Challenge G: An even more competitive and cost efficient railway
On the damaging effects of the ballast tamping operation
R. Perales1, G. Saussine1, N. Milesi2, Y. Descantes3
1
SNCF, Paris, France ; RFF, Paris, France2 ;
3
LCPC, Bouguenais, France
1. Introduction
The performance of ballasted railway tracks is largely controlled by the condition of the ballast bed. As
shown on figure 1, a typical active railroad cross-section is composed of a clean ballast layer, then a
transition layer of mostly clean ballast and finally a fouled ballast (or subballast) layer directly in
contact with the subgrade. Subballast is sand or gravel used to provide drainage and, along with
ballast, to distribute the dynamic loading applied by trains into the subgrade. The ballast and
subballast systems are commonly referred as the granular layer supporting track with a design
thickness of typically 45–75 cm.
Fig. 1: Typical active railroad ballast cross-section.
Over time, ballast is progressively contaminated by the fine-grained aggregate of the transition layer
and metal dust (Ionescu, 2005a; 2005b) that fill the void space between ballast stones. This
contamination is commonly referred as fouling. The accumulation of fine particles in the voids reduces
the porosity, drainage properties and elasticity of ballast layers, which in time may create the
conditions for undrained failure at certain depth below the track surface, when combined to heavy
rainfall. In this respect, it was reported that the primary cause of ballast layer clogging was the
degradation of aggregates, which accounts for up to 40% of the proportion of fouling materials.
Tamping is a maintenance procedure applied to the ballast bed when critical levels of deformation
and performance degradation of the ballasted railway track are reached. The tamping process
consists in compacting the ballast below the sleeper through vibrations and pressure squeezing
forces in order to restore the settlement of the substructure. This operation is carried out at each
sleeper:
the sleeper is lifted to a desired level,
the tamping tines are inserted into the ballast through vibrations in order to facilitate their
penetration,
then the tamping tines vibrate and squeeze ballast in order to fill the void underneath the lifted
sleeper,
Challenge G: An even more competitive and cost efficient railway
finally, the tamping tines vibrate and withdraw themselves from the ballast,
and the sleeper is released.
However, during this routine operation, significant quantities of ballast are suspected to be damaged
(Röthlisberger 2000). Indeed, the penetration of the vibrating tines into the ballast bed may cause
particle breakage, resulting in a loss of angularity of coarse particles together with an increase in the
fine particles content.
This paper focuses on the influence of the tamping process on ballast particle geometry. Indeed, this
work is devoted to evaluate at actual size the influence of the number of cycles applied by an
operational tamping machine on geometrical characteristics of ballast particles, such as the size
distribution and angularity.
2. Tamping program
This program consists in performing tamping at actual scale on a railway track located in Arras (North
of France). Figure 2 illustrates the initial poor condition state of the twenty meters long (thirty three
sleepers) railway section to be tested.
In order to prepare the test section, the first step consists in excavating a thickness of about twenty
centimeters of rubble-land mixture in order to introduce the ballast to be tested. 45 tons of LGV ballast
of fresh production were required for this experimentation (Fig. 3). The second step consists in
numbering 34 big-bags with dimensions 910 x 910 x 350 mm3 (see fig. 4), then in placing 28 big-bags
underneath the sleepers. In the third step, these bags are filled with ballast using a mechanical
shovel, while the remaining six bags are filled either mechanically or manually and stored aside to
serve as reference. Figure 5 illustrates the configuration of the track before the tamping operations
begin.
Then the tamping operation begins (Fig. 5 right), with the tamping machine operating at each sleeper
as shown on Figure 6 (Lim 2004, Paderno 2010). The tamping program described in Table 1 was
performed.
When the tamping is finished, the big-bags are removed from the railway track and packed on pallets
to be sent to LCPC in Nantes for detailed analysis.
Fig. 2: Railway track to be prepared for testing (Arras, France).
Challenge G: An even more competitive and cost efficient railway
Fig. 3: Ballast grains of fresh production.
Fig. 4: Big-bag.
Fig. 5: Big bags on the railway.
Challenge G: An even more competitive and cost efficient railway
Fig. 6: Tamping of the ballast contained in a big-bag located underneath the sleepers.
Tab. 1: Mode of filling big bags and loads applied.
Number of big-bag Fill mode Number of cycles Penetration velocity of
applied by tamping the tamping tines
machine
1 and 2 Hand 0 -
3, 4, 5 and 6 Shovel 0 -
7 and 8 Shovel 5 Normal
9 and 10 Shovel 10 Normal
11 and 12 Shovel 15 Normal
13 and 14 Shovel 20 Normal
15 and 16 Shovel 25 Normal
17, 18, 19, 20, 21, 22, Shovel 30 Normal
23 and 24
25 and 26 Shovel 30 Minimum
27 and 28 Shovel 30 Maximum
29 and 30 Shovel 35 Normal
31 and 32 Shovel 40 Normal
33 and 34 Shovel 45 Normal
3. Analysis program and test results
Once the 34 big-bags of ballast grains (unit mass between 300 and 500 kg) are received at LCPC in
Nantes, the first step consists in sampling three test portions of at least 40 kg each per bi-bag
according to clause 9.6 of European standard EN 932-1. These samples are subsequently tested for
the purpose of assessing the tamping damages in terms of grading and angularity shifts, with this
latter characteristic as well as the grading assessed using image processing techniques developed at
the LCPC (Descantes et al, 2006).
3.1 Sieve analyses
In order to identify the ballast size fraction, sieve analyses are carried out on the 12 test portions
sampled from the 4 reference big-bags filled mechanically. Figure 7 illustrates the average particle
Challenge G: An even more competitive and cost efficient railway
size distribution. The size fraction of the tested ballast is 31.5/50 mm and the proportion of fine
particles does not exceed 0.38%.
100
90
80
Cumulative passing (%)
70
60
50
Filling shovel
40
30
20
10
0
0 20 40 60 80 100
Sieve opening (mm)
Fig. 7: Average particle size distribution of the reference ballast contained in shovel filled bags.
The influence of tamping on the grading shift is assessed by comparing the previous grading curve to
that of ballast subjected to 45 tamping cycles. Fig. 8 shows that the number of tamping cycles has a
statistically significant influence on the ballast size evolution at a sieve opening of D = 50 mm. The
average size evolution may reach 6.4 points at this opening for 45 tamping cycles.
100
90
80
Cumulative passing (%)
70
60
average shovel - 0 cycle
average shovel - 45 cycles
50
40
30
20
10
0
0 20 40 60 80 100
Sieve opening (mm)
Fig. 8: Effect of 45 tamping cycles on the grading curve of ballast (by comparison with reference
big bags).
Figure 9 compares the average proportions of particles in each elementary class between worn-out
(45 cycles) and reference ballasts. Particles with a size greater than 50 mm are the most impacted
because they are the backbone that bears the stress during the tamping process. Fragmentation of
these particles leads to an increase in the proportion of particles in the size range of 35.5 to 50 mm
and the size of the resulting fragments seems to be about 25 to 28 mm (these are not fines).
Challenge G: An even more competitive and cost efficient railway
50
45
Non-cumulative passant (%)
40
35
30
25 average shovel - 45 cycles
average shovel - 0 cycle
20
15
10
0
0,08
0,13
0,16
0,32
0,63
1,25
3,15
0,1
0,2
0,5
0,8
1
1,6
2
2,5
4
5
6,3
8
10
12,5
16
20
25
28
31,5
35,5
45
50
56
63
90
Sieve opening (mm)
Fig. 9: Effect of 45 tamping cycles on particle sizes per class (by comparison with
reference big bags).
Knowing the grain diameter for which the wear is established (D = 50 mm), the influence of the
tamping cycle number on the wear of grains can be discussed. Figure 10 relates the grading shift at
D = 50 mm to the tamping cycle number. Wear seems to become statistically significant when the
number of cycles exceeds 20.
6
Wear à D (%) = 0.2295x(nb of cycles) - 4.7473
R2 = 0.832
4
Average wear at D (%)
-2
-4
-6
-8
0 5 10 15 20 25 30 35 40 45
Number of tamping cycles
Fig. 10: Evolution of wear at D = 50 mm as a function of the tamping cycle number
(penetration velocity of the tamping tines is set to “normal”).
Challenge G: An even more competitive and cost efficient railway
In order to study the influence of the tines penetration velocity on the wear of grains, 2 big-bags were
subjected to a 30-cycle-tamping process with the tines penetration velocity set to the minimum, 2 with
the velocity set to the maximum and 8 with velocity set to normal. Figure 11 shows that the minimum
velocity is significantly less aggressive than the other 2 (difference of 3.2 points).
100
90
80
Cumulative passing (%)
70
60
minimum velocity
50
normal velocity
40 maximum velocity
30
20
10
0
0 20 40 60 80 100
Sieve opening (mm)
Fig. 11: Average influence of the penetration velocity of the tamping tines of the
grading curve after 30 cycles.
3.2 Grading analysis using the videograder
The same samples as in the previous section were tested using the videograder.
Comparison between reference ballast and ballast subjected to 45 tamping cycles is shown on figure
12. The particle size analyses results obtained using the videograder do not show statistically
significant changes between these two ballasts. One explanation may be that sieve analysis is
sensitive to the real three-dimensional geometry of the particles, while the ellipsoid model
implemented in the videograder is less.
Yet, conclusion about the statistically significant influence of different penetration velocities is
confirmed by the videograder analysis as shown on figure 13.
Challenge G: An even more competitive and cost efficient railway
100
90
Cumulative proportion (%)
80
70
60
average shovel - 0 cycle
50 average shovel - 45 cycles
40
30
20
10
0
0 20 40 60 80 100
Size particle (mm)
Fig. 12: Effect of tamping on the videograder curves (by comparison with reference
big bags).
100
90
80
Cumulative proportion (%)
70
60
minimum velocity
50 normal velocity
40 maximum velocity
30
20
10
0
0 20 40 60 80 100 120
Size particule (mm)
Fig. 13: Average influence of the penetration velocity of the tamping tines of the
videogranulometer curve after 30 cycles.
3.3 Angularity analysis using the videograder
As shown on figure 14, tamping does not cause any significant loss of angularity to the ballast. This
conclusion is not surprising given the short duration of a tamping cycle, with stresses suffered by the
ballast being more like a shock than like a phenomenon spread over time such as attrition wear.
Challenge G: An even more competitive and cost efficient railway
0,1665
0,1660
0,1655
ANG (dimensionless)
0,1650
0,1645
0,1640
0,1635
0,1630
0,1625
0,1620
0,1615
0,1610
0 5 10 15 20 25 30 35 40
Tamping cycle number
Fig. 14: Evolution of the average angularity of ballast with tamping cycle number.
As before, the analysis of 36 samples collected from 12 big-bags was used to study the influence of
the penetration velocity of the tamping tines on the evolution of the ballast average angularity.
Examination of Figure 15 shows a marginally significant evolution from the minimum velocity to
maximum velocity, but it must be linked to the reduced interval of confidence.
0,1665
0,1660
0,1655
ANG (dimensionless)
0,1650
0,1645
0,1640
0,1635
0,1630
0,1625
0,1620
0,1615
Minimum
minimale Normal
moyenne Maximum
maximale
Penetration velocity
Fig. 15: Evolution of the average angularity of ballast angularity average on the
penetration velocity for 30 tamping cycle.
Challenge G: An even more competitive and cost efficient railway
4. Conclusions
This study investigates the effects of actual size tamping operations on the geometrical characteristics
of ballast of fresh production.
The results cover 34 big-bags of ballast grains with a unit mass between 300 and 500 kg. These were
subjected to different numbers of tamping cycles from 0 to 45. The conclusions can be drawn as
follows:
the particles of characteristic size greater than D = 50 mm are probably the most impacted
because they constitute the skeleton of the ballast layer, viz. the backbone that bears the
stresses applied by the tamping process,
Fragmentation of theses particles leads to an increase in the proportion of particles in the size
range of 35.5 to 50 mm and the resulting fragments appear to have a characteristic size of
about 25 mm, not in the fines range (the proportion of fine particles doesn’t exceed 0.38% per
big-bag),
the penetration minimum velocity of the tamping tines is less aggressive than the other two
normal and maximum velocities, the latter having no significant effect at D = 50 mm,
the tamping does not cause any statistically significant wear in terms of loss of ballast
angularity. In fact, the stresses suffered by the ballast are applied during a too short time.
Acknowledgments
The authors gratefully acknowledge RFF (Réseau Ferrée de France) who has funded this project.
Biography
Y. Descantes, Y. Fosse, F. Milcent, Automated measurement of railway ballast angularity. Journal of
Materials in Civil Engineering (ASCE), vol. 18, issue 4, p. 612-618, 2006.
D. Ionescu, Critical review of granular media with special reference to railway ballast, U. o. W. library,
ed., Wollongong, 86, 2005a
D. Ionescu, Ballast degradation and measurement of ballast fouling, RTR, 2, 12-19, 2005b.
W.L. Lim, Mechanics of railway ballast behaviour, PhD Thesis, Nottingham University, 2004.
C. Paderno, Comportement du ballast sous l’action du bourrage et du traffic ferroviaire, PhD Thesis,
EPDF, Suisse, 2010.
B. Röthlisberger, Stopfversuch in Ostermundigen – Teil I &II, 2000.
