Transport Proceedings of the Institution of Civil Engineers

Volume 165 Issue TR1 Transport 165 February 2012 Issue TR1
Pages 2737 http://dx.doi.org/10.1680/tran.8.00037
Design of embankmentstructure transitions Paper 0800037
for railway infrastructure Received 22/10/2008 Accepted 26/10/2010
Published online 18/11/2011
Gallego Giner, Lopez Pita, Vieira Chaves and Keywords: embankment/mathematical modelling/rail track design
Rivas Alvarez
ICE Publishing: All rights reserved

Design of embankment
structure transitions for
railway infrastructure
Inmaculada Gallego Giner PhD Eduardo W. Vieira Chaves PhD
Associate Professor, E.T.S. Ingenieros de Caminos, Canales y Puertos, Associate Professor, E.T.S. Ingenieros de Caminos, Canales y Puertos,
Universidad de Castilla- La Mancha, Ciudad Real, Spain Universidad de Castilla- La Mancha, Ciudad Real, Spain
Andres Lopez Pita PhD Ana Mara Rivas Alvarez PhD
Professor, E.T.S. Ingenieros de Caminos, Canales y Puertos, Universidad Associate Professor, E.T.S. Ingenieros de Caminos, Canales y Puertos,
Politecnica de Cataluna, Barcelona, Spain Universidad de Castilla- La Mancha, Ciudad Real, Spain

The introduction of a structure (bridge, viaduct) along a railway represents a transition point with an abrupt change in
the vertical stiffness from one transverse section to another. This condition is even more important in high-speed
railway, since the layout criteria are stricter than for conventional train lines. However, in many cases, it is nevertheless
necessary to construct a structure (viaduct, bridge). This paper presents the key ideas that explain the atypical
behaviour exhibited by embankmentstructure transitions and demonstrates the influence that longitudinal varia-
tions in vertical stiffness have on track deterioration. The deterioration problem remains unsolved and, at present,
there are no common design solutions employed by the different European railway administrations. In the present
study, three-dimensional numerical simulations of designs, which are currently employed in embankmentstructure
transitions, are generated using the finite-element method. Analysis of the results leads to design recommendations
that can be used by railway infrastructure designers when designing embankmentstructure transitions.

Notation The interaction between the dynamic loads, the tracks vertical
b variable related to the track defects and to the defects and the structure causes a rapid deterioration of the
vehicle defects railroad geometry and premature damage to the interaction site
c cohesion (N/m2 ) (Gallego Giner, 2006; Gallego Giner et al., 2005).
E elastic modulus (MPa)
GC cement-treated material with an elastic modulus of Detailed analysis of this topic, which is shown below, indicates
18 3 103 MPa that the damage is caused by variation in the vertical stiffness
K vertical track stiffness (kN/mm) from one section of the railway track to the next. This damage
m unsprung mass of the vehicle (t) gives rise to an increase in the dynamic overloads and hence to
MT cement-treated material the creation of differential settlement. The overloads and settle-
MGT cement-treated material with an elastic modulus of ments, in turn, are affected by factors that are related to the three
1.6 3 102 MPa elements that interact in those zones: the track, the embankment
QS1, QS2 material classifications determined by the Comite and the structure.
and QS3 D-117 (ORE, 1983a, 1983b)
SC cement-treated material with an elastic modulus of 1.1 The importance of variation in the vertical
5 3 103 MPa stiffness.
V running speed of the vehicle (km/h) Since railway companies began to consider the necessity of
Poisson9s ratio increasing overall train velocities, quantifying the vertical loads
r density (N/m2 ) exerted on the track by high-speed trains has become more
(QNS ) typical deviation of the dynamic overloads due to important. The loads are considered to be one of the main factors
the unsprung masses of the material (t) responsible for the deterioration of the track geometry. In this
internal friction angle (8) respect, the Prudhomme formulation (Prudhomme, 1970) is
j() damping of the track particularly relevant and is expressed as

1. Embankmentstructure transitions:
behaviour and design V p
Experience shows that embankmentstructure transitions are a 1:
QNS 0:45 b
m
K
j
100
source of frequent problems and require regular maintenance.

27
Downloaded by [ Centro Federal de Educacao T^echnologica de Minas Gerais] on [08/06/17]. Copyright ICE Publishing, all rights reserved.
 Transport Design of embankmentstructure
Volume 165 Issue TR1 transitions for railway infrastructure
Gallego Giner, Lopez Pita, Vieira Chaves and
Rivas Alvarez

where (QNS ) is the typical deviation of the dynamic overloads present study, an average maximum stiffness variation between
due to the unsprung masses of the material (t); V is the vehicle two consecutive sleepers of 20% has been established.
running speed (km/m); b is the variable related to the track and
vehicle defects; m is the unsprung mass of the vehicle (t); K is 1.2 Track embankmentstructure interaction
the vertical track stiffness (t/mm); and j() is the damping of the The behaviour of the transition is characterised by the differential
track. settlements that appear in adjacent structures and embankments.
Once created, these settlements can grow progressively. The
In Equation 1, K is defined as the relation between the value of process by which differential settlements are generated can be
the load applied on the rail, and the total settlement that this explained by the following succession of phases (Figure 1).
produces. Equation 1 introduces new criteria that are aimed at
reducing the influence of the track and the vehicle. In order to
(a) A difference in stiffness increases the dynamic load on the
avoid increasing the dynamic overloads due to the non-suspended
track.
masses (QNS ), these criteria suggest the importance of having a
(b) Due to the increase in the dynamic load, a differential
low value of both the track vertical stiffness (K) and the unsprung
settlement can form between the structure and the adjacent
mass of the vehicle (m). This influence becomes more relevant
embankment.
with increasing speed.
(c) This differential settlement will cause an increase in the
dynamic load the next time a train passes.
Theoretical and experimental studies have shown that the track
(d ) The dynamic load increase will cause an increase in the
vehicle interaction not only affects the absolute value of the
existing differential settlement. The last two phases will be
stiffness, but also the variation of the stiffness between adjacent
repeated for each load circulation along the embankment.
sections of track. This has been presented in other studies
(Esveld, 2001; Hettler, 1986; Hunt, 1997; Lopez Pita, 1983,
2001; Teixeira, 2003). These papers have shown that the vertical The dynamic overloads increase in value and are fundamentally
stiffness variations not only produce differential settlements, but affected by the following factors
also increase tension on the ballast, which gives rise to floating
sleepers. Lopez Pita (2001) has advised that a new quality (a) the variation in stiffness between two adjacent sections of
criterion should be included in the high-speed lines that involve a track (Lopez Pita, 2001), which is an inherent characteristic
control of the longitudinal heterogeneity in the track. In the of every embankmentstructure transition

Existence of a change of stiffness

Existence of
Value of track defects
K

Dynamic overloads increase when the train is moving

Existence of
Speed and expansion
load by axle devices

Longitudinal
movements of A larger settlement originates in the transition zone as Type of structure
the table compared to those in the structure or the adjacent transition: and foundation
A differential settlement results

Type of
abutment Geotechnical
This differential settlement generates larger dynamic
characteristics
overloads the next time a train passes of the backfill,
Factors related to the track embankment
Factors related to the and original
structure ground
Factors related to The differential settlement grows
geotechnical aspects

Figure 1. Diagram of the deterioration process of the track

geometry that originates in an embankmentstructure transition
(Gallego Giner, 2006)

28
Downloaded by [ Centro Federal de Educacao T^echnologica de Minas Gerais] on [08/06/17]. Copyright ICE Publishing, all rights reserved.
 Transport Design of embankmentstructure
Volume 165 Issue TR1 transitions for railway infrastructure
Gallego Giner, Lopez Pita, Vieira Chaves and
Rivas Alvarez

Ballast
(b) characteristics of the traffic, speed and load by axle
Sub-ballast
(Prudhomme, 1970)
Track bed Y
(c) the appearance of defects on the track, which is the main
cause of differential settlements (Prudhomme, 1970) X
Z
(d ) the existence of irregularities in the track, such as welded E1 E2 E3 E1
seams, joints, and expansion devices (Jenkins et al., 1974; PA
Kuroda, 1973). E2
Ltransition
PB
To summarise, it has been shown that the origin of settlement
E3 Beginning of the
problems is the variation in vertical stiffness between neighbour-
H - embankment transition block
ing sections of the track, which causes a dynamic overload
increase and subsequently, a differential settlement. The value of Original ground
End of the transition
the overloads and settlements, in turn, is affected by factors that block
are related to the track, the embankment and the structure.
Figure 2. General schematic of an embankment structure. The
1.3 Longitudinal design of currently employed transition is the zone adjacent to the structure (Gallego Giner,
transitions 2006)
In order to resolve these problems, solutions have been imple-
mented that primarily consist of arranging different materials in
order to achieve a gradual increase in stiffness from one sleeper Beginning of the
transition block
to another as it approaches the structure. Thus, different railway
administrations have developed different types of longitudinal
embankment sections, many of which have been included in Code Type 2
material Type 1
719R (UIC, 1994) and Code D-230.1/RP3 (ERRI, 1999). material

A review of the designs employed by the different European

railway administrations, addressed by Gallego Giner (2006), has Original ground
identified five typical types of embankments, which are listed here. End of the transition block
(a)
(a) Backfill behind the abutment that is composed of either End of the transition
highly compressed materials or granular materials treated block
with cement.
(b) Use of a transition slab that is built either with reinforced
concrete or with another material. Type 1
Type 2 material
(c) Insertion of horizontal layers made of different materials over material
the track formation.
(d ) Use of geotextiles and geosynthetics in order to reinforce the Original ground
backfill behind the abutment. Beginning of the
(e) Treatment of the track bed and sub-ballast with cement. transition block
(b)
In addition to these measures, the different materials behind the
abutment have been identified and categorised into three types. Figure 3. Diagram of the beginning and end of the technical
The two most often employed are those shown in Figures 2 and 3. blocks for slopes PB and PA (Gallego Giner, 2006)
In the present study they are referred to as slope types PA and PB.

In summary, a review of the diverse solutions employed has 2.1 The analysed domain and geometry of the model
shown a lack of homogeneity in the design solutions; each The domain analysis is shown in Figure 2, in which a generic
railway administration uses different designs for the longitudinal diagram of a transition composed of three different materials is
sections of the embankmentstructure transitions. included. The elastic modulus increases as the distance to the
structure decreases. The length of the transition can reach up to 80
2. Numerical model to 90 m if the embankment has a significantly elevated height.
In the present study, finite-element models in three dimensions Modelling the entire domain would involve a large computational
were used to simulate the behaviour of the embankmentstructure cost. Fortunately, this can be solved by the selection of a track
transitions. The aim was to quantify the vertical stiffness of the section that contains the basic characteristics of a transition. For
track according to the utilised materials and their disposition. both considered slope types (PA and PB), these characteristics occur

29
Downloaded by [ Centro Federal de Educacao T^echnologica de Minas Gerais] on [08/06/17]. Copyright ICE Publishing, all rights reserved.
 Transport Design of embankmentstructure
Volume 165 Issue TR1 transitions for railway infrastructure
Gallego Giner, Lopez Pita, Vieira Chaves and
Rivas Alvarez

when the material changes from one type to another at the 1 Y

beginning or end of the slope (Figure 2). In this way, the sections to Type1 material
be modelled are those included within the rectangular boxes shown
X
in Figure 3.
Z
With regard to the geometry of the model, the x-axis represents
the sleeper direction, the y-axis is the vertical direction and the z-
axis is the rail direction. As a simple track was considered, Type 2
symmetry with respect to the x 0 plane was applied. The material
dimensions of the transverse section used values similar to those
employed in the MadridSevilla high-speed train line (Barcones
Minguenza, 1991): 0.30 m for the ballast, 0.25 m for the sub-
ballast and 0.60 m for the track bed. The height of the embank-
Type PA31 slope
ment corresponded to the average value of the aforementioned
line (7 m) and the height of the original ground (3 m).
1 Y
Type 1 material
From the loaded sleeper, the important phenomena are apparent
up to the fourth sleeper. Therefore, four unloaded sleepers were X
introduced at each end of the model. The analysis was realised
with the central four sleepers under the railway load. Along the Z
z-axis, a domain of 12 sleepers was chosen.

2.2 Modelling elements of the track superstructure

In order to model the elements of the track superstructure (such
as the rails, the elastic bearing and the sleeper), data were taken
from the existing literature. For the rail, the main aim was to
simulate, as closely as possible, its resistance to flexure. For the
elastic bearings, the aim was to equalise their vertical stiffness.
Type 2 material Type PB11 slope
For the sleeper, since its section was not constant along its length,
it was assumed that the stiffness of each element was equal to the Figure 4. Finite-element models for the transitions with slope
stiffness of the sleeper in that zone. types PA31 and PB11 (Gallego Giner, 2006)

2.3 Material constitutive model

The constitutive models for the model materials were released to simulate real conditions. This was an improvement on
existing models of track formations.

(a) a classic isotropic, linear-elastic model was used for the rail,
In order to solve the tensional discontinuities that appear at the
sleeper, elastic bearing, and the materials treated with cement
interface of the ballast and the sleeper, the bounded degrees of
(b) a perfectly elasto-plastic model with a DruckerPrager
freedom were used in this mode to replicate a solution that has
surface was used for the remainder of the model (ballast, sub-
been adopted by the Comite D-117 (ORE, 1983a, 1983b). The
ballast, track bed, embankment, original ground).
use of bounded degrees of freedom required the introduction of
different nodes for each material at the contact surface. These
The different geomechanical characteristics were either those nodes must move equivalently in the direction perpendicular to
prescribed in the MadridSevilla line or included in the Comite the contact plane. However, these nodes can move at different
D-117 (ORE, 1983a, 1983b) record for the classified materials values in directions that are parallel to the contact plane.
(QS1, QS2, QS3 and rock).
The load applied to a sleeper is transmitted to two adjacent
The three-dimensional meshes that were created, which corre- sleepers. Consequently, in addition to solving for the load, the
spond to the beginning or to the end of the technical block of the history of the previous loads that have affected that sleeper must
slope types PA and PB, are shown in Figure 4. also be considered. In order to observe the behaviour of two
consecutive sleepers with different stiffness under load, while
2.4 Adopted hypothesis and load application taking into account their load history, four successively loaded
Movement perpendicular to the delimiting planes was limited at sleepers were considered. To carry out the calculations, the load
the boundary conditions. However, the boundary conditions for of only one axle was considered while assuming that the effects
the planes that constitute the slopes of different materials were of the remaining axles were negligible.

30
Downloaded by [ Centro Federal de Educacao T^echnologica de Minas Gerais] on [08/06/17]. Copyright ICE Publishing, all rights reserved.
 Transport Design of embankmentstructure
Volume 165 Issue TR1 transitions for railway infrastructure
Gallego Giner, Lopez Pita, Vieira Chaves and
Rivas Alvarez

As the behaviour of the material was not linear, it was necessary 3.2 Results of the first phase
to apply the load in several stages. As outlined below, the first The analysis of the case studies given in Table 1 shows that slope
stage only considered the weight of the materials until the tension PB31 (Figure 3) was the most suitable solution for making the
balance was reached, whereas later stages considered the loads transition from QS3 to MGT. This slope was the most accurate
over the four central sleepers that were due to the train. The type of geometry because it produced only a minor stiffness
values of the tension and displacement were calculated from the increase between the two successive sleepers and distributed the
difference between the values after the railway load passed and variation in stiffness over a greater length (Gallego Giner, 2006).
those under loaded conditions. As the calculated stiffness values in these case studies were
similar for the original ground (QS1 and QS2), QS3 and rock, it
was considered sufficient for the rest of the study to only
3. Presentation and analysis of results
calculate the cases corresponding to QS2 and QS3, which
3.1 Case studies and methodology reduced the number of cases.
Currently, the existing differences in designs used for embank-
mentstructure transitions are in the types of materials used and On the other hand, for the case studies that corresponded to
in the adopted construction methods. This justifies the employ- conventional embankments, the analysis of the results shows that
ment of different case studies by varying the geotechnical and the highest stiffness was obtained by laying granular material
geometric parameters. treated with cement on an original ground of rock; in this case,
the stiffness reached values near 87.8 kN/mm (Table 3). If it was
The geometrical parameters included the type and grade of slope assumed that the stiffness of structures with ballast could reach
(Table 1). In addition, the geotechnical parameters included the values between 130 and 150 kN/mm, then it was clear that the
modulus of elasticity of materials that compose the embankment abrupt change in stiffness near the structure had not been
fill and the original ground (Table 1). resolved in any of the analysed solutions. In each of the case
studies, this parameter changed dramatically within the first few
The geotechnical parameter values that were considered in the metres close to the abutment.
models are given in Table 2. For the modulus of elasticity, a range
of values was used that matched those adopted in the numerical To solve this problem, new numerical models were proposed that
model presented by the Comite D-117 (ORE, 1983a, 1983b). The utilised designs aimed at increasing the stiffness of the sections
elastic modulus values coincided with the lower limit values for closest to the abutment. To this end, two designs were proposed.
the following material types: QS1, QS2, QS3, and rock (Table 2). The first considered the use of materials that were treated with
cement, which had a high elastic modulus, to backfill the abutment
To make additional comparisons, the stiffness was calculated for (type A in Figure 5). The other design focused on increasing the
cases in which the fill corresponded to conventional embank- stiffness by substituting the track bed (type B in Figure 5) and the
ments that were made with the same type of material as the sub-ballast (type C in Figure 5) with the same treated material
transitions. found in the backfill behind the abutment (Figure 5).

In the first phase, all of these case studies were calculated The set of materials that were treated with cement and considered
numerically and the results were analysed. This analysis deter- in this study had the following principal characteristics.
mined whether a second phase was needed.
(a) Cement-treated granular material (MGT) having a long-term
elastic modulus of 1.6 3 102 MPa. MGT is a material made
Geometric parameters Geotechnical parameters with granular soil (similar to the track bed). The minimum
cement content was 3% by weight. The minimum value of
Type of design Slope value Type of material Original the modulus of elasticity demanded by the Spanish Railway
(Figure 1) (H : V) transition (type1/ ground Administration is 1.6 3 102 MPa (Gestor de Infraestructuras
type2) Ferroviarias, 2004) and so that value was adopted here.
(b) Cement-treated granular material (SC) with a long-term
PA 1:1 QS3/MGT* QS1 elastic modulus of 5 3 103 MPa was adopted here. SC is a
PB 3:1 QS2 material made with granulated soil. Its minimum cement
QS3 content was 3.5% by mass. The modulus of elasticity took
Rock values between 5 3 103 and 8 3 103 MPa (DGC, 2004; IEC,
*Cement-treated granular material at 3% by weight. 2003).
(c) Cement-treated granular material (GC) with a long-term
Table 1. Geometric and geotechnical parameters considered in elastic modulus of 18 3 103 MPa was adopted here. GC is a
the model (Gallego Giner, 2006) material made by dry crushing. The minimum cement content
was 3.5% by mass. The modulus of elasticity took values

31
Downloaded by [ Centro Federal de Educacao T^echnologica de Minas Gerais] on [08/06/17]. Copyright ICE Publishing, all rights reserved.
 Transport Design of embankmentstructure
Volume 165 Issue TR1 transitions for railway infrastructure
Gallego Giner, Lopez Pita, Vieira Chaves and
Rivas Alvarez

Material E: MPa v c: N/m2 : 8 r: N/m3

Rail steel 2.10 3 105 0.30 7.50 3 104

Elastic bearing 2.95 3 104
Sleeper element 1 8.01 3 104 0.25
Sleeper element 2 5.02 3 104 0.25
Sleeper element 3 3.68 3 104 0.25
Ballast 1.30 3 102 0.20 0 45 1.90 3 104
Sub-ballast 1.20 3 102 0.30 0 45 1.90 3 104
Track bed 80.00 0.30 0 35 2.00 3 104
Material QS1 12.50 0.40 15 000 10 2.00 3 104
Material QS2 25.00 0.30 10 000 20 2.00 3 104
Material QS3 80.00 0.30 0 35 2.00 3 104
Cement-treated granular material (MGT) .
1 60 3 102 0.25 2.30 3 104
Cement-treated material (SC) 5.00 3 103
Cement-treated material (GC) 18.00 3 103
Rock 3.00 3 103 0.20 2.70 3 104

Table 2. Values of the geotechnical parameters considered in the

model (Gallego Giner, 2006)

Embankment material Original ground QS1 Original ground QS2 Original ground QS3 Original ground rock

Full QS2: K(QS2): kN/mm 10.186 15.717 33.226 38.433

Full QS3: K(QS3): kN/mm 28.899 50.929 70.606 74.027
Full MGT: K(MGT): kN/mm 67.782 73.676 83.214 87.800

Table 3. Static stiffness values, K (kN/mm), for conventional

embankments (Gallego Giner, 2006)

Sub-ballast Sub-ballast Sub-ballast

Track bed Track bed Track bed

QS3 QS3 QS3

MGT, SC MGT, SC MGT, SC
or GC or GC or GC

Type A transition Type B transition Type C transition

Figure 5. Diagram of material locations: transition from QS3 to

materials treated with cement (Gallego Giner, 2006)

between 18 3 103 and 22 3 103 MPa (DGC, 2004; IEC, (a) Substituting the track bed, or the track bed and the sub-ballast
2003). with granular material that had been treated with cement
(MGT) did not solve the abrupt change in stiffness near the
3.3 Results of the second phase abutment. Therefore, MGT is not a suitable transition
The results of the numerical model mentioned in the previous material close to the structure.
paragraph are given in Table 4 and show the following beha- (b) If GC was used in the backfill of the abutment and in the
viours. formation of the track bed and sub-ballast, then very high

32
Downloaded by [ Centro Federal de Educacao T^echnologica de Minas Gerais] on [08/06/17]. Copyright ICE Publishing, all rights reserved.
 Transport Design of embankmentstructure
Volume 165 Issue TR1 transitions for railway infrastructure
Gallego Giner, Lopez Pita, Vieira Chaves and
Rivas Alvarez

Treated zone (see Figure 5) Original ground

QS2: kN/mm QS3: kN/mm

MGT (E 1.6 3 102 MPa) Embankments 73.68 83.21

Embankment + track bed 76.95 101.30
Embankment + track bed + sub-ballast 79.66 134.10
SC (E 5 3 103 MPa) Embankments 100.22 106.64
Embankment + track bed 115.78 116.43
Embankment + track bed + sub-ballast 245.26 292.16
GC (E 18 3 103 MPa) Embankments 101.30 107.40
Embankment + track bed 117.23 124.77
Embankment + track bed + sub-ballast 310.67 374.30

Table 4. Static stiffness values, K (kN/mm), for conventional

embankments with materials treated with cement (Gallego
Giner, 2006)

stiffness values were obtained when the original ground was created to calculate the stiffness values of the track at the
material was type QS3. This stiffness was even greater than beginning, KI , and the end, KF , of the transition from the QS3
the stiffness that could be reached by the structure alone. material to the SC or GC material (Figure 5). Only the proposed
(c) When the original ground was composed of QS2 material, slope type PB31 was analysed.
use of SC and GC resulted in high stiffness values (between
240 and 310 kN/mm) that were similar to those over the For the case studies (Figure 5), the static stiffness values obtained
abutment when the structure foundation was piled. for the end, KF , and beginning, KI , transitions are presented in
(d ) It was observed that the original ground had a strong Table 5.
influence on the stiffness in the designs that included
replacement of the track bed and the sub-ballast by a treated 3.4 Critical analysis of the results
material. The influence of the original ground was negligible For the cases in which the original ground material was type
in the other designs. QS2, using GC in the backfill nearest to the structure created a
zone (near the material) with a strong stiffness (Figure 6). When
In summary, the use of SC and GC to form the backfill behind SC was used in the backfill, the variation in the static stiffness,
the abutment and to replace the track bed and the sub-ballast KF , zone was less than the variation produced by using GC.
resulted in high (as well as comparable) stiffness values for the Hence, in such cases, use of SC is recommended.
track. The use of these materials as the backfill behind the
abutment would prove advantageous because they diminished the If the structure foundation is piled, then the stiffness above
abrupt change in the stiffness value in this transitional region. the abutment will be very high (likely closer to 310 than to
240 kN/mm). In this situation, it would be preferable to use GC
In order to estimate the value of the increases in stiffness, K, as the transition material. In this way, an abrupt change in the
when the material changed from one material to another, a model stiffness from one sleeper (over the treated material) to the next

Original ground QS2 Original ground QS3

Transition type Beginning (KI ) K: kN/mm End (KF ) K: kN/mm Beginning (KI ) K: kN/mm End (KF ) K: kN/mm:

QS3/MGT 58.07 99.68 70.34 118.73

QS3/SC 65.17 219.29 70.61 228.71
QS3/GC 65.40 245.26 70.61 311.67

Table 5. Static stiffness values, K (kN/mm), for transitions made

with QS3 and cement-treated materials, where the track bed and
the sub-ballast have been replaced by these treated materials
(Gallego Giner, 2006)

33
Downloaded by [ Centro Federal de Educacao T^echnologica de Minas Gerais] on [08/06/17]. Copyright ICE Publishing, all rights reserved.
 Transport Design of embankmentstructure
Volume 165 Issue TR1 transitions for railway infrastructure
Gallego Giner, Lopez Pita, Vieira Chaves and
Rivas Alvarez

K(SC) 2453 KF 2193 K(GC) 3107 KF 2453

KF 26 KF 654
(119%) (267%)

SC QS3
QS3 PB31
PB31 GC
KI 143 KI 145
(281%) (285%)

KI 652 K(QS3) 509 KI 654 K(QS3) 509

Original ground QS2
Original ground QS2

Average increase of K per

Average increase of K
sleeper 43 kN/mm (66 %)
per sleeper 5 kN/mm
*Values of K in Tables 3 and 4 (76%)
*Values of KI and KF in Table 5 Piled foundation

Figure 6. Analysis of the stiffness variations in the transitions

from material QS3 to cement-treated materials. In this case, the
original ground type is QS2 (Gallego Giner, 2006)

(over the abutment) can be avoided. There is still an abrupt K(SC) 2921 KF 22871
increase of 26.7% where the transition ends (Figure 6). However,
it is reasonable to assume that the increase is distributed over
KF 634 Sub-ballast
several sleepers (approximately four sleepers on each side of the
(277%) Track bed
most affected sleeper) and not abruptly from one section to the
next, as is the case near the abutment.
SC QS3
PB31
If the original ground material is type QS3, the most viable
option is the use of SC because the use of GC results in K(QS3) 706
excessively high values of stiffness (Table 5) that exceed those
over the structure. The variations in stiffness produced in this KI 706
situation are shown in Figure 7.
Original ground QS3
The different designs shown in Figures 6 and 7 present some
viable solutions for the stiffness variations in the sleepers closest
to the abutment. However, at the beginning and end of the Average increase of K per
sleeper 44 kN/mm (62%)
technical block of a transition from the material QS3 to the
materials treated with cement, a greater stiffness increase is
introduced as compared to cases that use granular material treated *Values of K in Tables 3 and 4
with cement (MGT). In this case, it would be possible to obtain a *Values of KI and KF in Table 5
viable solution if another material that has been treated with
cement and has a medium value of the elastic modulus was Figure 7. Diagram of the transition from material QS3 to the
introduced between material QS3 and the abutment. materials treated with cement. In this case, the original ground
type is QS3 (Gallego Giner, 2006)
Hence, to find a solution that improves the transition designs near
the abutment, it is necessary to establish a relationship between
4. Recommendations for design
the global vertical stiffness of the track and the deformability of
the material that forms the backfill behind the abutment. Along 4.1 Methodology for the design process
with this information, knowledge of the stiffness values at the The stiffness values that are obtained for the different materials
beginning and end of the technical block is essential to quantify treated with cement are related through logarithmic approxima-
the increase in stiffness produced both in the transitional zones tions and obtain very high coefficients of correlation (between
and from one sleeper to the next. These issues will be addressed 0.94 and 0.99), which are given in Figure 8. This figure shows
in the following section. that it is possible to extrapolate the vertical stiffness values of the

34
Downloaded by [ Centro Federal de Educacao T^echnologica de Minas Gerais] on [08/06/17]. Copyright ICE Publishing, all rights reserved.
 Transport Design of embankmentstructure
Volume 165 Issue TR1 transitions for railway infrastructure
Gallego Giner, Lopez Pita, Vieira Chaves and
Rivas Alvarez

track to values of the elastic modulus of the treated material with In Table 3, the stiffness values of the conventional embankment
a fair degree of accuracy. for the QS3 material are obtained by K(QS3) and consequently
the value KI KI  K(QS3). Conversely, it is possible to fix the
A method to define the total stiffness variation produced in the value of the stiffness variation to obtain the modulus of the
transition between the materials is shown in Figure 9. In the case material closest to the abutment and, consequently, the desired
of slope type PB31, the stiffness value produced at the beginning, stiffness value over the abutment.
KI , and at the end, KF , of the transition between the MT and QS3
materials is displayed (see Figure 9); it is possible to calculate 4.2 Example
the variation of this parameter using the formula KT KF  KI . With the purpose of clarifying the design methodology, an
This is presented according to the relationship between the elastic example is presented. This example is presented in Figures 8 and
modulus of material QS3 and that of the corresponding treated 9. In the case of QS2 as the original ground type, the track stiffness
material (MT). value over the abutment is set to 200 kN/mm. Applying Figure 8, a
value of 1900 MPa is obtained for the deformation modulus of the
The curves of Figure 9 were plotted by using a logarithmic treated material adjacent to the structure. Next, using Figure 9, the
approximation that relates the stiffness value at the beginning and stiffness variations that occur when this material is used as a
end of the transition to the elastic modulus values of the transition from material QS3 to the structure are ascertained. Later,
materials. Those approximations were obtained with coefficients the relation between the obtained modulus (1900 MPa) and the
of correlation between 0.93 and 1.0. value of the QS3 modulus (80 MPa) is calculated to be (1900/
80) 23.75. While analysing the intersection points of the
From the information provided in Figures 8 and 9, a solution to abscissa 28.73 with the corresponding graph (see Figure 9), it is
the transition between material QS3 and the structure can be possible to determine the values of KI and KF and consequently the
found. Starting with a given stiffness that is close in value to that values of KT , where KT KF  KI and KF K(MT)  KF . In
of the track over the structure abutment and by using the diagram this example, the values are the following: KI 62.46 kN/mm,
in Figure 8, the elastic modulus of the material behind the KF 180.26 kN/mm and K(MT) 200 kN/mm. Therefore,
abutment can be identified in order to achieve the aforementioned KT 115.87 kN/mm and KF 19.74 kN/mm. The obtained
stiffness. With the value of the elastic modulus, Figure 9 is used conventional stiffness value of the embankment for the QS3
to obtain the value of KI and KF and consequently the values of material is KI (QS3) 50.93 kN/mm and KI 13.49 kN/mm
KT and KF , where KT KF  KI and KF K(MT)  KF . (see Table 3). These values are presented in Table 6.

40000
GC (E 18 000 MPa)

35000 Embankment (Type A)

Value of the static vertical stiffness: kN/mm

SC (E 5000 MPa)

30000 Embankment Track

Original
bed (Type B)
ground QS2

25000 Embankment Track

bed Sub-ballast
(Type C)
20000 Embankment (Type A)

Original Embankment Track

15000 ground QS3 bed (Type B)

Embankment Track
10000 bed Sub-ballast
(Type C)
MGT (E 160MPa)
5000
0 5000 10 000 15 000 20 000
(E 1900 MPa) Elastic modulus: MPa

Figure 8. Values of the track vertical stiffness for embankments

made of cement-treated materials (Gallego Giner, 2006)

35
Downloaded by [ Centro Federal de Educacao T^echnologica de Minas Gerais] on [08/06/17]. Copyright ICE Publishing, all rights reserved.
 Transport Design of embankmentstructure
Volume 165 Issue TR1 transitions for railway infrastructure
Gallego Giner, Lopez Pita, Vieira Chaves and
Rivas Alvarez

350

300
Vertical stiffness value KI, KF: kN/mm

250
Beginning
Original (KI)
KF ground
200
QS2 End
(KF)
KF 18026
150 MT QS3
Beginning
KT Original (KI)
KI ground
100 KT QS3
End
(KF)
50 KI 6249
KT KF KI

0
0 50 100 150 200 250
Elastic modulus relation (E(MT)/E(QS3))

( 1900
80
2375)

Figure 9. Values of the track vertical stiffness at the beginning

and end of the transitions between material type QS3 and the
different materials treated with cement. In this case, the slope is
PB31 (Gallego Giner, 2006)

the abutment is near the value of stiffness on embankments with

Original ground materials treated with cement. On the other hand, Figures 8 and 9
are applicable for the cases in which QS2 and QS3 are the
Measures of stiffness QS2: kN/mm QS3: kN/mm original ground, otherwise extrapolating is necessary.

K(QS3) (see Table 3) 50.93 70.61

5. Conclusions
KI 62.46 70.56
An abrupt increase in the stiffness occurs when a granular
KI KI K(QS3) 13.49 0
material treated with cement (MGT at 3% by weight) is used near
KT KF  KI 115.87 140.08
the abutment. This problem can be solved with the use of SC and
KF 180.26 210.64
GC to form the backfill behind the abutment and to replace the
K(MT) (see Figure 8) 200.00 254.70
track bed and the sub-ballast because they diminish the abrupt
KF K(MT)  KF 19.74  44.06
change in the stiffness value in this transitional region.
Table 6. Proposed static stiffness values obtained for the transition
design from material QS3 to a cement-treated material with The original ground material was observed to have a strong
E 1900 MPa (Gallego Giner, 2006) influence on the stiffness when the embankment was constructed
with QS2, QS3 or MGT material. This influence of the original
ground material was negligible if the embankment was con-
structed with SC or GC material.
The case of the original ground, QS3 material was investigated in
the same way as type QS2; it is only necessary to extract the data The stiffness value over a structures abutment should be taken
from the corresponding graphs. The results for a treated material into account when designing transitions between the structure and
with a modulus of 1900 MPa are shown in Table 6. The data the abutment. Greater knowledge of the tracks vertical stiffness
indicate that the resulting stiffness variations were better than those along the structure is crucial in order for the transition design to
obtained in the case where SC or GC is used (see Figures 6 and 7). be adjusted to the specific characteristics of each structure. The
different designs (SC and GC) present some viable solutions with
The presented method assumes that the stiffness values (K) over regards to the stiffness variations in the sleepers closest to the

36
Downloaded by [ Centro Federal de Educacao T^echnologica de Minas Gerais] on [08/06/17]. Copyright ICE Publishing, all rights reserved.
 Transport Design of embankmentstructure
Volume 165 Issue TR1 transitions for railway infrastructure
Gallego Giner, Lopez Pita, Vieira Chaves and
Rivas Alvarez

abutment. However, at the beginning and end of the technical wheel/rail vertical dynamic forces. Railway Engineering
block of a transition from the QS3 material to the materials treated Journal 3(1): 216.
with cement, a greater stiffness increase is introduced in compari- Kuroda S (1973) Dynamics variation of wheel load, attributed to
son with use of granular material treated with cement (MGT). vertical deformation of rail end. Railway Technical Research
Institute Quarterly Reports 14(3): 143144.
To find a solution that improves the transition designs, it is Lopez Pita A (1983) La heterogeneidad resistente de una va y su
necessary to find a material to place behind the abutment that has incidencia en la evolucion de la nivelacion longitudinal: una
a modulus value which does not create an excessive increase in aproximacion al problema. Revista de Obras Publicas,
the stiffness close to the abutment and that, at the same time, October, pp. 719735.
does not cause a great stiffness increase in the critical points at Lopez Pita A (2001) La rigidez vertical de la va y el deterioro de
the beginning, end or material transitions. In this work, a las lneas de alta velocidad. Revista de Obras Publicas
methodology is presented that makes it possible to satisfy these 148(3415): 726.
two aims. The application of the proposed graphics improves the ORE (Office de Recherches et dEssais de lUnion Internationale
design of embankmentstructure transitions. des Chemins de Fer) (1983a) Adaptation optimale de la voie
classique au trafic de lavenir. Rapport no. 27: Comportement
REFERENCES des structures dassise de la voie sous charges repetees.
Barcones Minguenza J (1991) Infraestructura de alto Comite D-117 ORE, Paris, France (in French).
rendimiento. In Alta Velocidad Espanola Construyendo ORE (1983b) Adaptation optimale de la voie classique au trafic
Futuro. Servicio de publicaciones del Ministerio de Fomento, de lavenir. Rapport no. 28: Abaques de comportement du
Madrid, Spain, pp 3545 (in Spanish). systeme voie/fondation. Comite D-117 ORE, Paris, France (in
DGC (Direccion General de Carreteras del Ministerio de French).
Fomento) (2004) Pliego de Prescripciones Tecnicas Prudhomme MA (1970) La voie. Revue Generales des Chemins
Generales, 1st edn. Centro de Publicaciones del Ministerio de de Fer 1: 5272 (in French).
Fomento, Madrid, Spain (in Spanish). Teixeira PF (2003) Contribucion a la reduccion de los costes de
ERRI (European Rail Research Institute) (1999) Code D-230.1/RP mantenimiento de vas de alta velocidad mediante la
3. Bridge Ends Embankment. Structure Transition. ERRI, optimizacion de su rigidez vertical. Honours thesis,
Utrecht, the Netherlands. Polytechnic University of Cataluna, Barcelona, Spain (in
Esveld C (2001) Modern Railway Track, 2nd edn. MRT- Spanish).
Productions, Zaltbommel, the Netherlands. UIC (Union Internationale des Chemins de Fer) (1994) Code 719
Gallego Giner I (2006) Heterogeneidad resistente de las vas de R. Ouvrages de terre et couches dassise ferroviaires. UIC,
Alta Velocidad: Transicion terraplen-estructura. Honours Paris, France (in French).
thesis, University of Castilla-La Mancha, Ciudad Real, Spain
(in Spanish).
Gallego Giner I, Lopez Pita A, Ubalde L and Teixeira PF (2005)
Track deterioration in high-speed railway: the influence of
longitudinal variational stiffness in the embankment-structure
transitions. Congress on Railway Engineering, University of
Westminster, London, UK.
Gestor de Infraestructuras Ferroviarias (2004) Instrucciones y
Recomendaciones para redaccion de proyectos y plataformas,
Artculo 1.2. Recomendaciones sobre las cunas de transicion.
Ministerio de Fomento, Madrid, Spain (in Spanish).
Hettler A (1986) Modelluntersuchungen der Gleissetzungen am WH AT DO YO U T HI NK?

Ubergang Brucke. ETR Eisenbahntechnische Rundschau 35: To discuss this paper, please email up to 500 words to the
245250 (in German with Spanish translation by Fundacion editor at [email protected]. Your contribution will be
de Ferrocarriles Espanoles). forwarded to the author(s) for a reply and, if considered
Hunt HEM (1997) Settlement of railway track near bridge appropriate by the editorial panel, will be published as a
abutment. Proceedings of the Institution of Civil Engineers discussion in a future issue of the journal.
Transport 123(1): 6873. Proceedings journals rely entirely on contributions sent in
IEC (Instituto Espanol del Cemento y sus aplicaciones, CEDEX) by civil engineering professionals, academics and students.
(2003) Manual de firmes con capas tratadas con cemento, Papers should be 20005000 words long (briefing papers
2nd edn. Centro de Publicaciones del Ministerio de Fomento, should be 10002000 words long), with adequate illustra-
Madrid, Spain (in Spanish). tions and references. You can submit your paper online via
Jenkins HH, Stephenson JE, Clayton GA, Morland GW and www.icevirtuallibrary.com/content/journals, where you
Lyon D (1974) The effect of track and vehicle parameters on will also find detailed author guidelines.

37
Downloaded by [ Centro Federal de Educacao T^echnologica de Minas Gerais] on [08/06/17]. Copyright ICE Publishing, all rights reserved.