Accepted Manuscript

Numerical Study of Transition Zone Between Ballasted and Ballastless Railway

Track

Mojtaba Shahraki, Chanaka. Warnakulasooriya, Karl Josef Witt

PII: S2214-3912(15)00009-4
DOI: http://dx.doi.org/10.1016/j.trgeo.2015.05.001
Reference: TRGEO 42

To appear in:

Received Date: 26 January 2015

Revised Date: 17 April 2015
Accepted Date: 2 May 2015

Please cite this article as: M. Shahraki, Chanaka. Warnakulasooriya, K.J. Witt, Numerical Study of Transition Zone
Between Ballasted and Ballastless Railway Track, (2015), doi: http://dx.doi.org/10.1016/j.trgeo.2015.05.001

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Numerical Study of Transition Zone Between Ballasted and Ballastless
Railway Track

Mojtaba Shahraki1,*, Chanaka.Warnakulasooriya1, Karl Josef Witt1

1
Bauhaus-Universität Weimar, Department of Geotechnical Engineering, Coudraystrasse 11C, 99423 Weimar,
Germany

Abstract

Track stiffness plays an important role in design and construction stages of the railway track. Smoother
travel and longer life time of a track depend on the immutable track stiffness. Nevertheless, changing the
track stiffness is inevitable, especially in the transition zone, where the slab track connects to a
conventional ballasted track. In this zone track stiffness changes abruptly and it causes differential
settlement, which is the main cause for degradation of tracks and foundations at transition zones. A
number of remedies have been proposed or used to provide gradual stiffness transition, such as the use of
gradual pad stiffness, long sleepers and auxiliary rail in the transition zone. The emphasis of this study is
held on the assessment of the behavior of different types of the transition zone under the train moving
loads. For that reason, well-known commercial finite-element method package has been used to
investigate the dynamic behavior of the transition zone under the passage of high-speed trains. The results
of the dynamic analysis are presented and compared in two circumstances; one considering the common
improvement in the superstructure by installing auxiliary rails or gradually increasing the length of the
sleepers in transition area, and the other by enhancing the substructure through constructing two-part
transition section.

Keywords: Transition zone, High-speed railway embankment, Moving loads.

1 Introduction

Conventional ballasted railway tracks require periodical tamping due to uneven settlements of the ballast
during operation. The sleeper panel must be adjusted to guarantee a smooth run of the wheel sets. Based
on the existing experiences this kind of maintenance works are significantly increased for high-speed
tracks. Ballastless (slab) track constructions offer an alternative solution due to the enormous reduction of
maintenance work and the long service life with constant serviceability conditions (Audley and Andrews,
2013, Esveld and Markine, 2006). However, connection between ballastless and ballasted railway tracks is
unavoidable. Immediate change in the vertical support and discontinuity of support constituents material
cause stiffness variation that builds a stress concentrated part and raises irregularities of track level. These
changes make the connection area as one of the major sources of problems for the track. Therefore, a
special design or remedy has to contemplate in transition area between two different kinds of track system

*
Corresponding author.
Tel.: +49 36 43584569; fax: +49 36 43584564
E-Mail addresses: mojtaba.shahraki@uni-weimar.de (M. Shahraki), chanaka.warnakulasooriya@uni-weimar.de (C.
Warnakulasooriya), kj.witt@uni-weimar.de (K. J. Witt)

1
to reduce the discontinuity of track stiffness. This area is called transition zone, and its purpose is to bring
a gradual adjustment between the subgrade modulus of the slab track and the ballasted track. These zones
are the main concern to railway infra-managers, since often an additional maintenance is required to
preserve track level, and ride quality. This extra maintenance increases the exploitation costs, and
eventually causes delays in the operation of the railway track. Many railway engineers have suggested
different designs and recommendations for the construction of transition zones. However, it seems none of
available designs are ultimate solution for mitigation of track degradation. For this reason, it is essential to
understand efficiency of each approach, and choose the most suitable construction method or remedy for a
track under the passage of high-speed trains. Another important aspect of the transition zones is the
uneven settlement that causes other problems frequently, for example, hanging sleepers or unsupported
sleepers. These sleepers normally appear in the vicinity of the track with higher stiffness values, which
amplify the dynamic impact load on track. Hence, track degradation will rapidly increase. (Dahlberg,
2004, ShuGuang et al., 2007, Ferreira and López-Pita, 2013).

Finite-element method (FEM) offers a significant aid to determine the dynamic response of complex
geotechnical models, especially due to its capability of detailed simultaneous prediction of stress
distribution and displacements in the system without assuming any failure modes. The use of the finite-
element (FE) analysis has become widespread and popular in geotechnical practice for controlling and
optimizing engineering tasks. Numerical simulation can be useful to quantify long and short term behavior
of the track, besides estimation on required modifications and maintenance of whole track. For the
simulation, one should consider the model simplifications and the choice of numerical techniques. Those
simplifications help save time and money; however, there results need to be validated for practical
purposes.

To solve the problem of railway transition zone, the dynamic effect of the transition should be analyzed,
including the displacement, acceleration and stress distribution of each part of transitions. In this study,
3D finite-element models have been developed, wherein the transition is constituted by one section of
ballasted track and one section of slab track. The ballast track consists of the rails, sleepers, ballast and
soil, while the slab track consists of rails, concrete slab, mortar layer, support concrete and soil. In order to
reduce these dynamic effects on the track due to the abrupt changes in the track stiffness, different types of
remedies have been implemented. These methods are mostly designed for the transition zones in the
vicinity of bridges such as installing auxiliary rails, and gradually increasing the length of the sleepers in
transition zone. In fact, the basic theoretical aspects of changing the support stiffness at connection area
are the same; hence, most of the improvements can be used in the transition area of ballasted and
ballastless tracks. Consequently, the main objective of this study is the assessment of different transition
zone’s behavior under the train moving loads.

2 Methodology and material properties

2.1 Background
Change in the materials or structure of substructure (or superstructure) of the track cause abrupt change in
track stiffness. Therefore, in the vicinity of that area, differential elastic track deflections happen, which
result in increasing dynamic loads of the trains that can initiate localized degradation, for instance, where
the conventional ballasted track connects to modern slab track. Figure 1 illustrates abrupt variation in
track stiffness due to change of substructure material.

2
 Figure 1: Schematic representation of abrupt variations in track stiffness, after (Andre Paixao et al., 2013)

To reduce the abrupt change of the stiffness in the transition zone, research communities have introduced
different types of remedies such as installing auxiliary rails, gradually increasing the length of the sleepers
in transition area, and partially replacing the subgrade with stiffer soils. The remedies are based on the
basic technical guideline which given in literature (Lei and Mao, 2004, David Read and Dingqing Li,
2006, Shan et al., 2013). Normally having relatively high track stiffness is beneficial due to the fact that it
provides sufficient track resistance to applied loads and results in decrease of track deflections. A rail with
higher bending stiffness distributes the load over a greater number of sleepers, and as a result the
transferred force to the sleepers and rail fastening system evidently decrease. Nevertheless, high stiffness
is also a cause to increase in dynamic forces in the wheel – rail interface, sleepers and in ballast which is
ultimate reason for track wear and fatigue (Hunt, 2005, Berggren et al., 2011).

2.2 Geometry and material properties

In a 2D model, load distribution of a train in longitudinal direction cannot be modeled. Moreover, the rail
is discretely supported by the sleepers, while in a 2D plane strain model, it has been assumed that the
transversal profile of the track is invariable in the longitudinal direction with continues support. The 3D
model can model the discrete support for the rails as well as load distribution of a train in longitudinal
direction. Therefore, a 3D model has been created to study the behavior of a transition zone by
constructing different types of remedies.

In the model, the substructure of the ballasted track is composed of the ballast, the sub-ballast, capping
layer, while slab track is constructed of the Concrete Boundary Layer (CBL), Hydraulic Bonded Layer
(HBL) and Frost Protection Layer (FPL). Both, ballasted and ballastless, tracks have been supported by
three layers soil at the bottom. Depth of the upper layers and the last layer are 2 m and 20 m respectively.
This total depth of 24 m is noticeably higher than other layers which insure that no wave reflection
happens at the boundaries in the dynamic analysis. However, to minimize the computation time coarse
meshing has been applied to the last soil layer.

The superstructure of both tracks includes rails, pads, and sleepers. Normal UIC 60 rail has been modified
to rectangular section (Witt, 2008), dimensions selected in the way to maintain the properties of the
original rail satisfactory, see Figure 2. This modification reduces the computation time, and minimizes
results error due to the meshing of a complex geometry.

3
 Figure 2: Cross sectional detail of actual rail and the modeled rail [mm]

The rail pads with 10 mm thickness are provided between the sleepers, and rails. The length of the
ballasted track is 50 m including 83 standard concrete sleepers (B70), and center to center distance of
60 cm. The sleeper cross-section is rectangular 20 × 20 cm, and 2.5 m length, which is the modification of
the standard sleeper according to Witt (2008). Figure 3 illustrates the depth and the position of each layer
of ballasted track.

Figure 3: Ballasted track section from ANSYS, [m]

The slab track dimensions are adopted from the Rheda 2000 system, which is specifically introduced for
the high-speed intercity express (ICE) trains in Germany. This system is under development for a long
period of time, and has been widely used around the world. The Rheda system consists of double-block
sleepers embedded in concrete, and rail pads are located between the rail and each block of the sleeper.
Figure 4 illustrates arrangement of the sleepers and slab dimensions; all dimension are according to the
Rheda 2000 modifications (Michas, 2012). The slab track length is 30 m contains 46 double-block
sleepers embedded in concrete, and center to center distance of 65 cm. Figure 4 shows the arrangement of
the sleepers and dimensions of the slab track.

Figure 4: Modified sleeper arrangement (Michas, 2012)

4
Figure 5 illustrates the geometry of the slab track structure together with depth of each layer of modified
Rheda system.

Figure 5: Slab track section arrangement [m]

Table 1 presents the material properties of the ballasted track that have been obtained from Correia et al.
(2007), as well as the slab track material properties as stated in Michas (2012).

Density [Kg/m3] Young’s Modulus [GPa] Poisson’s Ratio [-]

Rail 6186 210 0.3

Rail pad 1950 0.21 0.3

Ballast sleeper 2054 30 0.2

Slab sleeper 2400 70 0.2

Ballast 1800 0.2 0.1

Sub-ballast 2200 0.3 0.2

Capping layer 2200 0.4 0.2

CBL 2400 34 0.2

HBL 2400 5 0.2

FPL 2400 0.12 0.2

Subgrade layer 1 2000 0.05 0.34

Subgrade layer 2 1950 0.087 0.3

Subgrade layer 3 1950 0.21 0.3

Table 1: Material properties of tracks

2.3 Transition zones

Ballasted and slab track systems represent very different dynamic behavior, and the places where two
tracks meet will have much more intriguing dynamic effect to both vehicle and the track. To mitigate the
effect of sudden support stiffness change, different types of remedies' transition zones have been
introduced. In this study, dynamic behavior of three common transition zones has been analyzed, and
compared them accordingly. These remedies are based on boosting the stiffness of the ballasted track
gradually before the slab track. Sleeper length increment and auxiliary rail are applicable for the
superstructure, and the improved subgrade approach can be constructed into the substructure. Geometry of
the first two remedies are the same, while for the last one some modifications have applied to minimize

5
the computation time. Figure 6 illustrates the zone with abrupt changes in stiffness without any transition
improvements.

Figure 6: Transition zone arrangement without any improvement

2.3.1 Increase track stiffness with longer sleepers

The particular transition zone is provided up to 7.5 m from the ballast - slab interface in the direction of
ballast based on AREMA Plan No 913-5 (David Read and Dingqing Li, 2006). The sleeper length has
gradually increased in three steps with each step having four sleepers of the same increased length. In each
step length of increments are 3.05 m, 3.35 m and 3.65 m, and no change has been made to the distanced
between sleepers or the geometry of embankment. Figure 7 illustrates general graphical description.

Figure 7: Transition zone with gradual sleeper length increment

2.3.2 Auxiliary rail

The system is originally designed for the German ICE train network. Over the years, it has performed
successfully, and it is still being used widely around the world. Auxiliary rails were used for the
prevention of growing irregularity and the reducing the variation of wheel load. Actually, auxiliary rails

6
enable for train vehicles to operate smoothly between slab track and ballasted track. In this method, the
additional rails improve the bending stiffness of the track in the area, and help to better transformation of
the loads to sleepers and consequently, reduce pressure on ballast. Here, the basic dimensions of the model
have been arranged according to (Plotkin and Davis, 2008). Nevertheless, the rail track is a complicated
structural system; therefore, some of the actual structural components have to be neglected or modified
accordingly. As an example the track base plates which are the auxiliary rail resting pads have been
neglected and it is assumed that the auxiliary rail is directly connected with sleepers without having an
intermediate medium. The provided additional rails are 30 m long, and 20 m of each lies on the ballast
side of the track and 10 m on the slab track. Additional rails have assumed to have the same properties as
main rail, and distance of each one to the near main rail is 450 mm, see Figure 8.

Figure 8: Discretized mesh of Auxiliary rail method

2.3.3 Improved subgrade transition zone

The last remedy that has been analyzed is based on adjustment of the substructure. As it shows in Figure
9, the transition zone provided with two different layers consists of Group A & B Filter and Graded
Broken Stones (GBS). Materials properties of these new subgrades have taken from study of the transition
zone between the ballasted track and the bridge abutment (Shan et al., 2013) revolved around Chinese
high-speed railway code, and those properties can be seen in Table 2. Important aspect to consider is new
introductions do not change any properties of superstructure in both ballasted and slab track, and it is
provided as an improvement for the top two subgrade layers.

Density [Kg/m3] Young’s Modulus [GPa] Poisson’s Ratio [-]

Group A & B Filter 2000 0.065 0.32

Graded Broken Stone 1950 0.075 0.3

Table 2: Material properties of introduced layers (Shan et al., 2013)

Figure 9: ANSYS models of improved subgrade system.

7
2.4 FEM specification
All elements (rail, ballast, etc.) are discretized using solid elements. SOLID186 element consists of 20
nodes having three degrees of freedom per node applied. Contact areas are modeled as homogeneous and
their properties do not change in time. The contact between each surface formulated by using CONTA174
which is the 3D surface to surface contacts with TARGE170 segments. Constitutive model of all
components considered as linear elastic material. The coordinate system is Cartesian in the connection of
the two tracks arrangements with the Z-axis positive increment to the ballasted track direction. X-axis is
representing the transverse direction, and Y-axis represents the vertical direction.

2.4.1 Loading procedure

In spite of many types of commercial software have been introduced to geotechnical engineering none of
them put the real effort for solving railway geotechnical problems. In order to analyze the transition zones
and their dynamic behavior, one of the main purposes is to provide with the moving load, and it can be
done with two kinds of procedures. One is frequency domain approach, to investigate the wheel response
to a “Moving Irregularity” and the other one is the time domain analysis, to use numerical time
interrogation process in order to investigate train - track dynamic properties, and Newmark integration
method can be identified as an example (Dahlberg, 2004). For this particular research time domain
approach is adopted (Zhao et al., 2006).

The moving load problem can be analyzed using the FEM in which the moving load is modeled by a
sequence of discrete pulses. For example, for a moving load with the constant speed, at any given time
that the load is passing one node of the rail, that node will be loaded by a force, and all other nodes will
gain the load of zero, see Figure 10. Therefore, the moving load is simulated through the definition of
stationary loads at the rail nodes with time dependent modulus. The loading configurations adopted from
the UIC code which includes four bogies of two passenger cars. The model simulates track-subgrade, and
vehicle-track model is not considered.

ANSYS transient dynamic analysis is used to determine the response of the track due to time dependent
loading conditions. Inertial and the damping effects have considered in this method. All simulations were
analyzed in the time domain using direct integration with implicit schemes. Sequential loading system by
selecting nodes in every 500 mm distance at the middle of the top rail surface has been applied (Jiang et
al., 2014). It has been assumed that train speed is 300 km/h (83.33 m/s) with the axel load of 180 kN (for a
symmetric case 90 kN). Time step has chosen 0.006 s which equals to the distance between two adjacent
loading nodes divided by the train speed.

Figure 10: Loading method of the moving load, after (Jiang et al., 2014)

8
2.4.2 Boundary conditions
The symmetric conditions have taken into account, assuming that the distribution of loads and stresses to
have symmetric distribution over the track. In order to prevent wave reflection from the boundaries, the
dimensions of subgrade provided with 15 m from the center of the track. This is to comply with the
assumption of the stress waves do not reach to the very end of the used subgrade layers. Open surfaces of
the ballast, sub-ballast, capping, CBL, HBL, and FPL layers set to free in every direction. In the case of
the transition zone with granular improvement, since the top two subgrade layers were sloped surfaces
with shorter width than the other two sections, the sloped surfaces are kept free in every direction as well.
The interface of the ballast and the slab track assumed to be having no interactions and horizontal
movement (Z-direction) has constrained. Extreme bottom surface of the subgrade layer provided with a
fixed constraint. The relative motions between all the connected surfaces were neglected to avoid other
adverse effects due to component individual behavior.

3 Results and discussion

3.1 Dynamical analyses results

Figure 11 and 12 show that when the load is traveling from lower stiffness region to the higher stiffness
region, there is a clear reduction of the deflection on top of the rail, and the subgrade surface. As expected
the lower deflection can be seen in the higher support stiffness section of the track. The auxiliary rail
clearly indicates a much improved smoother transformation to the deflection than that of the different
sleeper length case. Moreover, the peak of the diagrams at connection between two tracks for the auxiliary
rail is smaller than others that present less dynamic impact on the slab as well as passengers comfort. In
the point of modified subgrade improvement, it shows improvement for the transition. However, there is a
rapid decrease of the deflection at the interfaces where the subgrade changes to different materials with
different stiffness capacities.

Figure 11: Vertical deflection on top of the rail

9
 Figure 12: Vertical deflection on subgrade

On the point of acceleration, Figure 13 and 14 show a strong fluctuation at the interface of two types of
tracks. With the transition improvements it can be seen that auxiliary rail provides an acceptable
transformation between the two support stiffness regions, as well as strong reduction in fluctuation at the
interface. In the modified subgrade improvement case, even though the acceleration varies closer to the
auxiliary rail behavior, there is higher fluctuation of acceleration at the interfaces. This is due the combine
effects of high stiffness difference in two tracks above and the high stiffness variation between the graded
broken stone and the ordinary subgrade.

Figure 13: Maximum vertical acceleration on top of the rail

10
 Figure 14: Maximum vertical acceleration on subgrade

Figure 15 shows higher vertical stress on the subgrade in the ballast section than the slab track. Although,
there is sudden increase of the stress on the subgrade surfaces at the interface of the slab track. Auxiliary
rail system provides the step change for the stress variation. Furthermore, higher support stiffness and an
extra connection between the two track systems, reduces the sudden drastic variation of the stress on the
subgrade. Due to the stiffness changes in modified subgrade transition there are fluctuations of the stress
on the subgrade at the interface regions. The high fluctuations in the ballast and slab interface region
appear due to variation of the stiffness of the tracks as well as subgrade.

Figure 16 shows the reduction in the vertical stresses on ballast layer. Clearly, different sleeper length
method decreases the stress on the ballast more than other method because of providing higher interface
surface between bottom of sleepers and the ballast.

Figure 15: Vertical stress on subgrade

11
 Figure 16: Vertical stress on ballast

4 Conclusion

A three dimensional finite-element model has been developed to study the dynamical behavior of different
types of transition zones between ballasted and ballastless track under passage of high-speed trains. Two
widely known transitions systems, which are auxiliary rail system and variation of sleeper lengths in the
vicinity of the transition zone has been adopted for the comparison. Besides, the subgrade transition
system which mostly constructed in transition zones in bridge approaches has been used with intense
modification of cross section and material properties. A conclusion has been made on the basis of vertical
deflection, vertical acceleration, and vertical stress on the subgrade.

In general, the auxiliary rail installation shows a high rate of track dynamic improvement. It provides with
much-needed smoother transformation to reduce the impact causing due to the sudden abrupt change in
stiffness, while the modified subgrade system shows considerable effects to the dynamic characteristics in
the total system. It suggests that it has a high potential to improve to use in the track transition problems.
However, all the cases suggest that there is a sensitive zone generated when the load moving from lower
stiffness region to the higher stiffness region. This occurs within the first 5 m of the slab track from the
track interface. The strong attention has to put to increase the dynamic behavior in this region.

References

ANDRE PAIXAO, EDUARDO FORTUNATO & RUI CALCADA 2013. Design and Construction of
Backfills for Railway track Transition Zones. Proceedings of the Institution of Mechanical
Engineers, Part F: Journal of Rail and Rapid Transit, 28.
AUDLEY, M. & ANDREWS, J. 2013. The effects of tamping on railway track geometry degradation.
Proceedings of the Institution of Mechanical Engineers, Part F: Journal of Rail and Rapid
Transit, 227, 376-391.
BERGGREN, E., HOSSEINGHOLIAN, M., SAUSSINE, G., RODRIGUEZ, M., CUELLAR, V. &
VIALLETEL, H. 2011. Methods of track stiffness measurments. France patent application.
CORREIA, A. G., CUNHA, J., MARCELINO, J., CALDEIRA, L., VARANDAS, J., DIMITROVOVÁ,
Z., ANTÃO, A. & SILVA, M. G. D. Dynamic analysis of rail track for high speed trains. 2D
approach. In: SOUSA, L. R., FERNANDES, M. M. & TAYLOR, E. A. V. J., eds. 5th Intl

12
 Worksop on Application of Computational Mechanics on Geotechnical Engineering, 04-2007
2007 Portugal Taylor & Francis Group, 14.
DAHLBERG, T. 2004. Railway track settlements - a literature review. the EU project SUPERTRACK.
Linköping, Sweden.
DAVID READ & DINGQING LI 2006. Research Results Digest 79. USA: Transit Cooperative Research
Program (TCRP).
ESVELD, C. & MARKINE, V. 2006. Assessment of high-speed slab track design. European Railway
Review, 12, 55-62.
FERREIRA, P. A. & LÓPEZ-PITA, A. 2013. Numerical modeling of high-speed train/track system to
assess track vibrations and settlement prediction. Journal of Transportation Engineering, 139,
330-337.
HUNT, G. A. 2005. Review of the effect of track stiffness on track performance. In: WOOD, J. (ed.). Rail
Safety & Standards Board.
JIANG, H., BIAN, X., CHENG, C., CHEN, Y. & CHEN, R. 2014. Simulating train moving loads in
physical model testing of railway infrastructure and its numerical calibration. Acta Geotechnica,
1-12.
LEI, X. & MAO, L. 2004. Dynamic response analyses of vehicle and track coupled system on track
transition of conventional high speed railway. Journal of Sound and Vibration, 271, 1133-1146.
MICHAS, G. 2012. Slab Track System For High Speed Railways. Master Degree Project, Royal Institute
of Technology.
PLOTKIN, D. & DAVIS, D. 2008. Bridge Approaches and Track Stiffness. Washington, DC: U.S.
Department of Transportation, Federal Railroad Administration.
SHAN, Y., ALBERS, B. & SAVIDIS, S. A. 2013. Influence of different transition zones on the dynamic
response of track–subgrade systems. Computers and Geotechnics, 48, 21-28.
SHUGUANG, Z., WEIHUA, Z. & XUESONG, J. 2007. Dynamics of high speed wheel/rail system and its
modelling In: BULLETIN, C. S. (ed.). Springer.
WITT, S. 2008. The Infuence of Under Sleeper Pads on Railway Track Dynamics. Linköping University.
ZHAO, D., NEZAMI, E. G., HASHASH, Y. M. A. & GHABOUSSI, J. 2006. Three-dimensional discrete
element simulation for granular materials. Engineering Computations, 23, 749-770.

13