geosciences

Article

Numerical
Article Modelling of Double-Twisted Wire Mesh for
Numerical Modelling of Double-Twisted Wire Mesh for
Low-Energy Rockfall Catch Fences
Low-Energy Rockfall Catch Fences
Hassan Al-Budairi 1,1,*, Zhiwei Gao 2 and Andrew Steel 1
Hassan Al-Budairi * , Zhiwei Gao 2 and Andrew Steel 1
1 QTS Group Ltd., Glasgow ML10 6QJ, UK; andrewsteel@qtsgroup.com
21 James Watt School
QTS Group of Engineering,
Ltd., Glasgow University
ML10 6QJ, UK of Glasgow, Glasgow G12 8QQ, UK;
2 zhiwei.gao@glasgow.ac.uk
James Watt School of Engineering, University of Glasgow, Glasgow G12 8QQ, UK
** Correspondence:
Correspondence:hassanalbudairi@qtsgroup.com
hassanalbudairi@qtsgroup.com

Abstract:
Abstract: Low-energy
Low-energy rockfall
rockfall catch
catch fences
fences areare designed
designed to to protect
protect infrastructure
infrastructure suchsuch as
as railways
railways
and
and roads
roads wherein the kinetic
kinetic energy
energyof offalling
fallingrocks
rocksisisless
lessthan
than100100kJ.
kJ.The
Thetypical
typical design
design consists
consists of
of a double-twisted
a double-twisted steelwire
steel wiremesh
meshsupported
supportedby byground
groundpostsposts and
and strengthened by anchoring
anchoring wirewire
ropes.
ropes. The
The fence
fence stops
stops falling
falling rocks
rocks by
by dissipating
dissipating the the impact
impact energy
energy mainly
mainly through
through elastoplastic
elastoplastic
stretching
stretching of
of steel
steel wires
wires inin the
the mesh.
mesh. InIn this
this study,
study, aa three-dimensional
three-dimensional finitefinite element
element model
model forfor
double-twisted
double-twisted wirewire mesh
mesh waswas developed
developed in Abaqus/Explicit.
Abaqus/Explicit.The Themodel
modelhas has been
been verified
verified using
using
both
both quasi-static
quasi-static loading
loading andand impact
impact tests.
tests. ItIt was
was found
found thatthat proper
proper geometrical
geometrical representation
representation is is
essential for accurate simulation of wire deformation modes and the interaction
essential for accurate simulation of wire deformation modes and the interaction between double- between double-
twisted
twisted wires.
wires. The
Themodel
modelalsoalsoenables
enablesthetheapplication
application of ofthe
thereal
realstress–strain
stress–strain relationship
relationship of
of aasingle
single
steel
steel wire
wire in
in constitutive
constitutive models.
odels.

Keywords:
Keywords: rockfall; low-energy
rockfall; barriers;
low-energy double-twisted
barriers; wire mesh;
double-twisted wiredynamic
mesh; finite element
dynamic modelling
finite element
modelling

1.
1. Introduction
Introduction
Citation: Al-Budairi,H.;
Citation: Al-Budairi, H.;Gao,
Gao,Z.;
Z.; The
The records
records of the national landslide
landslide database
database provided
provided by by the British
British Geological
Geological
Steel, A. Numerical Modelling of
Steel, A. Numerical Modelling of Survey (BGS) forfor Great
Great Britain
Britain [1]
[1] show
show that
that there
there has
has been
been aa significant
significant rise
rise in
in the
thenumber
number
Double-TwistedWire
Double-twisted WireMesh
Meshfor
for
of rockfall
of rockfallincidents
incidentsininthe
theUK
UKinin recent
recent years
years (Figure
(Figure 1). many
1). In In many locations
locations alongalong UK
UK rail-
Low-Energy Rockfall Catch Fences.
Low-Energy Rockfall Catch Fences. railways, there is a significant risk of falling rocks reaching the infrastructure,
ways, there is a significant risk of falling rocks reaching the infrastructure, causing the causing the
Geosciences 2023,
Geosciences 13, x.
2023, 13, 180. https://
derailment of trains, and damage to tracks and signalling systems [2].
doi.org/10.3390/geosciences13060180
https://doi.org/10.3390/xxxxx

Academic Editors:
Academic Editors: Jesus
Martinez-Frias,
Jesus Daniele
Martinez-Frias, Peila,Peila,
Daniele
Stéphane Lambert
Stéphane Lambert and
Anna
and Giacomini
Anna Giacomini

Received: 16
Received: 16February
February2023
2023
Revised: 5 May 2023
Revised:5 May 2023
Accepted: 31
Accepted: 31 May
May 2023
2023
Published: 15
Published: dateJune 2023

Copyright: © 2023 by the authors.

Copyright: © 2023 by the authors.
Submitted for possible
Licensee MDPI, Basel, open access
Switzerland.
publication under
This article is the access
an open terms article
and
conditions of the Creative Commons
distributed under the terms and
Attribution (CCCreative
conditions of the BY) Commons
license
(https://creativecommons.org/license
Attribution (CC BY) license (https://
s/by/4.0/). Figure
creativecommons.org/licenses/by/ Figure 1.
1. Number
Number of
of major
major rockfalls
rockfalls in
in the
the UK,
UK, as
as recorded
recorded by the BGS
by the BGS [1].
[1].
4.0/).

Geosciences 2023, 13, x. https://doi.org/10.3390/xxxxx www.mdpi.com/journal/geosciences

Geosciences 2023, 13, 180. https://doi.org/10.3390/geosciences13060180 https://www.mdpi.com/journal/geosciences

Geosciences 2023, 13, 180 2 of 12
Geosciences 2023, 13, x FOR PEER REVIEW 2 of 13

Rockfall hazards can be mitigated by various means, including catch fences, which
Rockfall hazards can be mitigated by various means, including catch fences, which
consist of steel wire meshes, cables, and ground supporting posts, etc. (see Figure 2).
consist of steel wire meshes, cables, and ground supporting posts, etc. (see Figure 2). Com-
Compared to other rockfall mitigation methods, catch fences offer many advantages,
pared to other
including rockfall
a shorter mitigationtime,
construction methods,
reduced catch fences
visual offer many
impact, advantages,
and lower including
maintenance costs.
aCatch
shorter construction time, reduced visual impact, and lower maintenance
fences can be classified as high (or low)-energy dissipation systems based on costs. Catch
their
fences cantobedissipate
capacity classifiedtheas energy
high (oroflow)-energy
rockfalls abovedissipation systems
(or below) based energy
an impact on theirthreshold
capacity
to
ofdissipate
100 kJ. Inthe energyhigh-energy
general, of rockfalls above (or below)
catch fences an impact energy
are semi-rigid, permanentthreshold of 100and
structures, kJ.
In general, high-energy catch fences are semi-rigid, permanent structures,
are designed according to established guidelines [3,4]. These high-cost systems are not and are de-
signed
normallyaccording
requiredtoinestablished
the UK, whereguidelines [3,4]. These
most rockfall high-cost
incidents may besystems are as
classified notlow-energy.
normally
required
Under these conditions, a non-permanent catch fence system, as shown in FigureUnder
in the UK, where most rockfall incidents may be classified as low-energy. 2, can
these
offer conditions, a non-permanent
adequate protection at a lowercatch
cost.fence system, as shown
This lightweight systemincanFigure 2, can offer
be installed ade-
manually,
quate protection
without recourse at to
a lower cost. This
mechanised lightweight
plants. systemthe
It dissipates canimpact
be installed
energymanually,
of fallingwithout
objects
recourse
throughtoplastic
mechanised plants. It
deformation ofdissipates
the flexiblethewire
impact
meshenergy
andof falling
the posts,objects
which through plastic
must then be
deformation
replaced. of the flexible wire mesh and the posts, which must then be replaced.

Figure
Figure2.2.Components
Componentsof
ofaalow-energy
low-energyrockfall
rockfallcatch
catchfence.
fence.

The
The design low-energycatch
design of low-energy catchfences,
fences,which
which have
have received
received far far
lessless attention
attention [5–7][5–7]
than
than high-energy
high-energy systems,
systems, is based
is based primarily
primarily on fieldontests
fieldand
tests and engineering
engineering judgements.
judgements. Numer-
Numerical
ical modelsmodels for high-energy
for high-energy catchhave
catch fences fences have perforce
perforce paid particular
paid particular attentionattention
to materialto
material failure mechanisms
failure mechanisms [8–10] and [8–10] and dissipation
energy energy dissipation
capacities capacities
[11–14]. [11–14].
Although Although
similar
numerical
similar modelling
numerical approaches
modelling have been
approaches haveused
beentoused
simulate low-energy
to simulate systems
low-energy [15,16],
systems
severalseveral
[15,16], issues require further further
issues require investigation.
investigation.
First,in
First, inthe
thestudies
studiescited
citedabove,
above,thethemodelling
modellingof ofdouble-twisted
double-twistedwire wiremesh
meshhas hasbeen
been
simplified
simplified to to reduce the difficulties and computing time of geometrical
difficulties and computing time of geometrical modelling. Dou- modelling. Double-
twisted wire
ble-twisted mesh
wire meshconsists of symmetrically
consists of symmetrically woven
woven hexagonal
hexagonal cells, wherein
cells, whereineach cellcell
each has
fourfour
has sections (sides)
sections of single
(sides) wireswires
of single and two
andsections of twisted
two sections wires, as
of twisted shown
wires, as in Figurein3.
shown
The structure
Figure of the cellsofaffects
3. The structure the cellsthe affects
macroscopic strength and
the macroscopic flexibility
strength andofflexibility
the wholeofmesh. the
When mesh.
whole a wire When
mesh isa subjected
wire meshtoisan impact, the
subjected wires
to an in thethe
impact, double-twisted sections stretch
wires in the double-twisted
and twist
sections simultaneously.
stretch The impact energy
and twist simultaneously. The is dissipated
impact energy through frictionthrough
is dissipated betweenfric- the
double-twisted wires and through elastoplastic stretching of individual
tion between the double-twisted wires and through elastoplastic stretching of individual wires. If the impact
load exceeds
wires. the material
If the impact strength,
load exceeds the single-wire sections
material strength, become damaged,
single-wire which causes
sections become dam-
the unravelling of the adjacent wires, which itself degrades
aged, which causes the unravelling of the adjacent wires, which itself degrades the response of nearby thecells.
re-
This zone
sponse of degradation
of nearby cells. Thisincreases
zone ofwith increasingincreases
degradation impact load.withThese interactions
increasing impactcannotload.
be simulated
These interactionsaccurately
cannotifbe the finite element
simulated (FE) model
accurately of theelement
if the finite double-twisted
(FE) model section
of theis
represented using a single-beam element [8,15]. A similar issue
double-twisted section is represented using a single-beam element [8,15]. A similar issue occurs when representing
the terminal
occurs when ends of the twisted
representing sectionsends
the terminal withofdiscontinuous nodes, when
the twisted sections employing the
with discontinuous
discrete element approach [9,10,13,14,16].
nodes, when employing the discrete element approach [9,10,13,14,16].
Geosciences 2023, 13, x FOR PEER REVIEW 3 of 13
Geosciences 2023, 13, 180 3 of 12

Figure 3. A
Figure 3. A single
single cell
cell in
in aa double-twisted
double-twisted hexagonal
hexagonal wire
wire mesh.
mesh.

A second issue relates to the material modelling of the steel wires. Simplified models
A second issue relates to the material modelling of the steel wires. Simplified models
that define a failure criterion in terms of global measures, such as the maximum deforma-
that define a failure criterion in terms of global measures, such as the maximum defor-
tion of the mesh [3,4], risk failing to capture the important elements of system response,
mation of the mesh [3,4], risk failing to capture the important elements of system response,
particularly in cases wherein the impact scenario differs from the tests used for parame-
particularly in cases wherein the impact scenario differs from the tests used for parameter
ter calibration purposes [15]. A proper constitutive model for describing the mechanical
calibration purposes [15]. A proper constitutive model for describing the mechanical re-
response of a single wire is needed.
sponse of a single wire is needed.
In this paper, a detailed analysis of a double-twisted hexagonal wire mesh is presented.
In this paper, a detailed analysis of a double-twisted hexagonal wire mesh is pre-
A direct modelling approach in the three-dimensional space is used to create the wire mesh
sented. A direct modelling approach in the three-dimensional space is used to create the
model, which is then analysed using Abaqus/Explicit [17]. The material constitutive model
wire
is mesh
based on model, which is
experimental then
tests onanalysed usingcut
single wires Abaqus/Explicit [17].meshes.
out from the wire The material consti-
In adopting
tutive model is based on experimental tests on single wires cut out from the wire
this rigorous approach, we aim to better understand the response of low-energy catch meshes.
In adopting
fences this rigorous
to impact, and therebyapproach, we aimtotoimproving
to contribute better understand the response of low-en-
design methods.
ergy catch fences to impact, and thereby to contribute to improving design methods.
2. Numerical Modelling of the Double-Twisted Wire Mesh
2. Numerical Modelling of
A three-dimensional the Double-Twisted
model Wire
of a double-twisted Mesh wire mesh has been anal-
hexagonal
ysed A three-dimensional
using Abaqus/Explicit model
[17].ofFull
a double-twisted
advantage was hexagonal
taken ofwire mesh has to
its capability been ana-
handle
lysed using
highly Abaqus/Explicit
non-linear geometric and [17]. Full advantage
material responseswas under taken of itsloading
dynamic capability to handle
conditions.
highly non-linear geometric and material responses under dynamic loading conditions.
2.1. Geometrical Representation
2.1. Geometrical
An explicitRepresentation
model of Maccaferri [18] double-twisted hexagonal 2.7 mm diameter
wire mesh
An explicit model ofwith
was created, cell dimensions
Maccaferri of 80 mm ×
[18] double-twisted 100 mm, 2.7
hexagonal referred to as P8/2.7.
mm diameter wire
SolidWorks
mesh was created, with cell dimensions of 80 mm × 100 mm, referred to as P8/2.7. wire
CAD software [19] was used to create the geometry of each twisted
separately,
SolidWorkswhichCAD was then exported
software into Abaqus.
[19] was used to create the geometry of each twisted wire sep-
As each
arately, whichwire
wasconsists of straight
then exported and twisted sections, its geometry can be created
into Abaqus.
from As connected
each wire straight andoftwisted
consists straightsegments
and twistedto form a 3D sketch
sections, path. This
its geometry can simplified
be created
approach creates a discrete geometrical representation of the wires,
from connected straight and twisted segments to form a 3D sketch path. This simplified as shown in Figure ap-
4a.
This leads to convergence difficulties and inaccuracies in the calculation
proach creates a discrete geometrical representation of the wires, as shown in Figure 4a. Thisof the directional
components when thedifficulties
leads to convergence finite element
and model is created.
inaccuracies in theThis is because
calculation the directional
of the normal sections
com-
and cross-sections of neighbouring wire elements are not collinear
ponents when the finite element model is created. This is because the normal sectionsat the connection points.
and
Consequently,
cross-sections wires are created wire
of neighbouring as a single
elementscontinuous geometryatwith
are not collinear the no interconnection
connection points.
points, which represents
Consequently, wires are the wireas
created mesh length,
a single as showngeometry
continuous in Figurewith4b. no interconnection
In practice,
points, wires are the
which represents coated
wirewith
mesh a PVC layer
length, as to enhance
shown their weather
in Figure 4b. resistance; this
layer is not modelled geometrically, but its effect on frictional resistance can be considered.
When a double-twisted mesh panel is subjected to impact, its wires, particularly those
in the double-twisted sections, stretch, twist and bend. To reproduce these deformations in
the numerical model, the Timoshenko beam element was used. This element can simulate
three modes of deformation: axial stretching, bending (change in curvature), and twisting
in three-dimensional space. This element has an advantage over the other elements, having
simpler geometry and better computational efficiency.
Geosciences2023,
Geosciences 13,x180
2023,13, FOR PEER REVIEW 4 4ofof13
12

Geometricalrepresentations
Figure4.4.Geometrical
Figure representationsof
ofaadouble-twisted
double-twistedhexagonal
hexagonalwire
wire mesh.
mesh.

Inpractice,
In addition,wires
Abaqus
are has a powerful
coated and robust
with a PVC layer tobeam-to-beam
enhance their contact
weatheralgorithm
resistance;which
this
simulates the contact between the surfaces of beams rather than their centrelines. This is
layer is not modelled geometrically, but its effect on frictional resistance can be considered.
achieved by specifying the radius of the beam (wire). The algorithm uses penalty-based
When a double-twisted mesh panel is subjected to impact, its wires, particularly
contact constraints to enforce the contact behaviour between the surfaces. The penalty-
those in the double-twisted sections, stretch, twist and bend. To reproduce these defor-
based contact algorithm works by applying a penalty force to the nodes of the beam element
mations in the numerical model, the Timoshenko beam element was used. This element
that are in contact with the surface with which the beam is in contact. However, a modelling
can simulate three modes of deformation: axial stretching, bending (change in curvature),
difficulty is that Abaqus assumes that these beam elements are initially straight, which
and twisting in three-dimensional space. This element has an advantage over the other
complicates the representation of the double-twisted sections. To resolve this difficulty, a
elements, having simpler geometry and better computational efficiency.
curvature control feature in Abaqus, which increases the number of elements wherein the
In addition, Abaqus has a powerful and robust beam-to-beam contact algorithm
curvature is high, is used. It calculates the seed distribution based on the curvature of the
which simulates the contact between the surfaces of beams rather than their centrelines.
edge along with the target element size. With this tool, users can specify the deviation factor,
This is achieved by specifying the radius of the beam (wire). The algorithm uses penalty-
which is a measure of how much the element edges deviate from the original geometry. As
based contact constraints
the deviation to enforce
factor reduces, the contact
the number behaviour
of elements between the surfaces.
that Abaqus/CAE The pen-
create around the
alty-based
curve will contact
increase.algorithm worksstudy
A convergence by applying a penalty
was conducted to force to the
identify the nodes
optimumof the beam
deviation
element
factor forthat
theare in contact with
double-twisted themesh
wire surface
andwith which the(straight)
the optimum beam is in contact.
element However,
size.
a modelling difficulty is that Abaqus assumes that these beam elements are initially
Geosciences 2023, 13, x FOR PEER REVIEWA typical model of a hexagonal double-twisted wire mesh panel with an optimal 5 of 13
straight, which complicates the representation of the double-twisted sections.
element size and element distribution is shown in Figure 5. The edge wire (selvedge To resolve
wire)
this difficulty,
is also a curvature
presented, and has acontrol
larger feature
diameterinthan
Abaqus, whichitself.
the mesh increases the number of ele-
ments wherein the curvature is high, is used. It calculates the seed distribution based on
the curvature of the edge along with the target element size. With this tool, users can spec-
ify the deviation factor, which is a measure of how much the element edges deviate from
the original geometry. As the deviation factor reduces, the number of elements that
Abaqus/CAE create around the curve will increase. A convergence study was conducted
to identify the optimum deviation factor for the double-twisted wire mesh and the opti-
mum (straight) element size.
A typical model of a hexagonal double-twisted wire mesh panel with an optimal el-
ement size and element distribution is shown in Figure 5. The edge wire (selvedge wire)
is also presented, and has a larger diameter than the mesh itself.

Figure5.
Figure 5. Geometric
Geometric modelling
modelling of
of aa double-twisted
double-twistedhexagonal
hexagonalwire
wiremesh
meshpanel.
panel.

2.2. Constitutive Model for Single Steel Wires

An elastoplastic constitutive model is considered in Abaqus/Explicit to describe the
material behaviour of the double-twisted wire mesh. The model also accounts for material
failure, wherein damage initiation and damage evaluation criteria were implemented.
To determine the necessary parameters for the elastoplastic constitutive and material
Geosciences 2023, 13, 180 5 of 12
Figure 5. Geometric modelling of a double-twisted hexagonal wire mesh panel.

2.2. Constitutive Model for Single Steel Wires

2.2. Constitutive Model constitutive
An elastoplastic for Single Steel Wiresis considered in Abaqus/Explicit to describe the
model
material behaviour ofconstitutive
An elastoplastic the double-twisted
model iswire mesh. The
considered model also accounts
in Abaqus/Explicit for material
to describe the
failure, wherein
material behaviour damage initiation and damage
of the double-twisted wire mesh.evaluation criteria
The model alsowere implemented.
accounts for material
failure,
Towherein
determine damage initiationparameters
the necessary and damage forevaluation criteriaconstitutive
the elastoplastic were implemented.
and material
To determine
failure the necessary
model, a series of uniaxialparameters for the
tensile tests elastoplastic
were carried outconstitutive and material
on single wires using a
failure
uniaxialmodel,
testinga machine
series of (Instron
uniaxial3369)
tensileandtests were carried out on
a video-extensometer single wires
(LIMESS RTSS)using a
for op-
uniaxial testing machine (Instron 3369) and a video-extensometer (LIMESS
tical strain measurements. The preparation of the test specimens and the test setup are RTSS) for optical
strain
shown measurements.
in Figure 6. The Theeffective
preparation of the
length of test
the specimens
specimens andwasthe test setup
similar to theare shown
single in
wire
Figure
sections6. in
The effective
the length
mesh cells. Theoftest
the specimens was weresimilar to theasingle
taken from mesh wire
panelsections in the
and straight-
mesh
ened cells. The test
carefully, specimens
in order to avoidwere taken the
twisting from60amm mesh panel
test and(highlighted
section straightenedin carefully, in
blue). The
order to avoid were
bent sections twisting the 60by
gripped mmthetest section
tensile test(highlighted in blue).
machine, which The bent
increases thesections
contactwere
area
gripped
betweenby the
the tensile
wires andtest
themachine,
grips to which
prevent increases the contact
wire slippage and area between
material theoccurrence
failure wires and
the grips to
near the grips.prevent wire slippage and material failure occurrence near the grips.

Figure6.6.Uniaxial
Figure Uniaxialtensile
tensiletests
testson
onsingle
singlewire
wirespecimens
specimensof
ofdouble-twisted
double-twistedhexagonal
hexagonalmesh.
mesh.

Thematerial
The material damage
damage initiation
initiation criterion
criterion and
and the
thedamage
damageevolution
evolution law
lawdescribe
describethe
the
progressivedegradation
progressive
Geosciences 2023, 13, x FOR PEER REVIEW degradationinin material
material stiffness
stiffness up up to the
to the point
point of material
of material failure,
failure, as6 illus-
as illustrated
of 13
trated
by by the typical
the typical uniaxial
uniaxial stress–strain
stress–strain response
response shownshown in Figure
in Figure 7. The 7. The following
following ele-
elements
ments
are are required
required tothis
to define define thisin
model model in Abaqus:
Abaqus:
11 The effective
effective (undamaged) material response (a-b-c-d’). This can be extracted
extracted from
from
the uniaxial tensile test data of single wires.

Figure 7. Uni-axial elastic–plastic material response, with progressive damage.

Figure 7. Uni-axial elastic–plastic material response, with progressive damage.

2 The damage initiation criterion (Point c in Figure 7). Abaqus provides a phenomeno-
logical ductile damage criterion to predict the onset of damage due to nucleation,
growth, and the coalescence of voids. The model assumes that the equivalent plastic
strain at the onset of damage 𝜀 (η, 𝜀 ), is a function of stress triaxiality, η, and the
Geosciences 2023, 13, 180 6 of 12

2 The damage initiation criterion (Point c in Figure 7). Abaqus provides a phenomeno-
logical ductile damage criterion to predict the onset of damage due to nucleation,
growth, and the coalescence of voids. The
 model assumes that the equivalent plastic
. pl

strain at the onset of damage ε f η, ε , is a function of stress triaxiality, η, and the
. pl
equivalent plastic strain rate, ε . The stress triaxiality is the ratio of hydrostatic stress,
p, to the equivalent tensile (von Mises) stress, σv .

p (σ1 + σ2 + σ3 )/3
η= = q (1)
σv
[(σ1 − σ2 )2 + (σ1 − σ3 )2 + (σ2 − σ3 )2 ]/2
where σ1 , σ2 and σ3 are the principal stresses. For beam element, the principal stresses σ2
and σ3 are null; thus, η = 1/3 according to Equation (1).
Ductile damage takes place when the following condition is satisfied:

dε p
Z
wD =  .  =1
pl
(2)
ε f η, ε

where w D is a state variable that increases monotonically with plastic deformation. Exper-
. pl

imental test data provide the equivalent plastic strain at the onset of damage, ε f η, ε ,
whereas the equivalent plastic strain, ε p , is
r
p 2 p p
ε = ε :ε (3)
3 ij ij
p
where ε ij is the plastic strain tensor. The strain data should be obtained from tests that have
similar loading conditions to the intended application.
3 The damage evolution law (c–d in Figure 7). To reproduce the material response after
damage initiation, it is assumed that the damage is characterized by the progressive
degradation of the material stiffness and yield stress. Abaqus uses a damage variable,
D, to capture the combined effect of all damage mechanisms, as shown in Figure 7.
The damage variable has a range of values from zero (no damage) to unity (material
failure). The softening of the material stress tensor, σij , is given by

σij = (1 − D )σij (4)

where σij is the material stress tensor without material damage.
To alleviate the mesh dependency that arises from strain localisation during progres-
sive damage, Hillerborg’s fracture energy criterion [17] is used in Abaqus to determine
the response after damage initiation. This criterion makes use of the equivalent plastic
displacement, u pl (which is zero before damage initiation), where

u pl = Lε pl (5)

in which L is the characteristic length of the element, and ε pl is the equivalent plastic strain
without material damage. The characteristic length depends on the element geometry;
for beam elements, it is the length between the nodes along the element axis. In addition,
elements are removed from the model when the damage variable D is equal to unity.
Nominal stress–strain data from three uniaxial tests carried out at a 2 mm/s strain rate
are used, while the strain rate effects are not explicitly incorporated in this study. The mean
nominal stress–strain curve was calculated from these data and used to calculate the true
stress–true plastic strain data required by Abaqus. Thus,

σtrue = σnom (1 + ε nom ) (6)

 the true stress–true plastic strain data required by Abaqus. Thus,
𝜎 =𝜎 (1 + 𝜀 ) (6)

𝜀 = ln (1 + 𝜀 ) (7)
Geosciences 2023, 13, 180 7 of 12
where 𝜎 and 𝜀 are the true stress and the true plastic strain, respectively, while
𝜎 and 𝜀 are the nominal (engineering) stress and strain, respectively.
p
3. Model Calibration and Validation
ε true =Results
ln(1 + ε nom ) (7)
p
The
where numerical
σtrue and ε true model
are the was
true calibrated
stress and theand validated
true by undertaking
plastic strain, respectively,comparisons
while σnom with
the
andexperimental data reported
ε nom are the nominal in the
(engineering) literature,
stress alongside
and strain, full-scale tests conducted as
respectively.
part of this study.
3. Model Calibration and Validation Results
The numerical
3.1. Finite model wasofcalibrated
Element Modelling and validated by undertaking comparisons with
the Wire Response
the experimental data reported in the literature, alongside full-scale tests conducted as part
The
of this model’s predictions have been compared with the published experimental data
study.
[9,15] obtained from static in-plane tensile load tests on mesh panels of dimensions 320 ×
3.1.mm,
580 Finitewith
Element
80 ×Modelling
100 mmofhexagonal
the Wire Response
double-twisted cells and a 2.7 mm wire diameter.
In theseThetests,
model’s predictions
the wire have been
mesh panel compared to
was subjected with the published
a uniform tensileexperimental
load along the top
data [9,15] obtained from static in-plane tensile load tests on mesh
edge at a constant loading rate of 10 mm/min. The lower edge of the panel panels of dimensions
was fully re-
320 × 580 mm, with 80 × 100 mm hexagonal double-twisted cells and
strained, and an additional four points (near the vertical sides) were restraineda 2.7 mm wirefrom hor-
diameter. In these tests, the wire mesh panel was subjected to a uniform tensile load along
izontal movement. In the model, additional constraints were imposed at the ends of wires
the top edge at a constant loading rate of 10 mm/min. The lower edge of the panel was
tofully
prevent them from unravelling, as shown in Figure 8.
restrained, and an additional four points (near the vertical sides) were restrained from
horizontal movement. In the model, additional constraints were imposed at the ends of
wires to prevent them from unravelling, as shown in Figure 8.

Figure 8. Numerical simulation of tensile tests on wire mesh panels [9,15].

Figure 8. Numerical simulation of tensile tests on wire mesh panels [9,15].
The vertical reaction force at the lower edge is plotted against the displacements of
the upper edge for three experiments, as shown in Figure 9. The model overestimates
The vertical reaction force at the lower edge is plotted against the displacements of
the reaction force at the initial stage of the test, but this is almost certainly due to initial
the upper edge for three experiments, as shown in Figure 9. The model overestimates the
slackness before the full engagement of the mesh. The numerical model is in excellent
reaction
agreementforce
withatthe
the initial
data stage of the test, but this is almost certainly due to initial
thereafter.
A more challenging test of the model is provided by comparing its predictions with
those obtained from out-of-plane experimental tests reported in the literature [10,18,20].
In Ref. [10], a 3 × 3 m mesh panel with 80 × 100 mm hexagonal double-twisted cells
manufactured from 3 mm diameter wire (P8/3) was supported horizontally by a rigid
structure, and fully constrained on all four edges. The panel was subjected to vertical load
by a hemi-spherical plate which was placed underneath the centre of the panel and pulled
upwards at a constant velocity of 10 mm/s. The loading was continued until the onset of
material damage.
In Figure 10, the model prediction of the deformation of the panel is shown in a state
of maximum deformation, just before the onset of material failure.
Geosciences 2023, 13, x FOR PEER REVIEW 8 of 13

Geosciences 2023, 13, 180 slackness before the full engagement of the mesh. The numerical model is in excellent
8 of 12
agreement with the data thereafter.

Figure 9. Comparison between experimental data [9,15] and the numerical model for uniaxial load-
ing of a wire-mesh panel.

A more challenging test of the model is provided by comparing its predictions with
those obtained from out-of-plane experimental tests reported in the literature [10,18,20].
In Ref. [10], a 3 × 3 m mesh panel with 80 × 100 mm hexagonal double-twisted cells man-
ufactured from 3 mm diameter wire (P8/3) was supported horizontally by a rigid struc-
ture, and fully constrained on all four edges. The panel was subjected to vertical load by
a hemi-spherical plate which was placed underneath the centre of the panel and pulled
upwards at a constant velocity of 10 mm/s. The loading was continued until the onset of
material damage.
In Figure 10, the model prediction of the deformation of the panel is shown in a state
Figure 9. Comparison between experimental data [9,15] and the numerical model for uniaxial loading
Figure
of 9. Comparison
maximum between just
deformation, experimental data
before the [9,15]
onset ofand the numerical
material failure. model for uniaxial load-
of a wire-mesh panel.
ing of a wire-mesh panel.

A more challenging test of the model is provided by comparing its predictions with
those obtained from out-of-plane experimental tests reported in the literature [10,18,20].
In Ref. [10], a 3 × 3 m mesh panel with 80 × 100 mm hexagonal double-twisted cells man-
ufactured from 3 mm diameter wire (P8/3) was supported horizontally by a rigid struc-
ture, and fully constrained on all four edges. The panel was subjected to vertical load by
a hemi-spherical plate which was placed underneath the centre of the panel and pulled
upwards at a constant velocity of 10 mm/s. The loading was continued until the onset of
material damage.
In Figure 10, the model prediction of the deformation of the panel is shown in a state
Figure 10. Model simulation of punching test on a wire mesh panel [10].
of maximum deformation, just before the onset of material failure.
The experimental data and the model’s predictions of vertical reaction force to the
Geosciences 2023, 13, x FOR PEER REVIEW 9 of 13
vertical displacement at the top of the spherical plate are plotted in Figure 11. The results
Figure 10. Model
show that simulation
the model of punching
satisfactorily test onthe
predicts a wire mesh panel [10]. response of the panel.
load–deformation

The experimental data and the model’s predictions of vertical reaction force to the
vertical displacement at the top of the spherical plate are plotted in Figure 11. The results
show that the model satisfactorily predicts the load–deformation response of the panel.

Figure 10. Model simulation of punching test on a wire mesh panel [10].

The experimental data and the model’s predictions of vertical reaction force to the
vertical displacement at the top of the spherical plate are plotted in Figure 11. The results
show that the model satisfactorily predicts the load–deformation response of the panel.
Figure 11.
Figure Comparisonbetween
11. Comparison betweenexperimental
experimental[10]
[10] and
and predicted
predicted load–displacement
load–displacement data
data for
for aa
punching test.
punching test.

For comparison
For comparison purposes,
purposes, the
the model’s
model’s predictions
predictionsfor
foraa2.7
2.7mm
mmdiameter
diameterwire
wiremesh
meshis
also
is shown
also shown inin
Figure 11.11.
Figure AsAsexpected,
expected,this mesh
this responds
mesh responds more
moreflexibly toto
flexibly the load.
the load.
For this grade of wire, Table 1 shows that the model predictions of punching load
strength, and the mesh deformation at that point, are in excellent agreement with the ex-
perimental data reported in References [18,20].
Geosciences 2023, 13, 180 9 of 12

For this grade of wire, Table 1 shows that the model predictions of punching load
strength, and the mesh deformation at that point, are in excellent agreement with the
experimental data reported in References [18,20].

Table 1. Comparison of punching test results [18,20] for P8/2.7 wire mesh panel.

Maccaferri Wire
Mesh Specifications Model Prediction Difference
Data Sheet for P8/2.7
Punching test strength (kN) 70 62.3 11%
Punching displacement (mm) 514 523 1.7%

3.2. Dynamic Impact Loading Tests

In general, impact tests are conducted by dropping masses onto wire meshes that
are fully constrained on all four sides [5,8–10]. This test arrangement over-constrains the
wire mesh and cannot reproduce the response of wire meshes in the field, where they are
strongly constrained by the posts but only partially restrained longitudinally by wire ropes
(Figure
Geosciences 2023, 13, x FOR PEER REVIEW 2). To overcome this limitation, in this study, an impact testing rig was designed
10 of 13
and used to laterally test the constrained wire mesh panels under more realistic conditions,
as shown in Figures 12 and 13.

Figure 12.
Figure Schematic of
12. Schematic of the
the impact
impact test
test rig.
rig.

Various panel lengths and boulder masses were employed. The test rig has two
adjustable horizontal crossbeams which support the panel, and a high-speed camera
(500 fps) is used to capture the boulder velocity and the wire mesh deformation. Two
additional cameras are used to record the lateral deformation of the panel and the deflection
of the supporting beams.
In this paper, we report test data obtained from impact tests conducted using spherical
concrete “boulders” of 100 kg mass (test 1) and 200 kg mass (test 2) on 2 × 2 m wire mesh
panels. The boulders were lifted to a height of 2.5 m above the mesh, which produced a
velocity of 7 m/s at impact. The camera data were collected and analysed using Tracker, a
video analysis and modelling tool [21].
Geosciences 2023, 13, 180 10 of 12

Figure 12. Schematic of the impact test rig.

Geosciences 2023, 13, x FOR PEER REVIEW 11 of 13

produced a velocity of 7 m/s at impact. The camera data were collected and analysed using
Figure 13.aThe
Figure 13.
Tracker, video
The test rig
rig with
with front
testanalysis and and
and side
side views
frontmodelling toolof
views the
the boulder
[21].
of boulder after
afterimpact.
impact.
In
In the literature, self-weight deflection has been modelled usingpre-tension
the literature, self-weight deflection has been modelled using pre-tensionloads loads[9],
[9],
but Various panel lengths and boulder masses were employed.
but this may introduce unwanted side effects. Here, we impose the gravitational force
this may introduce unwanted side effects. Here, we impose The
the test rig
gravitational has two ad-
force in
in
justable
the model horizontal
prior to crossbeams
the impact which
load. support
The the
boulders panel,
are and a high-speed
modelled
the model prior to the impact load. The boulders are modelled as rigid bodies, and the as rigid camera
bodies, (500
andfps)
theis
used
Abaqus to capture the
contact algorithm
Abaqus contact boulder velocity
algorithm isis used
usedto and
tosimulatethe
simulatethewire mesh
theinteraction deformation.
interactionbetween Two
betweenrigid additional
rigidand
and cam-
deforma-
deformable
erasbodies.
ble
bodies.are used to record the lateral deformation of the panel and the deflection of the sup-
porting
Figurebeams.
Figure 14
14 shows
shows a comparison between between measured
measured and simulated
simulated displacements
displacements for for
these
these In
twothis paper,
tests. The we report
model test data
effectively obtained
reproduces from
the impact
vertical tests
model effectively reproduces the vertical response of the conducted
response of the usingduring
masses
masses spher-
dur-
icalthe
ing
the concrete
firstfirst “boulders”
impact
impact andand of 100
beyond.
beyond. Thekgaccuracy
The mass (test
accuracy ofof 1) and
the
the 200 is
model
model kg mass (testby
isreinforced
reinforced 2) the
by on
the2observation
× 2 m wire
observation
mesh
that
that the panels.
therebound, The
rebound,and boulders
andeveneventhe were lifted
thesubsequent to
subsequentimpact, a height
impact,are of
arealso
also 2.5
well
well m above
captured
captured the
byby mesh,
thethe which
algorithm.
algorithm.

Figure 14.
Figure Comparison between
14. Comparison between experimental
experimental and
and numerical
numerical data
data for
for vertical
vertical displacements.
displacements.

A comparison between the experimental data and the numerical predictions for the
A comparison between the experimental data and the numerical predictions for the
boulders’ velocities is shown in Figure 15a. Again, they are in excellent agreement. From
boulders’ velocities is shown in Figure 15a. Again, they are in excellent agreement. From
these data, the kinetic energy of the masses can be readily determined (Figure 15b), which
these data, the kinetic energy of the masses can be readily determined (Figure 15b), which
shows how the kinetic energy of the masses changes over time. Both masses are first brought
shows how the kinetic energy of the masses changes over time. Both masses are first
to rest in approximately 0.1 s. The results shown in Figures 14 and 15 are interesting in
brought to rest in approximately 0.1 s. The results shown in Figures 14 and 15 are inter-
esting in that they demonstrate that the mesh panels cannot be modelled as linear springs,
since otherwise the maximum deflections (and time to peak deflection) would differ by a
factor of √2.
 Figure 14. Comparison between experimental and numerical data for vertical displacements.

A comparison between the experimental data and the numerical predictions for the
boulders’ velocities is shown in Figure 15a. Again, they are in excellent agreement. From
Geosciences 2023, 13, 180 these data, the kinetic energy of the masses can be readily determined (Figure 15b), which
11 of 12
shows how the kinetic energy of the masses changes over time. Both masses are first
brought to rest in approximately 0.1 s. The results shown in Figures 14 and 15 are inter-
esting in that
that they they demonstrate
demonstrate that the that
meshthepanels
mesh cannot
panels cannot be modelled
be modelled as linear
as linear springs,
springs, since
since otherwise the maximum deflections (and time to peak deflection) would differ
otherwise the maximum deflections (and time to peak deflection) would differ by a factor by a
√ of √2.
factor
of 2.

Comparisonbetween
15. Comparison
Figure 15. betweenexperimental
experimental and
and numerical
numerical data
data forfor
(a) (a) vertical
vertical velocity
velocity andand
(b)
kinetic energy.
(b) kinetic energy.

4. Conclusions
Modelling double-twisted wire meshes with simplified geometries undermines the
deformation modes and interactions that occur between wires in double-twisted sections.
In addition, the constitutive models used for single steel wires cannot reproduce their real
stress–strain relations. Here, direct modelling approach to wire mesh is considered; it
obtained good agreement between the experimentally observed responses to static and
impact loadings and the model. This result offers the prospect of exploring alternative
catch fence designs in the search for improved efficiency. The model represents an initial
step toward studying the effect of interactions between wires in double-twisted sections on
the wire mesh’s response. In addition, further investigation into the material failure model,
based on destructive testing, is recommended.

Author Contributions: Writing—original draft, H.A.-B.; Writing—review & editing, Z.G.; Supervi-
sion, A.S. All authors have read and agreed to the published version of the manuscript.
Funding: This research was funded by The Knowledge Transfer Partnerships (KTP) programme
of the Innovate UK (Project number: 9980) and QTS Group Ltd., a leading railway infrastructure
services company in the UK (http://www.qtsgroup.com/ (accessed on 7 June 2023)).
Data Availability Statement: The data presented in this study are available on request from the
corresponding author. The data are not publicly available due to commercial restrictions.
Acknowledgments: We acknowledge the use of the laboratory and high-performance computing
facilities provided by the School of Engineering, University of Glasgow. The technical contribution to
this work by Simon Wheeler and Trevor Davies of the University of Glasgow is also acknowledged.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Foster, C.; Pennington, C.; Culshaw, M.G.; Lawrie, K. The national landslide database of Great Britain: Development, evolution
and applications. Environ. Earth Sci. 2011, 66, 941–953. [CrossRef]
2. Network Rail UK. West of Exeter Route Resilience Study. 2014. Available online: https://www.networkrail.co.uk/wp-content/
uploads/2016/11/West-of-Exeter-Route-Resilience-Study.pdf (accessed on 3 February 2023).
3. Etag, E.O.T.A. Guideline for European technical approval of falling rock protection kits. Eur. Organisat. Techn. Appr. 2008, 27.
4. Gerber, W. Guideline for the Approval of Rockfall Protection Kits; Swiss Agency for the Environment, Forests and Landscape (SAEFL);
Swiss Federal Research Institute WSL: Birmensdorf, Switzerland, 2001.
5. Buzzi, O.; Spadari, M.; Giacomini, A.; Fityus, S.; Sloan, S.W. Experimental Testing of Rockfall Barriers Designed for the Low
Range of Impact Energy. Rock Mech. Rock Eng. 2013, 46, 701–712. [CrossRef]
6. Muraishi, H.; Samizo, M.; Sugiyama, T. Development of a Flexible Low-Energy Rockfall Protection Fence. Q. Rep. Railw. Tech. Res.
Inst. 2005, 46, 161–166. [CrossRef]
Geosciences 2023, 13, 180 12 of 12

7. Al-Budairi, H.; Gao, Z.; Steel, A.; Wheeler, S.; Davies, T. Modelling and optimising of a light-weight rockfall catch fence system.
In Proceedings of the NAFEMS UK Conference, Telford, UK, 15–16 June 2016; pp. 129–132.
8. Gentilini, C.; Gottardi, G.; Govoni, L.; Mentani, A.; Ubertini, F. Design of falling rock protection barriers using numerical models.
Eng. Struct. 2013, 50, 96–106. [CrossRef]
9. Thoeni, K.; Lambert, C.; Giacomini, A.; Sloan, S.W. Discrete modelling of hexagonal wire meshes with a stochastically distorted
contact model. Comput. Geotech. 2013, 49, 158–169. [CrossRef]
10. Bertrand, D.; Nicot, F.; Gotteland, P.; Lambert, S. Discrete element method (DEM) numerical modeling of double-twisted
hexagonal mesh. Can. Geotech. J. 2008, 45, 1104–1117. [CrossRef]
11. Escallón, J.; Wendeler, C.; Chatzi, E.; Bartelt, P. Parameter identification of rockfall protection barrier components through an
inverse formulation. Eng. Struct. 2014, 77, 1–16. [CrossRef]
12. Van Tran, P.; Maegawa, K.; Fukada, S. Experiments and Dynamic Finite Element Analysis of a Wire-Rope Rockfall Protective
Fence. Rock Mech. Rock Eng. 2013, 46, 1183–1198. [CrossRef]
13. Bertrand, D.; Trad, A.; Limam, A.; Silvani, C. Full-Scale Dynamic Analysis of an Innovative Rockfall Fence Under Impact Using
the Discrete Element Method: From the Local Scale to the Structure Scale. Rock Mech. Rock Eng. 2012, 45, 885–900. [CrossRef]
14. Nicot, F.; Cambou, B.; Mazzoleni, G. From a constitutive modelling of metallic rings to the design of rockfall restraining nets. Int.
J. Numer. Anal. Methods Géoméch. 2001, 25, 49–70. [CrossRef]
15. Mentani, A.; Giacomini, A.; Buzzi, O.; Govoni, L.; Gottardi, G.; Fityus, S. Numerical Modelling of a Low-Energy Rockfall Barrier:
New Insight into the Bullet Effect. Rock Mech. Rock Eng. 2016, 49, 1247–1262. [CrossRef]
16. Lambert, S.; Bertrand, D.; Berger, F.; Bigot, C. Low energy rockfall protection fences in forested areas: Experiments and numerical
modelling. In Proceedings of the International Conference on Structures under Shock and Impact (SUSI Congress), Kyoto, Japan,
25 May 2009; pp. 133–138.
17. Abaqus Analysis User’s Manual; Version 6.14; Dassault Systems Simulia Corp.: Providence, RI, USA, 2014.
18. Maccaferri Technical Data Sheet for Rockfall Netting. 2015. Available online: https://www.geofabrics.co/sites/default/files/
technicaldata/tds-dtnet-pvc-gl-may-2015.pdf (accessed on 3 February 2023).
19. SolidWorks; Version 2016; Dassault Systèmes SolidWorks Corp.: Waltham, MA, USA.
20. UNI 11437:2012; Rockfall Protective Measures: Tests on Meshes for Slope Coverage. UNI Nazionale Italiano di Unificazione:
Milano, Italy, 2012.
21. Tracker Video Analysis and Modelling Tool. Version 4.96. Available online: http://physlets.org/tracker/ (accessed on 3
February 2023).

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.