Construction and Building Materials 92 (2015) 58–63

Contents lists available at ScienceDirect

Construction and Building Materials

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

Ballast drainage in Brazilian railway infrastructures

Cassio Paiva a,1, Mateus Ferreira a, Adelino Ferreira b,⇑
a
Department of Geotechnics and Transport, Universidade Estadual de Campinas, Campinas, São Paulo, Brazil
b
Department of Civil Engineering, University of Coimbra, Coimbra, Portugal

h i g h l i g h t s

 Comparing clean ballast samples with samples contaminated with different volumetric percentages of soil.
 Analyzing the reduction in permeability of the ballast layer due to the presence of soil through laboratory tests.
 Analyzing the hydraulic conductivity of the ballast for different contamination rates.
 Verifying the necessity of ballast cleaning maintenance works.

a r t i c l e i n f o a b s t r a c t

Article history: The main objective of this research work was to analyze the reduction in permeability of the ballast layer
Received 2 February 2014 due to the presence of soil through laboratory tests. Granulometry and permeability tests were done
Received in revised form 27 March 2014 comparing clean ballast samples with samples contaminated with 10%, 15%, 20%, 25% and 40% volumetric
Accepted 5 June 2014
percentages of soil. The hydraulic conductivity of the ballast for different contamination rates was also
Available online 4 July 2014
analyzed. Finally, an equation was defined that can be used to predict the hydraulic conductivity for
different ballast contamination rates.
Keywords:
Ó 2014 Elsevier Ltd. All rights reserved.
Railway ballast
Drainage
Ballast contamination
Granulometry
Permeability

1. Introduction
The construction of railways in Brazil began in 1854 in the
imperial period, known as the second Reign, linked essentially to
coffee production and other agricultural products. It reached its
peak in the year 1930, in the Republican period, when more than
30,000 km of railways were built all over the country. Recently,
in particular since the 1990s, many rural railways have begun
being privatized. Currently, all national rail freight transport in
operation is under private management. Most of the State Railways
were built in the 19th century. The oldest, São Paulo Railway (SPR),
was built by British engineers and opened in 1867, linking the port
of Santos to Jundiaí, passing through São Paulo. Nowadays, the cen-
tral segments that connect the municipalities that make up the
metropolitan region of São Paulo operate in mixed traffic with a
huge predominance of metropolitan trains and little cargo traffic.
⇑ Corresponding author. Tel.: +351 239797101.
E-mail addresses: celpaiva@fec.unicamp.br (C. Paiva), mateus.affer@gmail.com
(M. Ferreira), adelino@dec.uc.pt (A. Ferreira).
1
Tel.: +55 193521 2343.
The end sections of these railway lines were ceded to concession-
ary companies primarily interested in industrial cargo transport to
the port of Santos. These railways have undergone some modifica-
tions during their lifetime, such as, for example, the increase in
length and the substitution of rigid fastenings for elastic ones. At
no time, with the exception of one, have these railways been
rebuilt to improve their load capacity or to remove the contami-
nated ballast. This situation, coupled with the interests of railway
concessionaires in maximizing profits by increasing the axle loads
and by using a maintenance process which involves merely adding
new ballast on top of the existing ballast layer, has led to several
changes in the quality of the railway lines. The old ballast was
buried into the soil layer and the fine particles of soil climbed up
to the ballast layer, closing it and creating ‘‘mud pockets’’ in rainy
seasons. This process of ballast contamination has affected its
properties as well as railway safety.
A typical cross-section of this type of railways is presented in
Fig. 1, where we can see the layers of clean ballast, usually as a
result of leveling operations, and the contaminated ballast. The
water that falls on the railway track does not drain directly into
the drainage ditch. It infiltrates and remains inside the ballast,

http://dx.doi.org/10.1016/j.conbuildmat.2014.06.006
0950-0618/Ó 2014 Elsevier Ltd. All rights reserved.
 C. Paiva et al. / Construction and Building Materials 92 (2015) 58–63 59

Table 1
Ballast granulometry [4].

Sieve size (mm) Grading category

Percentage passing by mass
A B
76.2 – 100
63.5 100 90–100
50.8 90–100 –
38.0 35–70 25–60
Fig. 1. Typical railway cross-section.
25.4 0–15 –
19.0 – 0–10
facilitating the climbing process of the fine particles of soil up to 12.0 0–5 0–5
the ballast layer. This problem is very common in all Brazilian rail-
ways built before 1970. The low permeability of the contaminated
The response of ballast to the degree of saturation is quite var-
ballast can be seen in Fig. 2, which shows the amount of water
ied. Water influences the ballast settlement, disruption of particles
retained in the ballast and the state of the platform after water
and can lead to traffic problems. Soils in saturated conditions form
infiltration. These photos were taken one day after the day of the
mud which can contaminate the ballast layer. The degree of water
rainfall.
saturation of soils can lead to an increase of 40% of the railway
platform settlement [5]. What is more, the contamination of the
2. General characteristics of railway ballast ballast with clay and silt particles reduces its drainage. These fine
particles can combine with water to form an abrasive mud, creat-
The ballast layer consists of granular material that supports the ing so-called ‘‘mud pockets’’. This contaminant material with low
railway grid, which consists of rails and sleepers, preventing their humidity can harden inside the ballast, preventing proper penetra-
displacement and providing enough elasticity to the railway, tion of the maintenance equipment [6]. According to Selig and
reducing impacts and ensuring efficient drainage and aeration. Its Waters [7], ballast contamination can occur in different ways,
main functions are [1,2]: which can be divided into five categories:

(a) Distributing stresses transmitted by the sleepers.  Breaking of ballast (13%).

(b) Attenuating the majority of train vibrations.
(c) Resisting transverse and longitudinal track shifting.
(d) Facilitating rainwater drainage.
(e) Allowing track geometry to be restored and correcting track
defects by using track maintenance equipment.
The surface of the ballast should be as flat and uniform as pos-
sible. Generally, its optimum thickness is around 25–30 cm, mea-
sured from the bottom of the sleepers. Special attention should
be given in placing the ballast layer over the subgrade to ensure
uniform settlement. Differences in the vertical settlement must
not exceed 10 mm [3]. Ballast specifications are not the same all
over the world. In Brazil, the NBR 5564 standard [4] defines the
requirements for the ballast granulometry (Table 1).
Particle size distribution has a significant influence on the
deformation behavior of railways. The elastic stiffness of the ballast
increases with an increasing uniformity of the particle size. Density
and friction angle decrease with a greater uniformity of particles.
Ballast with a more varied granulometry contains fewer voids con-
tent, making it stronger than uniform ballasts. In conclusion, the
optimal gradation of ballast must be intermediate between uni-
form and widely varied. This gradation should provide sufficient
drainage, density, shear strength and elastic modulus acceptable
for the railway [5].
 Infiltration coming from the surface of the ballast (1%).
 Wear of sleepers (3%).
 Infiltration coming from the lower granular layers (7%).
 Infiltration coming from the platform (76%).
According to Al-Qadi et al. [8], the most important form of con-
tamination is the breaking of the ballast. Tests have determined
that after performing track maintenance works twenty times, par-
ticle size is reduced considerably, up to 50% for granite ballast and
5–10% for limestone ballast, on average [9]. Selig and Waters [7]
presented findings on the permeability reduction due to ballast
contamination. The permeability of reasonably clean ballast is
about 10% of that presented by clean ballast. As contamination of
the ballast increases, permeability is rapidly reduced [10].
For better comparison of results, the level of ballast contamina-
tion will be assessed according to two indexes, one proposed by
Selig and Waters [7] and another proposed by the South-African
Railway Spoornet [9]. The first index, called fouling index, or
ballast contamination index, is defined by Eq. (1).
F 1 ¼ P 4 þ P200 ð1Þ
where P4 is the percentage of material passing in sieve no. 4
(4.76 mm); P200 is the percentage of material passing in sieve no.
200 (0.075 mm).

Fig. 2. Water retained in the ballast (left) and state of the platform after water infiltration (right).
60 C. Paiva et al. / Construction and Building Materials 92 (2015) 58–63

Table 2 assessed taking into account a larger number of sieves (Eq. (2)).
Level of ballast contamination according to the fouling index [7]. According to this index, ballast cleaning should be performed if
Fouling index Level of ballast contamination Fv is greater than 80%.
F1 < 1 Clean ballast
1  F1 < 10 Reasonably clean ballast F v ¼ ð0:4F 19 Þ þ ð0:3F 6:7 Þ þ ð0:2F 1:18 Þ þ ð0:1F 0:15 Þ ð2Þ
10  F1 < 20 Moderately clean ballast
20  F1 < 40 Contaminated ballast
F1  40 Highly contaminated ballast ð% of material in mass passing the sieve of 19 mmÞ  100
F 19 ¼ ð3Þ
27

Table 3 ð% of material in mass passing the sieve of 6:7 mmÞ  100

Soil properties [11].
F 6:7 ¼ ð4Þ
18
Description Variable Value
Depth d 0.5–2.0 m ð% of material in mass passing the sieve of 1:18 mmÞ  100
F 1:18 ¼ ð5Þ
Density of the soil in situ cnat 14.2 kN/m3 11:5
Density of solid particles cs 30.8 kN/m3
Dry density of the soil cd 11.1 kN/m3
Moisture content w 27.9% ð% of material in mass passing the sieve of 0:15 mmÞ  100
Voids ratio e 1.78
F 0:15 ¼ ð6Þ
27
Porosity n 64.0%
Degree of saturation Sr 48.4%

3. Case study
Table 4
Soil consistency limits [11]. 3.1. Introduction
Description Variable Value
The main objective of this research was to analyze the reduction
Liquid limit wL (%) 44.3
in permeability of the ballast layer, caused by the presence of soil
Plastic limit wP (%) 31.8
Shrinkage limit wS (%) 24.4
in it, using laboratory tests. Granulometry and permeability tests
Plasticity index IP (%) 12.5 were done comparing clean ballast samples and contaminated
samples with 10%, 15%, 20%, 25%, and 40% volumetric percentages
of soil. We analyzed 4 kg of ballast for each sample and the soil
The different classifications according to the value of F1 are pre- used for ballast contamination was a typical soil of the Campus
sented in Table 2. The index proposed by the South-African Rail- of Unicamp, in Campinas (SP), collected at depths between 0.50
way Spoornet defines that the ballast contamination must be and 2.00 m.

Granulometric distribution of the ballast

100

90

80
Passing percentage (%)

70
Ballast with:
60 10% of soil

50 15% of soil
20% of soil
40 25% of soil

30 40% of soil

20

10

0
0,01 0,1 1 10 100
Sieve size (mm)

Fig. 3. Particle size distribution of the ballast contaminated with 10%, 15%, 20%, 25%, and 40% of soil.

Table 5
Percentage of material passing the sieves.

Sieve size (mm) Percentage of material passing the sieve (%)

Ballast with 10% of soil Ballast with 15% of soil Ballast with 20% of soil Ballast with 25% of soil Ballast with 40% of soil
0.075 7 8 11 14 19
0.15 8 10 13 15 20
1.18 10 12 16 19 25
4.76 10 12 16 20 26
6.7 10 12 16 20 26
19 14 17 20 23 30
 C. Paiva et al. / Construction and Building Materials 92 (2015) 58–63 61

Table 6 size distribution of the ballast contaminated with 10%, 15%, 20%,
Contamination level according to the fouling index [7]. 25%, and 40% of soil. Table 5 presents the percentages of material
Ballast contamination rate (%) F1 (%) Classification passing in each sieve necessary to calculate the contamination lev-
10 17 Moderately clean ballast els according to the Selig and Waters index [7] and the index of the
15 20 Moderately clean ballast South-African Railway Spoornet [9], using Eq. (1) and Eqs. (2)–(6),
20 27 Contaminated ballast respectively. According to the first index (Table 6), the ballast with
25 34 Contaminated ballast 10% and 15% of soil is classified as moderately clean ballast while
40 45 Highly contaminated ballast
the ballast with 20% and 25% of soil is classified as contaminated
ballast, indicating that in this second situation cleaning the ballast
is necessary. The ballast with 40% of soil is classified as highly con-
Table 7 taminated ballast, indicating that in this case cleaning the ballast is
Contamination level according to the index of the South-African Railway Spoornet [9].
mandatory. According to the second index (Table 7), as expected,
Ballast contamination rate (%) Fv (%) Conclusion only the ballast with 20% of soil or more requires cleaning mainte-
10 58 It is not necessary to clean the ballast nance work.
15 70 It is not necessary to clean the ballast
20 89 It is necessary to clean the ballast
25 106 It is necessary to clean the ballast 3.4. Permeability
40 139 It is necessary to clean the ballast
Permeability tests were done using a constant load permeame-
ter to analyze clean ballast samples and samples contaminated
with 10%, 15%, 20%, 25%, and 40% of soil in terms of volume in order
3.2. Physical soil properties and classification to understand the variation of permeability. Fig. 4 shows the con-
taminated ballast inside the permeameter, which is 15.2 cm in
Table 3 shows the main physical soil properties. The values of diameter and 18.0 cm tall. Initially, the bottom of the permeameter
the soil consistency limits are shown in Table 4. According to the was covered by industrial vaseline in order to prevent water from
plasticity charter, this soil belongs to the ML group, and it is escaping during the test. In a second step, the volume of soil corre-
classified as an inorganic silt with median compressibility. sponding to the percentage previously defined was inserted into
the permeameter. In a third step, the right volume of ballast was
inserted into the permeameter. Finally, the lid was sealed with
3.3. Soil granulometry and contamination level industrial vaseline and put on the permeameter. Fig. 4 also shows
the permeameter during the test. The test results were only regis-
Granulometry tests were conducted according to the methodol- tered after 24 h of water circulating through the sample. We car-
ogy defined by the NBR 6457 standard [12]. The ballast used for ried out three repetitions for each sample making five
analysis was collected from a railway segment located near measurements at different times and considering two different
Campinas (SP), which fulfilled the A grading category of the NBR loads, 20.8 cm and 14.6 cm as typical heights of water on the rail-
5564 standard [4], presented in Table 1. Fig. 3 shows the particle way. Table 8 presents the characteristics of each test while Table 9

Fig. 4. Contaminated ballast in the permeameter (left) and permeability measurement (right).

Table 8
Test characteristics.

Test Ballast contamination rate Diameter of the permeameter Area of cross section Measurement duration Sample height Head of water
# (%) (cm) (cm2) (min) (cm) (cm)
1 10 15.2 181.16 1 16.5 20.8
2 10 15.2 181.16 1 16.5 14.6
3 15 15.2 181.16 1 15.5 20.8
4 15 15.2 181.16 1 15.5 14.6
5 20 15.2 181.16 3 17.1 20.8
6 20 15.2 181.16 3 17.1 14.6
7 25 15.2 181.16 3 18.0 20.8
8 25 15.2 181.16 3 18.0 14.6
9 40 15.2 181.16 3 15.6 20.8
10 40 15.2 181.16 3 15.6 14.6
62 C. Paiva et al. / Construction and Building Materials 92 (2015) 58–63

Table 9
Test results.

Test # Repetition Temperature (°C) Measurement 1 (ml) Measurement 2 (ml) Measurement 3 (ml) Measurement 4 (ml) Measurement 5 (ml)
1 1 28 350 352 352 352 352
1 2 28 322 325 325 325 322
1 3 28 340 346 344 340 340
2 1 28 252 255 255 255 257
2 2 28 245 245 247 245 245
2 3 28 245 247 245 245 247
3 1 23 185 190 190 190 192
3 2 23 200 202 202 200 200
3 3 23 185 190 192 200 185
4 1 23 145 150 150 150 150
4 2 23 150 150 155 155 152
4 3 23 155 152 152 155 152
5 1 18 360 365 365 365 370
5 2 18 370 365 360 362 362
5 3 18 370 375 370 370 370
6 1 18 280 285 287 290 280
6 2 18 280 285 283 283 290
6 3 18 290 280 285 282 285
7 1 27 350 360 360 360 350
7 2 27 330 335 350 330 340
7 3 27 330 330 332 335 335
8 1 27 262 270 270 268 265
8 2 27 240 245 245 247 245
8 3 27 252 247 247 247 255
9 1 21 220 220 210 215 210
9 2 21 220 225 223 210 213
9 3 21 215 220 220 213 215
10 1 21 140 140 140 135 137
10 2 21 140 135 137 140 140
10 3 21 133 135 137 140 140

presents the results. Table 10 shows the average (X), the standard Table 11
deviation (S) and the characteristic global value of measurements Hydraulic conductivity values for each ballast contamination rate.

(Xp) considering a student distribution. Situation Ballast contamination rate (%) Hydraulic conductivity (cm/s)
The hydraulic conductivity of the ballast (k) was calculated 1 10 0.0250700
using Eq. (7). 2 15 0.0132700
3 20 0.0096865
VL 4 25 0.0092075
k¼ ð7Þ
Aht 5 40 0.0047170

where V is the volume of the collected liquid; h is water load; t is

time of the test duration; A is cross section of the permeameter
and L is height of the sample.
Table 11 shows the average hydraulic conductivity values of the
ballast for each contamination rate. The temperature of the water
that passes by the permeameter must also be annotated for com-
parison of results obtained from tests carried out at different tem-
peratures. Analyzing Table 11, we can see that the hydraulic
conductivity of the ballast decreases quickly as the contamination
rate increases. For example, the hydraulic conductivity of the bal-
last contaminated with 10% of soil is 2.5  102 cm/s (average
value for tests #1 and #2) while the hydraulic conductivity of
the ballast contaminated with 40% of soil is approximately
0.5  102 cm/s (average value for tests #9 and #10). Nevertheless,
it must be stated that the measurement duration in tests #9 and
#10 was 3 min, while in tests #1 and #2 it was only 1 min. These
hydraulic conductivity values are very low in comparison with the
hydraulic conductivity of clean ballast, which is about
3.0  101 cm/s [13]. For example, the hydraulic conductivity of
the ballast contaminated with 10% of soil, which does not require
cleaning maintenance work, according to the index of the South-
African Railway Spoornet [9], is about 1200 times lower than the
hydraulic conductivity of the clean ballast. These results
Hydraulic conductivity (cm/s)

Table 10
Average, standard deviation and characteristic global value of measurements.

Test # Ballast contamination rate (%) X (ml) S (ml) Xp (ml)

1 10 339.13 11.66 334.96

2 10 248.67 4.50 247.06
3 15 193.53 6.24 191.30
4 15 151.53 2.65 150.58
5 20 366.60 4.32 365.05
6 20 284.33 3.55 283.06
7 25 341.80 11.52 337.67
8 25 253.67 10.10 250.05
9 40 216.60 4.72 214.91 Ballast contamination rate
10 40 253.67 10.10 250.05
Fig. 5. Variation of the hydraulic conductivity with the ballast contamination rate.
 C. Paiva et al. / Construction and Building Materials 92 (2015) 58–63
demonstrate the problems of water flow caused by the contamina-
tion of the ballast by soil from the underlying layers.
Fig. 5 shows the variation of the hydraulic conductivity with the
ballast contamination rate. Considering an exponential variation,
Eq. (8) was defined to predict the hydraulic conductivity for differ-
ent ballast contamination rates. The correlation coefficient R2 was
0.9134, indicating a very good fit between the equation and the
data.
k ¼ 0:0324  e5:041bcr
where k is the hydraulic conductivity (cm/s) and bcr is ballast con-
tamination rate.
4. Conclusions
The main objectives of this research study were to verify the
necessity of cleaning maintenance work and to analyze the reduc-
tion in permeability of contaminated ballast in laboratory tests
comparing clean ballast samples and samples contaminated with
different volumetric percentages of soil.
According to the Selig and Waters index, the ballast with 10%
and 15% of soil is classified as moderately clean ballast while the
ballast with 20% and 25% of soil is classified as contaminated
ballast, indicating that in this second case cleaning the ballast is
necessary. The ballast with 40% of soil is classified as highly
contaminated ballast, indicating that in this case cleaning the
ballast is mandatory. According to the index of the South-African
Railway Spoornet, as expected, the ballast with 20% of soil or
more also requires cleaning maintenance work. According to the
permeability tests, the hydraulic conductivity values of the ballast
contaminated with soil are very low in comparison with the
hydraulic conductivity of clean ballast and these values decrease
63
quickly with the increase of the contamination rate. These results
demonstrate the problems of water flow caused by the contamina-
tion of the ballast by soil from the underlying layers. In conclusion,
special attention should be given to contaminated ballast because
it might perform at levels below that which is expected.
References
[1] Paiva C. Superestrutura ferroviária: estudo dos elementos componentes. SP,
ð8Þ Brazil: Universidade de Campinas; 2012.
[2] Profillidis V. Railway management and engineering. 3rd ed. Hampshire,
England: Ashgate Publishing Limited; 2006.
[3] Esveld C. Modern railway track. 2nd ed. Netherlands: MRT-Productions; 2001.
[4] ABNT. NBR 5564 – Via férrea – lastro ferroviário – requisitos e métodos de
ensaio. Rio de Janeiro, Brazil: Associação Brasileira de Normas Técnicas; 2011.
[5] Indraratna B, Salim W, Rujikiatkamjorn C. Advanced rail geotechnology –
ballasted track. Taylor & Francis Group, USA: CRC Press; 2011.
[6] Silva L. Fundamentos teórico-experimentais da mecânica dos pavimentos
ferroviários e esboço de um sistema de gerência aplicado à manutenção da via
permanente. Ph.D. thesis. Rio de Janeiro, Brazil: Universidade Federal do Rio de
Janeiro; 2002.
[7] Selig E, Waters J. Track geotechnology and substructure management. London,
England: Thomas Telford; 1994.
[8] Al-Qadi I, Xie W, Roberts R. Scattering analysis of ground-penetrating radar
data to quantify railroad ballast contamination. NDT & E Int 2008;41(6):441–7.
[9] Lichtberger B. Track compendium: formation, permanent way, maintenance,
economics. Hamburg: Eurailpress; 2005.
[10] Indraratna B, Shanin M, Salim W. Stabilisation of granular media and
formation soil using geosynthetics with special reference to railway
engineering. Wollongong, Australia: University of Wollongong; 2007.
[11] Gon F. Caracterização geotécnica através de ensaios de laboratório de um solo
de diabásio da região de Campinas. Dissertação de Mestrado. São Paulo, Brazil:
Faculdade de Engenharia Civil, Arquitetura e Urbanismo, Universidade
Estadual de Campinas; 2011.
[12] ABNT. NBR 6457 – Amostras de solo – preparação para ensaios de
compactação e ensaios de caracterização. Rio de Janeiro, Brazil: Associação
Brasileira de Normas Técnicas; 1968.
[13] Indraratna B, Nimbalkar S, Tennakoon N. The behaviour of ballasted track
foundations: track drainage and geosynthetic reinforcement. GeoFlorida 2010:
Advances in Analysis, Modelling & Design; 2010.