GSP 173 Advances in Measurement and Modeling of Soil Behavior

EFFECTS OF PARALLEL GRADATION ON STRENGTH PROPERTIES OF

BALLAST MATERIALS

Domenica Cambio1, and Louis Ge2

1
University of Naples Federico II, Department of Geotechnical Engineering, Via
Claudio, 21 80125, Napoli, Italy; E-mail: domenicacambio@virgilio.it
2
University of Missouri-Rolla, Department of Civil, Architectural, and Environmental
Engineering, 1870 Miner Circle, Rolla, MO 65409; PH (573) 341-7193; FAX (573)
341-4729; E-mail: geyun@umr.edu

ABSTRACT: The railroad ballast is used to fill in irregular surface topology,

distribute and transfer loads from a surface structure or system to the subgrade or
subsoils as uniformly and widely as possible in order to provide stable and stiff long-
term embankment support for railways. A typical ballast grain sizes range from 30 to
70 mm, which makes large-scale laboratory tests difficult to conduct. The parallel
gradation technique is to preserve the particle shape, particle surface roughness, and
particle mineralogy, and creates a parallel gradation of soil with a maximum particle
size for the available apparatus. This paper presents the result of a series of
monotonic direct shear tests for three ballast materials having parallel gradation
curves, which are served as background study for the ongoing research on validating
it under loading-unloading condition.

INTRODUCTION
Testing and modeling constitutive behavior of roadbeds and subgrade materials
under traffic loading has been a challenging task for geotechnical engineers. A
typical example include railroad ballast, usually comprised of highly coarse-graded
gravel-size particles, such as crushed or fractured rock or aggregates, with grain sizes
in the range of 30 to 70 mm. The ballast is used to fill in irregular surface topology,
distribute and transfer loads from a surface structure or system to the subgrade or
subsoils as uniformly and widely as possible in order to provide stable and stiff long-
term embankment support for railways. While ballast is typically deposited or placed
at variable packing densities, it is expected to behave elastically and exhibit minimal
stiffness and strength degradation over long time periods, and a large number of
repeated load cycles.
The mechanical behavior of ballast or rockfill materials have been studied for
decades (Marachi et al. 1972; Raymond and Diyaljee 1979; Janardhanam and Desai

1
GSP 173 Advances in Measurement and Modeling of Soil Behavior

1983; Indraratna et al. 1998). These material properties are found governed by the
factors including particle size, particle shape, surface roughness, parent rock strength,
particle crushing strength, particle size distribution, density, degree of saturation,
confining pressure, load history, and number of load cycles (Indraratna and Salim
2005). The ability of the parallel gradation physical analog model was investigated
and validated by Jernigan (1998) for Swedish railroad ballast, which is linearly
graded granular material, ranging from 32 to 64 mm in size. He concluded that the
use of parallel gradation method is to preserve the particle shape, particle surface
roughness, and particle mineralogy, and creates a parallel gradation of soil with a
maximum particle size for the available apparatus. Varadarajan et al. (2003) reported
that there are four techniques used to reduce the size of the large-sized crushed rock
materials, and the parallel gradation technique was found most suitable. All the
previous work on parallel gradation technique was done under monotonic loading
condition, but there are issues, such as attrition, and particle angularity, which have
not been addressed under the circumstances of cyclic loading. This paper presents
the result of a series of monotonic direct shear tests for three ballast materials having
parallel gradation curves, which are served as background study for the ongoing
research on validating it under loading-unloading condition.

EXPERIMENTAL SETUP
Direct Shear Apparatus
The direct shear apparatus has both an upper and lower shear boxes, and the
sample is sheared along the plane between them by pushing the lower shear box
horizontally with a normal (vertical) load applied to it. The shear force is measured
with a load cell that is attached between the normal load actuator and the top of the
shear box. The test was conducted with two stages, consolidation and shear loading.
The first one consists in the application, through normal actuator, of a normal stress
to that one investigated. The duration of the consolidation stage was 1 minute. The
second stage consists in the application, through shear actuator deformation, of a
shear displacement with a rate of 1 mm/min. The duration of the shear load stage was
defined by the maximum shear deformation which 15 minutes.

Modeled Ballast Materials

The ballast materials were shipped from the Iron Mountain Trap Rock Company,
MO, which provides 3 and 4A mainline ballast to railroad industry. Smaller size of
ballast materials were also available from the site and were used to manufacture 3
sets of materials (M1, M2, and M3) having parallel gradation curves, as shown in
Figure 1, to the prototype ballast.
Before manufacturing the M1, M2, and M3 materials, their parallel gradation
curves were determined and chosen carefully so that as many as sieving pans can be
used. This is critical for the material preparation to get a smooth and parallel
gradation curve as the prototype ballast.
A total of 18 direct shear tests are reported in the papers as listed in Table 1, where
3 constant vertical stress levels and 2 density states were chosen. The initial void
ratios for dense and loose specimens were determined by trial and error. Also, the
relative densities for both dense and loose specimens remain unknown due to the lack

2
GSP 173 Advances in Measurement and Modeling of Soil Behavior

of material information. The maximum and minimum void ratios will be determined
in the next phase of the project.

CLAY SILT SAND GRAVEL ROCK

100

90
M1
80

70

60
% Passing

50 M2

40

30

20
M3

10

0
0.0001 0.001 0.01 0.1 1 10 100 1000
Grain size [mm]

FIG. 1 Grain size distribution curves for M1, M2, and M3 materials.

TABLE 1. Direct shear testing program.

MONOTONIC DENSE TESTS MONOTONIC LOOSE TESTS

MATERIAL VERTICAL INITIAL MATERIAL VERTICAL INITIAL

FILENAME FILEMANE
DENSE STRESS VOID LOOSE STRESS VOID
[-] [kPa] [-] [-] [-] [kPa] [-] [-]
1 20 0.77 M1S20D 1 20 1.17 M1S20S
1 40 0.77 M1S40D 1 40 1.16 M1S40S
1 80 0.76 M1S80D 1 80 1.16 M1S80S
2 20 0.76 M2S20D 2 20 1.15 M2S20S
2 40 0.77 M2S40D 2 40 1.15 M2S40S
2 80 0.77 M2S80D 2 80 1.16 M2S80S
3 20 0.76 M3S20D 3 20 1.14 M3S20S
3 40 0.78 M3S40D 3 40 1.16 M3S40S
3 80 0.77 M3S80D 3 80 1.13 M3S80S

EXPERIMENTAL RESULTS
There are 9 normalized shear stress versus shear displacement curves for dense
material shown in Figure 2. The families of red, green, and blue curves represent M1,
M2, and M3 materials, respectively. For each family of curves, the thicker line
stands for the data with higher vertical stress. It is found that the stress ratio τ/σ
reaches the peak at about 1.8 mm shear displacement for both M1 and M2 while 2.5
mm for M3. For each material, the lower the vertical stress, the higher the stress ratio
at the peak. Figure 3 shows the normalized shear stress versus shear displacement

3
GSP 173 Advances in Measurement and Modeling of Soil Behavior

curves for loose materials. The same color scheme was adopted for M1, M2, and M3,
respectively. It is also found that for each material, the lower the vertical stress, the
higher the stress ratio at the ultimate state.
These 18 tests were performed by using the “fresh” materials, whose gradation
curves displayed in Figure 1. Three bowls (for M1, M2, and M3, respectively) were
prepared to collect the material after each test. Three sieve analyses were then
carried out to determine the grain size distribution curves. Figure 4 shows the
gradation curves for the M1, M2, and M3 materials. The red curves denote the
gradation curves before the direct shear tests while the blue curves represent the
curves after the direct shear tests. As seen in Figure 4, the grain size distribution
curves did not change much at all after the monotonic direct shear tests, and the
curves for M1, M2, and M3 remain parallel. Table 2 summarizes the friction angles
from the 18 tests. It is worth noting that the peak friction angles for the dense M1
and M2 are both about 36o although their ultimate friction angle differs 4o (17.4o and
12.8o). This somehow proves the parallel gradation technique valid. However, the
ultimate friction angle for loose M1 is 6o lower than the angle for loose M2, which
shows invalidation of the parallel gradation technique. Compared to M1 and M2
materials, M3 behaved quite differently in terms of peak and ultimate friction angles,
which can conclude that the parallel gradation technique is not working well.

1.2
M1S20D

1 M2S20D

M3S20D
0.8
τ/σ M1S40D

0.6 M2S40D

M3S40D
0.4
M1S80D

0.2 M2S80D

M3S80D
0
0 2 4 6 8 10 12
δ T [mm ]

FIG. 2 Normalized shear stress versus shear displacement curves for dense
materials.

4
GSP 173 Advances in Measurement and Modeling of Soil Behavior

1.2
M1S20S

1 M2S20S

M3S20S
0.8
τ/σ M1S40S

0.6 M2S40S

M3S40S
0.4
M1S80S

0.2 M2S80S

M3S80S
0
0 2 4 6 8 10 12

δ T [mm ]

FIG. 3 Normalized shear stress versus shear displacement curves for loose
materials.

CLAY SILT SAND GRAVEL ROCK

100

M1 AFTER
90

80
M1 BEFORE
70

60 M2 AFTER
% Passing

50

M2 BEFORE
40

30
M3 AFTER
20

10 M3 BEFORE

0
0.0001 0.001 0.01 0.1 1 10 100 1000
Grain size [mm]

FIG. 4 Grain size distribution curves for M1, M2, and M3 (before and after the
tests).

5
GSP 173 Advances in Measurement and Modeling of Soil Behavior

TABLE 2. Peak and ultimate friction angles for M1, M2, and M3 materials.
Dense Loose
Materials
φpeak φultimate φultimate
[-] [°] [°] [°]
M1 35.5 17.4 17.4
M2 35.9 12.8 22.9
M3 43.1 19.9 34.3

CONCLUSIONS
A total of 18 monotonic direct shear tests were conducted to validate the parallel
gradation technique. The materials, namely M1, M2, and M3, were prepared from
the railroad ballast manufacturer. The gradation curves for M1, M2, and M3 are
parallel to the prototype railroad ballast. From the test results, it is found that parallel
gradation technique works well for M1 and M2 materials. M3 behaved differently
than M1 and M2 although they all have parallel gradation curves.
Angularity and particle crushing (attrition) due to loading condition are believed to
be two major factors influencing the validity of the parallel gradation technique. A
digital imaging technique has been proposed to re-visit the M1, M2, and M3
materials. A quantified measure will be defined to account for the particle angularity
for the parallel gradation technique. Attrition can be assessed by comparing the
gradation curves. Since the gradation curves did not change much from the test
results, it is believed that the particle angularity is the main factor causing the parallel
gradation technique not working with M3 material.

REFERENCES

Indraratna, B., Ionescu, D., and Christie, H.D. (1998). “Shear behavior of railway
ballast based on large-scale triaxial tests.” Journal of Geotechnical and
Geoenvironmental Engineering, ASCE, 124(5), 439-449.
Indraratna, B., and Salim, W. (2005). Mechanics of Ballasted Rail Tracks – A
Geotechnical Perspective, Taylor & Francis.
Janardhanam, R., and Desai, C.S. (1983). “Three-dimensional testing and modeling
of ballast.” Journal of Geotechnical Engineering, ASCE, 109(6), 783-796.
Jernigan, R.L. (1998). The Physical Modeling of Soils Containing Oversized Particles.
Ph.D. thesis, University of Colorado at Boulder.
Marachi, N.D., Chan, C.K., and Seed, H.B. (1972). “Evaluation of properties of
rockfill materials.” Journal of Soil Mechanics and Foundation Engineering, ASCE,
98(1), 95-114.
Raymond, G.P., and Diyaljee, V.A. (1979). “Railroad ballast sizing and grading.”
Journal of Geotechnical Engineering, ASCE, 105(5), 676-681.
Sitharam, T.G., and Nimbkar, M.S. (2000). “Micromechanical modeling of granular
material: effect of particle size and gradation.” Geotechnical and Geological
Engineering 18, 91-117.

6
GSP 173 Advances in Measurement and Modeling of Soil Behavior

Varadarajan, A., Sharma, K.G., Venkatachalam, K., and Gupta, A.K. (2003).
“Testing and modeling two rockfill materials.” Journal of Geotechnical and
Geoenvironmental Engineering, ASCE, 129(3), 206-218.