International Journal of Mechanical Sciences 217 (2022) 106990

Contents lists available at ScienceDirect

International Journal of Mechanical Sciences

journal homepage: www.elsevier.com/locate/ijmecsci

Magnetic stray field measurements to identify and localise impact-induced

plastic deformation in a steel structure
P.C. Meijers ∗, C.T. Jolink, A. Tsouvalas, A.V. Metrikine
Delft University of Technology, Faculty of Civil Engineering, Stevinweg 1, 2628CN Delft, The Netherlands

ARTICLE INFO ABSTRACT

Keywords: To study the irreversible changes in the magnetic stray field surrounding a steel structure caused by impact-
Plastic deformation induced elastic and plastic deformations, a steel cylinder was repeatedly subjected to axial impacts of various
Magnetic stray field changes magnitudes. Due to impacts that induce elastic deformation, the measured magnetic stray field of the structure
Axial impact
converges to a global magnetic equilibrium. However, as soon as plastic deformation develops, a deviation from
Steel cylindrical structure
this trend is observed. From the spatial distribution of the stray field, the location of the plastic deformation is
determined. Subsequently, the underlying processes of the measured evolution of the stray field are discussed
and successfully incorporated into an elementary model of the structure’s magnetisation to simulate the results
from the experiment. It is expected that the reported observation is useful for a class of engineering applications
in which non-contact and non-collocated measurements can be utilised to identify structural damage under
dynamic loading.

1. Introduction that it reduces the total cost of wind turbine, as it requires less steel
than the grouted version [7]. Nevertheless, the design does require a
Driven by the ambitious climate goals set out by governments to perfect alignment between the pile head and the superstructure. Any
reduce the emission of greenhouse gasses, the demand for energy gen- plastic deformation of the pile head can potentially disturb this delicate
erated from renewable sources has soared during the past decade [1], alignment [8].
and it is expected that this trend will continue in this new decade [2]. To determine whether a hammer blow inflicted plastic deformation
One attractive method to generate energy is the offshore wind turbine.
at the pile head, a non-collocated method which relies on conventional
Despite the plethora of available foundation types for such a structure,
strain measurement devices has been proposed in [9]. However, this
steel monopiles are the preferred choice in the relatively shallow
approach requires the sensors to be mounted on the monopile, which
North Sea [3]. Monopiles are thin-walled cylindrical structures with
a diameter of several metres and a length of tens of metres, which makes them prone to damage as a result of energetic hammer blows.
are most commonly driven into the seabed by means of an hydraulic Hence, hammer-induced plastic deformation is preferably inferred from
impact hammer. Each hammer blow generates a compressive stress data collected by a non-contact sensor. Therefore, this paper focusses
wave, which propagates downwards along the pile and enables the on the strain-induced change in the magnetic stray field of the steel
pile’s progression into the seabed. To overcome the increasing soil re- pile, since this field permeates the space around the structure allowing
sistance at greater penetration depths, the input energy of the hammer for non-contact measurement. The aim of the stray field technique is
is raised accordingly. Especially at the pile head, high-energy impacts to detect plastic deformation in an early stage using a non-collocated
may induce stresses close to the material’s yield limit, increasing the and non-contact sensor. During pile driving, when a region of plastic
risk of plastic deformations there. deformation is detected, the driving process can be altered to avoid
Until recently, plastic deformations at the pile head were of little further damage to the pile. Moreover, plastic deformation is detrimen-
concern, as that part did not contribute significantly to the bearing tal for the service life of the support structure. Hence, by localising and
capacity of the pile. However, the industry has switched to a new
quantifying the damage, the implications for the structural integrity can
connection between the monopile and the superstructure that relies
be determined.
on a bolted flange connection [4]. This design change was necessary
Currently, several magnetic methods exist to detect plastic defor-
as undesired settlements of the previously-used grouted connections
mation in steel specimens. The first group of these methods actively
were registered [5,6]. A flanged connection has the additional benefit

∗ Corresponding author.
E-mail address: p.c.meijers@tudelft.nl (P.C. Meijers).

https://doi.org/10.1016/j.ijmecsci.2021.106990
Received 30 July 2021; Received in revised form 3 November 2021; Accepted 6 December 2021
Available online 29 December 2021
0020-7403/© 2021 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
P.C. Meijers et al. International Journal of Mechanical Sciences 217 (2022) 106990

employ an external magnetic field during the evaluation process. Plastic Table 1
Chemical composition of the steel cylinder.
deformation is subsequently detected by (i) analysing the characteris-
tics of measured magnetisation curves [10,11], (ii) determining the in- Element C Si Mn P S Cr

tensity of the Barkhausen noise [12,13], or (iii) measuring the magnetic % 0.14 0.180 0.79 0.11 0.007 0.17
stray field when the specimen is magnetically saturated [14,15]. The
latter technique has been applied to localise defects in pipelines [16–
18] and in wire ropes used in suspension bridges [19]. However, in between load cycles directly reflect the irreversible changes in the
these active methods require the generation of a substantial external magnetisation, which result from two different physical processes.
magnetic field in the vicinity of the structure. For an offshore monopile First, whenever the strain introduced by a load cycle remains below
installation, actively generating such a field is deemed impractical the elastic limit of the material, the strain energy supplied by the load
given the large structural dimensions. enables the magnetic domain walls to move towards a magnetic equi-
In contrast, passive magnetic methods operate using only the ambi- librium by overcoming pinning sites which prevented this movement in
ent magnetic field, e.g. the geomagnetic field. For lab-scale specimens, the first place [33,34]. Consecutive load cycles of the same magnitude
stress concentration zones, cracks and regions of plastic deformation do not irreversibly alter the magnetisation further. However, when a
have successfully been located [20–22]. However, large-scale cylindri- new loading cycle induces strains that surpass the previously endured
cal structures exhibit a substantial demagnetising field due to their peak strain, an irreversible change occurs towards a new magnetic
geometry [23], complicating the analysis of the magnetic stray field. equilibrium state [31]. Note that these processes, although related to
Thus far, little experimental data for such structures have been re- the elastic regime, may cause irreversible changes in the magnetisation.
ported. Normally, a specimen subjected to a passive magnetic eval- Second, as soon as plastic deformation develops in the structure, the
uation technique is fully demagnetised before a load is applied to dislocation density locally increases [35]. In conjunction, the number
erase the magnetic history, as the presence of an initial magnetisation of magnetic domain walls increases; since these walls are pinned by
frustrates the detection of defects [24–26]. For large-scale structures, the imperfections, the size of the average magnetic domain in the
fully demagnetising the whole structure is practically impossible, as material is effectively reduced. As a consequence, the bulk magneti-
this requires a carefully-controlled external magnetic field on a massive sation irreversibly decreases for the given external field. Moreover,
scale. the plastic deformation introduces residual strains in the material.
Passive magnetic methods record the irreversible changes in the Both effects resulting from plastic deformation irreversibly change the
magnetic stray field, since a mechanical stress applied to the system, structure’s magnetisation [36]. As a first load cycle that exceeds the
either below or above the material’s elastic limit, irreversibly alters its material’s yield limit simultaneously introduces a new peak strain,
magnetisation [27,28]. To date, all methods used to identify whether an both aforementioned irreversible changes of the magnetisation occur
irreversible change in the magnetisation is caused by elastic or plastic in conjunction, obscuring the direct physical process. Here, a method is
strains are developed under well-controlled conditions in which the developed to distinguish between the two causes of permanent changes
location of the plastic deformation is known a priori and the loading in the magnetic stray field, which in essence allows one to separate
is introduced gradually [29–31]. However, for large-scale structures plastic from elastic deformations in the structure.
this is generally not the case, prohibiting the application of the current
methods. In this paper, the irreversible magnetisation change caused 3. Experimental set-up
by plastic deformation is successfully identified in a steel cylinder
subjected to axial impacts without a priori knowledge of the exact cir- To assess whether the above-described two types of irreversible
cumferential location of the damage by considering the tendency of the changes can be discriminated when plastic deformation develops in
stray field to converge to a global magnetic equilibrium. Subsequently, an uncontrolled environment, an axial drop-weight experiment has
the observed behaviour is simulated by applying a simple empirical been conducted on a thin-walled steel cylinder, for which the set-up is
model for the structure’s magnetic susceptibility. presented in Fig. 1a. The length of the cylinder is 𝐿 = 1500 mm; its outer
This paper is structured as follows. Section 2 provides general diameter equals 𝐷 = 406.4 mm; and its wall thickness is ℎ = 2.5 mm.
background regarding the magnetic stray field and the causes for The steel grade is S275J2H with tensile yield stress of 300 N/mm2 ,
irreversible change in that field. In Section 3, the set-up of the axial- ultimate strength of 484 N/mm2 and Young’s modulus 𝐸 = 210 GPa.
impact experiment is presented, of which the results are discussed in Table 1 presents the chemical composition of the material.
Section 4. Section 5 presents the modelling of the measured evolution A 403 kg concrete mass, which can fall freely along an aluminium
of the stray field. Finally, conclusions are drawn in Section 6. guiding system from different heights 𝓁, provides the means to intro-
duce an axial impact and thus compressive strain of varying amplitude
2. Irreversible changes of the magnetic stray field in the structure. Axial strain gauges (type: UFLA-5-11) located at 𝑧 =
−200 mm measure the structure’s deformation during impact to quan-
Surrounding the steel structure shown in Fig. 1a, the total magnetic tify these strains, which are recorded using a 16-bit data acquisition
field 𝐁𝑡 (𝐩) at point 𝐩 consists of a uniform background field 𝐁0 and the system (National Instruments USB-6343) and a sampling frequency of
magnetic stray field of the structure 𝐁(𝐩): 50 kHz. Note that, even though the structure is axially symmetric, an
impact does not induce a perfectly axially-symmetric strain state due
𝐁𝑡 (𝐩) = 𝐁0 + 𝐁(𝐩) . (1)
to the fact that it is practically impossible to assure perfect symmetry
The former represents, for instance, the Earth’s magnetic field, which— of the impact load. Hence, the strain field is expected to have an
for the current purpose—is assumed space and time invariant. Changes unknown circumferential distribution. Moreover, the strain gauges are
in the latter field, 𝐁(𝐩), are directly related to changes in the total not positioned in the axial plane where plastic deformation is expected
remanent magnetisation of the structure 𝐌(𝐫) since these two quantities to develop. However, the strain gauges are merely used to obtain an
are related as follows [32]: indication of magnitude of the impact-induced strains and will not be
[ ] directly compared to the magnetic stray field measurements.
𝜇 3 (𝐌(𝐫) ⋅ 𝐬) 𝐬 𝐌(𝐫)
𝐁(𝐩) = 0 − d𝛺, (2) In between impacts, the magnetic field around the top of the cylin-
4𝜋 ∫𝛺 ‖𝐬‖5 ‖𝐬‖3 der is measured with a tri-axial fluxgate magnetometer (type: FLC3-70),
in which 𝐫 is the position vector that characterises all points within which has a measuring range of ±200 μT, a sensitivity of 0.035 μT/mV.
the structure’s volume 𝛺, 𝐬 = (𝐩 − 𝐫) is the separation vector, and The sensor is mounted on a ring which can rotate and be positioned
𝜇0 is the magnetic constant. Consequently, measured changes of 𝐁𝑡 (𝐩) at various heights and offsets relative to the structure. By revolving

2
P.C. Meijers et al. International Journal of Mechanical Sciences 217 (2022) 106990

Fig. 1. (a) Schematic of the set-up of the thin-walled cylinder, including the background field 𝐁0 . (b) Measured magnetic stray field in the vicinity of the top of the cylinder after
five impacts from 𝓁 = 1000 mm (test 2e).

Table 2 with other low-velocity drop-weight experiments [37], it is expected

Detailed description of the tests in consecutive order. Suffixes i, e and p denote initial that plastic deformation will develop close to the impacted end. Hence,
state, elastic and plastic deformation, respectively. The drop height is represented by
𝓁, the (average) measured peak strain is 𝛿𝜀𝑧 and the number of impacts needed to
Fig. 2 shows the data measured on a ring at 𝑧 = −20 mm and 𝑟 =
reach a magnetic equilibrium is denoted by #. 223.2 mm, i.e. at 20 mm from the cylinder’s surface. In the circum-
id 𝓁 [mm] 𝛿𝜀𝑧 [μm/m] # ferential profile, Fig. 2a, four distinct regions are shaded, indicating
1i – – –
the positions and time moments at which visual plastic deformation
2e 1000 832 5 developed. Figs. 2a and 2c show the plastically deformed surface after
3e 1500 1200 5 test 7p and 5p, respectively. At each region, the evolution of 𝐵𝑟 with
4e 2000 1510 8 each test is presented in Fig. 2d, in which the error bars indicate the
5p 2500 1574 1 uncertainty in the measured value corresponding to 2% of ‖𝐁‖.
6p 2500 1550 1
7p 2500 1544 1
Initially (test 1i), the structure is close to the demagnetised state
due to the partial demagnetisation procedure. Nonetheless, since the
external geomagnetic field is not controlled in the set-up, the mag-
netisation in the sample still adheres to some extent to the direction
the sensor around the cylinder and subsequently applying Eq. (1), a of this external field, which results in the measured inhomogeneous
spatial map of the magnetic stray field is identified. During rotation, a initial stray field profile of 𝐵𝑟 . The observed skewness of the initial (and
data point is recorded every 1.5◦ , and afterwards the stray field data subsequent) profiles results from the geometry-induced demagnetising
is presented unfiltered. As an example, Fig. 1b shows the stray field field caused by the transverse component of the geomagnetic field,
close to the top of the cylinder measured after five impacts from a drop which points approximately towards 𝜃 = 325◦ .
height of 1000 mm (test 2e). As the field lines converge towards the top Subsequently, the axial impacts introduce sufficient energy to mag-
of the cylinder, this end acts as an effective magnetic south pole; given netise the structure further; its magnetisation aligns with the external
the fact that the magnetic stray field is divergence free, the lower end of field 𝐁0 . As this field has a predominate downward component, it
the cylinder serves as the effective north pole (not shown in the figure). drives the magnetisation vector towards the negative 𝑧-axis, resulting
To eliminate the influence of the magnetothermomechanical history in the observed direction of 𝐵𝑟 towards the top of the cylinder. This
of the cylinder, the structure is initially demagnetised using a hand-held alignment, nevertheless, is not perfect due to the obstructing effect
degaussing device. Nevertheless, due to the presence of the external of internal pinning sites. Each impact supplies strain energy to the
field, the magnetisation will not be completely reduced to zero. The structure, which causes some domain walls to break away from these
effect of this imperfect demagnetisation on the validity of the results pinning site and settle on a stronger pinning site, resulting in an
and the proposed method for plasticity detection is addressed later in irreversible change in the structure’s magnetisation. After a few impacts
the discussion of the experimental results. After the demagnetisation, from the same impact height, the domain configuration no longer
the structure is subjected to an axial impact, which is repeated until a changes, i.e. the magnetisation has reached a metastable equilibrium.
magnetic equilibrium is reached, i.e. the remanent magnetic stray field When a larger amount of strain energy is provided to the system,
no longer changes. After mapping the stray field, the drop height is e.g. by increasing the impact height, this additional energy allows
increased. This procedure is repeated until visible plastic deformation some domain walls to overcome their pinning site and settle into
starts to develop. another configuration, which represents a new metastable equilibrium.
Fig. 3 shows the evolution of the radial component of the stray field
4. Experimental results and discussion including the intermediate values after the impacts that comprise the
first three tests, displaying this convergence to an equilibrium value
An overview of the set of experiments is shown in Table 2, including with repetition of the load. Similar behaviour under elastic loads has
the measured peak strain 𝛿𝜀𝑧 and the number of impacts needed to previously been reported [31].
reach a magnetic equilibrium. The tests are numbered in consecutive The large difference between the initial state (1i) and the stray field
order and have suffixes i, e and p to denote initial state, elastic after 5 impacts of 1000 mm (2e) stems from the fact that the initial state
and plastic deformation, respectively. Rather than depicting the three- is close of the demagnetised state, in which the magnetic domains are
dimensional field as presented in Fig. 1b, Fig. 2 focuses on the dominant oriented more or less randomly. Due to the energy supplied to the mate-
component of the stray field; the radial component 𝐵𝑟 . In accordance rial by the impacts, the magnetic domain walls are reordered such that

3
P.C. Meijers et al. International Journal of Mechanical Sciences 217 (2022) 106990

Fig. 2. (a) Circumferential profile of the radial component of the magnetic stray field 𝐵𝑟 measured at 𝑧 = −20 mm and 𝑟 = 223.2 mm, which is an offset of 20 mm from the cylinder’s
surface. Grey areas indicate regions where visible plastic deformation developed; darker shaded regions were inflicted earlier. (b) Top view of the deformed cylinder after test 7p.
The yellow dashed line indicates the position of the initial surface of the cylinder. (c) Plastic deformation at the top of the cylinder after test 5p. (d) Evolution of 𝐵𝑟 in between
tests along four selected circumferential positions.

the overall magnetisation of the structures aligns better to the external

magnetic field. For the subsequent tests within the elastic regime, the
differences between successive stray field values are noticeably smaller,
since the domain wall configuration is already much closer to the global
equilibrium state. Nonetheless, the observed evolution of the stray field
is still significantly larger than the measurement error. An exception to
this trend appears between 2e and 3e at 𝜃 = 330◦ (Fig. 2d). However,
this difference lies within the error margin; hence, the converging trend
appears to continue as well.
In contrast, as soon as plastic deformation develops, e.g. test 5p
at 𝜃 = 330◦ , the direction of the change in the stray field deviates
from this trend. This behaviour is a consequence of the increased
pinning site density, leading to a decrease in the bulk magnetisation.
For circumferential positions where no visible damage developed in
test 5p (𝜃 = 30◦ , 𝜃 = 60◦ and 𝜃 = 150◦ ), the trend towards global
equilibrium simply continues until plastic deformation does develop,
i.e. during test 6p and 7p at 𝜃 = 30◦ and 𝜃 = 60◦ , respectively.
Therefore, it seems that the departure from this trend permits not only
detecting, but also localising the damage.
During the development of plastic deformation, the remanent stray Fig. 3. Measured evolution of the radial component of the remanent stray field at
𝑧 = −200 mm, 𝑟 = 223.2 mm and 𝜃 = 120◦ for the first three tests showing the
field irreversibly changes as a result of three physical processes of
intermediate values after each impact.
which the effects are superimposed:

(i) deterioration of magnetic properties due to the formation of

plastic deformation. This always results in a local decrease of (ii) irreversible domain wall motion due to energy provided by elas-
strength of the magnetisation, and reciprocally a demise of the tic deformation (Fig. 3). For the presented results, this leads to a
stray field’s amplitude in the vicinity of the damage; relatively small increase of the magnitude of the magnetisation,

4
P.C. Meijers et al. International Journal of Mechanical Sciences 217 (2022) 106990

Nonetheless, for an initial magnetisation in the unshaded area, both

elastic and plastic strain tend to decrease the magnetisation, obscuring
the cause of the permanent shift in the magnetisation. Hence, to suc-
cessfully identify the source of an irreversible magnetisation change,
the initial magnetic state of the structure must lie within the shaded
area. This specific initial state can be achieved by simply demagnetising
the area of interest. As seen from the presented experimental data,
perfect demagnetisation is not required.

5. Modelling the irreversible stray field changes

Based on the presented results, quantifying the plastic deformation

is, at present, not yet possible. However, by including a dependency
on plastic strain in a model for the magnetisation 𝐌(𝐫) in Eq. (2),
this should ultimately be feasible. In the following, the stray field of
the cylinder is simulated using a rudimentary model for the magnetic
Fig. 4. Simplified representation of the energy 𝐸 associated with the attained susceptibility to ascertain whether such a model is able to replicate
magnetisation 𝑀, in which 𝑀eq denotes the global magnetic equilibrium.
the observed irreversible stray field changes under elastic and plastic
deformation.

which is clear from the evolution of the stray field during tests 5.1. The stray field in the presence of an external field
2e to 4e;
(iii) permanent change in the geometry due to the damage, which Even though the external field 𝐁0 is assumed to be time and space
brings the surface of the cylinder closer to the sensor (see the invariant, the magnetisation of the cylinder will not be uniform due
inset in Fig. 2a). Normally, the strength of the measured stray to the geometry of the structure, which leads to a significant demag-
field increases when a ferromagnetic material comes closer to a netising field [38]. To compute the magnetic field generated by the
magnetic field sensor. structure’s magnetisation, a numerical approach applicable to steel
sheets is adopted [32], which assumes that the magnetisation compo-
Of these three processes, only the first reduces the stray field’s mag-
nent in the normal direction of the sheet (the radial component 𝑀𝑟 in
nitude. The latter two increase the measured strength of the stray
the current situation) is negligible.
field. Given that, due to the development of the plastic zone, the
By evenly subdividing the pile’s volume 𝛺 into 𝑁𝜃 elements in the
amplitude of the measured stray field still diminished, one is inclined
circumferential and 𝑁𝑧 elements in the axial direction, the structure
to conclude that the sustained damage significantly reduced the local
is discretised, and the total number of elements equals 𝑁 = 𝑁𝜃 𝑁𝑧 .
magnetisation, and that this decline can only be attributed to the
Under the assumption that the magnetisation is constant over each
development of a region of substantial plastic deformation. ( )
element and is concentrated at the element’s barycentre 𝐫𝑖 , i.e. 𝐌 𝐫𝑖 ,
Although the magnetisation’s tendency towards a global equilibrium
state resulting from elastic load cycles is pronounced, the experimental the magnetic field at 𝐩 is given by a summation of the contributions
data show that the exact value of this magnetic equilibrium state cannot from all elements:
𝜇0 ∑
𝑁
easily be determined a priori. Fortunately, an exact value is unneces- ( ) 𝐩 − 𝐫𝑖
sary to detect and localise plastic deformation, since the development 𝐁(𝐩) = 𝐌 𝐫𝑖 ⋅ 𝐧 𝑖 d𝛤𝑖 , (3)
4𝜋 𝑖=1 ∮𝛤𝑖 ‖𝐩 − 𝐫𝑖 ‖3
of permanent strain violates the described trend. However, for some ‖ ‖
initial magnetic states, this transgression is not observed when plastic in which 𝜇0 denotes the magnetic constant, and 𝐧𝑖 is the outward
deformation forms. This becomes apparent by a closer examination of normal to the element’s boundary 𝛤𝑖 . Note that Eq. (3) is simply the
a simplified representation of the energy 𝐸 associated with each value discretised counterpart of the continuous function given by Eq. (2). By
of the structure’s magnetisation 𝑀 (Fig. 4). In reality, the presented evaluating the integrals, the expression above is rewritten as follows:
line is a hypersurface, which depends on a plethora of parameters: ∑
𝑁
( )
external field strength and orientation, elastic strains, temperature, and, 𝐁(𝐩) = 𝜇0 𝐆𝑖 𝐌 𝐫𝑖 = 𝜇0 𝐆𝐩 𝐌, (4)
most importantly, the distribution of pinning sites for the magnetic 𝑖=1
domain walls. The (local) minima of the hypersurface, indicated by the in which 𝐆𝑖 contains the values of the evaluated integrals at each
vertical dotted lines, denote the admissible values for the structure’s element. In the latter part of the expression, these contributions are
magnetisation. condensed into a single matrix 𝐆𝐩 , and 𝐌 is a vector incorporat-
Assuming that the dislocation distribution does not change signifi- ing the magnetisation components from all elements. Since the ra-
cantly due to the induced elastic strains, the additional strain energy dial magnetisation component is deemed insignificant, 𝐌 contains 2𝑁
introduced by a new peak strain provides the system with sufficient entries.
energy to overcome the barrier that originally prevented it from reach- To express the magnetisation in terms of the external magnetic field
ing a lower energy state. Hence, induced elastic strains 𝜀𝑒 move the 𝐁0 , an appropriate constitutive equation is required. For a (locally)
magnetisation closer to the global equilibrium 𝑀eq , which is illustrated isotropic material, a scalar magnetic susceptibility 𝜒 suffices. Hence,
by the arrows at the top of Fig. 4. When plastic strain 𝜀𝑝 develops, the following implicit constitutive equation is employed:
the dislocation density increases. These newly-formed pinning sites ( )
𝐁0
disintegrate the existing magnetic domains, reshaping the hypersurface, 𝐌=𝜒 + 𝐆𝐫 𝐌 , (5)
𝜇0
and shifting the global minimum towards zero magnetisation. The
direction of the magnetisation change caused by plastic deformation in which 𝐆𝐫 is a 2𝑁 × 2𝑁 matrix representing the non-local interaction
is marked by the arrow below the figure. of the structure’s magnetisation, which is obtained by substituting 𝐩 =
When the structure’s initial magnetic state is in the grey-shaded 𝐫𝑖 for each element in Eq. (4). Assuming that 𝜒 is not a function of the
area shown in Fig. 4, the remanent stray field displays the behaviour magnetisation itself, rearrangement of Eq. (5) yields
described above: the deviation from the trend allows one to discrim- ( )−1 𝐁0
inate between the two causes for irreversible magnetisation changes. 𝐌 = 𝜒 𝐈 − 𝜒𝐆𝐫 , (6)
𝜇0

5
P.C. Meijers et al. International Journal of Mechanical Sciences 217 (2022) 106990

Table 3 of the stray field for areas that remain in the elastic regime (e.g. 2e to
Dimensions and discretisation parameters used to simulate the evolution of the stray
6p at 𝜃 = 60◦ ). Moreover, the deviation of the trend is captured by the
field.
localised decrease of 𝜒 due to the development of plastic deformation.
Parameter Value
Of course, currently, numerical values for the modelled susceptibility
𝐷 406.4 mm
𝐿 1500 mm are calibrated to fit the measurement data, since the exact amount
ℎ 2.5 mm of plastic deformation is initially unknown; the total strain was not
𝑁𝜃 64 measured at the actual location were that damage occurred. However,
𝑁𝑧 80
the fact that the stray field’s evolution can be simulated using such an
elementary isotropic model seems to show that the processes discussed
in Section 4 govern the observed irreversible magnetisation changes.
in which 𝐈 is the 2𝑁 × 2𝑁 identity matrix and (⋅)−1 denotes a matrix The proposed modelling approach provides a promising starting point
inversion. When the susceptibility in known, the structure’s magnetisa- for further research into the application of the described observation to
tion is computed by means of Eq. (6); subsequently, the stray field at identify, localise and potentially quantify regions of plastic deformation
𝐩 is determined by employing Eq. (4). as a result of repeated impact loads.

5.2. Model for the magnetic susceptibility

6. Conclusions
The model for the magnetic susceptibility should reflect the ob-
servations mentioned in the discussion above. To accommodate these,
Magnetic stray field measurements of a steel cylinder subjected
the following empirical expression for the magnetic susceptibility 𝜒 is
proposed: to axial impacts from various heights demonstrate that irreversible
( ) changes in the structure’s magnetisation caused by plastic deformation
‖ ‖
𝜒 𝜃, 𝑧, 𝛿𝜀𝑧 , 𝜀𝑝 = 𝜒0 + 𝛼𝑒 ‖ ‖
‖𝛿𝜀𝑧 (𝜃, 𝑧)‖ − 𝛼𝑝 ‖ 𝜀 (𝜃, 𝑧)‖ (7) can be distinguished from irreversible changes due to elastic deforma-
‖ 𝑝 ‖
in which 𝜒0 denotes the susceptibility in the partially demagnetised ini- tion. The former changes result in a deviation from the tendency of
tial state and ‖⋅‖ represents the absolute value. The term 𝛼𝑒 ‖ ‖
‖𝛿𝜀𝑧 (𝜃, 𝑧)‖
the magnetisation to converge towards a global equilibrium when the
reflects the tendency of the magnetisation to converge towards a structure is initially close to the demagnetised state. These results are
metastable magnetic equilibrium whenever a new peak strain 𝛿𝜀𝑧 is obtained in an experiment in which the exact location of the plastic
sustained, with 𝛼𝑒 a positive real-valued constant. Since the magnetic deformation is unknown a priori and the loading is not well-controlled.
properties of the material deteriorate with increasing plastic defor- With the aid of an empirical isotropic model for the structure’s mag-
‖ ‖
mation, the term −𝛼𝑝 ‖𝜀𝑝 (𝜃, 𝑧)‖ is added to the model, in which 𝛼𝑝 netic susceptibility, the measured data is successfully reproduced. The
‖ ‖
is positive model constant. More advanced empirical models for the approach based on evolution of the structure’s magnetic stray field
inclusion of the effect of plastic deformation have been proposed in presented in this paper is able to detect, localise, and potentially quan-
literature [36,39]; however, effectively, these models simply reduce tify regions of plastic deformation in dynamically loaded large-scale
the susceptibility with increasing plastic deformation. Hence, the basic structures.
form given by Eq. (7) is adopted in this paper.
Although the definition of a magnetic susceptibility is common in
the treatment of magnetoelastic problems [40,41], these normally do CRediT authorship contribution statement
not account for time-dependent irreversible changes of the magnetic
properties. As the latter are essential for the description of the evolution
P.C. Meijers: Conceptualization, Methodology, Software, Valida-
of the magnetisation caused by repeated impacts, the formulation of 𝜒
tion, Formal analysis, Investigation, Writing – original draft, Visu-
in Eq. (7) implicitly includes the mechanical loading history via 𝛿𝜀𝑧 and
𝜀𝑝 . alisation, Data curation. C.T. Jolink: Conceptualization, Methodol-
Naturally, the proposed model for 𝜒 is a function of the spatial ogy, Investigation, Writing – original draft. A. Tsouvalas: Conceptu-
coordinates, as the plastic deformations develops in certain confined alization, Writing – review & editing, Supervision. A.V. Metrikine:
regions. Fig. 5a presents the applied susceptibility at the four circum- Conceptualization, Writing – review & editing, Supervision.
ferential positions which are identical to the locations in Fig. 2. Since
it is assumed that almost the entirety of the structure remains in the
elastic regime, the susceptibility is a monotonic function that increases Declaration of competing interest
with increasing drop height. The corresponding values are represented
by the line for 𝜃 = 150◦ (Fig. 5a). At the other three locations, plastic The authors declare that they have no known competing finan-
deformation does develop; and it is assumed that this damage is con- cial interests or personal relationships that could have appeared to
fined to a small region close to the top of the cylinder, i.e. between 0 ≥ influence the work reported in this paper.
𝑧 ≥ −50 mm. The circumferential location of the damage corresponds
to the grey-shaded areas in Fig. 2a. To complete the model description,
the external field 𝐁0 = [𝐵0𝑥 𝐵0𝑦 𝐵0𝑧 ]𝑇 = [14.3 −9.5 −35.3]𝑇 μT, and the Acknowledgements
numerical values for the remaining relevant parameters are listed in
Table 3.
This research is part of the EUROS programme, which is supported
Fig. 5b presents the computed evolution of the radial component
of the stray field 𝐵𝑟 at 𝑧 = −20 mm, 𝑟 = 223.2 mm and the four by NWO domain Applied and Engineering Sciences, Netherlands and
circumferential positions. Apart from the initial state (1i), the corre- partly funded by the Dutch Ministry of Economic Affairs, Netherlands.
spondence between the simulation and the measurements is excellent. Furthermore, the authors would like to thank the technical staff at the
The discrepancy for the initial state could be attributed to intricate Stevin II laboratory and DEMO for their assistance and valuable ideas:
magnetic behaviour close the demagnetised state, which cannot be cap- Kees van Beek, Fred Schilperoort, Ruben Kunz, Peter de Vries, Giorgos
tured by a uniform isotropic magnetic susceptibility. On the contrary, Stamoulis, John Hermsen, Léon Roessen, Marten van der Meer, Paul
the scalar susceptibility is sufficient to model the observed evolution Vermeulen, Ton Blom and Maiko van Leeuwen.

6
P.C. Meijers et al. International Journal of Mechanical Sciences 217 (2022) 106990

Fig. 5. (a) Modelled magnetic susceptibility 𝜒 in the four regions. (b) Simulated evolution of 𝐵𝑟 using the proposed model for 𝜒. As a reference, the dashed lines present the
measured evolution of 𝐵𝑟 .

References [18] Ege Y, Coramik M. A new measurement system using magnetic flux leakage
method in pipeline inspection. Measurement 2018;123:163–74. http://dx.doi.
[1] Perveen R, Kishor N, Mohanty SR. Off-shore wind farm development: Present org/10.1016/j.measurement.2018.03.064.
status and challenges. Renew Sustain Energy Rev 2014;29:780–92. http://dx. [19] Christen R, Bergamini A, Motavalli M. Influence of steel wrapping on magneto-
doi.org/10.1016/j.rser.2013.08.108. inductive testing of the main cables of suspension bridges. NDT & E Int
[2] WindEurope, Offshore Wind in Europe. Key trends and statistics 2019. Tech. 2009;42(1):22–7. http://dx.doi.org/10.1016/j.ndteint.2008.08.003.
rep., WindEurope; 2020. [20] Bao S, Jin P, Zhao Z, Fu M. A review of the metal magnetic memory method. J
[3] Doherty P, Gavin K. Laterally loaded monopile design for offshore wind farms. Nondestruct Eval 2020;39(1):11. http://dx.doi.org/10.1007/s10921-020-0652-z.
Proc Inst Civ Eng - Energy 2012;165(1):7–17. http://dx.doi.org/10.1680/ener. [21] Dong L, Xu B, Dong S, Song L, Chen Q, Wang D. Stress dependence of the spon-
11.00003. taneous stray field signals of ferromagnetic steel. NDT & E Int 2009;42(4):323–7.
[4] Jensen KS, r. J. Petersen S, Pedersen RR. European offshore wind engineering– http://dx.doi.org/10.1016/j.ndteint.2008.12.005.
past, present and future. Proc Inst Civ Eng - Civ Eng 2018;171(4):159–65. [22] Wang ZD, Gu Y, Wang YS. A review of three magnetic NDT technologies. J
http://dx.doi.org/10.1680/jcien.17.00040. Magn Magn Mater 2012;324(4):382–8. http://dx.doi.org/10.1016/j.jmmm.2011.
[5] Chen T, Wang X, Gu X, Zhao Q, Yuan G, Liu J. Axial compression tests of grouted 08.048.
connections in jacket and monopile offshore wind turbine structures. Eng Struct [23] Viana A, Rouve L-L, Cauffet G, Coulomb J-L. Magneto-mechanical effects under
2019;196:109330. http://dx.doi.org/10.1016/j.engstruct.2019.109330. low fields and high stresses - application to a ferromagnetic cylinder under
[6] Dallyn P, El-Hamalawi A, Palmeri A, Knight R. Experimental investigation on pressure in a vertical field. IEEE Trans Magn 2010;46(8):2872–5. http://dx.doi.
the development of wear in grouted connections for offshore wind turbine org/10.1109/TMAG.2010.2043825.
generators. Eng Struct 2016;113:89–102. http://dx.doi.org/10.1016/j.engstruct. [24] Guo P, Chen X, Guan W, Cheng H, Jiang H. Effect of tensile stress on
2015.11.028. the variation of magnetic field of low-alloy steel. J Magn Magn Mater
[7] Gollub P, Jensen JF, Giese D, Güres S. Flanged foundation connection of 2011;323(20):2474–7. http://dx.doi.org/10.1016/j.jmmm.2011.05.015.
the Offshore Wind Farm Amrumbank West - Concept, approval, design, tests [25] Leng J, Xu M, Zhou G, Wu Z. Effect of initial remanent states on the variation
and installation: Flanged foundation connection of the Offshore Wind Farm of magnetic memory signals. NDT & E Int 2012;52(Supplement C):23–7. http:
Amrumbank West - Concept, approval, design, tests and installation. Stahlbau //dx.doi.org/10.1016/j.ndteint.2012.08.009.
2014;83(8):522–8. http://dx.doi.org/10.1002/stab.201410178. [26] Ren S, Ren X, Duan Z, Fu Y. Studies on influences of initial magnetization
[8] Janele TO, Landskröner S, Klose M. Berechnung der Rammschädigung an state on metal magnetic memory signal. NDT & E Int 2019;103:77–83. http:
Flanschen von Monopile-Gründungen. Stahlbau 2015;84(7):494–500. http://dx. //dx.doi.org/10.1016/j.ndteint.2019.02.002.
doi.org/10.1002/stab.201510283.
[27] Brown WF. Irreversible magnetic effects of stress. Phys Rev 1949;75(1):147–54.
[9] Meijers PC, Tsouvalas A, Metrikine AV. A non-collocated method to quantify
http://dx.doi.org/10.1103/PhysRev.75.147.
plastic deformation caused by impact pile driving. Int J Mech Sci 2018;148:1–8.
[28] Jiles DC, Atherton DL. Theory of the magnetisation process in ferromagnets
http://dx.doi.org/10.1016/j.ijmecsci.2018.08.013.
and its application to the magnetomechanical effect. J Phys D: Appl Phys
[10] Makar JM, Atherton DL. Effect of uniaxial stress on the reversible and irreversible
1984;17(6):1265. http://dx.doi.org/10.1088/0022-3727/17/6/023.
permeabilities of 2% Mn pipeline steel. IEEE Trans Magn 1994;30(4):1380–7.
[29] Yao K, Wang ZD, Deng B, Shen K. Experimental research on metal mag-
http://dx.doi.org/10.1109/20.305536.
netic memory method. Exp Mech 2012;52(3):305–14. http://dx.doi.org/10.1007/
[11] Takahashi S, Kobayashi S, Tomáš I, Dupre L, Vértesy G. Comparison of magnetic
s11340-011-9490-3.
nondestructive methods applied for inspection of steel degradation. NDT & E Int
2017;91:54–60. http://dx.doi.org/10.1016/j.ndteint.2017.06.001. [30] Leng J, Liu Y, Zhou G, Gao Y. Metal magnetic memory signal response to plastic
[12] Maylin MG, Squire PT. The effects of stress on induction, differential permeability deformation of low carbon steel. NDT & E Int 2013;55(Supplement C):42–6.
and Barkhausen count in a ferromagnet. IEEE Trans Magn 1993;29(6):3499–501. http://dx.doi.org/10.1016/j.ndteint.2013.01.005.
http://dx.doi.org/10.1109/20.281209. [31] Li Z, Dixon S, Cawley P, Jarvis R, Nagy PB, Cabeza S. Experimental studies of
[13] Kleber X, Vincent A. On the role of residual internal stresses and dislocations on the magneto-mechanical memory (MMM) technique using permanently installed
Barkhausen noise in plastically deformed steel. NDT & E Int 2004;37(6):439–45. magnetic sensor arrays. NDT & E Int 2017;92(Supplement C):136–48. http:
http://dx.doi.org/10.1016/j.ndteint.2003.11.008. //dx.doi.org/10.1016/j.ndteint.2017.07.019.
[14] Wang Y, Liu X, Wu B, Xiao J, Wu D, He C. Dipole modeling of stress-dependent [32] Chadebec O, Coulomb J-L, Janet F. A review of magnetostatic moment method.
magnetic flux leakage. NDT & E Int 2018;95:1–8. http://dx.doi.org/10.1016/j. IEEE Trans Magn 2006;42(4):515–20. http://dx.doi.org/10.1109/TMAG.2006.
ndteint.2018.01.004. 870929.
[15] Babbar V, Bryne J, Clapham L. Mechanical damage detection using magnetic [33] Atherton D, Welbourn C, Jiles D, Reynolds L, Scott-Thomas J. Stress-induced
flux leakage tools: Modeling the effect of dent geometry and stresses. NDT & E magnetization changes of steel pipes–Laboratory tests, Part II. IEEE Trans Magn
Int 2005;38(6):471–7. http://dx.doi.org/10.1016/j.ndteint.2004.12.002. 1984;20(6):2129–36. http://dx.doi.org/10.1109/TMAG.1984.1063572.
[16] Atherton DL, Teitsma A. Detection of anomalous stresses in gas pipelines by [34] Makar JM, Atherton DL. Effects of isofield uniaxial cyclic stress on the magne-
magnetometer survey (invited). J Appl Phys 1982;53(11):8130–5. http://dx.doi. tization of 2% Mn pipeline steel - behavior on minor hysteresis loops and small
org/10.1063/1.330316. major hysteresis loops. IEEE Trans Magn 1995;31(3):2220–7. http://dx.doi.org/
[17] Li Z, Jarvis R, Nagy PB, Dixon S, Cawley P. Experimental and simulation methods 10.1109/20.376241.
to study the Magnetic Tomography Method (MTM) for pipe defect detection. NDT [35] Gilman JJ. Microdynamical theory of plasticity. In: McMahon CJ, editor.
& E Int 2017;92:59–66. http://dx.doi.org/10.1016/j.ndteint.2017.07.018. Advances in materials research, Vol. 2. New York: Interscience Publishers; 1968.

7
P.C. Meijers et al. International Journal of Mechanical Sciences 217 (2022) 106990

[36] Sablik MJ, Yonamine T, Landgraf FJG. Modeling plastic deformation effects in [39] Wang ZD, Deng B, Yao K. Physical model of plastic deformation on magnetization
steel on hysteresis loops with the same maximum flux density. IEEE Trans Magn in ferromagnetic materials. J Appl Phys 2011;109(8):083928. http://dx.doi.org/
2004;40(5):3219–26. http://dx.doi.org/10.1109/TMAG.2004.832763. 10.1063/1.3574923.
[37] Karagiozova D, Jones N. Dynamic elastic–plastic buckling of circular cylindrical [40] Liu LL, Feng WJ, Ma P. A penny-shaped magnetically dielectric crack in a
shells under axial impact. Int J Solids Struct 2000;37(14):2005–34. http://dx. magnetoelectroelastic cylinder under magnetoelectromechanical loads. ZAMM -
doi.org/10.1016/S0020-7683(98)00343-6. J Appl Math Mech / Z Angew Math Mech 2015;(2):12.
[38] Beleggia M, Vokoun D, De Graef M. Demagnetization factors for cylindrical [41] Khurshudyan AZ. Min(max)imization of horizontal and vertical displacements of
shells and related shapes. J Magn Magn Mater 2009;321(9):1306–15. http: a fibre-reinforced magneto-elastic cantilever rod. ZAMM - J Appl Math Mech
//dx.doi.org/10.1016/j.jmmm.2008.11.046. / Z Angew Math Mech 2018;98(11):1924–9. http://dx.doi.org/10.1002/zamm.
201800192.