infrastructures

Article
Sleepers Spacing Analysis in Railway Track Infrastructure
Roberto Sañudo 1, *, Marina Miranda 2 , Borja Alonso 1 and Valeri Markine 3

1 SUM+LAB Group, Universidad de Cantabria, 39005 Santander, Spain; [email protected]

2 Geotechnical Group, Universidad de Cantabria, 39005 Santander, Spain; [email protected]
3 Department of Engineering Structures, Delft University of Technology, 2628 CN Delft, The Netherlands;
[email protected]
* Correspondence: [email protected]; Tel.: +34-94-2201759

Abstract: Sleeper spacing has been a taboo subject throughout the railway’s history. Safety concerns
related to the structural integrity have been the main causes of not addressing this matter. There
are no specific and clear recommendations or guidelines in relation to this matter and the distances
do not go more than 0.8 m. In order to go beyond this current situation, the following research
paper analyses the influence of the spacing between sleepers on the behaviour of ballasted tracks
by performing a dynamic simulation with finite elements in two dimensions for different track
configurations, different elements, geometries, and separations within the frame of the ODSTRACK
project. The variables studied are the vertical displacements, the forces and stresses on the most
important elements of the superstructure, as well as the vertical accelerations in the sleepers and
the train. The values obtained from the numerical simulations were compared with the maximum
permitted values according to the guidelines. To limit this distance to the most restrictive variable
among those analysed, it is necessary to make important assumptions, such as the permissible values
and effective support contact areas between the sleepers and the ballast. The preliminary analyses
carried out shed light on a possible increment of the spacing between sleepers’ axes up to more than
0.8 m. This suggests that important savings in railways construction costs can be achieved, and they
Citation: Sañudo, R.; Miranda, M.; will help to develop the next stage of the ODSTRACK project.
Alonso, B.; Markine, V. Sleepers
Spacing Analysis in Railway Track Keywords: sleepers spacing; finite element modelling; dynamic design; vertical accelerations;
Infrastructure. Infrastructures 2022, 7, stresses; vertical displacements
83. https://doi.org/10.3390/
infrastructures7060083

Academic Editor: Pedro

Arias-Sánchez 1. Introduction
Railway management administrations around the world provide the distance between
Received: 5 May 2022
sleepers based on their experience, which seem to be mainly derived from questions of
Accepted: 9 June 2022
Published: 12 June 2022
historical nature. In Europe, the most common distance between sleeper axes is 600 mm.
However, this distance varies in other countries around the world, and it depends on
Publisher’s Note: MDPI stays neutral several factors, such as track gauge, the thickness of the ballast layer, or the dimension
with regard to jurisdictional claims in and materials of the sleepers. Design manuals for railway projects in the United States
published maps and institutional affil-
recommend the spacing between sleepers depending on the material of the sleeper and
iations.
the thickness of the ballast [1]. The recommended separation between supports (sleepers)
can vary between 0.546 m (wooden cross sleepers and a 0.152 m thick section of ballast)
and 0.711 m (concrete sleepers and a 0.203 m thickness of ballast), and in these cases, the
Copyright: © 2022 by the authors.
continuous welded rail is recommended. The distance between sleepers can also be 0.61 m
Licensee MDPI, Basel, Switzerland.
when there are steel sleepers in a ballast section of 0.203 m thickness [1].
This article is an open access article
Sleeper density is the number of sleepers per track length, usually given per km.
distributed under the terms and Sometimes it is better to define the number of sleepers per distance because the sleeper
conditions of the Creative Commons spacing is variable depending on their location along the track, as in cases of turnouts,
Attribution (CC BY) license (https:// switches and crossings, and rail joints, where sleeper spacings are variable. Previous
creativecommons.org/licenses/by/ works [1–3] studied the effects of increasing the support position between the slab track
4.0/). and ballast track. The results showed that the best option was to separate the first sleeper,

Infrastructures 2022, 7, 83. https://doi.org/10.3390/infrastructures7060083 https://www.mdpi.com/journal/infrastructures

Infrastructures 2022, 7, 83 2 of 20

at 0.6 m, from the axle of the last support on the slab track side. Normal values of sleeper
density are around 1660 sleepers per km [4]. It must be considered that a reduction in
the distance between sleepers causes an increase in the stiffness of the track. This effect is
similar to the increase in the size of the sleeper [5]. Normally, a high stiffness is desirable
and better than a low stiffness. However, both extremes are detrimental. High stiffness
produces accelerated wear, fatigue, and cracking problems [6]. On the other hand, the
appearance of differential settlements, loss, and the wear of the ballast that can result in the
hanging sleepers are common problems in tracks with a low stiffness [7]. It is important to
note that the poor quality of a track may lead to a decrease in passenger comfortability and
safety [8], and the deterioration of the vehicle–overhead system interaction performance [9].
It is necessary to study the distance between supports (sleepers) in the ballast track, as
it can reduce considerably the initial construction costs, and because it implies the reduction
of the material. Therefore, it is desirable to carry out this analysis, but always without
jeopardising safety and good quality conditions. It is important to save materials in the
short term (during construction, estimated from 15 to 40% cost savings, depending on
distance) and in the long term [10].
All track elements are important to have an overall understanding of the track perfor-
mance. It is known that cost of the track can be assumed proportional to the number of
these elements. This cost can vary depending on different factors, such as orography or the
structures and punctual elements that can be found along the track, the latter of which being
the ones that most affect the cost. The track cost in railways it is usually given per km. To
have an idea of this, the total cost of the first high speed line (Madrid–Sevilla) in Spain was
EUR 2.704 M, which means a cost of EUR 5.4 M per km [11]. Infrastructure costs (ICt) can be
divided into construction costs (ICC ) and maintenance costs (ICM ) (ICt = ICC + ICM ) [12].
All elements of track structure are inside these two components. Sleepers in the track are
included in the construction costs, as they are a track material necessary to build a proper
railway track, but they also need to be maintained, and therefore they are also included
in the maintenance costs. The infrastructure costs are fixed costs that evolve linearly with
the length of the corridor. Construction cost is proportional to the number of kilometres
of the railway line, and as sleepers are also proportional to the track length, it can be said
that the number of elements is proportional to the construction cost. A reduction in the
number of track elements per track length can help to reduce this part of the equation ICC.
Therefore, a reduction in ICC may cause a reduction of ICt. Figure 1 shows the savings in
relation to the increase in the sleeper spacing. In order to achieve this, it is necessary to
analyse the options to increase the sleeper spacing under the safety limits to the circulations
(moving material, passengers, and users) as well as the track structure. This figure was
taken from [13], where there is an important discussion about this matter. It is important
to consider that track construction costs will change, but so will track maintenance and
renewal. The distance between sleepers will affect the frequency of track maintenance [13].
Up until now, there have been some studies showing that it is possible to increase the
commonly used sleeper distance in Spain and most European countries, which is 0.6 m
between axes. Several countries use longer distances, as with UIC (Union Internationale de
Chemins de Fer). For example, the sleeper spacing varies depending on the track gauge [14],
reaching values up to 0.73 m. According to [15], the spacing between sleepers depends
on several factors, such as the resistance of the rail, the type of the sleeper, the ballast
thickness, the bearing capacity of the subbase, the axle loads, the volume, and the speed of
the traffic. The ORE (Office of Recherches et Essais de L’union de Chemins de Fer), today
known as UIC, recommendations consider values between 450 and 650 mm as acceptable
for the distance between sleepers. In addition, sleeper spacing in the range of 630 and
790 mm had a negligible influence on the vertical stress level in the subgrade for a unit
loading applied to the sleeper [14]. Some authors, like Wu and Thompson (2000), study the
optimal arrangement [15] by the repetition of a track unit composed of five sleeper bays of
0.6 m spacing and five bays of 0.7 m. Other authors [16] have recommended to choose the
optimum sleeper spacing between 0.54 and 0.67 m. American railroads present an increase
Infrastructures 2022, 7, x FOR PEER REVIEW 3 of 22

Infrastructures 2022, 7, 83 3 of 20
study the optimal arrangement [15] by the repetition of a track unit composed of five
sleeper bays of 0.6 m spacing and five bays of 0.7 m. Other authors [16] have recom-
mended to choose the optimum sleeper spacing between 0.54 and 0.67 m. American rail-
of just 10%
roads of rail
present anstresses
increaseand no increase
of just on the
10% of rail bearing
stresses pressure
and between
no increase sleepers
on the andpres-
bearing the
ballast when the
sure between sleepers
sleepers andthat are
the 12 inches
ballast when(0.305 m) widethat
the sleepers are spaced at 30 inches
are 12 inches (0.305(0.762
m) widem)
instead of the conventional distances (0.6 m) [17]. Most railway administrators
are spaced at 30 inches (0.762 m) instead of the conventional distances (0.6 m) [17]. Most limit this
spacing
railwaydepending
administratorson the track
limit loads
this and train
spacing speed. on
depending Forthe
example, Indian
track loads railways
and have
train speed.
three main distances between sleepers’ axes, and depending on the type of
For example, Indian railways have three main distances between sleepers’ axes, and de-line and sleeper
spacing,
pendingthey cantype
on the be increased
of line and upsleeper
to 1.0 m.
spacing, they can be increased up to 1.0 m.

Figure1.1.Cost
Figure Costsavings
savingsestimation
estimationin
interms
termsof
ofsleepers
sleepersspacing.
spacing.

Fromdata
From dataobtained
obtainedin inprevious
previousstudies,
studies,ititseems
seemsthat thatthe
thedistance
distancebetween
betweensleepers
sleepers
can
canbebeincreased
increasedbut butthere
thereis is
nono maximum
maximum clear value
clear valueor recommendations
or recommendations for itfor
[13]. In the
it [13]. In
research of [13], the authors talk about the transversal direction in relation
the research of [13], the authors talk about the transversal direction in relation to sleeper to sleeper spacing.
The theoretical
spacing. results show
The theoretical a more
results thana 40%
show morereduction
than 40%inreduction
the track lateral
in the resistance
track lateral from
re-
asistance
spacingfromof 600 mm toof
a spacing a 600
spacingmm to ofa1000 mm.of Therefore,
spacing it will be important
1000 mm. Therefore, to keep
it will be important
studying the effects
to keep studying of effects
the changing the sleepers’
of changing the distance
sleepers’and providing
distance more datamore
and providing to clarify
data
the possibilities
to clarify and give some
the possibilities and give guidelines to increase
some guidelines to the distance
increase between between
the distance sleepers.sleep-
This
work presents
ers. This worka presents
numerical a 2D study to2D
numerical analyse
study the performance
to analyse of the ballastoftrack
the performance when
the ballast
the distance of the sleepers is varied. Several parameters, such as
track when the distance of the sleepers is varied. Several parameters, such as the sleepers’the sleepers’ material,
dimensions, rail type, and
material, dimensions, train velocity,
rail type, and trainare analysed.
velocity, This paper
are analysed. is focused
This paper ison sleepers
focused on
spacing
sleepers spacing in classical railway tracks (ballasted tracks), while the analysissupport
in classical railway tracks (ballasted tracks), while the analysis of slab track of slab
spacing is out of
track support the scope
spacing of of
is out this
thepaper.
scope of this paper.
Recent
Recent studies have analysed thestructural
studies have analysed the structuralresponse
responseof ofmonoblock
monoblockrailway railwayconcrete
concrete
sleepers
sleepers and fastening systems under vertical and lateral loads [18]. Here,the
and fastening systems under vertical and lateral loads [18]. Here, theresearchers
researchers
present
presentaathree-dimensional
three-dimensionalfinite finiteelement
elementrailway
railwaymodel modelto tostudy
studythe thestructural
structuralresponse
response
of concrete sleepers and the fastening system. As a complement,
of concrete sleepers and the fastening system. As a complement, its authors its authors study
study thethe
in-
influence
fluence of sleeper spacing. They conclude that the influence of sleeper spacing on thethe
of sleeper spacing. They conclude that the influence of sleeper spacing on be-
behaviour
haviour ofofrailway
railwaytracks
tracksisisonly
onlytangible.
tangible. This
This previous
previous parametric
parametric analysis
analysis sheds
shedssomesome
light
lightover
overthe thesleeper
sleeper spacing.
spacing. Here, some
Here, someof the study’s
of the magnitude
study’s magnitude includes an experiment
includes an experi-
with a slight increase with the sleeper spacing, for example, the lateral contact force at the
ment with a slight increase with the sleeper spacing, for example, the lateral contact force
rail angled guide plate interface, the maximum compressive stress of concrete material,
at the rail angled guide plate interface, the maximum compressive stress of concrete ma-
or the compressive stress in the rail pad; however, other parameters experiment with
terial, or the compressive stress in the rail pad; however, other parameters experiment
increasing the sleeper spacing, and the vertical rail-seat contact force. The results presented
with increasing the sleeper spacing, and the vertical rail-seat contact force. The results
here go further and offer more solutions in relation to sleeper spacing.
presented here go further and offer more solutions in relation to sleeper spacing.
The manuscript is structured as follows. After this introduction, the next section de-
scribes the numerical model, where the train and track models are described. The next section
presents all the results obtained in the numerical analysis. After that, a parametric analysis
Infrastructures 2022, 7, 83 4 of 20

of the results is carried out to find a suitable sleeper spacing. An extended discussion of the
results is found in Section 4. To finish the study, the conclusions section resumes the most
important findings of this research, providing a range of potential sleeper spacings.

2. Numerical Analysis
Finite elements analyses, using the 2D finite element software [3], were performed to
study the track behaviour and the effect on all the elements of the railway superstructure
when the distance between sleepers is modified. The feasibility of increasing the sleeper dis-
tance without jeopardising the safety of the track will be analysed by modelling a reference
case, followed by parametric studies to analyse the effects on various track configurations,
on the track displacements, stresses, and strains. Numerical models have helped in nu-
merous applications to railways, in track transition structures analysis [1–3,7,18–22], in
evaluating the structural analysis of the variable railway gauge [23], or to evaluate the slab
track (asphalt), as in [24]. Although the model uses real parameters and the mechanical
properties of the materials, a similar model was validated in [25].
It is important to highlight that this model is a simplified model (compared with
others like the one presented in [8,9] in the preliminary analysis (within the ODSTRACK
project)). Thanks to these initial results, researchers can manage and develop laboratory
tests in the next phase of the project in order to have an experimental understanding of
the phenomena.

Train and Track Model

A
Infrastructures 2022, 7, x FOR PEER REVIEWballasted track is chosen to perform the analysis. The model represents a 200 5 of m
22
length straight track. This distance is considered enough to obtain reliable results given the
characteristics of the program and the length of the trains in the simulation [3]. The finite
element size of the track is 0.05 m, which means that there is a total of 4000 elements. To
− In the horizontal plane, the structural properties are symmetric with respect to the
optimise the computational effort and time, central symmetry with respect to the vertical
track axis.
plane that passes through the centre of the track is assumed. Rail surface defects are not
− No track transversal stability has been analysed.
considered, and constant longitudinal stiffness is assumed. Elements of the track are formed
− The contributions of the axial forces in the layers are neglected.
by an inertial system of masses, springs, and dumpers. As with other models [19], this uses
− Bending length of deformations of the stiff layer is bigger than the thickness of the
a combination of spring and dumpers to represent the track and vehicle. Their properties
stiff layer.
are given in Table 1 and shown in Figure 2. The bending of the sleeper was not considered,
− Thermal effects have not been considered in the study.
as this will be the next step within the investigation for future work. Rail–wheel interaction
follows the Hertz contact theory.
Table 1. Mechanical parameters of the track superstructure elements used in the track model.
Table 1. Mechanical parameters of the track superstructure elements used in the track
Ballast model.
Track
RAIL UIC Track
Ballast 60
K1 (kN/m) 100,000
Rail pads (ud long) RAIL UIC 60
C1
K1(KN·s/m)
(kN/m) 35
100,000
Rail pads (ud long)
C1E(KN
(kN/m)
·s/m) 80,000,000
35
Sleepers
EPoisson
(kN/m) 80,000,000
0.2
Sleepers
K2Poisson
(kN/m) 0.2
200,000
Ballast K2 (kN/m) 200,000
Ballast C2 (KN·s/m)
C2 (KN·s/m)
24,000,000
24,000,000
E (kN/m2 ))
E (kN/m 180,000
2
Foundation 180,000
Foundation
ʋ (KN
(KN·s/m)
·s/m) 0.25
0.25
K (kN/m)
K (kN/m) 1,000,000
1,000,000
Formation
Formation C (KN·s/m) 20
C (KN·s/m) 20

The train model used in the analysis is an ICE3 train (Siemens Velaro model). It consists
of 5 cars modelled as a suspended mass and an inertial system of springs and dampers,
whose mechanical characteristics are shown in Table 2 and in Figure 2. The simulations
are carried out for different train speeds, varying from 40 km/h to 350 km/h, to cover
high-speed velocities. Transient time of calculation was between 2 and 4 s depending on
train speed and finite element size.
 K2 (kN/m) 200,000
Ballast
C2 (KN·s/m) 24,000,000
E (kN/m2) 180,000
Foundation
ʋ (KN·s/m) 0.25
Infrastructures 2022, 7, 83 K (kN/m) 1,000,000 5 of 20
Formation
C (KN·s/m) 20

Figure
Figure2.2.Simplified
Simplifiedmodel.
model.Train
Trainmodel
model(left).
(left).Couple
Couplemodel
modelof
oftrain
trainand
andtrack
trackas
asaacombination
combinationof
of
springs and dumpers (right).
springs and dumpers (right).

Table 2. Mechanical properties of the train used in the simulation.

Table 2. Mechanical properties of the train used in the simulation.
Mechanical Characteristics for the Train
Mechanical Characteristics for the Train
K1 (kN/m) 400
Bogie C1K1 (kN/m)
(KN·s/m) 400
50
Bogie C1 (KN·s/m) 50
MM(KN)
(KN) 3.625
3.625
K2
K2(kN/m)
(kN/m) 300
300
Carbody
Car body (KN·s/m)
C2(KN·s/m)
C2 2424
MM(kN)
(kN) 48.642
48.642
Rw (m) 0.42
Rw (m)
dw (m)
0.42
2.50
Wheel
Wheel dw (m)
Mw (KN) 2.50
1.0135
Mw (KN) 1.0135
The sleeper distance is varied between 0.6 m and up to 1 m in 0.1 m intervals. To char-
The model represents a conventional high-speed track with the typical elements com-
acterizes the performance of the track in each case, the rail, sleepers, ballast displacement,
monly used in Spanish railways lines (conventional and high-speed). This is the reference
shear forces, and bending moments were also studied. To complement these variables,
case, and no elements are considered under the ballast layer. This initial configuration is
vertical accelerations in the sleepers and the train are also analysed. These variables have
been chosen in the study from a physical/degradation point of view (of the track).
It is important to consider initial assumptions in the used model:
- The model analyses a straight track section.
- The structure is loaded by vertical loads only.
- In the horizontal plane, the structural properties are symmetric with respect to the
track axis.
- No track transversal stability has been analysed.
- The contributions of the axial forces in the layers are neglected.
- Bending length of deformations of the stiff layer is bigger than the thickness of the
stiff layer.
- Thermal effects have not been considered in the study.
The model represents a conventional high-speed track with the typical elements com-
monly used in Spanish railways lines (conventional and high-speed). This is the reference
case, and no elements are considered under the ballast layer. This initial configuration is
chosen because the authors wanted to see the performance of the different superstructure
elements in first place and to have an initial model to compare with the rest of the cases.
Infrastructure layers are considered as a case itself. To broaden the analysis to other track
configurations, materials, and train conditions, a parametric study varying the train speed,
dimensions, and the materials of the sleepers and the infrastructure layers are analysed.
All the different studied cases are summarized in Table 3.
Infrastructures 2022, 7, 83 6 of 20

Table 3. Resume of the different analysed cases in this preliminary approach.

Description Rail Different Speeds (km/h) Sleeper Type Infrastructure

1 Initial case UIC 60 40, 80, 120, 180, 210, 250, 300, 350 AI04 prestressed High stiffness
2 Speed increasing UIC 60 40, 80, 120, 180, 210, 250, 300, 350 AI04 prestressed High stiffness
3 Use of different rails UIC 60/UIC 54 40, 210, 350 AI04 prestressed High Stiffness
4 Different sleeper material UIC 60/UIC 54 40, 210, 350 See Table 4 High Stiffness
5 Different sleeper size UIC 60 40, 210, 350 See Table 5 High Stiffness
Influence of the
6 UIC 60 40, 210, 350 AI04 prestressed See Table 6
infrastructure layers

Table 4. Material properties for the different sleepers used in the analysis.

Density Sleeper
Material E (KN/m2 ) Poisson
(KN/m3 ) Dimensions (m)
Monobloc (Prestressed) 80,000,000 0.2 25 2.6 × 0.3 × 0.22
Wood 385,000 0.4 7.7 2.6 × 0.3 × 0.22

Table 5. Values of the sleepers’ dimensions used in the simulation.

Difference per
Variation Length (m) Width (m) Altura (m) Area Base (m2 ) Volume (m3 )
Sleeper (m3 )
Reference case (initial) 2.60 0.30 0.22 0.78 0.1716 0
Longitudinal length (L) 3.00 0.30 0.22 0.9 0.198 0.022
Cross length (A) 2.60 0.40 0.22 1.04 0.2288 0.0572
Cross and longitudinal
3.00 0.40 0.22 1.2 0.264 0.0924
variation (T)

Table 6. Material properties for the analysed subbases.

Definition E (kN/m2 ) Poisson Density (Tn/m3 )

No subbase (Initial case) - - -
ODS (Subbase I) 20,000,000 0.25 2.00
ODS1 (Subbase II) 210,000 0.30 1.95

This model has been used and checked in other previous analysis perfor-
mance [1,7,10,11,20–22]. The model was also used before to solve the track transition
problem. It only considered vertical loads and symmetry with respect to the track axis. The
initial assumptions considered straight track and track axe symmetry. According to [26],
the investigation shows a new approach to evaluate the lateral resistance associated with
extended sleeper spacing. This research shows laboratory tests and investigates the effect
of the interference between adjacent sleepers on the lateral resistance per sleeper. Here, the
researchers proposed a new method to estimate the lateral resistance of a ballast with an
extended sleeper spacing.

3. Results
Results of the parametric study are presented in this section. As was said previously,
the classical ballasted track has been used in this study. For each case, the vertical displace-
ments of the rail and the sleepers, the shear forces on the rail and sleepers together with
bending moments, and the contact pressure between the sleeper and ballast are presented.
In some cases, the vertical accelerations in the train and sleepers are also presented.

3.1. Conventional High-Speed Superstructure (Reference Case)

Results obtained in terms of vertical displacements, shear forces, bending moments,
and vertical stresses are presented in Figure 3 for the different spacings between sleepers.
Infrastructures 2022, 7, 83 7 of 20

The positive values of vertical displacements represent the rail deflection. The positive
values of forces and stresses are tension, and the negative ones refer to the compressive
Infrastructures effects. InPEER
2022, 7, x FOR order to obtain the stresses on the ballast, recommendations from [27] are
REVIEW 8 of 22
followed, using a half sleeper area. This will be discussed in depth in Section 4.

Figure 3. Case 1. From left to right, and from top to bottom. Maximum vertical displacements of the
Figure 3. Case 1. From left to right, and from top to bottom. Maximum vertical displacements of the
rail and sleeper, vertical forces on rail and sleeper, rail bending moments, and stresses on ballast.
rail and sleeper, vertical forces on rail and sleeper, rail bending moments, and stresses on ballast.
3.2. Speed Variation
As expected, there is an increase of the maximum vertical displacements in the rails when
As presented before, speeds from 40 km/h to 350 km/h are analysed to cover all val-
the distance between the sleepers increases (see Figure 3). The maximum vertical displacement
ues up to high-speed lines. Due to the assumption of considering a constant vertical stiff-
of the rail is around 1 mm for the conventional sleepers’ distance of 0.6 m, and it increases
ness along the railway track, and not considering rail surface defects, there are very small
to slightly more variations
than 1.6 mm forvalues
in the the separations
analysed with between the sleepers
the increasing speed.of 1
Tom. Thisthe
show means
track perfor-
that attending the mance when the train speed increases, in this case, the authors focused on the
current rail maximum vertical displacements recommended values, the vertical
sleeper distance could not beon
accelerations increased because
the sleepers and thethevertical
rail vertical displacement
accelerations goes(body
on the train up more
and bogies),
than 1 mm [28]. However,
as shown in this limit4.ofThe
Figure 1 mm is too conservative,
frequencies of the wheeland forceinshow
mostacases
linearthis valuewith the
increase
goes up to more than 1 mm [29]. Therefore, it might be possible to
train speed and decrease when sleeper spacing increases (Figure 4).increase it.
Vertical accelerations
The maximum vertical displacement onofthe
thesleepers
sleeperincrease wheninthe
is affected thesleeper spacing
same way. increases.
It is
The smallest differences are found in the separations of
0.5 mm when the distance between the sleepers is 0.6 m, and it increases up to 0.9 mm for 0.8 and 0.9 m. Accelerations go
from 3 m/s 2 to almost 30 m/s2 and up to 40 m/s2 for speeds of 350 km/h. Admissible values
the maximum distance analysed of 1 m. These values are all below the limits proposed, so
can go in
there are no problems from
this10sense
m/s2 [31,32]
[30]. to more than 100 m/s [33,34] according to different authors.
2

Vertical accelerations on the train, both in the bogie and in the body, initially increase
The rail shear force does not exceed 80 kN in any case. It is not noticeably affected by
and then, when reaching 120 km/h, they are more or less invariant when increasing the
the separation of the sleepers. The rail bending moments are not significantly affected by the
distance. In the case of the bogie, they do not exceed more than 0.8 m/s2. The accelerations
sleepers’ distances. Their values are between 21 and 23 kN · m. The last variable denotes
measured in the passenger train go from less than 0.002 m/s2 to less than 0.06 m/s2. It is
the
stresses under the sleeper,
curious to and theythat
observe are in
highly affected
the case of 0.9bym the distance
distance, the between the sleepers.
accelerations are much higher
They increase bythan almost 200 kN/m 2
in the other cases,when
even forthethe
distance
case of goes
the 1 from 0.6 tobetween
m distance 1 m. Looking at axes.
the sleepers’
Figure 3, the trendThefor the variables
motive considered
of these anomalous can be
values appreciated.
is not clear. Vertical displacements
and contact stresses between the sleeper and the ballast are the most sensitive variables to the
increase of the sleeper spacing. This initial conclusion derived from the reference case needs
to be checked in the rest of the analysed cases, so a parametric study is developed.

3.2. Speed Variation

As presented before, speeds from 40 km/h to 350 km/h are analysed to cover all
values up to high-speed lines. Due to the assumption of considering a constant vertical
Infrastructures 2022, 7, 83 8 of 20

stiffness along the railway track, and not considering rail surface defects, there are very
small variations in the values analysed with the increasing speed. To show the track
performance when the train speed increases, in this case, the authors focused on the vertical
accelerations on the sleepers and the vertical accelerations on the train (body and bogies),
as shown in Figure 4. The frequencies of the wheel force show a linear increase with the
train speed and decrease when sleeper spacing increases (Figure 4).
Vertical accelerations on the sleepers increase when the sleeper spacing increases. The
smallest differences are found in the separations of 0.8 and 0.9 m. Accelerations go from
3 m/s2 to almost 30 m/s2 and up to 40 m/s2 for speeds of 350 km/h. Admissible values
can go from 10 m/s2 [31,32] to more than 100 m/s2 [33,34] according to different authors.
Vertical accelerations on the train, both in the bogie and in the body, initially increase
and then, when reaching 120 km/h, they are more or less invariant when increasing the
distance. In the case of the bogie, they do not exceed more than 0.8 m/s2 . The accelerations
measured in the passenger train go from less than 0.002 m/s2 to less than 0.06 m/s2 . It is
curious to observe that in the case of 0.9 m distance, the accelerations are much higher than
in the other cases, even for the case of the 1 m distance between the sleepers’ axes. The
motive of these anomalous values is not clear.
Vertical accelerations in the train (bogies and body) seems not to vary greatly when
increasing the train speed. However, the vertical accelerations in sleepers increase with
the train speed. This increase follows an almost a linear variation from 50 km/h onwards
for all analysed sleeper distances. As presented before, the limits in the literature vary
from 10 to 100 m/s2 , so considering an average value, it allows for sleeper spacings of
1.0 m. However, this is the most restrictive variable for the ones presented in Figure 4. The
simulation time is enough, and as can be seen in Figure 4, the initial disturbances were
stabilized in a steady state solution (train bogie and body). The track irregularities have
Infrastructures 2022, 7, x FOR PEER
not REVIEW 9 ofreason
been considered in the analysis’ vertical accelerations, and this is the main 22 why
the obtained values are low, but indicative for the trend of the vertical accelerations.

Figure 4. Case 2. Maximum and minimum vertical accelerations in the sleepers (up, left), in the
Figure 4. Case 2. Maximum and minimum vertical accelerations in the sleepers (up, left), in the
bogie (up, right), and in the train (down) for the speeds studied and the defined spacings. Frequency
bogiewith
(up,train
right), and
speed andinsleeper
the train (down)
spacing for the
variation speeds
(down, studied and the defined spacings. Frequency
right).
with train speed and sleeper spacing variation (down, right).
Vertical accelerations in the train (bogies and body) seems not to vary greatly when
increasing the train speed. However, the vertical accelerations in sleepers increase with
the train speed. This increase follows an almost a linear variation from 50 km/h onwards
for all analysed sleeper distances. As presented before, the limits in the literature vary
from 10 to 100 m/s2, so considering an average value, it allows for sleeper spacings of 1.0
Infrastructures 2022, 7, 83 9 of 20

3.3. Case 3. Rail Type

Two commonly used rails in Spain are used in the simulation, namely UIC 54 (54 kg
per lineal metre) and UIC 60 (60 kg per lineal metre). High-speed railways use UIC
60 nowadays, and the UIC 54 rail is disappearing in most conventional tracks. Only narrow-
gauge tracks in Spain keep this type of rail. Figure 5 shows the vertical displacements in
the sleepers and rails, as well as the forces and stresses for the two considered types of rails.
The differences between both types of rails are negligible, as, for example, the shear
force on the rail, which increases when the distance between the sleepers increases, and
as expected, the lighter rail suffers more. In the same way, the stresses transmitted to the
ballast are lower when using the UIC 60 rail (heavier rails). The vertical displacements
of the rail are around 7.5% less in the case of the heavier rail, as expected. All the values
obtained are in the range of admissible values, presented later in Section 4. When the
sleeper distance increases, the vertical displacement in the sleepers increases as well,
and the trend is similar in both cases. The differences are around 3.5%.

3.4. Sleepers’ Material

The influence of the sleepers’ material on the track performance is presented here.
Infrastructures 2022, 7, x FOR PEER REVIEW of sleepers commonly used in Spanish railway lines are considered,
Two types 10 the
of 22classic
wooden one made by coniferous oak and the conventional prestressed monobloc concrete
sleeper (type AI-04). Their properties are presented in Table 4. Sleeper material in this
model onlyare
the rail resulted
around in sleeper
7.5% mass
less in variation.
the case of the heavier rail, as expected. All the values
In the same way, as for the two types
obtained are in the range of admissible values, of rails, the behaviour
presented of the sleepers
later in Section 4. When is thesimilar
from one material
sleeper to the other.
distance increases, There displacement
the vertical are negligible in differences in the shear
the sleepers increases forces
as well, andon the
rails
theand sleepers,
trend and
is similar the stresses
in both ondifferences
cases. The the ballast.
are(Figure
around6).3.5%.

Figure 5. Case 3. Maximum vertical displacements of the rail and sleeper, vertical forces on rail and
Figure 5. Case 3. Maximum vertical displacements of the rail and sleeper, vertical forces on rail and
sleeper, rail bending moments, and stresses on ballast.
sleeper, rail bending moments, and stresses on ballast.
3.4. Sleepers’ Material
The influence of the sleepers’ material on the track performance is presented here.
Two types of sleepers commonly used in Spanish railway lines are considered, the classic
wooden one made by coniferous oak and the conventional prestressed monobloc concrete
sleeper (type AI-04). Their properties are presented in Table 4. Sleeper material in this
model only resulted in sleeper mass variation.
In the same way, as for the two types of rails, the behaviour of the sleepers is similar
from one material to the other. There are negligible differences in the shear forces on the
 Two types of sleepers commonly used in Spanish railway lines are considered, the classic
wooden one made by coniferous oak and the conventional prestressed monobloc concrete
sleeper (type AI-04). Their properties are presented in Table 4. Sleeper material in this
model only resulted in sleeper mass variation.
In the same way, as for the two types of rails, the behaviour of the sleepers is similar
Infrastructures 2022, 7, 83 from one material to the other. There are negligible differences in the shear forces on 10 of 20
the
rails and sleepers, and the stresses on the ballast. (Error! Reference source not found.).

Infrastructures 2022, 7, x FOR PEER REVIEW 11 of 22

Figure 6. Case 4. Maximum vertical displacements of the rail and sleeper, shear forces on rail and
Figure 6. Case 4. Maximum vertical displacements of the rail and sleeper, shear forces on rail and
sleeper, rail bending moments, and stresses on ballast.
sleeper, rail bending moments, and stresses on ballast.

3.5. Sleeper Dimensions and Size

The dimensions
The dimensions of of the
the sleeper
sleeperconsidered
consideredin inthe
thereference
referencecase
caseare
are2.6
2.6mmlong
long×× 0.3 m
wide ×
wide × 0.22
0.22 m mhigh,
high,which
whichcorresponds
corresponds to atoconventional
a conventional prestressed
prestressed concrete
concrete sleeper
sleeper used
used in Spain in cases of an international gauge of 1435 mm. In this section,
in Spain in cases of an international gauge of 1435 mm. In this section, the effect of varying the effect of
varying
one one dimension
dimension in the longitudinal
in the longitudinal direction anddirection
anotherandinanother in the transverse
the transverse direc-
direction will be
tion will be
analysed analysed
(Table 5). (Table 5).
The idea is not to increase considerably the size because the costs of the sleepers will
increase as well.
well. Therefore,
Therefore,the thefirst
firsttwo
twocases
cases
areareto to keep
keep oneone dimension
dimension thethe
samesame as
as the
the
old old sleeper,
sleeper, the the AI-04.
AI-04. TheThe lastlast
casecase increases
increases both both dimensions
dimensions (length
(length andand width)
width) to
to in-
increase the support area of the sleeper (over the ballast). The
crease the support area of the sleeper (over the ballast). The sleeper length sleeper length was chosen to
be increased
increased up up to 3 m, as longer lengths can create structural problems of sleeper bending
stresses
stresses and
and can
can be problematic
problematic during
during tamping
tamping works.
works. The second case increases sleepers
up
up to 0.4 m. This value gives a distance between the sleepers of 0.2 m in normal conditions
(considering sleeperspacing
(considering a sleeper spacingofof0.6 0.6m).
m).TheThe
lastlast
casecase considered
considered increases
increases the length
the length and
and width at the same time, which gives the biggest support area of
width at the same time, which gives the biggest support area of all. Obviously, the last all. Obviously, the
last case is the best of all considered. Figure 7 shows the most important
case is the best of all considered. Figure 7 shows the most important values for the varia- values for the
variables obtained.
bles obtained.
The most important variable is the stress on the ballast (Figure 7). The influences on
the vertical displacements, shear forces, and rail bending moments are negligible. This
might be explained because the support area is really important, especially for vertical
displacements and the stress under the sleeper.
The effective area of the sleeper is the contact area of the sleeper base with the ballast
layer. This effective area is usually defined as F, and there are several references which
give different values for this “Fi”, i.e., half-sleeper or less (depending on if it is prestressed
monobloc, or a reinforced twin block sleeper) [27], to more than 5900 cm2 [35]. This is
needed to make a hypothesis of the contact area between the ballast and the sleeper.
 be increased up to 3 m, as longer lengths can create structural problems of sleeper bending
stresses and can be problematic during tamping works. The second case increases sleepers
up to 0.4 m. This value gives a distance between the sleepers of 0.2 m in normal conditions
(considering a sleeper spacing of 0.6 m). The last case considered increases the length and
Infrastructures 2022, 7, 83 width at the same time, which gives the biggest support area of all. Obviously,11the of 20last
case is the best of all considered. Figure 7 shows the most important values for the varia-
bles obtained.

Figure 7. Case 5. Maximum vertical displacements of the rail and sleeper, shear forces on rail and
sleeper, rail bending moments, and stresses on ballast (A—change the sleeper width, L—change the
sleeper length, T—change both width and length, I—initial size of the sleeper).

As expected, it is better to increase the total area of the sleeper pad. By increasing
the sleeper pad area, there is a reduction on the vertical displacements on the rail and the
sleepers, and there is a reduction on the stresses over the ballast.

3.6. Infrastructure Layers

This case analyses the influence of the infrastructure layers. In order to see the effect
of the type of infrastructure layers on the track performance when the distance between the
sleepers is increased, three cases for subgrades have been simulated. The first case shows a
track over a nondeformable ground (no subbase) considering a support on a material of
infinite stiffness, the second case uses a stiff material, “subbase I”, and the third case uses
a soft material, “subbase II”. Table 6 shows the values for the adopted materials, always
using ballast track (classical track), as the slab track is out of the scope of this paper.
It can be appreciated that the uses of a better or worse infrastructure layer slightly affects
the vertical displacements of both the rail and the sleeper (Figure 8). The rest of variables are
slightly influenced by this change. When focusing on the vertical displacements (rail and
sleeper), bending moments, and stresses on the ballast, all of them tend to increase as the
separation between the sleepers increases. If the shear forces on the rails and the sleepers
and the bending moments are analysed, the trend is to keep more or less constant.
In relation to the stress on the ballast, there is no big difference between the stiffer
and softer subbase, and there are small differences when considering an infinite stiffness
subbase. The same happens with the shear forces in the rails. There is no difference between
the considered cases. Stresses on the ballast increase with the spacing, as happened in the
previous cases. Stresses on the ballast are slightly higher when the subbase is softer, as the
stiffer layers resist the majority of the force. Differences in the stress over the ballast are
minimum, but slightly worse for the deformable layers such as subbase II.
 It can be appreciated that the uses of a better or worse infrastructure layer slightly
affects the vertical displacements of both the rail and the sleeper (Error! Reference source
not found.). The rest of variables are slightly influenced by this change. When focusing
on the vertical displacements (rail and sleeper), bending moments, and stresses on the
ballast, all of them tend to increase as the separation between the sleepers increases. If the
Infrastructures 2022, 7, 83 12 of 20
shear forces on the rails and the sleepers and the bending moments are analysed, the trend
is to keep more or less constant.

Infrastructures 2022, 7, x FOR PEER REVIEW 13 of 22

Figure 8. Case 6. Maximum vertical displacements of the rail and sleeper, shear forces on rail and
Figure 8. Case 6. Maximum vertical displacements of the rail and sleeper, shear forces on rail and
sleeper, rail bending moments, and stresses on ballast.
sleeper, rail bending moments, and stresses on ballast.

In relation
3.7. Rail to theand
Pad Stiffness stress on the ballast, there is no big difference between the stiffer
Dimensions
and softer subbase, and there
The material and dimensions are small
of the differences
under-railwhen
padsconsidering an infinite
are also analysed. Thestiffness
rail pad
subbase.
material is characterised by its stiffnesses. They are the three conventional rail pad typesbe-
The same happens with the shear forces in the rails. There is no difference in
tween
Spain: the considered
60 kN/mm, 80cases.
kN/mm,Stresses on the
and 100 ballast Fastening
kN/mm. increase with
clipsthe spacing,
have as happened
not been studied in
in theanalysis,
this previous cases.
and Stressesrefer
the authors on the
to ballast
[36] forare slightly
relevant higher when
conclusions thefailure
of the subbase is softer,
mechanism
as the stiffer layers resist the majority of the
regarding high-frequency excitation (580–680 Hz). force. Differences in the stress over the ballast
are minimum, but slightly worse for the deformable layers such as subbase
The initial model considers 100 kN/mm stiffness pads. Simulations have also been II.
carried out for pads of 60 kN/mm and pads of 80 kN/mm. The results showed that when
3.7.
the Rail Pad Stiffness
stiffness andpad
of the rail Dimensions
increases, the vertical displacements of the rail decrease, but
The material
the vertical and dimensions
displacements of the under-rail
on the sleepers pads is
increase. There arenoalso analysed.
difference The rail
in shear pad
stresses
material is characterised
on the rails, but the stifferbypads
its stiffnesses.
generate a They
slightare the three
increase conventional
of shear stresses onrailthe
pad types
sleepers,
which
in Spain:means a higher
60 kN/mm, 80stress
kN/mm, overandthe100
ballast layer.Fastening
kN/mm. Rail bending
clips moments have studied
have not been a slight
increase
in (the biggest
this analysis, difference
and the authorsisrefer
less than 4.5%
to [36] forwhen the pad
relevant stiffness decreases
conclusions (Figure
of the failure 9)).
mech-
anismThe variation
regarding with pad dimensions
high-frequency excitationis(580–680
also analysed.
Hz). Simulations were performed
by changing one
The initial of theconsiders
model dimensions 100ofkN/mm
the pad,stiffness
the longitudinal direction have
pads. Simulations of thealso
pad,been
and
keeping the transverse direction constant. Normal pad dimensions
carried out for pads of 60 kN/mm and pads of 80 kN/mm. The results showed that when are 150 × 180 mm. For
the stiffness
the analysis,ofthethe
longitudinal values were
rail pad increases, varied from
the vertical 100 mm and
displacements of 200
the mm to 300 mmbut
rail decrease, so
to see
the clearly
vertical their influence
displacements onin
thethe results.increase. There is no difference in shear stresses
sleepers
on theGenerally
rails, but speaking, the results
the stiffer pads generateareabetter
slightfor the larger
increase pads.
of shear The vertical
stresses displace-
on the sleepers,
ments, the shear stress on the sleepers, and the moments on rails decrease.
which means a higher stress over the ballast layer. Rail bending moments have a slight However, on
the other hand, the forces on the pad are considerably greater in the case
increase (the biggest difference is less than 4.5% when the pad stiffness decreases (Error! of the smaller-
dimensionsource
Reference pads. Regarding the increase in the spacing between the sleepers, an increasing
not found.)).
trend can be considered with an increasing distance (Figure 9).
The distance between axle bogies in the train is also considered and simulated. The
results showed that increasing the distance between the axles in the bogies from 1.8 m to
2.5 m is positive for all variables except one, the bending moments in rails. Values for rail
 in this analysis, and the authors refer to [36] for relevant conclusions of the failure mech-
anism regarding high-frequency excitation (580–680 Hz).
The initial model considers 100 kN/mm stiffness pads. Simulations have also been
carried out for pads of 60 kN/mm and pads of 80 kN/mm. The results showed that when
the stiffness of the rail pad increases, the vertical displacements of the rail decrease, but
Infrastructures 2022, 7, 83 13 of 20
the vertical displacements on the sleepers increase. There is no difference in shear stresses
on the rails, but the stiffer pads generate a slight increase of shear stresses on the sleepers,
which means a higher stress over the ballast layer. Rail bending moments have a slight
bending
increasemoments have
(the biggest a slight increase
difference in this
is less than case,
4.5% but the
when there arestiffness
pad always values under
decreases the
(Error!
maximum allowed ones (see
Reference source not found.)). Table 7).

Infrastructures 2022, 7, x FOR PEER REVIEW 14 of 22

Figure 9. Case study with different pad materials (up) and different geometry (down).
Figure 9. Case study with different pad materials (up) and different geometry (down).

The variation with pad dimensions is also analysed. Simulations were performed by
Table 7. Maximum values and allowed values to the analysed variables according to references.
changing one of the dimensions of the pad, the longitudinal direction of the pad, and
Track Element keeping
Variable the transverse directionLimits
constant. Normal pad dimensions are
(Units) 150 × 180 mm. For
Reference
the analysis, the longitudinal values were varied from 100 mm and 200 mm to 300 mm so
Rail Displacements Conventional track 1.0 (mm) Maynar (2008) [28]
to see clearly their influence in the results.
High-speed track 0.75 (mm)
Generally speaking,UIC
Stress the60 results are better
Compressive, for the larger pads.
72 (MPa) The vertical
Martínez displace-
(1992) [37]
ments, the shear stress on the sleepers, and the
UIC 60 Axial, 92 (MPa) moments on rails decrease. However, on
the other hand, the forces on the pad are considerably greater in the case of the smaller-
52 kg/mL Rail, 180 (MPa) Singh (2016) [38]
dimension pads. Regarding60the kg/mL Rail,in
increase 225 (MPa)
the spacing between the sleepers, an increasing
60 kg/mL Rail (Break),
trend can be considered with an increasing distance 880 (MPa) Lichtberger
(Error! Reference (2011)
source not[35]
found.).
Bending stress 130 (MPa) Eisenman & Leykauf (1993) [39]
The distance between axle bogies in the train
UIC 60 (Corrosion-CWR) 200 (MPa)
is also considered and simulated.
Doyle (1980) [29]
The
results showed that increasing the distance between
UIC 60 (New-CWR) 282 (MPa) the axles in the bogies from 1.8 m to
2.5 m is positive for allUIC
variables except one, the
60 (Corrosion-jointed) 210bending
(MPa) moments in rails. Values for rail
bending moments haveUIC 60 (New-jointed)
a slight increase in 320this(MPa)
case, but there are always values under
the maximum allowed ones fatigue (Jointed)
(see Table 78,000
7). (kN/m2 ) Doyle (1980) [29]
fatigue (CWR) 176 (MPa)
Sleeper Displacements 5.0 (mm) Bisht (2015) [22]
Stress 6.5 (MPa) Doyle (1980) [29]
Compressive resistance 45 (MPa) Doyle (1980) [29]
Softwood sleepers 1000–1500 (MPa) Esveld (2001) [27]
Contact pressure between rail
Hardwood sleepers 1500–2500 (MPa)
and sleeper
Concrete supports ≤ 4000 (MPa)
Bending stress Tensile stress 5.5 Doyle (1980) [29]
Ballast Stress 0.3 (MPa) Lichtberger (2011) [35]
0.5 (MPa) Esveld (2001) [27]
Infrastructures 2022, 7, 83 14 of 20

4. Discussion
This section establishes a range of possible distances between the sleepers’ axes by
comparing the maximum allowed values of the different values found in the guidelines,
standards, and literature with the results obtained in the analyses performed. A summary
of the maximum admissible values for displacement and stresses is presented in Table 7.
One of the most restrictive variables is the vertical displacement of the rail. Nowadays,
it is limited to 1 mm of vertical displacement. This value is overcome when the sleeper
spacing is higher than 0.6 m or 0.65 m. However, there are several works providing limiting
values higher than 1 mm, as it is increased to 2 mm [28,29]. If this limit is increased to 2 mm,
all distances between the sleepers’ axes analyses give values lower than this (Figure 10).
Increasing the limit to just 1.5 mm, the distance between the sleepers can be increased up
to 0.9 m.
Shear stresses on rails are for all analysed cases are lower than the ones allowed. The
same happens with the shear stress and bending stress on the sleepers (Figure 11). In the
same way, the shear forces in rails, sleepers, and rail bending moments are well within
the maximum allowed values. Therefore, this preliminary analysis shows that vertical
displacements in the rail seems to be the most restrictive variable of all analysed, together
with the stresses on the ballast.
Stresses on the ballast are a clear limitation for the sleeper’s distance, although there is
not a clear limit, as it can be 300 kN/m2 , 500 kN/m2 , or 555 kN/m2 according to [27,30,35].
These stresses are directly affected by the support area considered; therefore, it is important
to consider a suitable effective support area of the sleeper. In this case, the assumption
taken to calculate the support area is the one proposed by [27], which is to consider half of
the sleeper area. This corresponds to the case F5 (see Table 8). As other authors consider
different assumptions, in this study several values have been considered. Other lower
Infrastructures 2022, 7, x FOR PEER REVIEW
values are chosen, as they give more restrictive results and with just one being bigger 16 of 22
(Table 8).

Figure10.
Figure 10.Maximum
Maximum allowed
allowed limits
limitsfor
forvertical
verticaldisplacements
displacements on
onrails
railsand
andsleepers
sleepersaccording
accordingto
to
Table 7 values.
Table 7 values.

Shearthe
From stresses on rails
analysed are there
cases, for allisanalysed cases are
no possibility to lower
have thethandesirable
the ones stress
allowed. The
under
same
300 kN/m 2 . However,
happens with thewhen
shear considering
stress and bending
500 kN/m 2 ason
stress thelimiting
the sleepersvalues,
(Error!four
Reference
of the
source not
support found.).
options In the
(F3, F4, F5, same way,
and F6) the shearthe
accomplish forces
valueinfor
rails, sleepers,
different and spacings
sleeper rail bending
up
moments are well within the
to 0.85 m in some cases (Figure 12). maximum allowed values. Therefore, this preliminary anal-
ysis It
shows that vertical
is important displacements
to consider that, many in the railthe
times, seems to be thefor
limitations most
the restrictive variable
track elements are
of all analysed,
difficult to reach together withtimes,
and at many the stresses on the
the values areballast.
over these values [28,29], such as for the
example of the vertical displacements of the rail.
Vertical accelerations obtained in the train body and the bogies are low if they are
compared, for example, to the limits of 0.7·g (6.86 m/s2 ) in Chinese railways [33], or
 Figure 10. Maximum allowed limits for vertical displacements on rails and sleepers according to
Infrastructures 2022, 7, 83 15 of 20
Table 7 values.

Shear stresses on rails are for all analysed cases are lower than the ones allowed. The
0.22 m/s 2 [1] or 0.20 m/s2 –0.30 m/s2 [34]. In all simulated cases, the vertical body acceler-
same happens with the shear stress and bending stress on the sleepers (Error! Reference
ations obtained
source are far
not found.). In away fromway,
the same thesethe
values.
shearFor wheelsets,
forces in rails,some authors
sleepers, give
and railstandard
bending
deviationarevalues 2 and 0.47 m/s2 for a train speed of 137 km/h,
moments well(STD
withinvalues) of 0.18 m/s
the maximum allowed values. Therefore, this preliminary anal-
andshows
the bogie 2 [40]. Values for comfort in trains are
ysis thatframe reaches
vertical values upin
displacements tothe
1.83rail
m/sseems to be the most restrictive variable
under 2
1 m/s . together with the stresses on the ballast.
of all analysed,

Figure 11. Maximum allowed shear stress and bending stress in rails according to Table 7 values.
Figure 11. Maximum allowed shear stress and bending stress in rails according to Table 7 values.
Infrastructures 2022, 7, x FOR PEER REVIEW 17 of 22

Stresses on the ballast are a clear limitation for the sleeper’s distance, although there
Table 8. Effective support area Fi considered of sleepers to calculate stress on ballast (Figure 12).
is not a clear limit, as it can be 300 kN/m2, 500 kN/m2, or 555 kN/m2 according to [35, 27,
F5 0.39 (Half sleeper) <0.60 m 0.80–0.85 m 0.85–0.90 m
30]. These stresses are directly
Effective Sleeper affected by Limit
the support area considered;
Limit therefore,
Limitit is im-
Fi F6 2 0.4 2 <0.602m 0.80–0.85 m 20.85–0.90 m
portant to consider a suitable
Support (m ) effective(300 support
kN/m area) of (500
the kN/m
sleeper.) In this (555case,
kN/m the) as-
sumption
F1 taken to calculate the support area
From the analysed cases,
0.25 is the
<0.60there one proposed
m is no possibility by [27], which
<0.60tomhave the desirable is
m con-
to
<0.60stress under 300
sider half of the sleeper area. This corresponds to the case F5 (see Table 8). As other
F2 kN/m 2. However, when considering
0.2756 <0.60 m 500 kN/m 2 as the
<0.60 mlimiting values, four
0.60–0.65 of
mtheau-
support
thors consider different assumptions, in this study several values have been considered.to 0.85
F3 options
0.30 (F3, F4, F5, and F6) accomplish
<0.60 m the value for
0.60–0.650 different
m sleeper spacings
0.65–0.70 m up
OtherF4lower values m arein some cases (Error! Reference
0.35
chosen, as they give<0.60 m source
more not0.70–0.75
restrictivefound.). m
results and with 0.80–0.85
just onem be-
F5 0.39 (HalfItsleeper)
is important to consider
<0.60that,
m many times, the limitations
0.80–0.85 m for the trackm
0.85–0.90 elements are
ing bigger
F6 (Error! Reference
0.4 to source
difficult reach andnot
at found.).
many
<0.60times,
m the values are over
0.80–0.85 m these values [28,29],
0.85–0.90 m such as for
the example of the vertical displacements of the rail.
Table 8. Effective support area Fi considered of sleepers to calculate stress on ballast (Figure 12).

Limit (300 Limit (500 Limit (555

Fi Effective Sleeper Support (m2)
kN/m2) kN/m2) kN/m2)
F1 0.25 <0.60 m <0.60 m <0.60 m
F2 0.2756 <0.60 m <0.60 m 0.60–0.65 m
F3 0.30 <0.60 m 0.60–0.650 m 0.65–0.70 m
F4 0.35 <0.60 m 0.70–0.75 m 0.80–0.85 m

Figure 12. Importance of effective area considered under sleeper. It is important in order to calculate
Figure 12. Importance of effective area considered under sleeper. It is important in order to calculate
the stress in the ballast surface just under the sleeper.
the stress in the ballast surface just under the sleeper.
Vertical accelerations obtained in the train body and the bogies are low if they are
compared, for example, to the limits of 0.7·g (6.86 m/s2) in Chinese railways [33], or 0.22
m/s2 [1] or 0.20 m/s2–0.30 m/s2 [34]. In all simulated cases, the vertical body accelerations
obtained are far away from these values. For wheelsets, some authors give standard de-
viation values (STD values) of 0.18 m/s2 and 0.47 m/s2 for a train speed of 137 km/h, and
the bogie frame reaches values up to 1.83 m/s2 [40]. Values for comfort in trains are under
2
Infrastructures 2022, 7, 83 16 of 20

Values of vertical acceleration on sleepers measured in the field are between 11·g
(107.8 m/s2 ) and 16·g (156.8 m/s2 ) when the sleeper is unsupported [33]. Other authors
present limiting values up to 80 m/s2 and impact values up to 100 m/s2 [34]. Normal
vertical acceleration values for conventional speeds are around 10–11 m/s2 [31,32] and
1.8·g (17.64 m/s2 ) or 2·g (19.6 m/s2 ) [38].
If the most restrictive values of vertical acceleration are chosen as the limit 10 m/s2 , it
is overcome when the sleepers spacing is 0.8 m with train speeds from 110 km/h onwards
(Figure 13). However, with lower distances between the sleepers’ axes, such as 0.7 m,
trains can reach speeds higher than 200 km/h. On the other hand, if this limiting value is
increased up to 20 m/s2 , which is still much lower than the accelerations measured in the
field, there are no problems with train speeds up to 275 km/h or sleeper spacings.
Vibrational analysis and fatigue are not considered in this initial study. In the men-
tioned article by Zou et al. [16], in their vibrational study for urban tracks, they recom-
mended to choose the optimum sleeper spacing between 0.54 m and 0.67 m. Other distance
arrangements can be found in [15], by the repetition of a track unit composed of five
sleepers’ bays of 0.6 m spacing and five bays of 0.7 m.
Maximum values for the vertical acceleration in bogies [40] and the train body [33] are
under the allowable limits (please notice that track irregularities have not been considered
in the simulation). The most sensible parts are the sleepers, but depending on the limits
imposed, as maximum values can go from 10 m/s2 [31,32] to more than 150 m/s2 [33]. By
considering vertical accelerations up to 100 m/s2 , there is no problem with increasing the
sleeper spacing. Considering lower values, such as 20 m/s2 , a distance of 0.8 m between
Infrastructures 2022, 7, x FOR PEER REVIEW 18 of 2
sleepers can be achieved for train speeds of more than 220 km/h (Figure 13). Therefore, it
can be said that for a distance over 0.8 m, normal train speeds (more than 200 km/h) will
not be a problem.
According to increased up tonumerical
all previous 20 m/s2, which is still much
simulations, the lower than the accelerations
most restrictive variables aremeasured
ver- in the
field, there are no problems with train speeds up to 275 km/h or sleeper spacings.
tical displacements on rails and the stress in the ballast. These two parameters are defined
Vibrational analysis and fatigue are not considered in this initial study. In the men
by their limits, in the case of vertical displacements they go from 3 mm in conventional
tioned article by Zou et al. [16], in their vibrational study for urban tracks, they recom
tracks to 1 mm in high-speed tracks, while the maximum admissible ballast stress depends
mended to choose the optimum sleeper spacing between 0.54 m and 0.67 m. Other dis
mainly on the maximum allowable limit for this value (300 or 500 kN/m2 ) and the effective
tance arrangements can be found in [15], by the repetition of a track unit composed of five
support of the sleeper.
sleepers’ bays of 0.6 m spacing and five bays of 0.7 m.

Figure 13. Limits for vertical acceleration on sleepers.

Figure 13. Limits for vertical acceleration on sleepers.

Maximum values for the vertical acceleration in bogies [40] and the train body [33
are under the allowable limits (please notice that track irregularities have not been con
sidered in the simulation). The most sensible parts are the sleepers, but depending on the
limits imposed, as maximum values can go from 10 m/s2 [31,32] to more than 150 m/s
[33]. By considering vertical accelerations up to 100 m/s2, there is no problem with increas
Infrastructures 2022, 7, 83 17 of 20

Results for these two most restrictive variables are summarized in Table 9. Looking at
it, it can be said that it is possible to increase the sleeper spacing from 0.6 m to 0.65 in the
most restrictive cases and to intervals between 0.8 m and 0.85 m in other cases.

Table 9. Sleeper spacing gaps according to maximum values of most important variables in the analysis.

Variable Values Range of Distance

1 0.6–0.65 m
Vertical displacements (Rail) (mm) 1.5 0.9 m
3 No problem
300 Impossible
F1
500 Impossible
300 Impossible
F2
500 Less than 0.6 m
300 Impossible
F3
500 0.6–0.65 m
300 Impossible
Stress on ballast (kN/m2 ) F4
500 0.7–0.8 m
300 Impossible
F5
500 0.8–0.85 m
300 Impossible
F6
500 0.8–0.85 m
300 0.65–0.7 m
F7
500 No problem

Sleeper spacing is important in the design and optimization of railway superstructures.

This parametric analysis shows an initial study of how it can affect the track performance.
Results have demonstrated that it is possible to increase this distance from the current
600 mm commonly used in different railway administrations around the world. It is
important from two points of view. The economic, as saving sleepers per km of track means
saving money, and it also involves environmental savings (as it is possible to reduce energy
consumption in the process and reduce the carbon footprint as well). If one concrete sleeper
produces 0.83 KgCO2/sleeper [41,42], if the sleeper spacing changes from 0.6 m to 0.8 m, it
is possible to save 346 Kg/CO2 per km of the track. Therefore, environmental implications
also need to be considered.

5. Conclusions
Sleeper spacing analysis is an important issue in railways from an economic, environ-
mental, and safety point of view. Railway track costs can be reduced if these elements of the
track superstructure are reduced. Not only the cost, but the safety values are reduced when
the sleeper spacing is increased. Therefore, in order to check the feasibility of increasing
this distance and to reduce some sleepers per km, a dynamic numerical analysis has been
carried out. This study is within the frame of the ODSTRACK project, which aims to
analyse the track performance under different sleeper spacings.
The aim of this initial analysis was to study the effect produced by the distance between
sleepers in several elements of the railway structure. The distances analysed go from the
conventional 0.6 m (in Europe) to 1.0 m between sleeper axes every 0.1 m. The elements
analysed of the track structure were the rail, sleepers, and ballast. The parametric study
carried out served to determine the most sensitive variables to sleeper spacing changes,
which are the rail displacements and the ballast stresses. The used software is a powerful
tool for such parametric analysis.
This preliminary analysis determined that it is possible to increase the distance be-
tween sleepers beyond the conventional distance of 0.6 m, and therefore it is possible to
Infrastructures 2022, 7, 83 18 of 20

save costs on these track elements, not only during new track construction, but during
track maintenance and renewal.
Results showed that the distance between the sleepers’ axes between 0.8 and 0.85 m is
feasible (Table 9), but more studies should be performed to validate it. With this increase in
the distance between sleepers, it is possible to save between 20 and 30% of the construction
costs [10,13]. This must be compared with the expected degradation rate of the ballast. In
this case, it is important to analyse the ballast contact stress limits and the effective area
of the sleeper support. Thermal effects were considered in [13], resulting an increase of
three times the longitudinal rail axial forces when the sleeper spacing increased. These
initial results will help to develop, with accuracy, a laboratory test for a deeper knowledge
of the sleeper spacing.
Following this premise, more numerical and experimental studies are being performed
to validate further the limits of sleeper spacings. The bending of sleepers has to be consid-
ered. Vibrational analysis and fatigue analysis should be performed to have a complete
knowledge of the consequences of increasing the distance between the sleepers’ axes. More
complex numerical analysis should be carried out, as well as a 3D analysis so to have further
knowledge of the influence on the track structure (with a focus on rail fastenings and rail
joints). In addition, experimental analysis should also be performed to fully understand
the impact of increasing the sleepers’ spacing. Future lines of research should follow these
mentioned steps.

Author Contributions: Conceptualization, R.S. and M.M.; methodology, R.S.; software, R.S.; valida-
tion, R.S., V.M., and B.A.; formal analysis, M.M.; investigation, R.S., B.A., and V.M.; resources, R.S.;
data curation, M.M.; writing—original draft preparation, R.S.; writing—review and editing, M.M.;
visualization, B.A.; supervision, V.M.; project administration, R.S. and B.A.; funding acquisition, R.S.
All authors have read and agreed to the published version of the manuscript.
Funding: This research was funded by MCIN/AEI/10.13039/501100011033 and by “ERDF A way
of making Europe”, grant number RTI2018-096809-J-I00, and the APC was funded by MCIN/AEI/
10.13039/501100011033 and by “ERDF A way of making Europe”.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: Not applicable.
Acknowledgments: This research has been possible thanks to Grant RTI2018-096809-J-I00, funded
by MCIN/AEI/10.13039/501100011033, and by “ERDF A way of making Europe”.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Sañudo, R.; Cerrada, M.; Alonso, B.; dell’Olio, L. Analysis of the influence of support positions in transition zones. A numerical
analysis. Constr. Build. Mater. 2017, 145, 207–217. [CrossRef]
2. Cerrada, M. Análisis Dinámico y Numérico de la Vía en las Zonas de Transición: La Influencia de la Posición de los Apoyos.
Master’s Thesis, University of Cantabria, Santander, Spain, 2016.
3. Sañudo, R.; dell’Olio, L.; Casado, J.A.; Carrascal, I.A.; Diego, S. Track transitions in railways: A review. Constr. Build. Mater. 2016,
112, 140–157. [CrossRef]
4. García_Díaz de Villegas, J.M. Apuntes de Ferrocarriles; Universidad de Cantabria: Santander, España, 1998.
5. Namura, A.; Kohata, Y.; Miura, S. Effect of sleeper size on ballasted track settlement. Q. Rep. RTRI 2004, 45, 156–161. [CrossRef]
6. Li, D.; Davis, D. Transition of railroad bridge approaches. J. Geotechn. Geoenviron. Eng. 2005, 131, 1392–1398. [CrossRef]
7. Sañudo, R.; Markine, V.L.; Dell’Olio, L. The Effect of Increasing Train Speed on Track Transition Performance; Paper 13.200
from CCP: 110. In Proceedings of the Thrid International Conference on Railway Technology: Research, Development and
Maintenance, Calgari, Italy, 5–8 April 2016; ISBN 978-1-905088-65-2. [CrossRef]
8. Zhai, W. Vehicle–Track Coupled Dynamics; Springer: Singapore, 2020. [CrossRef]
9. Song, Y.; Wang, Z.; Liu, Z.; Wang, R. A spatial coupling model to study dynamic performance of pantograph-catenary with
vehicle-track excitation. Mech. Syst. Signal Process. 2021, 151, 107336. [CrossRef]
Infrastructures 2022, 7, 83 19 of 20

10. Sañudo, R.; Cerrada, M.; Sainz, J.A.; Carrascal, I.; Casado, J.A.; Diego, S. Optimal distance between sleepers, a numerical analysis.
In Proceedings of the Fourth International Conference on Railway Technology: Research, Development and Maintenance
Railways, Sitges, Spain, 3–7 September 2018.
11. De Rus, G.; Inglada, V. Análisis Coste-Beneficio del Tren de Alta Velocidad en España. Rev. Econ. Apl. 1993, 1, 27–48.
12. Campos, J.; de Rus, G.; Barron, I. The Cost of Building and Operating a New High-Speed Rail Line. 2007. Available online:
https://ideas.repec.org/p/pra/mprapa/12396.html (accessed on 4 May 2022).
13. Ortega, R.S.; Pombo, J.; Ricci, S.; Miranda, M. The importance of sleepers spacing in railways. Constr. Build. Mater. 2021,
300, 124326. [CrossRef]
14. Comité D-117; ORE. Adaptation Optimale de la Voie Classique au Traffic de l’Avenir; Question D117, Rapport, 1; Office de Recherches
et d’Essais de l’Union Internationale des Chemins de Fe: Paris, France, 1983.
15. Wu, T.X.; Thompson, D.J. The influence of random sleeper spacing and ballast stiffness on the vibration behaviour of railway
track. Acta Acust. United Acust. 2000, 86, 313–321.
16. Zou, Y.; Wen, Y.; Sun, Q. Study on the Urban Rail Transit Sleeper Spacing Considering Train System. MATEC Web Conf. 2019,
296, 01008. [CrossRef]
17. Weber, J.W. Concrete crossties in the United States. Pre-Stressed Concr. Inst. J. 1969, 14, 46–57. [CrossRef]
18. El–Sayed, H.M.; Zohny, H.N.; Riad, H.S.; Fayed, M.N. Structural response of monoblock railway concrete sleepers and fastening
systems subject to coupling vertical and lateral loads: A numerical study. Structures 2021, 34, 995–1007. [CrossRef]
19. Ngamkhanong, C.; Ming, Q.Y.; Li, T.; Kaewunruen, S. Dynamic train-track interactions over railway track stiffness transition
zones using baseplate fastening systems. Eng. Fail. Anal. 2020, 118, 104866. [CrossRef]
20. Sañudo, R.; Markine, V.; Dell’Olio, L. Optimizing track transitions on high speed lines; IAVSD2011. In Proceedings of the 22nd
International Symposium on Dynamics of Vehicles on Roads and Tracks, Manchester, UK, 14–19 August 2011.
21. Sañudo, R.; Markine, V.; Pombo, J. Study on different solutions to reduce the dynamic impacts in transition zones for high-speed
rail. J. Theor. Appl. Vib. Acoust. 2017, 3, 199–222. [CrossRef]
22. Sañudo, R.; Miranda, M.; Markine, V. The influence of train running direction and track supports position on the behaviour of
transition zones. Transp. Res. Procedia 2016, 18, 281–288. [CrossRef]
23. Usamah, R.; Kang, D.; Ha, Y.D.; Koo, B. Structural Evaluation of Variable Gauge Railway. Infrastructures 2020, 5, 80. [CrossRef]
24. Atalan, M.; Prendergast, L.J.; Grizi, A.; Thom, N. A Review of Numerical Models for Slab-Asphalt Track Railways. Infrastructures
2022, 7, 59. [CrossRef]
25. Sañudo, R.; Jardí, I.; Martínez, J.C.; Sánchez, F.J.; Miranda, M.; Alonso, B.; dell’Olio, L.; Moura, J.L. Monitoring Track Transition
Zones in Railways. Sensors 2021, 22, 76. [CrossRef]
26. Koyama, E.; Ito, K.; Hayano, K.; Momoya, Y. A new approach for evaluating lateral resistance of railway ballast associated with
extended sleeper spacing. Soils Found. 2021, 61, 1565–1580. [CrossRef]
27. Esveld, C. Modern Railway Track; MRT-Productions: Zaltbommel, Netherlands, 2001; Volume 385.
28. Maynar, M.M. Apuntes de Introducción a la Dinámica Vertical de la Vía ya Las Señales Digitales en Ferrocarriles: Con 151 Programas en
Matlab, Simulink, Visual C++, Visual Basic y Excel; Ingenieria De Ferrocarriles Metros Y Tuneles Sl: Madrid, Spain, 2008.
29. Doyle, N.F. Railway Track Design: A Review of Current Practice; National Academies of Science Engineering and Medicine:
Washington, DC, USA, 1980.
30. Bisht, S. Heavy Haul Track Structure Design—An opportunity to save cost. In Proceedings of the IHHA 2015 Conference, Perth,
WA, Australia, 24–25 June 2015.
31. Cardona, D.A.R.; Benkahla, J.; d’Aguiar, C.; Sofia, N.; Calon, A.; Robinet, H.; Di Benedetto, H.; Sauzéat, C. High-Speed Ballasted
Track Behavior with Sub-Ballast Bituminous Layer. In Proceedings of the 2nd International Symposium on Railway Geotechnical
Engineering, Georail, Marne la Vallée, France, 6–7 November 2014; pp. 139–148.
32. Insa, R.; Salvador, P.; Inarejos, J.; Roda, A. Analysis of the influence of under sleeper pads on the railway train/track dynamic
interaction in transition zones. Proc. Inst. Mech. Eng. Part F J. Rail Rapid Transit. 2012, 226, 409–420. [CrossRef]
33. Shi, J.; Chan, A.H.; Burrow, M.P. Influence of unsupported sleepers on the dynamic response of a heavy haul railway embankment.
Proc. Inst. Mech. Eng. Part F J. Rail Rapid Transit 2013, 227, 657–667. [CrossRef]
34. UNE-EN 12299:2009; Aplicaciones Ferroviarias. Comodidad de Viaje para los Pasajeros. Medición y Evaluación; AENOR
Guidelines; Asociación Española de Normalización y Certificación (AENOR): Madrid, Spain, 2009; p. 46.
35. Lichtberger, B. Manual de Vía; Eurailpress, DVV Media Group: Hamburg, Germany, 2011.
36. Gao, X.; Wang, A.; Liu, L.; He, Y.; Ju, L. Analysis of failure mechanism of W1-type fastening clip in high speed railway and
structure study of damping composite. Eng. Fail. Anal. 2020, 118, 104848. [CrossRef]
37. Martínez, Á.H. El montaje de la vía de alta velocidad. Inf. Constr. 1992, 44, 49–66. [CrossRef]
38. Singh, D. Computational Stress Analysis of Rail-Wheel Model of Indian Railways. Ph.D. Thesis, Delhi Technological University,
New Delhi, India, 2016.
39. Eisenmann, J.; Leykauf, G. The effect of head checking on the bending fatigue strength of railway rails. In Rail Quality and
Maintenance for Modern Railway Operation; Springer: Dordrecht, Germany, 1993; pp. 425–433.
40. Dumitriu, M.; Fologea, D.; Cruceanu, I.C. Effects analysis of vertical track irregularities on bogie vibration-method based on
bogie modelling and wheelsets accelerations measurement. IOP Conf. Ser. Mater. Sci. Eng. 2021, 1018, 012001. [CrossRef]
Infrastructures 2022, 7, 83 20 of 20

41. ICE; MacDonald, M.; BRE. CESMM3: Carbon & Price Book 2011, 2010/2011 ed.; CESMM3 Series; ICE Publishing: London, UK;
Thomas Telford: London, UK, 2010.
42. Rempelos, G.; Preston, J.; Blainey, S. A carbon footprint analysis of railway sleepers in the United Kingdom. Transp. Res. Part D
Transp. Environ. 2020, 81, 102285. [CrossRef]