link to the Creative Commons license and
indicate if changes were made.
The images or other third party material in this chapter are included in the chapter’s Creative
Commons license, unless indicated otherwise in a credit line to the material. If material is not
included in the chapter’s Creative Commons license and your intended use is not permitted by
statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder.
Design and Experiment Research
of Oblique Crack Detection System
for Rail Tread Based on ACFM
Technique

1 Introduction

Railway transportation, as one of the main transportation methods in China, plays

an important role in the entire transportation system. In the past decade, China has
implemented six large-scale train speed increases, which have greatly improved
railway transportation capacity while also presenting challenges for safe train oper-
ations [1, 2]. Due to the action of wheel-rail cyclic loading, the rail tread is prone to
micro-cracks. These cracks are typical rolling contact fatigue cracks, with the crack’s
horizontal angle usually at 45° to the direction of travel. Initially, they form at an
angle of 10–15° perpendicular to the rail tread surface. After reaching a certain depth,
these cracks rapidly propagate towards the interior of the rail head at a larger angle,
leading to surface stripping, spalling, or even rail breakage. This poses a serious
threat to train safety [3–5]. Therefore, conducting inspections for oblique cracks on
the rail tread surface is crucial for ensuring safe railway operations and preventing
accidents.
At present, the commonly employed non-destructive testing techniques for rail
tread surface defects include ultrasonic testing, magnetic flux leakage testing, eddy
current testing, and visual inspection. However, the existing detection methods still
have limitations in effectively identifying surface defects on the rail tread surface
[6–11]. Ultrasonic testing has been widely used for the detection and assessment of
internal defects in rails, but it requires a coupling agent during testing and is subject
to blind spots in surface and near-surface inspection [12]. Magnetic flux leakage
testing requires magnetization of the rail, and it is not sensitive to small open surface
and near-surface cracks [13]. Eddy current testing can effectively detect surface and
near-surface defects, but the detection results are greatly influenced by the probe lift-
off [14]. Visual inspection techniques can accurately identify large-scale defects and
corrosion damage on rail tread surfaces. However, it is challenging to detect micro-
cracks on the surface and near-surface areas, and it cannot provide information about
the depth of the cracks [15].

© The Author(s) 2025 47

W. Li et al., Alternating Current Field Measurement Technique for Detection and
Measurement of Cracks in Structures, https://doi.org/10.1007/978-981-97-7255-1_4
48 Design and Experiment Research of Oblique Crack Detection System …

Compared to the aforementioned non-destructive testing techniques, Alternating

Current Field Measurement (ACFM) technology combines the uniform current field
of alternating current voltage drop detection and the non-contact detection of eddy
current testing. It has the advantages of precise mathematical modeling, insensi-
tivity to lift-off, simultaneous qualitative and quantitative assessment, and non-
retrospective detection [16–18]. Currently, some researchers have applied ACFM
technology in the field of rail transportation. Nicholson et al. [19] utilized finite
element software to establish a simulation model for rolling fatigue crack ACFM,
investigating the relationship between rolling fatigue cracks and characteristic
signals, and conducting experimental verification. Rowshandel et al. [20] combined
artificial neural networks with ACFM technology to achieve size assessment of
fatigue cracks in steel rails and wheel-rail systems. Shen et al. [21] investigated the
relationship between complex-shaped rolling fatigue cracks and ACFM character-
istic signals, and proposed a crack profile reconstruction method. The University of
Birmingham in the UK has established a high-speed rail measurement platform, using
ACFM detection instruments and pencil probes from TSC company, and employing
B-spline functions to reconstruct the response curve of characteristic signals at high
speed, achieving the detection of rail defects at 121.5 km/h [22]. Ge et al. [23]
added a “slipper” structure to the ACFM probe to avoid collision with steel rails and
switches, analyzed the impact of speed on detection signals, and achieved high-speed
detection of surface cracks on steel rails. From the above, it can be seen that current
research on the detection of rail surface defects mainly focuses on the processing
of characteristic signals. The experimental systems used generally employ common
AC electromagnetic field detection probes and equipment, lacking specially designed
detection systems tailored to the structure and defect distribution characteristics of
rail surfaces. This limitation has constrained the improvement of rail surface defect
detection effectiveness.
The principle of AC electromagnetic field detection for oblique cracks on rail
surfaces is shown in Fig. 1. When a sinusoidal AC signal is applied to the excitation
coil, the coil induces a uniform induced current on the rail surface through a U-shaped
magnetic core. If there is a crack on the rail surface, the uniform induced current
bypasses the crack from both sides and the bottom, causing spatial magnetic field
distortion. The magnetic sensor arranged above the rail surface picks up the distorted
magnetic field signal, enabling the detection and quantification of cracks.
This paper focuses on the structural characteristics of rail surfaces and the detec-
tion requirements for oblique cracks, applying AC electromagnetic field detection
technology to detect oblique cracks on rail surfaces. By establishing a finite element
simulation model of oblique cracks on rail surfaces using ACFM, the disturbance
mechanism of oblique cracks on induced electromagnetic fields and the influence of
scanning paths on characteristic signals were analyzed. Based on the characteristics
of rail surfaces, a rail surface detection probe and inspection scanning frame were
designed, and the development of detection instruments and software was completed.
Finally, a complete detection system was formed, which achieved effective detection
of oblique cracks on rail surfaces.
2 Finite Element Method Model 49

Fig. 1 ACFM rail surface

oblique crack detection
principle diagram

2 Finite Element Method Model

2.1 Simulation Model

Based on the characteristics of oblique cracks on rail surfaces, this paper estab-
lishes an ACFM simulation model for oblique cracks on rail surfaces using the finite
element simulation software COMSOL. Considering the trade-off between solution
time and accuracy, the simulation model is simplified as shown in Fig. 2. The model
mainly includes a simulated rail specimen, cracks, and air. The simulation analysis
is conducted by applying an excitation current density of 100 A/m with a frequency
of 1 kHz perpendicular to the crack length direction. The simulated rail specimen
is made of carbon steel. The dimensional parameters and material properties of the
simulation model are summarized in Tables 1 and 2, respectively.
Since cracks on rail surfaces usually originate from the gauge corner, the horizontal
angle of the crack is typically 45° with respect to the direction of travel, and it forms
a certain angle with the vertical direction and the rail surface. Therefore, in the
simulation model, the horizontal angle of the crack is set to 45°, the vertical angle is
set to 30°, the length of the crack is 12 mm, and the depth is 5 mm, as shown in Fig. 3.
The established finite element model is used to simulate and analyze the crack, and
the distribution of induced currents around the crack is extracted as shown in Fig. 4.
From the figure, it can be observed that in the absence of a crack, the surface induced
current is uniformly distributed and perpendicular to the crack direction. The cross-
sectional induced current follows the skin effect distribution pattern, with higher
current density on the crack surface and gradually decreasing in the depth direction.
When a crack exists, the induced current is disturbed, and the surface current bypasses
the two endpoints of the crack, with one endpoint in a clockwise direction and the
other endpoint in a counterclockwise direction. The cross-sectional induced current
bypasses the bottom of the crack. In order to simulate the scanning process of the
detection probe and analyze the response characteristics of the characteristic signals
at the center position of the crack, the magnetic flux density values at a position
50 Design and Experiment Research of Oblique Crack Detection System …

Fig. 2 Finite element

simulation model of ACFM
for rail surface oblique
cracks. a Overall view of the
simulation model. b Local
view of the oblique crack
region

Table 1 Model dimensions

Model Length/mm Width/mm Depth/mm
Air layer 1100 1100 100
Simulated steel rail specimens 500 500 10
Excitation current loading region 750 750 10
Crack evaluation region 80 50 7
Data extraction region 80 50 3

Table 2 Model material parameter properties

Model Model material Electrical Relative Relative dielectric
conductivity/S permeability constant
m−1
Air layer Air 50 1 1
Excitation current Air 50 1 1
loading region
Simulated steel rail Steel of the rail 5 × 106 150 1
specimens
Inclined crack Ai 50 1 1
2 Finite Element Method Model 51

Fig. 3 Graphical
representation of oblique
cracks on the rail surface

0.5 mm above the crack are extracted along the scanning path, as shown in Fig. 5.
In the figure, Bx represents the magnetic flux density component along the scanning
path, and Bz represents the magnetic flux density component normal to the rail
surface. From Fig. 5, it can be seen that the magnetic flux density value above the
defect-free rail surface remains constant, while at the crack location, the Bx signal
shows a trough, and the Bz signal exhibits peaks and troughs sequentially.

Fig. 4 Induced current. a Surface current. b Internal current

52 Design and Experiment Research of Oblique Crack Detection System …

Fig. 5 Feature signal

2.2 The Influence of the Scanning Path on the Characteristic

Signals

During the process of rail surface inspection, the position of the crack is unknown,
so it is necessary to explore the influence of the scanning path on the characteristic
signals. The magnetic field signals from different scanning paths are extracted as
shown in Fig. 6, with the legend distance representing the perpendicular distance
from the center of the skewed crack to the scanning path; when the distance is 0 mm,
the scanning path passes through the center of the crack. In order to more intuitively
express the spatial magnetic field variation above the crack, the Bx magnetic field
signal is uniformly baseline processed, and the uniformly baselined Bx signal is
defined as the NBx signal, as shown in Eq. 1:

NBx = Bx/Bx0 × 100% (1)

In the equation, Bx0 represents the baseline magnetic field signal value at the
location without a crack.
From Fig. 6a, it can be observed that when the offset distance is 0 mm, the NBx
signal is a trough signal; as the vertical distance gradually increases to 8 mm, the
trough decreases gradually, and a new peak appears on the right side of the trough,
reaching its maximum value at 7 mm. In Fig. 6b, as the scanning path deviates from
the crack center, the left trough of the Bz signal gradually decreases, while the right
peak continuously increases, reaching its maximum value when deviating by 6 mm.
This is due to the induced current bypassing the crack from both ends, resulting in
the accumulation and increase in current density at the ends of the crack, thereby
increasing the spatial disturbance of the magnetic field signal.
2 Finite Element Method Model 53

Fig. 6 The influence of scanning paths on the magnetic field signals a Bx magnetic field signal.
b Bz magnetic field signal ion

In order to provide a more precise description of the magnetic field characteristic

signals of the crack, the distortion variables ΔBx for the Bx signal and ΔBz for the
Bz signal are defined as shown in Eqs. 2 and 3, respectively.

ΔBx = NBxmax − NBxmin (2)

ΔBz = Bzmax − Bzmin (3)

54 Design and Experiment Research of Oblique Crack Detection System …

Fig. 7 Influence of scanning path on feature signal distortion

In the equations, NBxmax and NBxmin represent the maximum and minimum
values of the NBx signal, while Bzmax and Bzmin represent the maximum and
minimum values of the Bz signal, typically corresponding to the peak and trough
values of the Bz signal.
Figure 7 shows the variation curves of the distortion variables of the Bx and Bz
signals as the scanning path deviates from the crack center. From the graph, it can
be observed that the distortion variable of the Bx signal remains relatively small
within the range of 5 mm, continuously increases between 5 and 7 mm, reaching
its maximum value at 7 mm. This is because the magnetic leakage field rapidly
increases at the endpoint positions of the crack. Beyond 7 mm, the Bx distortion
variable starts to decrease. On the other hand, the distortion variable of the Bz signal
continuously increases from 0 to 6 mm and then decreases from 6 to 8 mm. For
different scanning paths, the distortion variables of the characteristic signals vary.
The distortion variable is smallest at the crack center position and largest near the
crack endpoints. Therefore, the extremum points of the characteristic signals can be
utilized to measure the length of the crack.

3 Testing System

3.1 Probe Design

To achieve rapid detection of the entire rail tread surface, a probe for detecting oblique
cracks in rail tread surface was designed. First, the magnetic core was simulated and
analyzed as shown in Fig. 8. A custom-shaped magnetic core was designed based
on the special structure of the rail gauge angle. The two legs of the magnetic core
have different lengths, and the bottom is designed with angled surfaces of different
3 Testing System 55

Fig. 8 Current distribution

on the rail surface

angles to match the rail surface, which minimizes the air gap. While ensuring that the
induced magnetic field covers the entire rail tread surface area, the design achieves a
higher induction current intensity and a more uniform magnetic flux density on the
rail tread surface.
Taking into account the detection accuracy of the array probe and the sensitivity of
the magnetic field sensor, TMR (Tunneling Magnetoresistance) sensors were chosen
as the array-type magnetic field sensors. The spacing between the sensors was set
to 5 mm. Oblique cracks on the rail tread surface develop from the rail gauge angle
towards the rail tread surface and typically have a length of around 6–30 mm, forming
a 45° angle with the direction of travel. Therefore, setting up 5 array TMR sensors is
sufficient to cover the oblique cracks. The distance between adjacent sensors is set
to 5 mm. The distribution schematic of sensors 1–5 at the rail gauge angle is shown
in Fig. 9.
According to the distribution characteristics of cracks on the rail tread surface,
a rail-shaped array detection probe was designed as shown in Fig. 10. The probe
utilizes a specially shaped magnetic core that matches the rail gauge angle and is
inclined at a 45-degree angle. The induced current direction is perpendicular to the
oblique cracks. The outer shell of the array probe is designed to closely fit with the
rail, and the array sensors are located at the bottom of the probe, maintaining a small
separation distance from the rail surface.
56 Design and Experiment Research of Oblique Crack Detection System …

Fig. 9 Distribution of array

sensors on the rail surface

Fig. 10 Rail inspection

array probes. a Design
diagram of the conformal
array probe. b Photo of the
conformal array probe
4 Experimental Research 57

3.2 Detection System

To meet the requirements of stability, portability, and efficiency in rail inspection,

an ACFM (Alternating Current Field Measurement) detection system for rail tread
surface oblique cracks is designed as shown in Fig. 11. The overall solution consists of
a hardware system and a software system. The hardware system includes a detection
probe, a rail inspection scanning frame, and a rail crack ACFM detection instrument.
The rail inspection scanning frame is made of aluminum profiles and is equipped
with wheels at the bottom to ensure stability during the detection process. The probe
fixture on the scanning frame enables the fixation and compression of the detec-
tion probe, reducing probe vibration and improving detection stability. The probe
fixture can accommodate different types of detection probes. The rail oblique crack
ACFM detection instrument is fixed on the bracket of the scanning frame. It mainly
consists of a power module, a signal generator, connectors, effective value processing
circuitry, data acquisition card, and an industrial-grade PC. The highly integrated
components enable miniaturization and integration of the detection instrument. The
instrument is connected to the detection probe and encoder through connectors and
signal cables. It outputs excitation signals from the signal generator while receiving
detection signals from the probe and position signals from the encoder. The soft-
ware system is developed using LabVIEW and MATLAB, enabling data display and
analysis of the rail tread surface oblique crack detection signals.

4 Experimental Research

4.1 Different Depth Oblique Crack Detection Experiments

In order to analyze the influence of crack depth on the characteristic signals, detection
experiments were conducted on rail specimens with different crack depths. The
excitation frequency of the detection instrument was set at 1 kHz, and the material
of the specimens was 60 kg/m rail. The crack depths were 1, 3, 5, and 7 mm, with
a length of 18 mm and a width of 0.3 mm. The cracks were inclined at a horizontal
angle of 45° and a vertical expansion angle of 50°, as shown in Fig. 12.
The detection probe was used to scan different depth cracks, and C-scan images
were generated based on the position relationship and the characteristic signal Bz
of different sensors, as shown in Fig. 13. The position at a lateral distance of 0 mm
represents the center position of sensor 3, and the centers of adjacent sensors have a
lateral distance difference of 5 mm. From the image, it can be observed that the cracks
cause distortion in the induced magnetic field signal. The red region represents the
peak of the Bz signal, and the blue region represents the trough of the Bz signal. The
central regions of the red and blue areas represent the two endpoints of the crack.
As the crack depth increases, the distortion of the Bz signal also increases. Among
them, the 1 mm deep crack has the smallest distortion amplitude, measuring 36
58 Design and Experiment Research of Oblique Crack Detection System …

Fig. 11 ACFM detection

system for oblique cracks on
the rail surface. a Schematic
diagram of the detection
system. b Photo of detection
system

Fig. 12 Schematic diagram of steel rail specimens with cracks at various depths. a Front view of
the specimen. b Top view of the specimen. c Left view of the specimen
4 Experimental Research 59

mV, while the 7 mm deep crack has the largest distortion amplitude, measuring 209
mV. For different sensor positions, sensor 2 and sensor 5 have the largest distortion
amplitudes, indicating that the crack endpoints are close to the scanning path of
these two sensors. Additionally, there is a difference in the peak and trough values
of the signals from these two channels in the direction of travel, indicating a certain
horizontal angle between the crack and the scanning direction. Based on the C-scan
image, the preliminary determination of the horizontal angle of the crack is around
45°, with a length of approximately 20 mm, which is close to the actual value of 18
mm. Therefore, it can be concluded that the detection instrument can detect cracks
with a minimum depth of less than 1mm and provide preliminary quantification of the
crack’s horizontal angle and length. This meets the requirements for array imaging
and large-scale detection of oblique cracks at rail gauge corners.
In order to analyze the influence of cracks of different depths on the character-
istic signals, the characteristic signals along the scanning path at the center of the
cracks were extracted, as shown in Fig. 14. Simultaneously, the curves of the distor-
tion amplitudes of the Bx and Bz signals for different crack depths were plotted,
as shown in Fig. 15. From the figures, it can be observed that as the crack depth
increases, the depth of the Bx signal troughs also increases, but the range of distortion
remains relatively constant. The peaks and troughs of the Bz signal show an increasing
trend, while the horizontal spacing between them remains relatively constant. This is
because as the crack depth increases, the distortion amplitude of the induced electric
field increases, resulting in an increase in the distortion amplitude of the induced
magnetic field.

4.2 Detection Experiment of Inclined Cracks of Different

Lengths

In order to analyze the influence of crack length on the characteristic signals, detection
experiments were conducted on rail specimens with cracks of different lengths. The
crack lengths were 6, 12, 18, 24, and 30 mm, with a depth of 5 mm and a width of 0.3
mm. The cracks were inclined at a horizontal angle of 45° and a vertical expansion
angle of 30°, as shown in Fig. 16.
Using array probes to detect cracks of different lengths separately, the C-scan
images of different crack lengths shown in Fig. 17 were obtained. From the images,
it can be observed that as the crack length increases, the distance between the peak
and trough of the Bz signal becomes larger, approaching different sensors. At the
same time, as the crack length increases, the overall distortion of the signal also
increases. This is because longer cracks induce a higher current density around their
ends, resulting in a stronger magnetic field signal in the Z direction. The length of
the crack can be obtained by measuring the distance between the peak and trough of
the Bz signal, and through five measurements, the average measurement error of the
array probe for crack length is less than 2 mm.
60 Design and Experiment Research of Oblique Crack Detection System …

Fig. 13 Different depth

crack array probe results.
a Crack depth of 1 mm.
b Crack depth of 3 mm.
c Crack depth of 5 mm.
d Crack depth of 7 mm
4 Experimental Research 61

Fig. 14 Different depth crack single probe results. a BX signal. b BZ signal

Fig. 15 The impact of crack depth on characteristic signalsl

62 Design and Experiment Research of Oblique Crack Detection System …

Fig. 16 Schematic diagram of steel rail specimens with cracks of different lengths. a Front view
of the specimen. b Top view of the specimen. c Left view of the specimen

In order to analyze the influence of crack length on the characteristic signals,

the characteristic signals along the scanning path at the center of the cracks were
extracted, as shown in Fig. 18. Simultaneously, the distortion amplitudes of the Bx
and Bz signals were plotted, as shown in Fig. 19. From the figures, it can be observed
that as the crack length increases, the distortion amplitude of the Bx signal initially
increases and then decreases, with an increasing range of distortion. Similarly, the
distortion amplitude of the Bz signal also increases and then decreases, with an
increasing horizontal spacing between the peaks and troughs. This is because as the
crack length increases, the range and density of the induced electric field distor-
tion increase, leading to an increase in the distortion amplitude of the characteristic
signals. However, as the crack length continues to increase, the distance between the
crack center and the endpoints increases, weakening the influence of the distortion
current at the endpoints on the central region, resulting in a decrease in the distortion
amplitude of the characteristic signals.

5 Conclusions

This paper focuses on the characteristics of rail surface inspection and designs an
ACFM (Alternating Current Field Measurement) system for detecting oblique cracks
on the rail surface. Through simulation models, the response mechanism between
the distorted electromagnetic field and oblique cracks is investigated. The influence
of scanning paths on the characteristic signals is analyzed. Additionally, a detection
probe, scanning frame, and detection instrument for detecting oblique cracks on the
rail surface are designed. Ultimately, a complete AC electromagnetic field-based
system for detecting oblique cracks on the rail surface is established. The system is
used to detect oblique cracks with different lengths and depths on the rail surface
5 Conclusions 63

Fig. 17 Results of crack

array probes of different
lengths. a Crack length of 6
mm. b Crack length of 12
mm. c Crack length of 18
mm. d Crack length of 24
mm. e Crack length of 30
mm
64 Design and Experiment Research of Oblique Crack Detection System …

Fig. 18 Results of single probe for cracks of different lengths. a BX signal. b BX signal

Fig. 19 The influence of crack length on characteristic signals

References 65

(1) The AC electromagnetic field detection method can be used to detect oblique
cracks on the rail surface. When an oblique crack occurs, the induced current
flows around the two endpoints of the crack, one in a clockwise direction and the
other in a counterclockwise direction. The cross-sectional current flows around
the bottom of the crack. As a result, the Bx signal at the center of the crack
exhibits a trough, while the Bz signal exhibits peaks and troughs.
(2) The distortion amplitudes of the Bx and Bz signals vary depending on the
scanning path. Since the induced current tends to concentrate at the endpoints
of the crack, the distortion amplitudes of the Bx and Bz signals are minimal at
the center of the crack. On the other hand, the distortion amplitudes of the Bx
and Bz signals are maximal near the endpoints of the crack.
(3) The detection system can effectively detect oblique cracks with a depth of 1
mm on the rail surface and measure the length of the crack. For C-scan images,
the distortion amplitude of a crack with a depth of 1 mm is 36 mV, while that
of a crack with a depth of 7 mm is 209 mV. The measurement error of the
crack length is less than 2 mm. The deeper the crack, the larger the distortion
amplitude of the Bx and Bz signals at the center of the crack. Additionally, the
distortion amplitudes of the Bx and Bz signals at the center of the crack increase
and then decrease as the length of the crack increases.

References

1. Jia X (2020) Research on contact fatigue damage of rail based on wheel-rail transient rolling
contact model. China Academy of Railway Sciences, Beijing
2. Xing L (2008) Research on defect characteristics and classification of higher speed rails. China
Academy of Railway Sciences, Beijing
3. Zerbst U, Lundén R, Edel KO et al (2009) Introduction to the damage tolerance behaviour of
railway rails: a review. Eng Fract Mech 76(17):2563–2601
4. Zhou Q, Zhang J, Guo Z et al (2010) Research on the rail damages and the preventive
countermeasures in heavy haul railways. China Railway Sci 31(1):27–31
5. Shen J, Zhou L, Rowshandel H et al (2020) Prediction of RCF clustered cracks dimensions
using an ACFM sensor and influence of crack length and vertical angle. Nondestruct Test Eval
35(1):1–18
6. Tian G, Gao B, Gao Y et al (2016) Review of railway rail defect non-destructive testing and
monitoring. Chin J Sci Instr 37(8):1763–1780
7. Li G, Huang P, Chen P et al (2011) Application of eddy current testing in the quantitative
evaluation of the rail cracks. J Zhejiang Univ 45(11):2038–2042, 2049
8. Lao J, Lu C (2013) Application of portable ultrasonic phased array instrument for rail welds
ultrasonic inspection. Adv Mater Res 717(1):384–389
9. Gao Y, Wang P, Tian G et al (2013) Velocity effect analysis of dynamic magnetization in high
speed detection for rail crack using MFL method. Nondestruct Test 35(10):53–58
10. Antipov A, Markov A (2014) Evaluation of transverse cracks detection depth in MFL rail NDT.
Russ J Nondestr Test 50(8):481–490
11. Yu H, Li Q, Tan Y et al (2019) A coarse-to-fine model for rail surface defect detection. IEEE
Trans Instrum Meas 68(3):656–666
12. Li T, Shi Y, Chen F et al (2021) Quantitative method of rail flaws based on ultrasonic phased
array and total focusing DAC mappings. J Mech Eng 57(18):32–41
66 Design and Experiment Research of Oblique Crack Detection System …

13. Li J (2017) Research on rail flaw detection based on time-frequency and space feature. South
China University of Technology, Guangzhou
14. Ma W, Liu D, Zhao W (2010) A portable eddy current flaw detection for railway testing. Mach
Des Manuf 2:88–90
15. Li N (2021) Research on rail surface defect detection system based on machine vision. Beijing
Jiaotong University, Beijing
16. Zhao J, Li W, Ge J et al (2020) Coiled tubing wall thickness evaluation system using pulsed
alternating current field measurement technique. IEEE Sens J 20(18):10495–10501
17. Shen J, Zhou L, Rowshandel H et al (2015) Determining the propagation angle for non-vertical
surface-breaking cracks and its effect on crack sizing using an ACFM sensor. Measur Sci
Technol 26(11):1–11
18. Yuan X, Li W, Yin X et al (2020) Visual reconstruction of irregular crack in austenitic stainless
steel based on ACFM technique. J Mech Eng 56(10):27–33
19. Nicholson G, Davis C (2012) Modelling of the response of an ACFM sensor to rail and rail
wheel RCF cracks. NDT&E Int 46(1):107–114
20. Rowshandel H, Nicholson GL, Shen JL et al (2018) Characterisation of clustered cracks using
an ACFM sensor and application of an artificial neural network. NDT&E Int 98:80–88
21. Shen J, Liu M, Dong C et al (2022) Analysis on asymmetrical RCF cracks characterisation
using an ACFM sensor and the influence of the rail head profile. Measurement 194:111008
22. Chacon Muñoz JM, García Márquez FP, Papaelias M (2013) Railroad inspection based on
ACFM employing a non-uniform B-spline approach. Mech Syst Signal Process 40(2):605–617
23. Ge J, Yan C, Hu B et al (2021) High-speed detection of surface crack on rail using alternating
current field measurement technique. J Mech Eng 57(18):66–74

Open Access This chapter is licensed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits use, sharing,
adaptation, distribution and reproduction in any medium or format, as long as you give appropriate
credit to the original author(s) and the source, provide a link to the Creative Commons license and
indicate if changes were made.
The images or other third party material in this chapter are included in the chapter’s Creative
Commons license, unless indicated otherwise in a credit line to the material. If material is not
included in the chapter’s Creative Commons license and your intended use is not permitted by
statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder.
Design and Testing of High-Resolution
Probe Arrays Using Alternating Current
Field Measurement Technique

Alternating Current Field Measurement (ACFM) technology is a new type of elec-

tromagnetic non-destructive testing technology, which can detect surface and near-
surface defects of metal parts [1, 2]. This technique is based on the principle of electro-
magnetic induction. When sinusoidal excitation current is loaded into the excitation
coil, a local uniform current field will be induced on the specimen surface. When the
uniform current encounters a defect, the current will generate a disturbance, resulting
in a disturbance of the magnetic field around the defect. The detection and evalua-
tion of the defect can be realized by picking up the spatially-distorted magnetic field
signal [3–5]. Due to the characteristics of non-contact measurement, small removal
effect and high quantitative accuracy, ACFM technology has been widely used in
petrochemical, railway transportation, nuclear power and other fields [6–8].
From the current research, the visualization methods of the existing ACFM tech-
nology and devices are sometimes based on magnetic field signal graphics such as
scan map, butterfly map and isoline color map [9]. Compared with the previous two,
isoline color map of magnetic field signal can not only determine the existence of
defects, but also invert the appearance of defects. At present, there are two methods
to obtain the isoline color map of magnetic field signals. One is to use a single probe
to conduct two-dimensional scanning of the detected object. For example, Yuan [10]
uses a three-axis bench to drive a single probe to conduct raster scanning of the irreg-
ular cracks on the surface of austenitic stainless steel to obtain the vertical magnetic
field Bz image, and uses the gradient field visualization reconstruction method of Bz
image to reconstruct the irregular cracks. However, the efficiency of this detection
method is low, if you want to get a higher resolution magnetic field image, you need
to conduct multiple scans. The other is to scan through array detection probe. Hu
et al. [11] used LKACFM detector and 8 array detection probe to detect aluminum
inner liner of high-pressure hydrogen cylinder, and obtained Bz images of round
holes and cracks at different angles. Wu [9] designed 7 array detection probes to
obtain three-dimensional magnetic field signals around rectangular slots and cylin-
drical corrosion pits, and used gradient fusion algorithm to invert surface profiles. Li

© The Author(s) 2025 67

W. Li et al., Alternating Current Field Measurement Technique for Detection and
Measurement of Cracks in Structures, https://doi.org/10.1007/978-981-97-7255-1_5
68 Design and Testing of High-Resolution Probe Arrays Using Alternating …

[12] designed an orthogonal U-shaped array probe to detect aluminum plate speci-
mens with surface corrosion defects, and proposed a defect surface shape inversion
algorithm. Topp et al. [13] used the 8-array detection probe and detection system of
TSC company to detect the railway, obtained the Bz scan image of the crack, and
determined the Angle and length of the crack. At present, scholars and companies
at home and abroad have designed relatively mature AC electromagnetic field array
probes, but the sensor spacing of the existing probes is mostly 5 mm or above [14–
17], resulting in low spatial resolution of the measured magnetic field image, which
can only be used to determine the existence of defects and cannot accurately obtain
the morphology of defects.
To solve this problem, an ACFM probe based on high resolution TMR sensor
array is proposed in this paper, and the testing system is designed and built. The
finite element software Comsol is used to establish an ACFM simulation model,
analyze the electromagnetic field distortion rule around different types of defects,
explore the relationship between the defect surface profile and magnetic field images,
design a high-resolution ACFM array detection probe and test system, and carry out
defect detection tests.

1 Finite Element Simulation

1.1 Model Building

The ACFM simulation model established by finite element software Comsol is

shown in Fig. 1. The main components include: air region, excitation coil, U-shaped
magnetic core and test piece to be tested. The entire model is surrounded by an air
domain. The excitation coil is wound on the U-shaped magnetic core, the number
of turns is 500, and the excitation signal uses a sinusoidal signal with a frequency
of 1000 Hz and a current size of 150 mA. The sample is aluminum plate, located
directly below the magnetic core, with a relative permeability of 1 and an electrical
conductivity of 3.774 × 107 S/m. The extracted induced current on the specimen
surface is shown in Fig. 2. The excitation coil generates a local uniform current field
on the specimen surface, which meets the requirements of ACFM.

1.2 Defect Disturbance Law

In order to explore the disturbance law of induced electromagnetic field on the spec-
imen surface which is caused by different types of defects, three defects are set on
the specimen surface aiming at common cracks and area defects.
The cracks are 20 mm × 0.5 mm × 4 mm (length × width × depth), the spherical
corrosion pit Φ20 mm × 4 mm (diameter × depth) and the square groove 20 mm ×
1 Finite Element Simulation 69

Fig. 1 Finite element

simulation model of ACFM

Fig. 2 Surface current

density

4 mm (side length × depth).The induced currents near the three defects are extracted
as shown in Figs. 3, 4 and 5. It can be seen that the uniform current field is disturbed
due to the existence of the defects. As shown in Figs. 3a and 5a, surface current
accumulates at the end points of the crack and the square groove, resulting in a
maximum current density at the end points. Similarly, although the surface profile
of the spherical corrosion pit is circular with no obvious tip, the surface current still
bypasses both sides of the arc, as shown in Fig. 4a, so the current density also presents
a maximum value on both sides of the arc. At the same time, due to the discontinuity
of the material, the current flows from the edge to the depth of the defect, as shown
in Figs. 3b, 4b and 5b, so the surface current density troughs in the defect region.
The distorted current causes the spatial magnetic field disturbance above the
defects. The simulation software adopts parametric scanning mode to simulate the
detection process of the actual probe, and extract the magnetic fields signal Bx and
Bz above 1 mm on the three defects. The scanning results are shown in Figs. 6, 7 and
8. Bx signal shows obvious troughs and Bz signal shows peaks and troughs, which
conform to the principle of ACFM. Moreover, it can be seen from Figs. 7a and 8a
that the shape of the trough in the Bx magnetic field image is roughly round and
square. As shown in Figs. 6b and 8b, the peak and valley distortion position in the
Bz image is the same as the current accumulation position at the defect endpoint.
70 Design and Testing of High-Resolution Probe Arrays Using Alternating …

Fig. 3 Disturbance law of

current around crack

The simulation results show that the induced current accumulates at the end of the
defect, and the accumulation current causes the distortion of the magnetic field Bz in
the Z direction. So the magnetic field image of Bz contains the position information
of the defective endpoint. At the same time, the current deflects from the edge of the
defect to the bottom, causing a disturbance in the X direction magnetic field Bx, and
the Bx magnetic field image can reflect the edge profile information of the defect.

2 Probe Design

2.1 Sensor Array Design

According to the simulation results, the amplitudes of the disturbed magnetic field
Bx and Bz above the defect are in the order of 10–3 T and 10–4 T respectively, and
the magnetic fields are relatively weak. Currently, commonly used magnetic field
sensors include coil, Hall, AMR, GMR and TMR sensors. Compared with other
types of magnetic field sensors, TMR sensors are more suitable for weak magnetic
field detection because of their large linear range, high precision and high sensitivity
2 Probe Design 71

Fig. 4 Disturbance law of

current around corrosion pit

[18, 19]. In addition, the small package size of TMR sensors can be made very
compact and dense, thereby improving the resolution in a small range [20], so TMR
sensors are very suitable for array applications with high spatial resolution, weak
magnetic field signal and high sensitivity.
Therefore, the TMR sensor is selected as the detection sensor in this paper. In
order to ensure that the measured magnetic field image has a high spatial resolution
while taking into account the manufacturing and processing cost, the sensor spacing
is set to 1 mm. In order to ensure that the designed array probe has a certain detection
range, the sensor array is composed of 64 TMR sensors, and the linear arrangement is
adopted, so the effective detection range of the sensor array is 64 mm. The designed
high-resolution sensor array is shown in Fig. 9. The black area at the bottom is the
location of the sensor array, and the red arrow represents the direction of the sensor’s
sensitive axis. Each sensor and three resistors form a Wheatstone bridge to measure
the magnetic field by measuring the change of resistance value. At the same time,
the entire circuit board is supplied with power, and three through-holes are arranged
in the middle of the circuit board to facilitate the sensor to be fixed when making the
probe.
72 Design and Testing of High-Resolution Probe Arrays Using Alternating …

Fig. 5 Disturbance law of

current around square groove

Fig. 6 Magnetic field image above the crack

2 Probe Design 73

Fig. 7 Magnetic field image above the corrosion pit

Fig. 8 Magnetic field image above the square groove

Fig. 9 TMR sensor array

74 Design and Testing of High-Resolution Probe Arrays Using Alternating …

2.2 Excitation Module Design

The excitation module includes an excitation coil and a U-shaped magnetic core.
The excitation coil is wound by a enamel-coated wire with a diameter of 0.15 mm
on the beam of the magnetic core, with a number of turns of 500. The material
of the magnetic core is manganese zinc ferrite, which can effectively enhance the
excitation effect. The width of the uniform current field on the specimen surface
is directly proportional to the width of the U-shaped magnetic core. In order to
ensure that the uniform current field stimulated by the excitation module can cover
the effective detection range of the sensor array, this paper adopts multiple magnetic
cores placed side by side to increase the width of the uniform current field. The width
of the selected single magnetic core is 16.1 mm, which is simulated by the simulation
software Comsol. Finally, 5 magnetic cores are selected to form an exciting magnetic
core array, and the induced current under the magnetic core array is extracted, as
shown in Fig. 10. The red area in Fig. 10a is the uniform current field, and Fig. 10b
is the induced current density at the location of the sensor array (directly below the
magnetic core array, X = 0 mm). 90% or more of the peak current is defined as the
uniform current, and the width of the uniform current area is obtained to be 66.3 mm,
which can cover the detection range of the sensor array.

2.3 Multiplex Module Design

To process and collect the signals picked up by the sensor array, a multi-channel signal
processing circuit and an AD acquisition channel are needed. However, too many
processing circuits will cause interference between signals, and too many acquisi-
tion channels will also cause the sampling rate to slow down, which will affect the
detection effect. Therefore, multiplex method is used to pre-process the multi-channel
signals output by sensor array. According to the time division transmission mode and
frequency division transmission mode, the multiplexing technology is divided into
time division multiplexing and frequency division multiplexing. In this paper, time
division multiplexing technology is adopted [21], and ADG1607 multiplexing chip
is selected to achieve multiplexing. As shown in Fig. 11, the chip has a single-chip
iCMOS® analog multiplexer with 8 differential channels. One of the eight differen-
tial inputs can be switched to a common differential output at an address determined
by the 3-bit binary address lines (A0, A1, and A2), and the chip conversion time
is 143 ns. The delay time during switching is negligible, and there is no significant
difference between the output and input signals, so there is almost no impact on the
final detection result.
A single ADG1607 chip can switch 16 differential inputs to 2 common differential
outputs, as shown in Fig. 11b. In order to realize the multiplexing of the output signals
of the sensor array, four multiplexing chips are set in the middle of the circuit board
to switch the 64 sensor output signals to 8 outputs, as shown in Fig. 12a. The design
2 Probe Design 75

Fig. 10 Simulation analysis

of five U-shaped cores

Fig. 11 ADG1607 multiplexer chip

76 Design and Testing of High-Resolution Probe Arrays Using Alternating …

Fig. 12 Signal multiplexing

module

of the multiplex module is shown in Fig. 12b. The upper two 40-pin connectors are
connected to the sensor array circuit board through the FPC SoftBank cable, and
the right side is the power supply interface and the analog switch control port. At
the same time, the EN is always set to high level, and the switch is switched by the
binary digital address line control chip output by the digital output channel of the
acquisition card, and the 64 signals are finally converted to 8 signals. This process
simplifies the hardware processing circuit.

2.4 Design of Probe Structure

The array detection probe is mainly composed of high-resolution TMR sensor array,
U-core array, excitation coil, multiplexing module, top cover, housing and Remo
socket. The structure of the detection probe is shown in Fig. 13. A groove is arranged
at the bottom of the interior of the housing for fixing the sensor circuit board, which
can be used to detect Bx signals. The sensor circuit board is placed vertically through
the fixing block, which can be used to detect Bz signals. The magnetic core array is
located directly above the sensor array, and a card slot is arranged inside the housing
to ensure that the magnetic core array will not move during detection. The top cover
is provided with a hole according to the size of the Remo socket to fix the Remo
socket, to achieve stable internal signal transmission and power supply for the probe.
The top cover is fixed to the housing by countersunk screws.
3 Experimental Study 77

Fig. 13 Detection probe

array

3 Experimental Study

3.1 Experimental System

The whole test system, as shown in Fig. 14, mainly consists of array detection probe,
detection instrument (hardware chassis and industrial PC integration) and defect
detection software. The power module is powered by the PC battery and supplies
power to the entire detection system through the voltage regulator and voltage conver-
sion modules. The excitation module of the detector loads the sinusoidal excitation
signal with frequency of 1 kHz and amplitude of 10 V onto the excitation coil
of the array probe, and the excitation coil induces a uniform electric field on the
surface of the specimen. When the probe passes through the defect, the uniform
electric field on the surface of the specimen is disturbed due to the difference in the
conductivity between the air and the specimen, resulting in the disturbance of the
magnetic field above the specimen. The sensor array in the detection probe picks up
the distorted spatial magnetic field signal, preprocesses the signal through the multi-
plexing module, and then passes the signal conditioning (amplification and filtering),
and finally transmits the signal to the signal acquisition module for AD conversion
and transmission to the PC. The software on the PC is developed based on Labview
to collect, process, store and display the magnetic field signal around the defect.

3.2 Experimental Test

The developed test system is used to detect the artificial defects of the prefabrication.
The test specimen was aluminum plate, and the defect type was consistent with the
simulation, as shown in Fig. 15. The crack size is 12 mm × 0.5 mm × 3 mm (length
× width × depth), and the Angle between the crack and the probe scanning direction
is 0°, 30° and 60°, respectively. The depth of the three pits is 3 mm, and the diameters
78 Design and Testing of High-Resolution Probe Arrays Using Alternating …

Fig. 14 Test system

are 10 mm, 14 mm and 18 mm, respectively. The depth of the three square slots is 3
mm and the side lengths are 6 mm, 10 mm and 15 mm respectively. The probe was
placed on the surface of the specimen, and the three types of defects were scanned
successively with a scanning step length of 0.5 mm. After the scan is completed,
the magnetic field images of Bx and Bz are obtained, as shown in Figs. 16, 17 and
18. The X-direction magnetic field Bx of the three defects showed troughs, and the
Z-direction magnetic field Bz showed peaks and troughs, which was consistent with
the simulation results. It can be seen that the ACFM array probe and system designed
in this paper can obtain the distorted magnetic field signal around the defect.

3.3 Defect Inversion Imaging

According to the simulation results, it can be seen that the Bz image can reflect the
endpoint position of the defect, and the Bx image has the edge information of the
defect, so the surface profile of the defect can be retrieved through the images of the
magnetic field signals Bx and Bz. The crack width is very narrow and there is no
obvious surface profile, so the end point of the crack can be determined using the Bz
3 Experimental Study 79

Fig. 15 Schematic diagram

of atificial defects

Fig. 16 Magnetic field

image of cracks

Fig. 17 Magnetic field

image of corrosion pits
80 Design and Testing of High-Resolution Probe Arrays Using Alternating …

Fig. 18 Magnetic field

image of square grooves

image, and the length and angle of the crack can be determined by connecting the
end point, as shown in Fig. 16b. The evaluation results of the three cracks are 13 mm
× 0°, 13.4 mm × 21.8°, 13.3 mm × 55.7°, respectively. The maximum evaluation
error of the length was 1.4 mm, and the maximum evaluation error of the Angle is
8.2°.
For area-type defects of the corrosion pit type, the surface profile is relatively
obvious but there is no obvious tip. The contour of the defect can be retrieved
according to the Bx image, as shown in Fig. 17a. It can be seen that the Bx image is
consistent with the surface profile of the corrosion pit, and the distance between the
peak and valley in the Bz image is the diameter of the surface profile, as shown
in Fig. 17b. The diameter quantization results are 9.5 mm, 12 mm and 15 mm
respectively, and the maximum quantization error is 3 mm.
For square slot area defects with relatively obvious tip and surface profile, surface
topography can be inverted by Bx images. As shown in Fig. 18a, the Bx image of a
square slot basically presents a square. The peak and valley distortion positions in
the Bz image are connected, which can simultaneously invert the surface topography
and quantify the side length of the square slot. As shown in Fig. 18b, the side length
quantization results are 7 mm × 8 mm, 9.5 mm × 9 mm, 12.5 mm × 17 mm,
respectively, and the maximum quantization error is 2.5 mm.

4 Conclusion

(1) The finite element simulation results of AC electromagnetic field show that the
induced current accumulates at the end of the defect, resulting in the distortion
of the magnetic field Bz in the Z direction, and the Bz image can reflect the end
References 81

information of the defect. The induced current bypasses the edge of the defect
to the bottom, causing the disturbance of the Bx signal of the magnetic field
in the X direction, and the Bx image can reflect the contour information of the
defect.
(2) The ACFM array detection probe based on high-resolution TMR sensor array
designed in this paper can obtain horizontal (X) and vertical (Z) magnetic field
images of different types of defects, and the magnetic field images have a spatial
resolution of 1 mm.
(3) The obtained high-resolution magnetic field image can accurately invert the
surface profile of the defect. Bz image can reconstruct the length and Angle
of the crack, and Bx image can invert the surface profile of the area defect.
The maximum evaluation error of crack length was 1.4 mm. The maximum
evaluation error of Angle is 8.2°. The maximum quantization error of corrosion
pit diameter is 3 mm. The maximum quantization error of square slot side length
is 2.5 mm. These results show that the system has high evaluation accuracy.

References

1. Salemi AH, Sadeghi SHM, Moini R (2004) Thin-skin analysis technique for interaction of
arbitrary-shape inducer field with long cracks in ferromagnetic metals. NDT&E Int 39:471–479
2. Wei FAN (2014) Design and test of the AC field detection system. Nanchang Hangkong
University, Nanchang
3. Lewis AM, Michael DH, Lugg MC et al (1998) Thin-skin electromagnetic fields around surface-
breaking cracks in metals. J Appl Phys 64(08):3777–3784
4. Yuan XA, Li W, Chen GM et al (2018) Two-step interpolation algorithm for measurement of
longitudinal cracks on pipe strings using circumferential current field testing system. IEEE
Trans Industr Inf 14:394–402
5. Zhao J, Li W, Yuan X et al (2021) Lift-off suppression algorithm for dual-frequency alternating
current field measurement. Nondestruct Test 43(04):5–9
6. Li W, Yuan X, Chen G et al (2015) Research on in-service detection for axial cracks on drill
pipe using the feed-through alternating current field measurement. J Mech Eng 51(12):8–15
7. Shen JL, Zhou L, Rowshandel H et al (2015) Determining the propagation angle for non-
vertical surface-breaking cracks and its effect on crack sizing using an ACFM sensor. Meas
Sci Technol 26(11):1–11
8. Smith M, Laenen C (2007) Inspection of nuclear storage tanks using remotely deployed
ACFMT. Insight 49(01):17–20
9. Wu Y (2017) Research on three-dimensional visualization of defects based on ACFM. China
University of Petroleum (East China), Qingdao
10. Yuan X, Li W, Yin X et al (2020) Visual reconstruction of irregular crack in austenitic stainless
steel based on ACFM technique. J Mech Eng 56(10):27–33
11. Hu D, Li D, Chen Z et al (2021) Detection technology of ac electromagnetic field in aluminum
liner of high-pressure hydrogen cylinder. Low Temp Spec 39(02):42–45
12. Li W (2007) Research on ACFM based defect intelligent recognition and visualization
technique. China University of Petroleum, Dongying
13. Topp D, Smith M (2005) Application of the ACFM inspection method to rail and rail vehicles.
Insight 47(06):354–357
14. Li W, Yuan X, Chen G et al (2017) Research on the detection of surface cracks on drilling riser
using the chain alternating current field measurement probe array. J Mech Eng 53(08):8–15
82 Design and Testing of High-Resolution Probe Arrays Using Alternating …

15. Zhao JM, Li W, Zhao JCH et al (2020) A novel ACFM probe with flexible sensor array for
pipe cracks inspection. IEEE Access 8:26904–26910
16. Mirshekar-Syahkal D, Mostafavi RF (2002) 1-D probe array for ACFM inspection of large
metal plates. IEEE Trans Instrum Meas 51(02):374–382
17. Topp D (2005) Recent developments and applications of the ACFM inspection method and
ACSM stress measurement method. Non-Destruct Test Austr 42(5):141–147
18. Zhang N, Peng L, Wu Y et al (2020) ECT probe based on magnetic field imaging with a high
resolution TMR sensor array. Chin J Sci Instrum 41(07):45–53
19. Gao J, Wang J, Qiu F (2017) Magnetic sensitive automobile detector based on TMR sensor.
Chin J Sci Instrum 38(08):2039–2046
20. Tao Yu, Lv K, Peng L et al (2020) Design of probe for heat exchange tubes detection of steam
of steam generator based on TMR sensor array. Instrum Techn Sensor 05:37–47
21. Liu H, Xiao B, Wang L (2007) Parameter detection of a certain missile control casing based
on virtual instrument technology. Chin J Sci Instrum 28(04):257–259

Open Access This chapter is licensed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits use, sharing,
adaptation, distribution and reproduction in any medium or format, as long as you give appropriate
credit to the original author(s) and the source, provide a link to the Creative Commons license and
indicate if changes were made.
The images or other third party material in this chapter are included in the chapter’s Creative
Commons license, unless indicated otherwise in a credit line to the material. If material is not
included in the chapter’s Creative Commons license and your intended use is not permitted by
statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder.
Design and Experimental Study of Inner
Uniform Electromagnetic Probe
in Stainless Steel Pipe

Petroleum and petrochemical industry is a pillar industry of the national economy,

which occupies a pivotal position in the national economy. In petroleum and
petrochemical, stainless steel pipelines are widely used in pipeline transportation
systems, including high-pressure furnace pipes, piping, petroleum cracking pipes,
fluid delivery pipes, heat exchange pipes, etc. But due to the corrosive medium in the
pipeline and the alternating cooling and heating environment, fatigue cracks are easy
to occur on the inner wall of the pipeline [1–3], usually, fatigue cracks aggregate and
grow in the longitudinal direction [4], which eventually leads to leakage and failure
of the pipeline, so it is necessary to regularly detect the pipeline.
At present, conventional detection technologies mainly include magnetic particle,
penetration, ultrasound, magnetic flux leakage and eddy current detection technolo-
gies [5]. Magnetic particle and penetration detection technology is currently the most
reliable non-destructive detection technology, but magnetic particle and penetration
detection need to be in direct contact with the tested material, the surface require-
ments are relatively high, and only crack detection can be realized, cracks cannot be
quantified, and detection in the pipeline is difficult to achieve [6]; ultrasonic detection
technology needs to add a coupler between the detection tool and the tested pipeline,
if it is not thoroughly cleaned [7], it is difficult to apply a coupler on the inner wall of
the pipeline; magnetic flux leakage detection technology can only detect ferromag-
netic materials [8]; eddy current detection technology is easily affected by lifting
and jitter.
Uniform field disturbance nondestructive testing methods, such as alternating
current electromagnetic field detection technology, alternating current potential drop
detection technology and circumferential current field detection, have good prospects
for detecting and evaluating conductive materials [9–11]. At present, a series of
studies have been carried out on the form of uniform electromagnetic field detection
probes, Li Wei and others have established a U-shaped probe simulation model and
built an experimental system [12]; Yuan Xin’an and others have used double U-
shaped probes to detect cracks in any direction [13]; Ge Haojiu and others have used

© The Author(s) 2025 83

W. Li et al., Alternating Current Field Measurement Technique for Detection and
Measurement of Cracks in Structures, https://doi.org/10.1007/978-981-97-7255-1_6
84 Design and Experimental Study of Inner Uniform Electromagnetic …

external penetrating probes to achieve the detection of cracks on the outer wall of
the pipeline [14, 15].
However, the U-shaped probe is not suitable for pipe surface detection. Due to
the skin effect of the external penetrating probe, it is difficult to realize the detection
of the inner wall of the pipeline. In order to realize the detection of the inner wall of
the pipeline, the internal detection probe is also required. However, the research on
the detection probe in the uniform electromagnetic field pipeline is still blank.
Aiming at the shortage that the current uniform electromagnetic field detection
probe is not suitable for in-pipe detection, this paper proposes a Hermholtz coil-type
aluminum alloy in-pipe detection probe and builds an experimental system. Through
the finite element software Comsol, establish a simulation model for in-pipe detection
of stainless steel pipes, analyze the distribution and change law of electromagnetic
fields in stainless steel pipes, extract characteristic signals, analyze the influence of
crack size on characteristic signals, design a stainless steel pipe in-pipe detection
probe, build an in-pipe detection experimental system, and detect cracks on the inner
wall of the pipeline.

1 Finite Element Emulation

The detection probe in the uniform electromagnetic field pipeline adopts the
Hermholtz coil structure. The Hermholtz coil is a pair of parallel and coaxial coils
with the same current-carrying coil. When the current in the same direction is passed
to the coil and the distance between the two coils is equal to the coil radius, the total
magnetic field of the coil will be uniformly distributed near the center of the axis.
Based on the characteristics of the Hermholtz coil, a detection model in the pipeline
is established to explore the distribution and change law of the electromagnetic field
in the pipeline [15].

1.1 Distribution Law of Electric Field in Pipeline

According to the structure of the Helmholtz coil, the finite element software
COMSOL is used to establish a simulation model for in-pipe detection as shown
in Fig. 1. The specific model parameters are shown in Tables 1 and 2.
Extracting the current density of the inner surface of the pipeline is shown in
Fig. 2 from the results, it can be seen that the maximum value of the induced current
is near the excitation coil, and a uniform electric field is induced in the middle of the
two coils, which meets the requirements of uniform electromagnetic field detection.
1 Finite Element Emulation 85

Fig. 1 Simulation model

Table 1 Sizes of model

Composition Outer diameter (mm) Inner diameter (mm) Length (mm)
Pipeline 65 45 200
Excitation coil 40 36 10
Air – – –

Table2 Parameters of model

Wire Number of Pipe Current Magnetic Conductivity Excitation
diameter turns material size (A) permeability (S/m) frequency
(mm) (Hz)
0.8 200 Stainless 0.5 1 3.7e7 2000
steel

Fig. 2 Surface current

density

1.2 Crack Disturbance Law

In order to explore the disturbance mechanism of cracks, the electromagnetic field

distribution law in the aluminum alloy pipeline when there are cracks on the inner
surface of the emulation pipeline. The length of the axial crack on the inner wall of
the pipeline is 20 mm, the width is 1 mm, and the depth is 5 mm.
86 Design and Experimental Study of Inner Uniform Electromagnetic …

Extracting the current density near the axial crack on the inner wall of the pipeline
is shown in Fig. 3. The results show that when a uniform current field passes through
the crack, as shown in Fig. 3a, the current on the surface will bypass the tip of the
crack. As shown in Fig. 3b, the internal current will bypass from the bottom of the
crack, which conforms to the principle of uniform electromagnetic field detection.
Using parametric scanning to simulate real probe detection, the origin of the coil
is scanned from − 60 mm to 60 mm in the Z coordinate, and the step size is 1 mm.
Considering the probe structure and the size of the sensor, the axial magnetic field
signal and the radial magnetic field signal directly above the crack are extracted at
7 mm as shown in Fig. 4. It can be seen from the figure that when there is no crack,
the signal is a fixed value, and the signal is 0. When passing through the defect, the
signal appears two smaller troughs and a larger crest, and the signal appears a crest

(a) Surface current density

(b) Internal current density

Fig. 3 Current dens

1 Finite Element Emulation 87

Fig. 4 Axial direction signal Bz and radial direction signal Br above crack

trough. Therefore, the axial magnetic field signal and the radial magnetic field signal
are used as characteristic signals for crack detection.

1.3 Effect of Crack Size on Characteristic Signal

In order to explore the relationship between the extracted feature signal and the crack
size, finite element emulation analysis is carried out on cracks of different depths
and different lengths. First, keep the length of the crack (20 mm) unchanged, and
emulation analysis is carried out on cracks of different depths (3, 4, 5, 6, 7 mm).
The axial magnetic field signal at the 7 mm lift-off above the crack is extracted. The
emulation result is shown in Fig. 5a. From the results, it can be seen that with the
increase of the crack depth, the height of the crest also increases. It is defined as
the abnormal variable of the axial magnetic field signal caused by the crack, and the
background value is the difference value between the maximum value of the crest
and the background value. The relationship between the obtained crack depth and
the abnormal variable is shown in Fig. 5b. It can be seen that within a certain range
of the crack depth, the abnormal and depth variable of the magnetic field maintain a
relatively good linear relationship, indicating that the characteristic signal contains
the crack depth information.
Keep the crack depth (5 mm) unchanged, emulation analysis of cracks of different
lengths (10, 20, 30, 40, 50 mm), and extract the radial magnetic field signal at the
7 mm lift-off above the crack. The results of emulation are shown in Fig. 6a, b. As
88 Design and Experimental Study of Inner Uniform Electromagnetic …

(a) BZ signals with cracks of different depths

(b) The amplitude of cracks at different depths is variable

Fig. 5 Relationship between BZ and the crack depth

shown, when the crack length is small, the crack length will affect the crest trough
spacing and amplitude of the signal. The longer the crack, the greater the crest trough
spacing and the distortion amplitude; when the crack is long, the crack length hardly
affects the amplitude of the distortion, and the crest trough The spacing will increase
with the increase of the crack length, so the characteristic signal contains the length
information of the crack, especially for the detection effect of long cracks will be
better.
2 Detection System Construction and Experiment 89

(a) Relationship between Br and the crack length

(b) Br peak-to-valley spacing of cracks of different lengths

Fig. 6 Relationship between Br and the crack length

2 Detection System Construction and Experiment

2.1 Detection Probe Design

The structure of the detection probe is shown in Fig. 7, which mainly includes an
excitation coil, a non-magnetic skeleton, a cable protection device, a connecting
device and a detection sensor. A 0.8 mm enameled wire is wound on both sides of
the non-magnetic skeleton for 200 turns to form a Hermholtz excitation coil, which
is used to generate an excited magnetic field; the cable protection device is used to
pass and protect the input and output of the excitation coil and the detection sensor;
the connecting device is used to connect with the driving device in the pipeline to
complete the automatic scanning of the pipeline.
90 Design and Experimental Study of Inner Uniform Electromagnetic …

Fig. 7 Structure of testing

probe

Ten detection sensors are evenly distributed in the circumferential direction of the
middle position of the excitation coil, which can realize the comprehensive detection
of cracks on the inner wall of the pipeline at one time. The installed circuit board is
shown in Fig. 8, and the selected sensor is a small volume, large linear range, and
high sensitivity TMR sensor. The sensor is installed near the edge of the center of
the circuit board, and the TMR for detecting the axial characteristic signal is on the
A side of the circuit board, as shown in Fig. 8a, and the TMR for detecting the radial
characteristic signal is on the B side of the circuit board, as shown in Fig. 8b. The
AD620 for amplifying the signal is installed on the circuit board, and the sum signal
is amplified 100 times each. With the help of the co-ground capacitance, the ability
of filtering, noise reduction and anti-interference is achieved [16, 17].

2.2 Detection System

The entire detection system as shown in Figs. 9 and 10, comprising a bench, pipeline
under test, detection probe, hardware integrated chassis, PC. The signal generator in
the hardware integrated chassis generates a sinusoidal AC signal of 2000 Hz, which
is loaded on the excitation coil of the detection probe. The excitation coil induces
a uniform electric field and magnetic field in the pipeline. The bench drives the
detection probe to run at a uniform speed of 10 mm/s in the pipeline. When there
is a defect in the pipeline, it will cause the electric field and magnetic field in the
pipeline. Distortion, the circumferential direction of the intermediate position of the
detection probe 10 evenly distribute the detection sensor picks up the magnetic field
signal, and the acquisition card in the hardware integrated chassis is collected into
the PC. The PC processes the signal based on the software identification program
jointly developed by Matlab and Labview, and finally identifies the defect.
2 Detection System Construction and Experiment 91

Fig. 8 TMR PCB

(a) A-side of TMR PCB

(b) B-side of TMR PCB

Fig. 9 Test system

2.3 Test Results

The cracks of different depths and the same length on the inner wall of the pipeline
are detected, and the pipe size is shown in Fig. 11. The obtained sum signal is shown
in Fig. 12. From the results, when there are cracks, the characteristic signal will crest,
and the characteristic signal will crest and trough. For cracks of the same length and
different depths, the deeper the crack, the greater the abnormal variable of the signal.
Observe each characteristic signal, the abnormal variable of the No. 3 sensor is the
largest, and it can be preliminarily determined that the defect exists near the No. 3
sensor. Extract the signal of the No. 3 sensor to obtain the relationship between the
abnormal variable of the signal and the crack depth as shown in Fig. 14. The results
92 Design and Experimental Study of Inner Uniform Electromagnetic …

Fig. 10 Experiment system structure chart

Fig. 11 Pipes with different depth cracks

show that the depth of the crack has a good linear relationship with the abnormal
variable of the signal, which is consistent with the emulation result (Fig. 13).
For cracks of different lengths and the same depth on the inner wall of the pipeline,
the size of the pipeline is shown in Fig. 14. The obtained sum signal is shown in
Fig. 15. For cracks of different lengths and the same depth, the longer the crack, the
greater the crest trough spacing of the signal. Observing each signal, the abnormal
variable of sensor 3 is the largest, so it can be preliminarily concluded that the crack
exists near sensor 3. Extract the signal of sensor 3 to obtain the relationship between
the signal crest trough spacing and the crack length as shown in Fig. 16. From the
results, it can be seen that the length of the crack and the spacing of the signal crest
trough have a good linear relationship, which is consistent with the emulation result.
2 Detection System Construction and Experiment 93

(a) Bzwith different crack depths

(b) Br with different crack depths

Fig. 12 Crack detection results at different depths

94 Design and Experimental Study of Inner Uniform Electromagnetic …

Fig. 13 Bz amplitude distortion of cracks with different depths

Fig. 14 Pipes of different length crack

2 Detection System Construction and Experiment 95

(a) Bz with different crack lengths

(b) Br with different crack lengths

Fig. 15 Crack detection results at different lengths

96 Design and Experimental Study of Inner Uniform Electromagnetic …

Fig. 16 Br peak-to-valley spacing of cracks of different lengths

3 Conclusion

According to the characteristics of in-pipe detection, based on AC electromagnetic

field detection technology, this paper proposes a new type of Hermholtz coil-type
stainless steel in-pipe detection probe. Through the finite element software Comsol, a
uniform electromagnetic field stainless steel in-pipe detection model is established.
The distribution and change law of the electromagnetic field in the stainless steel
pipeline are analyzed, and the influence of the length and depth of the crack on the
characteristic signal is discussed. An experimental system is built to detect the axial
cracks of different lengths and depths on the inner wall of the pipeline. The emulation
and experimental results show that:
(1) The Hermholtz coil stainless steel pipeline internal detection probe can excite
a uniform electric field and magnetic field on the inner surface of the pipeline.
When encountering defects, it will cause distortion of the current field, which
will further cause distortion of the space magnetic field.
(2) The abnormal variable of the signal contains the depth information of the crack.
The deeper the crack depth, the greater the abnormal variable of the signal; the
crest trough spacing of the signal contains the length information of the crack.
When the crack length is small, the crack length will affect the amplitude and
crest trough of the signal. When the crack length is longer, the influence of
the crack length on the amplitude of the signal decreases, mainly affecting the
spacing of the signal crest trough.
References 97

(3) The construction of the Hermholtz coil-type stainless steel pipeline internal
detection probe and its experimental system can realize the full-circumferential
rapid scanning of the axial cracks on the inner surface of the stainless steel
pipeline.

References

1. Zarrinzadeh H, Kabir MZ, Deylami A (2017) Experimental and numerical fatigue crack growth
of an aluminium pipe repaired by composite patch. Eng Struct 133:24–32
2. Taherishargh M, Vesenjak M, Belova IV et al (2016) In situ manufacturing and mechanical
properties of syntactic foam filled tubes. Mater Des 99:356–368
3. Malekan M, Barros FB, Sheibani E (2016) Thermo-mechanical analysis of a cylindrical tube
under internal shock loading using numerical solution. J Braz Soc Mech Sci Eng 38(8):2635–
2649
4. Malekan M, Cimini CA Jr (2018) Finite element analysis of a repaired thin-walled aluminum
tube containing a longitudinal crack with composite patches under internal dynamic loading.
Compos Struct 184:980–1004
5. Canadian Energy Pipeline Association (2007) Stress corrosion cracking recommended
practices. Canadian Energy Pipeline Association, Alberta
6. Chen CL (2018) Application of nondestructive testing technology in pressure pipeline. Technol
Equip 44(7):80–81
7. Wen Y, Feng Q, Li Y (2014) The application of ultrasonic testing in pipeline welds corrosion.
NDT 36(2):50–52
8. Yang LJ, Ge H, Gao SW (2018) Study on high-speed magnetic flux leakage testing technology
based on multistage magnetization. Chin J Sci Instrum 39(6):148–156
9. Li W (2007) Research on ACFM based defect intelligent recognition and visualization
technique. China University of Petroleum, Dongying
10. Lewis AM, Michael DH, Lugg MC et al (1988) Thin-skin electromagnetic fields around surface-
breaking cracks in metals. J Appl Phys 64(8):3777–3784
11. Chacón Muñoz JM, García Márquez FP, Papaelias M (2013) Railroad inspection based on
ACFM employing a non-uniform B-spline approach. Mech Syst Signal Process 40(2):605–617
12. Hu XC, Luo FL, He YZ et al (2011) Pulsed alternating current field measurement technique
for defect identification and quantification. J Mech Eng 47(4):17–22
13. Li W, Yuan XA, Chen G et al (2016) High sensitivity rotating alternating current field
measurement for arbitrary-angle underwater cracks. NDT&E Int 79(2):123–131
14. Li W, Ge J, Wu Y et al (2017) An electromagnetic Helmholtz-coil probe for arbitrary orientation
crack detection on the surface of pipeline. Mater Trans 58(4):641–645
15. Ge J, Li W, Chen G et al (2017) Multiple type defect detection in pipe by Helmholtz
electromagnetic array probe. NDT&E Int 91:97–107
16. Li W, Yuan XA, Chen GM et al (2017) Research on the detection of surface cracks on drilling
riser using the chain alternating current field measurement probe array. J Mech Eng 53(8):8–15
17. Gao YH, Zhang GX, Huang PJ et al (2009) Research on key technology of eddy current testing
system based on GMR transducer. Transduc Microsyst Technol 28(11):31–36
98 Design and Experimental Study of Inner Uniform Electromagnetic …

Open Access This chapter is licensed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits use, sharing,
adaptation, distribution and reproduction in any medium or format, as long as you give appropriate
credit to the original author(s) and the source, provide a link to the Creative Commons license and
indicate if changes were made.
The images or other third party material in this chapter are included in the chapter’s Creative
Commons license, unless indicated otherwise in a credit line to the material. If material is not
included in the chapter’s Creative Commons license and your intended use is not permitted by
statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder.
Research on the Detection of Surface
Cracks on Drilling Riser Using the Chain
Alternating Current Field Measurement
Probe Array

1 Introduction

In the offshore oil and gas industry, the drilling riser is an essential component that
is mostly utilized to link the drilling platform and the seabed wellhead, creating a
mud return channel. Stress corrosion cracking (SCC) is a prevalent issue on drilling
risers in service because of the complex alternating stress and corrosive conditions.
The primary cause of stress corrosion cracking is the collection of microcracks on
the riser’s surface. These cracks then progressively grow into longer longitudinal
cracks when external pressures engage on them, rupturing or leaking the riser and
finally resulting in catastrophic drilling mishaps. Therefore, conducting research on
the detection of early axial cracks on the surface of the riser is of great significance
for preventing riser failure accidents.
Due to the harsh environment of marine drilling and the complex dynamic loads
borne by drilling risers, as well as the accumulation of a large amount of debris on the
surface, traditional magnetic particle inspection techniques are difficult to detect the
surface cracks of the risers. Saturation magnetization is necessary for magnetic flux
leakage testing, which has significant requirements for lift-off height, riser surface
cleaning, and post-test demagnetization. However, cumbersome processes are not
acceptable for high-cost offshore operations.
In this paper, a chain array detection probe based on ACFM is proposed to
detect stress corrosion cracks on the outer surface of marine drilling risers. By using
multiple U-shaped current-carrying coils, a large uniform field region is excited on
the riser pipe surface. The distorted magnetic field above the cracking is extracted
by the array GMR sensors, and the cracking characteristic signals are displayed on
the computer. The use of a chain structure to encapsulate the excitation coils and
sensor arrays enables the detection of metal structures with any diameter, signifi-
cantly improving detection efficiency and applicability. Moreover, this provides an
efficient and universal method for detecting surface cracks on drilling risers.

© The Author(s) 2025 99

W. Li et al., Alternating Current Field Measurement Technique for Detection and
Measurement of Cracks in Structures, https://doi.org/10.1007/978-981-97-7255-1_7
100 Research on the Detection of Surface Cracks on Drilling Riser Using …

2 Principle of Alternative Current Field Measurement

The ACFM technique uses an excitation probe to form a localized uniform current
field on the surface of a metallic structure. When the current passes near a defect,
it will be deflected, and the deflection current will cause a spatial magnetic field
perturbation in which the X-direction flux density Bx appears as a trough in the
center of the crack, and the Z-direction flux density Bz appears as opposite peaks at
both ends of the crack.
The crack can be quantitatively evaluated by measuring the perturbed magnetic
field, with the technical principle as shown in Fig. 1a [1]. Due to the advantages of
non-contact measurement (no or little surface cleaning), no calibration, and quan-
titative assessment, ACFM is very suitable for the detection of surface cracks on
underwater structures [2].
The U-type magnetic core current-carrying coil can significantly improve the
effect of probe excitation current on metal surfaces [3]. Therefore, in this paper, a
rectangular current-carrying coil and a U-shaped Mn–Zn ferrite core are selected to
form a U-shaped excitation probe. As shown in Fig. 1b, an alternating electromagnetic
field is created within the solenoid when a sinusoidal signal is delivered through the
rectangular coil.In the low-frequency excitation state, the Maxwell equations can be
used to express the relationship between the electromagnetic fields in space:

∂B
∇ ×E =− (1)
∂t
∂D
∇ × H = J+ (2)
∂t

∇ ·B=0 (3)

∇ ·D=ρ (4)

∇ differential operator.
E electric field strength.
J Current density.
B Magnetic flux density.
H Magnetic field strength.
D Electrical flux density.
ρ volumetric charge density.

According to Maxwell Eqs. (1–4), the Helmholtz equation for the magnetic flux
density inside the solenoid can be obtained:

∇2B + k 2B = 0 (5)
2 Principle of Alternative Current Field Measurement 101

Fig. 1 ACFM schematic diagram. a Schematic diagram of ACFM. b Schematic diagram of the
U-type excitation probe detection

where k = ω(u0 ε0 )1/ 2 = ω/ c, u0 is the vacuum permeability, ε0 is the vacuum

capacitance.
In order to facilitate the solution of the governing equation of electromagnetic field
at the crack of ACFM, two quantities, vector magnetic potential A and scalar electric
−
→ −
→
potential B = ∇ × A , are introduced to separate the magnetic field variables from
the electric field variables, thus making it simple and easy to carry out the numerical
solution [4]. The two definitions are as follows:
102 Research on the Detection of Surface Cracks on Drilling Riser Using …

vector magnetic potential:

−
→ −
→
B =∇× A (6)

scalar electric potential:

−
→
E = −∇φ (7)

The vector magnetic potential and the scalar potential defined by (6) and (7) can
automatically satisfy the Faraday’s law of electromagnetic induction and Gauss’s
law of magnetic flux. And the partial differential equations of the magnetic field and
the electric field, respectively, are obtained after derivation:

−
→ →/
2− −
→
∇ 2 A − με∂ A ∂t 2 = −μ J (8)

→/
2− /
∇ 2 φ − με∂ φ ∂t 2 = −ρ ε (9)

where A(X , Y , Z) = AO (X , Y , Z) + AP (X , Y , Z) is the Laplace operator.

Due to the disturbance of the electromagnetic field caused by the defects, the
vector magnetic potential A, which represents the distribution of the magnetic field,
can be divided into two parts for ease of calculation:

A(X , Y , Z) = AO (X , Y , Z) + AP (X , Y , Z) (10)

where, A0 is the vector potential function for detecting current induction, and AP
is the vector potential function for electric field disturbance induction caused by
defects.
According to electromagnetic induction, the vector potential functions A0 and AP
both satisfy the Laplace equation [5]:

∂ 2A ∂ 2A ∂ 2A
+ + =0 (11)
∂X 2 ∂Y 2 ∂Z 2
In the formula, A0 satisfies the boundary condition of the defect-free state:

∂ 2 A0 ∂ 2 A0 m ∂ 2 A0
+ + = 0 |Z = 0 (12)
∂X 2 ∂Y 2 μr ∂Z 2

AP satisfies the boundary condition of the defect-free state:

∂ 2 Ap ∂ 2 Ap m ∂ 2 Ap
+ +
∂X 2 ∂Y 2 μr ∂Z 2
( )
cm ∂Ap
= 2+ δ(Y )|Z = 0 (13)
ur ∂Z
3 Finite Element Method Model 103

where m2 = 2i/δ 2 , δ is the skin depth, c is the defect width [6]:

δ = (μr μ0 π σ f )−1/2 = (2/σ μω)1/2 (14)

where μr is the relative permeability of the material, μ is the permeability of the

material, μ = μr μ0 , σ represents the conductivity, f represents the current frequency,
and ω represents the angular frequency of the current.
When the uniform current field excited by the U-type current-carrying coil passes
through the defective region, the current gathers at the two ends of the crack, which
causes the pipe radial spatial magnetic field Bz to produce peak and valley due to the
opposite direction of the current deflection, and the distance between the peak and
valley reflects the length of the crack. At the same time, due to the effect of crack
depth, the current density gradually decreases in the depth direction of the crack
center, which will produce a decrease in the magnetic flux density Bx in the axial
direction of the pipe, forming a wave valley. The larger the crack depth is, the larger
the Bx trough depth (the distortion rate of the amplitude in the background magnetic
field) is, so the Bx signal contains crack depth information. Owing to the accurate
ACFM mathematical model, the crack length and depth information can be obtained
by applying the uniform electromagnetic field to detect and evaluate the defects.

3 Finite Element Method Model

3.1 Modeling Design

The riser [7] that is frequently used in the Nanhai 8 platform, with an outside diameter
of 533.4 mm (21 inches) and a wall thickness of 15.875 mm (5/8 inch), is chosen
as the research object for this paper. Table 1 displays the precise dimensions and
specifications.
The ACFM simulation model for the axial cracks on the riser surface is established
by using ANSYS finite element software, as shown in Table 2. In order to reduce
the number of computational units, the model is built only in the upper part with
the cracked region. The dimensions of the simulation model are shown in Table 2.
The excitation coil is a U-shaped core current-carrying coil [8] and the excitation
signal is a sinusoidal signal with a frequency of 6 kHz and an amplitude of 1 V
[9, 10]. The current density around the crack is extracted in the simulation model
as shown in Fig. 2b. It can be clearly seen that the U-shaped magnetic core excites
two eddy current zones on the surface of the riser, and the middle of the vortex

Table 1 Dimensions of riser

Model Diameter/mm Wall thickness/mm Length/mm Material
Dimensions 533.4 15.875 18.21 X80 steel
104 Research on the Detection of Surface Cracks on Drilling Riser Using …

Table 2 Dimensions of simulation model

Model Diameter/ Wall Length/mm Length of Width of Depth of
mm thickness/ crack/mm crack/mm crack/mm
mm
Dimensions 533.4 15.875 2000 45 0.8 5

region approximately forms a uniform electric field region. There is a high current
density at both ends of the crack because the uniform electric field gathers at both
ends as it travels through the crack. At the same time, there will be a sparse current
density inside the middle crack air gap as a result of the current being diverted
to the depth direction when it passes through the crack’s center. It is stated that
perturbing the electric field will result in a change in the spatial magnetic field based
on the Maxwell classical electromagnetism relations. A path is defined along the
top of the crack with the path step set to 100, and the magnetic flux densities Bx
(axial direction) and Bz (radial direction) above the crack are extracted. As seen
in Fig. 2c, Bz produces a peak and valley at the ends of the crack, respectively,
and its peak and valley starting position reflects the crack length. Bx produces a
wave peak at the center of the crack (the negative sign indicates the direction, and
its absolute value is the valley). Bx contains the information of crack depth due
to the relationship between the flux density attenuation and the crack depth. The
aforementioned research demonstrates that the simulation model developed in this
work may achieve the quantitative detection of cracks on the riser surface while
adhering to the ACFM detection law.

3.2 Excitation Probe Spacing Simulation Analysis

Determining the distance between the row of excitation coils is necessary for
designing a chain array probe. Excitation coil spacing should be such that the column
surface in the surrounding area is fully covered, but it shouldn’t be too small to prevent
superposition of the excitation zone and space wastage. With the use of the simula-
tion model, as indicated in Fig. 3a, the core is offset from the crack by a specific arc
length L. The Bx and Bz flux densities above the crack are then extracted to provide
the Bx and Bz flux density distortions at various offset distances, as indicated in
Fig. 3b. The extracted Bx and Bz flux density distortions are shown in Table 3.
Figure 3 shows that both the Bx and Bz amplitudes decrease significantly as
the offset distance of the excitation core increases. The Bz magnitude distortion
decreases linearly when the core is shifted by 20 mm; it tends to decrease slowly
when the core is shifted by more than 20 mm; and when the core is offset by 50
mm, the Bz magnitude distortion is 11.23% of the Bz magnitude aberration when
the core is located directly above the crack (offset angle of 0°). The Bx amplitude
distortion falls sharply and then tends to decrease steadily when the core is offset
by 10 mm. When the offset distance reaches 50 mm, the Bx amplitude distortion
3 Finite Element Method Model 105

Fig. 2 Simulation result.

a Simulation model of the
riser surface axial crack.
b Distortion of surface
current in the crack region.
c Extractions of the crack
characteristic signal
106 Research on the Detection of Surface Cracks on Drilling Riser Using …

Fig. 3 Analysis of the influence region of an excitation probe. a Excitation core offset distance
diagram. b Bx and Bz flux density distortion plots

Table 3 Bx and Bz flux density distortions

L/mm 0 10 20 30 40 50
ΔBz/ × 10−4 T 9.44 6.28 2.95 2.11 1.77 1.06
ΔBx/ × 10−4 T 3.85 0.74 0.72 0.65 0.42 0.24

is 6.23% of that when the core is located directly above the crack (offset 0°), and
the distortion amplitude is less than 10% of the normal detection position. Such
distortion amplitude makes it easy to form an interfering signal, which results in
leakage detection and false detection. Therefore, according to the simulation results,
the maximum distance of the excitation core offset from the crack can be set to 40
mm, i.e., the maximum design spacing between the excitation cores of the chain
array probe is 80 mm.
4 Experimental Research 107

3.3 Sensor Spacing Simulation Analysis

After determining the chain excitation core spacing, it is also necessary to determine
the design spacing of the sensors below the core. With the help of the simulation
model, the spatial flux densities Bx and Bz on the surface of the riser are extracted at
a lift-off height of 4 mm above the crack, as shown in Fig. 4a, b. It can be seen that
the flux density distortion region above the crack is distributed between − 5 and 5
mm in the crack width direction.
As shown in Fig. 5a, the excitation core is located directly above the crack, and
the Bx and Bz flux densities are extracted at the positions of the sensor with different
offset crack distances (0–6 mm). Figure 5b plots the Bx and Bz flux density distortions
at various offset positions, and Table 4 displays the flux density distortions. It is
evident that the Bx and Bz amplitude distortions decrease with the increase of the
sensor offset crack distance. When the offset distance reaches 6 mm, the amplitude
distortion of Bz reaches the order of 10–5 T, and the characteristic signal is already
very weak. In order to obtain the characteristic signals of the defects, the maximum
offset crack distance of the detection sensor is 5 mm, or the maximum design spacing
of the sensor below the core is 10 mm.

4 Experimental Research

4.1 Design of the Testing System

Based on the principle of ACFM, a drilling riser chain array detection system is
designed as shown in Fig. 6. The current-carrying coil of the U-shaped magnetic core
receives a sinusoidal excitation signal from the signal generator with an amplitude
of 1 V and a frequency of 6 kHz. The riser’s surface is excited by a uniform current
field from the excitation core, and when this excitation electric field gets through
the surface cracks, the perturbation current distorts the spatial magnetic field. After
signal conditioning (amplification and filtering), the spatial magnetic field distortion
signals are detected by the array sensors placed at the bottom of the chain probe and
sent to the acquisition card. The array probe data is transferred by the acquisition
card to a computer for defect analysis and signal display.

4.2 Sensor Circuit Design

The sensitivity, large linear range, small size, and other characteristics of the Giant
Magneto Resistance (GMR) magnetic field sensor make it an excellent choice for
weak magnetic field detection [11]. GMR magnetic field sensors are used in the
drilling riser chain array ACFM probe sensors designed in this paper. In order to
108 Research on the Detection of Surface Cracks on Drilling Riser Using …

Fig. 4 The spatial flux densities Bx and Bz around the crack. a Bx space flux density. b Bz space
flux density

measure the Bx and Bz signals, the GMR sensors are orthogonally welded on both
sides of the PCB circuit board. Besides, the common ground capacitance is utilized
for filtering, noise reduction, and anti-interference, while the AD620 amplifies the
signals [12]. The front and back sides of the sensors’ circuit are shown in Fig. 7a, b.
4 Experimental Research 109

Fig. 5 Area of flux density distortion in the cracked area. a Diagram of the detecting sensor offset
distance. b Bx and Bz flux density distortions

Table 4 Magnetic flux density distortions around crack

L/mm 0 1 2 3 4 5 6
ΔBz/ × 10−4 T 4.23 3.69 3.10 2.46 1.41 1.01 0.05
ΔBx/ × 10−4 T 2.51 2.30 2.22 2.08 1.81 1.20 1.02

4.3 Chain Probe Design

The excitation core is fixed in the center of a single link, and the GMR detection
circuitry is distributed on the bottom plate of the single link, keeping the lift-off
height of the GMR sensors at 4 mm. The individual links can be freely bent to
110 Research on the Detection of Surface Cracks on Drilling Riser Using …

Fig. 6 Diagram of the testing system design

each other to ensure that the arc crack detection of variable pipe diameter can be
realized, and the link can be freely disassembled according to the detection area and
space restrictions. Due to the limitations of the acquisition card, the experimental
system adopts three single links, each of which has a current-carrying excitation core
inserted. 21 uniformly numbered detecting circuit boards are dispersed beneath the
links (No. 0–20 sensor). Figure 8a depicts the probe design, and Fig. 8b shows the
chain probe after it has been assembled and processed.

4.4 Test Experiment

The test specimen is a riser with an outer diameter of 533.4 mm and a wall thickness
of 15.875 mm. Its outer surface has an EDM-machined axial crack (Crack No. 0) with
dimensions of 45 mm in length, 0.8 mm in width, and 5 mm in depth. Furthermore,
there is consistency between the computer model and the experimental parameters.
Figure 9a displays the specimen and the crack, and Fig. 9b shows the entire inspection
devices. The chain probe is mounted on the spindle of the bench, and the computer
controls the bench to telescope or rotate via PLC. A signal generator provides a sinu-
soidal signal with a frequency of 6 kHz and an amplitude of 1 V to the chain probe.
The defect signals measured by the sensors inside the chain probe are processed and
captured by the signal processor and delivered to the computer. Intelligent visualiza-
tion software based on LabVIEW and MATLAB inside the computer displays the
characteristic signals of the defects.
The chain probe is placed close to the surface of the riser. Using the computer to
control the bench to extend slowly at a uniform speed of 5 mm/s, the bench drives
the chain probe to sweep uniformly across the arc along the axial direction of the
4 Experimental Research 111

Fig. 7 Circuit design for detection sensors. a Front side of PCB (Bz sensor). b Backside of PCB
(Bx sensor)

riser, and the GMR sensors pick up the Bx and Bz magnetic flux densities above
the crack and implement the A/D conversion by the acquisition card. The intelligent
visualization software inside the computer stores and recalls the Bx and Bz signals
of each channel for display. Finally, the magnetic flux densities above the crack of
the inspection board test are shown in Fig. 10.
Similarly, another axial crack (crack No. 1) on the surface of the laboratory riser
machined by EDM of 40 mm long, 0.8 mm wide and 5 mm deep is detected using
a chain probe at a rate of 5 mm per second. The test results with the most obvious
amplitude of the characteristic signal are selected, as shown in Fig. 11.
112 Research on the Detection of Surface Cracks on Drilling Riser Using …

Fig. 8 Chain probe design. a Design drawing of the chain probe. b Physical drawing of the probe
design

4.5 Analysis of Experimental Results

The Bx and Bz measured by sensors No. 9–11 show some distortions, with a trough
in Bx and two peaks in Bz. Since the GMR sensor measurement value is the absolute
value of the magnetic field at the location, the internal program of the software
adopts the root-mean-square algorithm for the collected data, so the collected Bz
signal changes from the original peak and valley to two peaks. The distance between
the two peaks of Bz reflects the length of the crack, and the depth of the valley of
Bx reflects the depth of the crack. The aberration pattern of the Bx and Bz signals
is in agreement with the results of the simulation model, which is in line with the
principle of ACFM detection.
At the same time, it can be seen that the Bx and Bz distortions measured by sensor
No. 10 have the strongest signals, while the Bz signal measured by sensor No. 11 and
No. 9 adjacent to sensor No. 10 has weak characteristics and distortion phenomena.
Moreover, the signals measured by the No. 8 and No. 12 sensors are cluttered, and
no characteristic signal of cracks can be seen. From this, it can be judged that the
crack appears near the No. 10 sensor, so as to realize the crack location detection.
4 Experimental Research 113

Fig. 9 Specimen and testing system. a Riser specimen and axial crack. b The testing system

From Fig. 10c, it can be obtained that the peak sampling point interval of the
characteristic signal Bz of crack No. 0 is 1255–1755, and the number of sampling
points is 500. Figure 11 shows the results of another crack No. 1 of the same depth
and width tested by the chain probe at the same test speed. The characteristic signal
Bz interval for this crack is 1095–1520 with 425 sampling points. Since the same
speed is used, the predicted crack length can be obtained from the ratio of the number
of sampling points of the Bz characteristic signal of crack No. 1 to the number of
sampling points of crack No. 0 as shown in Eq. (15).

L1 = L0 × N1 /N0 (15)

L0 is the length of crack No. 0, N0 is the number of peak sampling points of the
feature signal of crack No. 0, L1 is the predicted crack length, and N1 is the number
of peak sampling points of the crack sign signal of crack No. 1.
As predicted by formula (15), the length of crack No. 1 is 38.25 mm, and the
quantitative error of crack length is 4.36%, which means the detection accuracy
meets the engineering requirements.
114 Research on the Detection of Surface Cracks on Drilling Riser Using …

Fig. 10 Test result of 45 mm crack detection of drilling riser. a Sensor no. 8 test results. b Sensor
no. 9 test results. c Sensor no. 10 test results. d Sensor no. 11 test results. e Sensor no. 12 test results
5 Conclusions 115

Fig. 10 (continued)

Fig. 11 Test result of 45 mm crack detection of riser

5 Conclusions

In this paper, a chain array probe based on the ACFM technology principle is created
for the characteristics of stress corrosion cracks on the drilling riser surface. This
probe can accomplish rapid, quantitative, and location-based crack surface measure-
ment. The ACFM simulation model of a U-shaped current-carrying coil is estab-
lished with the aid of ANSYS software in order to analyze the characteristic signals
of the surface cracks of the riser, and optimize the distances of the excitation cores
and sensors. Based on the simulation results and GMR sensors, a chain-array ACFM
probe is constructed, and further tests are carried out to find the riser’s surface cracks.
Results from experiments and simulations indicate that:
(1) U-type ACFM current-carrying coil excitation probe can excite a uniform elec-
tric field region on the surface of drilling riser.Axial stress corrosion crack on
the riser’s surface can disturb the electric field, which in turn causes the spatial
magnetic field to become aberrated due to the perturbation current. Accordingly,
the length and depth of the crack is contained in the aberration magnetic fields
Bx and Bz.
(2) In order to realize the thorough detection of cracks on the surface of the drilling
riser, the design spacing of the excitation cores should be no more than 80 mm,
and the design spacing of the detection sensors should be no more than 10 mm.
116 Research on the Detection of Surface Cracks on Drilling Riser Using …

(3) The chain array probe designed by using GMR sensors can effectively detect
cracks on the surface of the riser, realize rapid, quantitative and location detec-
tion, and provide a set of effective methods for early crack discovery and integrity
management of drilling riser.

References

1. Li W, Chen G, Zhang C et al (2013) Simulation analysis and experimental study of defect

detection underwater by ACFM probe. Chin Ocean Eng Soc 27(2):277–282
2. Amineh RK, Ravan M, Sadeghi SHH et al (2007) Using AC field measurement data at an
arbitrary lift off distance to size long surface-breaking cracks in ferrous metals. NDT&E Int
40(7):537–544
3. Li W, Chen G (2013) Simulation analysis of U-shape inducer for ACFM. J Syst Simul
40(6):738–742
4. Li W (2007) Research on ACFM based defect intelligent recognition and visualization
technique. Dongying China Univ Petrol
5. Kang Z, Luo F, Hu Y et al (2004) The mathematical modelling of alternating current field
measurement and the establishment of approximately uniform field. NDT 26(11):546–550
6. Tao M, Chen D, Lu Q et al (2012) Eddy current losses of giant magnetostrictors: modeling and
experimental analysis. J Mech Eng 48(13):146–151
7. Liu X, Chen G, Chang Y et al (2013) Analyses and countermeasures of deepwater drilling riser
grounding accidents under typhoon conditions. Petrol Explor Dev 40(6):738–742
8. Li W, Chen G (2009) Defect visualization for alternating current field measurement based on
the double U-shape inducer array. J Mech Eng 45(9):233–237
9. Li W, Chen G, Li W et al (2011) Analysis of the inducing frequency of a U-shaped ACFM
system. NDT&E Int 44:324–328
10. Li W, Chen G, Yin X et al (2013) Analysis of the lift-off effect of a U-shaped ACFM system.
NDT&E Int 53:31–35
11. Gao Y, Zhang G, Huang P et al (2009) Research on key technology of eddy current testing
system based on GMR transducer. Transduc Microsyst Technol 28(11):31–36
12. Liu Y, Yang S, Yu B (2011) Detection of crack om nonferromugnetic metal using eddy current
test based on GMR. J Vibr Measur Diagn 31(6):747–753
References 117

Open Access This chapter is licensed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits use, sharing,
adaptation, distribution and reproduction in any medium or format, as long as you give appropriate
credit to the original author(s) and the source, provide a link to the Creative Commons license and
indicate if changes were made.
The images or other third party material in this chapter are included in the chapter’s Creative
Commons license, unless indicated otherwise in a credit line to the material. If material is not
included in the chapter’s Creative Commons license and your intended use is not permitted by
statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder.
An Electromagnetic Helmholtz-Coil
Probe for Arbitrary Orientation Crack
Detection on the Surface of Pipeline

1 Introduction

Pipelines provide the safest and most economical form of transportation of crude
oil, natural gas, and other petrochemical commodities compared to truck, rail cars,
and tankers [1]. Suffering from the degeneration of materials, for example crack,
corrosion, it is important to carry out nondestructive testing (NDT) to maintain high
reliability of pipeline transportation.
Surface cracks, especially the stress corrosion crack (SCC), are one of the most
harmful degradations considering their effect on structural integrity [2]. Subjected to
hoop stress and land movement load, the orientation of surface cracks may be axial
direction, circumferential direction or the others. Ultrasonic testing is a widely used
technique in pipeline detection. However, the need of couplant limits the detection in
gas transportation pipeline [3, 4]. As to magnetic flux leakage, the leaking magnetic
flux is proportional to the opening of the crack, so it is not sensitive to axial tight
SCC [5−7]. Magnetic particle inspection (MPI) is the most reliable NDT method in
pipeline detection. While MPI is a very effective inspection technique, its field use
can be costly when one considers the surface cleaning and operating time [5]. The
eddy current (EC) could not make accurate size assessment of cracks [8].
The aim of this paper is to propose an electromagnetic Helmholtz-coil probe for
oriented cracks mapping and sizing on the surface of pipeline. This paper is organized
as follows. In Sect. 2, the model of the Helmholtz-coil probe is presented and the
perturbation of the magnetic field above the crack is analyzed through the finite
element software COMSOL. In Sect. 3 the electromagnetic Helmholtz-coil probe is
set up and the oriented crack in the range of 0°–90° are mapped through experimental
testing.

© The Author(s) 2025 119

W. Li et al., Alternating Current Field Measurement Technique for Detection and
Measurement of Cracks in Structures, https://doi.org/10.1007/978-981-97-7255-1_8
120 An Electromagnetic Helmholtz-Coil Probe for Arbitrary Orientation …

Fig. 1 FEM model the electromagnetic Helmholtz-coil probe

2 Simulation of the Electromagnetic Helmholtz-Coil Probe

An electromagnetic Helmholtz-coil probe consists of two critical components was

presented in this paper: a Helmholtz-coil excitation and a detecting sensor array. The
Helmholtz-coil excited by alternating current (AC) was coaxial with the pipeline to
induce a uniform current field along the circumference of the pipeline and a uniform
magnetic field along the axis of the pipeline [9]. The tunnel magneto resistive (TMR)
detecting sensor array was used to measure the magnetic field signals.

2.1 Finite Element Model of the Electromagnetic

Helmholtz-Coil Probe

The finite element model of the electromagnetic Helmholtz-coil probe was built
through the COMSOL software, as shown in Fig. 1. The “magnetic field physics”
was selected and the “impedance boundary condition” was applied on the surface of
the pipeline. The dimensions of the model are shown in Table 1 and the characteristic
parameters are shown in Table 2. The electric current and magnetic flux density
distribution on the surface of pipeline without crack are shown in Fig. 2. As shown
in Fig. 2, the uniform electric current field and magnetic field are induced on the
surface of pipeline [10, 11].

2.2 Result Analysis

In the simulation, the oriented cracks in the range of 0°–90°, as shown in Fig. 3, were
detected. The position of the probe was moved along the axial direction (z-axis) to
2 Simulation of the Electromagnetic Helmholtz-Coil Probe 121

Table 1 The size of the FEM model

Model Diameter/mm Length/mm Width/mm Depth/mm Interval/mm
Pipeline (D/d) 65/45 300
Helmholtz-coil 80 4 40
Crack 20 1 5

Table 2 The characteristic parameter of the FEM model

Number of Pipeline Conductivity/S/ Permeability Excitation Frequency/Hz
turns material m current/A
500 Carbon 1.12e7 4000 1 1000
steel

Fig. 2 Electric current and magnetic flux density distribution on the surface of the pipeline

simulate the probe scanning process along a pipeline during experimental studies.
The response magnetic field of the probe was measured at a 1 mm lift-off. Two
components of the magnetic field, axial direction (z-axis) Bz and radial direction
(y-axis) By, were measured as the characteristic signals, as shown in Fig. 4.
As shown in Fig. 4a, b, with the increasing of the angle, the Bz signal changes
from dip to peak gradually and the By signal remained a peak and dip. However, the
order of the peak and dip in By signals varies with the increasing of crack orientation.
The electric current density around the 0° orientation crack on the surface of
pipeline and magnetic flux density around the 90° orientation crack in the x-y plane
of the model are shown in Fig. 5.
122 An Electromagnetic Helmholtz-Coil Probe for Arbitrary Orientation …

Fig. 3 Oriented cracks in

the range of 0°–90°

Fig. 4 Bz and By signals, a Bz, b By

It can be seen from the Fig. 5a that the electric current concentrating in the tips of
the crack, which is expressed as arrow, when the crack of 0° orientation is detected
is the electric current perturbation effect caused by discontinuous conductivity. As
shown in Fig. 5b, the magnetic flux leaks in the edge of the crack, which is expressed
as line, when the crack of 90° orientation is detected is the magnetic flux leakage
effect caused by discontinuous permeability. These phenomenon also explain the
variations of Bz and By signal in Fig. 4 that the Bz and By signals of the crack of 0°,
15° orientation are caused by the electric current perturbation effect and the signals
3 Experiments 123

Fig. 5 The electric current distribution and magnetic flux density around the crack, a electric
current density around the 0° orientation crack on the surface of pipeline, b magnetic flux density
around the 90° orientation crack in the x-y plane of the model

of the crack in the range of 30°–90° orientation are caused by the magnetic flux
leakage effect.
An implication of this is the possibility that the combination effect of the elec-
tric current perturbation and magnetic flux leakage around the crack may have the
advantage of detecting the oriented crack on the surface of pipeline.

3 Experiments

3.1 System Set Up

The structure of the Helmholtz probe is shown in Fig. 6. The excitation coils were
wound on the polymer frame. Detecting sensors were equally-spaced installed in the
polymer frame. Supports were used to fix the probe and keep the constant lift-off
to different pipeline. The extension-type tape was used to fix the sensor array. A
sensor array containing Tunnel Magneto Resistive (TMR) was employed to scan
the full circumference of the pipeline. Considering the space of the Helmholtz-coil
electromagnetic probe and actual manufacturing difficulty, a 24 equal-spaced sensor
array was selected. An electromagnetic Helmholtz-coil detection system was built,
as shown in Fig. 7. The excitation source produced an alternating current signal with
the frequency of 1 kHz and the magnitude of 10 V. The turns of the coil were 1000 in
total. The current was transferred to the excitation coil through the power amplifiers.
The detecting sensor array picked up the magnetic field and translated it into electric
signal. The signals were amplified and filtered in signal processing module. And
then, the signals were converted into digital signal by an A/D convertor and sent
to PC for signal processing. A detection software was developed to achieve defects
recognition. The Bz and By signals were shown in the display screen of computer.
The Helmholtz-coil probe was fixed in an axial scan table, as shown in Fig. 7.
124 An Electromagnetic Helmholtz-Coil Probe for Arbitrary Orientation …

Fig. 6 Structure of Helmholtz-coil probe

Fig. 7 Experimental setup

3.2 Result Analysis

The pipeline made from carbon steel was detected to test the detectability of
Helmholtz-coil probe to crack. The axial cracks, which are 1 mm width and 30 mm
length, with the depth in the range of 2–10 mm were machined on the surface of
pipeline by Electric Discharge Machining (EDM), as shown in Fig. 8.
The pipe string was moved along the axial direction at a speed of 10 mm/s and
the Bz and By signals were measured and shown in Fig. 9. It can be seen from the
figures that all of the cracks on the surface of pipeline could be detected obviously.
3 Experiments 125

Fig. 8 a Schematic diagram of pipeline and b actual cracks in pipe strings

Furthermore, the length of cracks can be obtained accurately. The amplitudes of the
signal are proportional to the depth of the cracks.
Oriented cracks with the orientation in the range of 0°–90° were machined on the
surface of pipeline by EDM, as shown in Fig. 10, and the parameters of the cracks
were 45 mm × 1 mm × 5 mm (length × width × depth). The pipeline was scanned

Fig. 9 Bz and By signals

126 An Electromagnetic Helmholtz-Coil Probe for Arbitrary Orientation …

Fig. 10 Pipeline with the cracks in the range of 0°–90° orientation

through the system in Fig. 7 at a speed of 10 mm/s. The data of the sensor array
were obtained and drawn in the MATLAB. The C scan results are shown in Fig. 11.
It can be seen from the results that the orientation of the cracks can be recognized
clearly. Moreover, with the increasing of the crack orientation, the Bz signal varies
from dip to peak gradually. The Bz and By signals of the crack of 0°, 15° and 45°
orientation are caused by the electric current perturbation effect and the signals of
the crack of 60°, 75° and 90° orientation are caused by the magnetic flux leakage
effect. Comparing with the results in simulation, we can see that the numerical and
experimental results have the similar variation law and mechanism that using the
combination effect of electric current perturbation and magnetic flux leakage, the
arbitrary orientation crack can be detected.

4 Conclusions

In this paper, a Helmholtz-coil probe has been proposed to detect oriented cracks
on the surface of the pipeline. A FEM model of this probe is put forward through
the COMSOL. Based on the model, the relationship between the crack orientation
and Bz and By signal is analysis. Finally, the structure of the Helmholtz-coil probe
with a sensor array is designed and the experimental studies are carried out to test
the detectability of oriented crack. The results of the simulations and experiments
show that: using the combination effect of the current perturbation and magnetic flux
leakage, the oriented crack on the surface of pipeline can be recognized clearly. The
length of the cracks can be measured through the By signals.
Although, the detection and length measurement of the oriented cracks through
the probe in this paper have been proved, the depth of the oriented crack can’t be
sized directly from the Bz signals. In the future work, we will focus on the depth
sizing algorithm of oriented cracks through the Helmholtz-coil probe. Moreover, in
the experiment, the artificial cracks were detected. In the next stage, some actual
cracks or closed SCC will be tested.
4 Conclusions 127

Fig. 11 C scan results of the oriented crack ranging from 0° to 90° with 15° interval, a, c, e, g, i,
k, m are the Bz signals, b, d, f, h, j, l, n are the By signals
128 An Electromagnetic Helmholtz-Coil Probe for Arbitrary Orientation …

Fig. 11 (continued)

References

1. Cheng YF (2013) Stress corrosion cracking of pipelines

2. Yusa N, Hashizume H, Urayama R et al (2014) NDT&E Int 61:29–34
3. Beller M (2015) Pipeline Gas J 242(11):30–36
4. Mak DK (1985) Ultrasonics 23(5):223–226
5. Smith M, Sutherby R (2005) Insight: non-destructive testing & condition monitoring.
47(12):765–768
6. Jarvis R, Cawley P, Nagy PB (2016) NDT&E Int 81:46–59
7. Wu J, Sun Y, Kang Y et al (2015) IEEE Trans Magn 51(1):1–7
8. Li W, Yuan X, Chen G et al (2014) NDT&E Int 67(8):17–23
9. Trout SR (1988) IEEE Trans Magn 24(4):2108–2111
References 129

10. Hatsukade Y et al (2007) IEEE Trans Appl Supercond 17(2):780–783

11. Wu B, Wang Y J, Liu X C et al (2015) Smart Mater Struct 24(7)

Open Access This chapter is licensed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits use, sharing,
adaptation, distribution and reproduction in any medium or format, as long as you give appropriate
credit to the original author(s) and the source, provide a link to the Creative Commons license and
indicate if changes were made.
The images or other third party material in this chapter are included in the chapter’s Creative
Commons license, unless indicated otherwise in a credit line to the material. If material is not
included in the chapter’s Creative Commons license and your intended use is not permitted by
statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder.
Circumferential Current Field Testing
System with TMR Sensor Array
for Non-contact Detection and Estimation
of Cracks on Power Plant Piping

1 Introduction

Piping is a critical component in power plant, such as the steam piping, hot reheat
steam piping, cold reheat steam piping, feed water piping, cooling water piping and
auxiliary piping. The piping serves in a harsh environment due to the noticeable
cyclical pressure and thermal stress [1–3]. The pressure can reach 30 MPa and the
temperature is more than 500 °C inside the piping. The high pressure and temperature
can cause initial thermal fatigue cracks on the outer surface of piping [4, 5]. What’s
more, the initial thermal fatigue cracks gather and grow quickly by the cyclical stress,
which cause the leakage and failure of the piping [6, 7]. Therefore, it is a critical
issue to detect and estimate cracks on power plant piping before the failure happens
[8, 9].
There are various non-destructive testing (NDT) methods for the detection of
cracks on piping. Because of the closed piping system, the inner detection techniques,
such as pipeline pigs, cannot be applied for the detection of cracks on power plant
piping during in-service time. The ultrasonic testing (UT) needs coupling medium,
which makes it an inadequate method in high temperature environment [10]. The
infrared testing is a good method for surface cracks detection. However, the temper-
ature changes constantly on the surface of piping. Thus it is hard to perform the
infrared testing on power plant piping [11]. The magnetic flux leakage testing (MFL)
is not sensitive to the narrow thermal fatigue cracks because of the little leakage of
magnetic field on stainless steel piping [12]. The eddy current testing (ET) is sensi-
tive to lift-off, which cannot penetrate thick coatings [13]. The microwave waveguide
imaging technique is mainly used for corrosion inspection on plate [14]. What’s more,
most of the methods as mentioned above need multiple scans to achieve a full 360°
inspection of the piping surface.
The magnetic particle testing (MT) and penetrant testing (PT) are traditional
NDT methods for surface cracks detection on power plant piping. However, most
processes of MT and PT needs manual operation in the harsh environment. These

© The Author(s) 2025 131

W. Li et al., Alternating Current Field Measurement Technique for Detection and
Measurement of Cracks in Structures, https://doi.org/10.1007/978-981-97-7255-1_9
132 Circumferential Current Field Testing System with TMR Sensor Array …

methods just give surface information of the crack but no depth information which is
more important to estimate the residual life of the piping. What’s more, the coatings
need to be removed or the attachments on the piping should be cleaned before NDT
operation. The magnetic particles and penetrants should be cleaned and the coatings
should be re-painted after the inspection, which is time-consuming and costly [15].
The current field perturbation NDT methods, such as alternating current field
measurement (ACFM), alternating current potential drop (ACPD) technique, are
promising techniques to inspect defects on conductive material [16, 17]. For the
detection of cracks on piping, the coaxial solenoid loaded with sinusoidal excitation
signal can induces the current field on the surface of piping in circumferential direc-
tion. In the center of the coaxial solenoid, the induced circumferential current field on
piping can be regard as an approximate uniform field. When the crack is presented,
the circumferential current field will be disturbed. The space magnetic field can be
picked up by sensors around the piping without contact [18, 19]. The defects can be
detected and evaluated by the distorted magnetic field.
The circumferential uniform current field testing system is presented for non-
contact detection and estimation of surface cracks on the power plant piping in
this paper. The circumferential current is induced on the surface of piping by a
coaxial encircling excitation coil. The inducing frequency is optimized to balance
the penetration depth and detection sensitivity. The circumferential uniform current
field testing system can cover a full 360° area on the surface of piping by TMR sensor
array in a one pass scan. All the cracks on the surface of piping can be identified and
evaluated visually and efficiently by the space magnetic field without contact using
the circumferential uniform current field testing system.
This paper is organized as follows. In Sect. 2, the finite element method (FEM)
model is set up and the inducing frequency is optimized by the simulation model.
In Sect. 3, the circumferential uniform current field testing system is developed and
the crack detection experiments are carried out. In Sect. 4, the cracks are estimated
by the distorted space magnetic field. In Sect. 5, the conclusion and further work are
delivered.

2 Circumferential Current Field FEM Model

2.1 Model Set up

The 3D FEM model of the circumferential current field is set up by ANSYS software,
as shown in Fig. 1a. The model consists of excitation coil, piping and axial crack. The
excitation coil is coaxial with the piping. The axial crack lies on the outer surface of
piping symmetrically. To simulate the detection process, a dynamic FEM model is
developed. The excitation coil is driven at 0.1 mm step-size to scan along the piping.
The material of excitation coil is copper and the piping is stainless steel. As shown
in Fig. 1b, the lift-off of the excitation coil is 10 mm and the lift-off of sensors is
7 mm, which gives enough space for the coatings. The size of the crack is 20 mm in
2 Circumferential Current Field FEM Model 133

length, 0.3 mm in width, 2 mm in depth. A sinusoidal excitation signal (frequency

20 kHz, amplitude 2 V) is loaded on the excitation coil.
When the excitation coil is above the crack, the current field around the crack
is extracted on the surface of pipe string. As shown in the bottom right corner of
Fig. 1a, the circumferential current field is uniform when the crack is not presented.
The uniform current field gathers and turns around at the tips of the axial crack. The
disturbed uniform current field perturbs the space magnetic field above the crack
[20].

Fig. 1 3D FEM model of

circumferential current field
testing probe. a FEM model.
b Size of the FEM model
134 Circumferential Current Field Testing System with TMR Sensor Array …

Fig. 2 Bx and Bz

To get the distorted magnetic field caused by disturbed uniform current field, the
magnetic field in axial direction (Bx) and radial direction (Bz) are picked up in the
center of the excitation coil at the lift-off of 7 mm. Thus, as the excitation coil moves
one step, the Bx and Bz are plotted once at each position, as shown in Fig. 2. The
Bx shows a trough in the center of the crack due to the decrease of current density
in the depth of the crack, which contains the depth information. Meanwhile, the Bz
plots a positive and negative peak at the tips of the crack. The distance between the
two peaks reflects the length of the crack.

2.2 Frequency Optimization

Due to the skin effect, inducing frequency determines the penetration depth on piping,
which has a great significant influence on detection sensitivity of crack depth [21, 22].
The skin thickness is given in Eq. (1), where ur is the relative magnetic permeability,
u0 is the magnetic permeability of free space, and σ is the electrical conductance,
f is the inducing frequency. When applying a high frequency excitation signal on
the excitation coil, the current field tends to concentrate in a thin layer flowing in
circumferential direction, which goes against sizing crack depth [23]. When the
inducing frequency is low, the current field has a large penetration depth, which is
good for sizing crack depth. However, when the inducing frequency is too low, the
induced current density is very weak on the surface of piping. Thus, the characteristic
signals of crack are weak, which can be covered by noise easily. Therefore, it is
2 Circumferential Current Field FEM Model 135

necessary to optimize the inducing frequency to balance the detection sensitivity and
penetration depth [24].
/
δ = 1 (π ur u0 σ f ) /
1 2
(1)

As mentioned above, the Bx contains the depth information of crack. Hence, the
effect of inducing frequency on Bx is analyzed using the FEM model. The background
magnetic field is extracted at different frequencies (50 Hz–60 kHz), as shown in
Fig. 3a. The background magnetic field rises steeply as the frequency increases from
50 Hz to 20 kHz and becomes a slow climb after 20 kHz. It suggests that when the
inducing frequency reaches 20 kHz, the current field saturates nearly on the surface
of piping. So to get a deeper penetration depth, the inducing frequency should be
lower.
The Bx with different inducing frequency is simulated, as shown in Fig. 3b.
Initially the troughs of Bx are flat and becomes obvious after 1 kHz. The sensi-
tivity of Bx is a critical parameter for the estimation of crack depth, which is defined
as follows [25].

Fig. 3 Frequency optimization. a Background of Bx with different frequencies. b Bx with different

frequencies. c Sensitivity of Bx with different frequencies
136 Circumferential Current Field Testing System with TMR Sensor Array …

SBx = Bxmax /Bx0 (2)

where ΔBxmax is the maximum distortion of Bx, Bx0 is the background of Bx.
A high sensitive Bx signal will greatly improve signal to noise ratio and carry
more crack depth information. As shown in Fig. 3c, the sensitivity of Bx increases
steeply initially and reaches top at 20 kHz and then drops at last. Above all, the
simulation results suggest that 20 kHz is the appropriate inducing frequency for high
sensitive detection of crack depth.

3 Circumferential Uniform Current Field Testing System

3.1 Testing System

In practice, the piping is fixed and the probe is driven by mechanical devices or robots
to scan the piping. In the laboratory, for convenience the probe is fixed on the scanner
and the piping is driven to pass through the probe. The circumferential uniform
current field testing system includes probe, excitation module, signal processing
module, acquisition card, scanner and personal computer (PC), as shown in Fig. 4.
The scanner is controlled by the programmable logic controller (PLC) in control
cabinet. The piping is driven by the scanner to pass through the probe at a constant
speed. The displacement information of the piping is recorded by the encoder. The
probe induces a uniform current field on the surface of piping by the excitation coil.
When a crack is presented, the uniform current field will be disturbed. The disturbed
current field perturbs the space magnetic field. The magnetic sensors on the probe
pick up the distorted magnetic field which is sent to signal processing box for signal
processing and acquisition. The software in the computer shows the imaging of space
magnetic field visually and estimates the size of the crack.
The probe consists of sensor module array, excitation coil and detachable nylon
yoke, as shown in Fig. 5. The sensor module is made of two high-precision tunnel
magneto resistance (TMR) magnetic sensors, whose operating temperature can reach
125 °C [26–28]. The two TMR sensors are sealed on each side of one common
printed circuit board (PCB) and the sensitive axis of the sensors is orthogonal, which
is used to measure the Bx and Bz. On the PCB, there are two primary amplifier chips
(AD620) for Bx and Bz amplification (Bx 5 times and Bz 10 times). To achieve a full
360°detection of the piping surface, the sensor modules are installed on the yoke with
an equal space as sensor array. In this paper, the sensor array is with 20° to each other
in the circumferential direction of the piping [29]. There are 5 close TMR sensor
arrays in the probe. The space magnetic field can be measured visually for imaging
under the sensor array [30]. To avoid removing the coatings on the piping, the lift-off
of sensor modules is 7 mm. Because the piping is closed, the yoke is separated in
half symmetrically. Thus the yoke and sensors can be disassembled and installed to
avoid the elbow pipes and flanges. When the two part yokes are encapsulated, the
4 Experiments and Discussions 137

Fig. 4 Block diagram of circumferential uniform current field test system

excitation coil (copper wire whose diameter is 1 mm) is wound on the yoke with 50
turns.
There are four main parts: power system, excitation module, signal processing
module and acquisition card in the signal processing box. The power system is a
rechargeable lithium ion battery which provides power for each module. The excita-
tion model transmits sinusoidal signal with 20 kHz frequency and 2 V amplitude, as
shown in Fig. 6a. The signal processing module includes second amplifying circuit
(Bx 10 times and Bz 10 times), band-pass filtering (10–30 kHz) and zeroing circuit,
as shown in Fig. 6b. Thus the Bx and Bz is amplified 50 times and 100 times respec-
tively as a whole. The Bx and Bz are filtered by the band-pass filtering and then
calibrated by the zeroing circuit to keep the same zero point and scale. The Bx and
Bz are captured by the acquisition card and then sent to the PC. The circumferential
uniform current field testing system is set up, as shown in Fig. 6c.

4 Experiments and Discussions

4.1 Crack Depth Estimation

The samples are two stainless steel pipes (external diameter 65 mm, inner diameter
47), as shown in Fig. 7. On No.1 sample, the length of cracks is the same (30 mm)
and the depth is different (2, 4, 6, 8 mm and a through crack). On No. 2 sample, the
138 Circumferential Current Field Testing System with TMR Sensor Array …

Fig. 5 Probe of circumferential uniform current field testing system. a TMR sensor module.
b Dvided yokes and sensors. c Photo of probe

length of cracks is different (55, 50, 45, 40, 35, 30 mm) and the depth is the same
(4 mm). The width of all cracks is 0.3 mm.
The first sample (No. 1 simple) is driven by the scanner to pass through the probe
at the speed of 10 mm/s (from 2 mm depth crack to through crack). The space
magnetic field Bx around the crack on the No. 1 sample is shown in PC, as shown in
Fig. 8. There are five troughs in Bx, while there are five opposite peaks in Bz at the
same location. The characteristic of the Bx and Bz are in accord with the simulation
results. The Bx and Bz are highlighted in the background magnetic field at the lift-off
of 7 mm, which helps to recognize the cracks easily without contact. Especially, the
maximum distortion values of Bx troughs and Bz peaks both appear in channel 4.
Thus the location of the cracks is confirmed on piping according to the position of
sensors in channel 4 o in the probe. Because the background of Bz approximates to
zero, the distortion of Bz is more smooth and outstanding than Bx.
The Bx and Bz are selected in channel 4, as shown in Fig. 9. The troughs of Bx
increase as the crack depth grows and the peaks of Bz goes up at the same time.
4 Experiments and Discussions 139

Fig. 6 Circumferential uniform current field testing system. a Excitation module. b Signal
processing module. c Photo of testing system

The SBx with different depth cracks are shown in Fig. 10a. The maximum distortion
values of Bz (ΔBz) with different depth cracks are shown in Fig. 10b.
The SBx and ΔBz almost increase linearly as the crack depth grows from 2 to
8 mm. However, there is a distortion point at the through crack. This is due to the
material discontinuity in the through crack and the uniform current could not flow
at the bottom of the crack. The current density drops sharply in the X direction and
gathers more seriously at the tips of the crack.
Because there is a so well linear relationship between SBx and the crack depth
(and the similar liner relationship between ΔBz and the crack depth), the calibration
method is proposed to estimate the crack depth. The first two shallower cracks are
140 Circumferential Current Field Testing System with TMR Sensor Array …

Fig. 7 Samples. a No. 1 sample with different depth cracks. b No. 2 sample with different length
cracks

set as the calibrated crack to predict the last three cracks. According to the first two
cracks, the depth of the other three cracks can be calibrated by Eqs. (3) and (4) fitted
by Fig. 10a, b respectively:

Ds = 119SBx − 0.2432 (3)

DΔ = 0.070098ΔBz + 0.8661 (4)

where Ds is the crack depth estimated by SBx , DΔ is the crack depth estimated by
ΔBz.
Thus the last three cracks can be estimated by Eqs. (3) and (4). The predicted
depths (PD) and relative errors are shown in Table 1. The 6 mm depth crack can be
estimated by SBx and ΔBz. The relative errors (RE) are 13.0% and 2.6% respectively.
The 8 mm depth and through crack are regarded as a break according the estimation
by SBx . The 8 mm depth is estimated by ΔBz and the relative error is 8.6%. The
through crack is identified as a break by SBx and ΔBz. If the first three cracks are set
as calibrated crack, the relative errors of 8 mm depth crack are both less than 5% by
SBx and ΔBz. We can make a conclusion that the crack depth can be estimated by SBx
and ΔBz using the calibration method. Thus the residual thickness of power plant
piping can be evaluated by periodic detection using circumferential uniform current
field testing system.
4 Experiments and Discussions 141

Fig. 8 Space magnetic field Bx and Bz of cracks on no. 1 sample

4.2 Crack Length Estimation

The second sample (No. 2 sample) is driven to pass through the probe at the same
speed (from 55 mm length crack to 30 mm length crack). As shown in Fig. 11, the
space magnetic field around the cracks on the No. 2 sample is plotted. There are 6
troughs in Bx and 6 peaks in Bz at the same location. Because the depth of the cracks
on No.2 sample is the same, there are no obvious changes in the troughs of Bx and
the peaks of Bz. Similarly, the troughs and peaks are located at channel 4.
142 Circumferential Current Field Testing System with TMR Sensor Array …

Fig. 9 Signals Bx and Bz in channel 4 of no. 1 samples. a Bx. b Bz

The Bx and Bz are selected in channel 4 and the encoder records the displacement
information in the X coordinate, as shown in Fig. 12. Because the peaks of Bz locate
at the tips of crack, the distance between the two opposite peaks of Bz (ΔL) reflects
the length of cracks. As shown in Fig. 12b, the ΔL becomes narrow as the crack
length decreases. However, the current field deflects clockwise at one tip of the crack
and deflects anticlockwise at the other tip of the crack. The maximum magnetic field
locates in the center of the deflected current field, which makes the ΔL less than the
crack length.
4 Experiments and Discussions 143

Fig. 10 a SBx with different depth cracks. b ΔBz with different depth cracks

Table 1 Estimated results of crack depth

Depth/ mm 2 4 6 8 Through
SBx /% 1.886 3.567 5.906 8.133 14.357
PD/mm – – 6.78 9.44 16.84
RE/% – – 13.0% Break Break
ΔBz/mv 15.974 44.150 70.930 111.550 212.900
PD/mm – – 5.84 8.69 15.79
RE/% – – 2.6% 8.6% Break

The ΔL with different length cracks is shown in Fig. 13. ΔL goes up linearly as
the crack length grows. ΔL is less than the actual length of the crack and the relative
errors become unacceptable for short crack, as shown in Table 2. The calibration
method is also proposed to estimate the length of the cracks. The first two cracks
are set as calibrated crack to predict the last four cracks. The calibration equation of
crack length (L) is given by Eq. (5) fitted by Fig. 13.

L = 1.042ΔL + 2.917 (5)

As shown in Table 2, the relative errors of the crack length drop sharply by the
calibration equation. The calibrated crack length equals the actual length of the crack
except the 45 mm length crack which has a tiny relative error. Thus the length of
the crack on piping can be estimated by the distance between the peaks of Bz using
the calibration method. What’s more, the development of the crack length on power
plant piping can be monitored and warned by periodic detection and estimation.
144 Circumferential Current Field Testing System with TMR Sensor Array …

Fig. 11 Testing results of no. 2 sample

5 Conclusion and Further Work

This paper presents a novel circumferential uniform current field testing system for
surface cracks detection and estimation on power plant piping. The dynamic FEM
model is developed to extract the characteristic signals of the crack. The inducing
frequency is optimized by the FEM model for sensitive and accurate detection of
the crack. In the end, the circumferential uniform current field testing system is
5 Conclusion and Further Work 145

Fig. 12 Signals Bx and Bz in channel 4 of no. 2 samples

set up and the crack detection experiments are carried out. The results show that
the cracks on piping can be detected and estimated visually and efficiently without
contact by the circumferential uniform current field testing system with TMR sensor
array using calibration method in a one pass scan. The depth of the crack can be
estimated by the sensitivity of Bx and the maximum distortion of Bz. Meanwhile, the
length of the crack can be evaluated by the distance between the opposite peaks of
Bz. What’s more, the circumferential uniform current field testing system provides a
new method for non-contact, visual and efficient detection and estimation of cracks
on power plant piping as an alternative technique to MT and PT. Further work will
focus on monitoring the development of the crack using circumferential uniform
current field testing system. Thus the propagation of the crack and the residual wall
thickness can also be monitored and estimated.
146 Circumferential Current Field Testing System with TMR Sensor Array …

Fig. 13 ΔL with different length cracks

Table 2 Testing results of crack length

Length/mm 55 50 45 40 35 30
ΔL/ mm 49.6 45.0 40.3 35.6 30.8 26.0
RE/% 9.8 10.0 10.4 11.0 12.0 13.3
L – – 44.9 40.0 35.0 30.0
RE/% – – 0.2 0 0 0

References

1. Jahromi AA, Firuzabad MF, Parvania M (2012) Optimized midterm preventive maintenance
outage scheduling of thermal generating units. IEEE Trans Power Syst 27:1354–1365
2. Zhang X, Zhang JS, Gockenbach E (2015) Estimation of leak rate through circumferential
cracks in pipes in nuclear power plants. Nucl Eng Technol 47:332–339
3. Amjady N, Daraeepour A (2011) Midterm demand prediction of electrical power systems using
a new hybrid forecast technique. IEEE Trans Power Syst 26:755–765
4. Locatelli ML, Khazaka R, Diaham S, Pham CD, Bechara M, Dinculescu S, Bidan P (2014)
Compact thermal failure model for devices subject to electrostatic discharge stresses. IEEE
Trans Power Electron 29:2281–2288
5. Zhou YZ, Miao M, Salcedo JA, Hajjar JJ, Liou JJ (2015) Compact thermal failure model for
devices subject to electrostatic discharge stresses. IEEE Trans Electron Devices 62:4128–4134
6. Saboonchi H, Ozevin D, Kabir M (2016) MEMS sensor fusion: Acoustic emission and strain.
Sens Actuat A 247:566–578
7. Guedri A (2013) Reliability analysis of stainless steel piping using a single stress corrosion
cracking damage parameter. Int J Press Vessels Pip 111:1–11
References 147

8. Gao YL, Tian GY, Li KJ, Ji J, Wang P, Wang HT (2015) Multiple cracks detection and visu-
alization using magnetic flux leakage and eddy current pulsed thermography. Sens Actuat A
234:269–281
9. Huang H, Mawby PA (2013) A lifetime estimation technique for voltage source inverters. IEEE
Trans Power Electron 28:4113–4119
10. Day P (2003) An automated ultrasonic inspection method for thickness checking of dryer
cylinders for paper making machines. Insight 45:628–631
11. Altet J, Gomez D, Perpinyà X, Mateo D, González JL, Vellvehi M, Jordà X (2013) Efficiency
determination of RF linear power amplifiers by steady-state temperature monitoring using
built-in sensors. Sens Actuat A 192:49–57
12. Afzal M, Udpa S (2002) Advanced signal processing of magnetic flux leakage data obtained
from seamless gas pipeline. NDT&E Int 35:449–457
13. Zhou DQ, Wang J, He YZ, Chen DW, Li K (2016) Influence of metallic shields on pulsed eddy
current sensor for ferromagnetic materials defect detection. Sens Actuat A 248:162–172
14. Zhang H, He Y, Gao B, Tian GY, Xu L, Wu R (2016) Evaluation of atmospheric corrosion on
coated steel using K-band sweep frequency microwave imaging. IEEE Sens J 16:3025–3033
15. Xie SJ, Chen ZM, Takagi T, Uchimoto T (2015) Quantitative non-destructive evaluation of
wall thinning defect in double-layer pipe of nuclear power plants using pulsed ECT method.
NDT&E Int 75:87–95
16. Li W, Yuan XA, Chen GM, Ge JH, Yin XK, Li KJ (2016) High sensitivity rotating alternating
current field measurement for arbitrary-angle underwater cracks. NDT&E Int 79:123–131
17. Nativel EL, Talbert T, Martiré T, Joubert C, Daudé N, Falgayrettes P (2015) Near-field electro-
magnetic tomography applied to current density reconstruction in metallized capacitors. IEEE
Trans Ind Inform 20:11–16
18. Zhu F, Zhang ZC, Luo JR, Zhang YW (2010) Investigation of the failure mechanism for an
s-band pillbox output window applied in high-average-power klystrons. IEEE Trans Electron
Devices 57:946–951
19. Qiu JZ, Sullivan CR (2013) High-frequency resistivity measurement method for multilayer
soft magnetic films. IEEE Trans Power Electron 28:3581–3590
20. Li W, Chen GM, Yin XK, Zhang CR, Liu T (2013) Analysis of the lift-off effect of a U-shaped
ACFM system. NDT&E Int 53:31–35
21. Fan MB, Wang Q, Cao BH, Ye B, Sunny AI, Tian GY (2016) Frequency optimization for
enhancement of surface defect classification using the eddy current technique. Sensors 16:1–16
22. Sunny AI, Tian GY, Zhang J, Pal M (2016) Low frequency (LF) RFID sensors and selective
transient feature extraction for corrosion characterisation. Sens Actuat A 241:34–43
23. Phan KL, Boeve H, Vanhelmont F, Ikkink T, Jong FD, Wilde HD (2016) Influence of metallic
shields on pulsed eddy current sensor for ferromagnetic materials defect detection. Sens Actuat
A 248:162–172
24. Brink EA, Hofsajer IW (2013) Analytical approach for determining the frequency dependent
characteristics of multipath conductive structures. IEEE Trans Power Electron 29:5835–5845
25. Noroozi A, Hasanzadeh RPR, Ravan M (2013) A fuzzy learning approach for identification of
arbitrary crack profiles using ACFM technique. IEEE Trans Magn 49:5016–5027
26. Arias SIR, Muñoz DR, Cardoso S, Ferreira R, Freitas PJPD (2015) Total ionizing dose (TID)
evaluation of magnetic tunnel junction (MTJ) current sensors. Sens Actuat A 225:119–127
27. Dabek M, Wisniowski P (2015) Dynamic response of tunneling magneto resistance sensors to
nanosecond current step. Sens Actuat A 232:148–150
28. Lim H, Lee S, Shin H (2015) Advanced circuit-Level model for temperature-sensitive read/
write operation of a magnetic tunnel junction. IEEE Trans Electron Devices 62:666–672
29. Li W, Yuan XA, Chen GM, Yin XK, Ge JH (2015) A feed-through ACFM probe with sensor
array for pipe string cracks inspection. NDT&E Int 67:17–23
30. Stefanov KD (2014) A statistical model for signal-dependent charge sharing in image sensors.
IEEE Trans Electron Devices 61:110–115
148 Circumferential Current Field Testing System with TMR Sensor Array …

Open Access This chapter is licensed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits use, sharing,
adaptation, distribution and reproduction in any medium or format, as long as you give appropriate
credit to the original author(s) and the source, provide a link to the Creative Commons license and
indicate if changes were made.
The images or other third party material in this chapter are included in the chapter’s Creative
Commons license, unless indicated otherwise in a credit line to the material. If material is not
included in the chapter’s Creative Commons license and your intended use is not permitted by
statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder.