Conference PaperPDF Available

Bearing capacity evaluation of a subgrade in a heavy haul railway in Brazil

June 2017

June 2017

DOI:10.1201/9781315100333-280

Conference: The 10th International Conference on the Bearing Capacity of Roads, Railways and Airfields (BCRRA 2017)

Authors:

Robson Costa

University of São Paulo

Rosangela Motta

University of São Paulo

Liedi Bernucci

University of São Paulo and Institute of Technological Research

Show all 6 authorsHide

Soil sample between sections I and II. Figure 2. Use of LWD equipment in section I.

…

Use of DCP equipment in section I. Figure 4. Discs with soil samples used during one of the MCT methodology test stages.

…

Soil sample in the triaxial test chamber.

…

Results of dynamic deformation modulus with LWD.

…

CBR results through DCP measurements. Figure 8. Grain size distribution of the evaluated material.

…

Figures - uploaded by Robson Costa

Content may be subject to copyright.

Content uploaded by Robson Costa

Content may be subject to copyright.

1977

Bearing Capacity of Roads, Railways and Airelds – Loizos et al. (Eds)

Bearing capacity evaluation of a subgrade in a heavy haul railway

inBrazil

R. Costa, R. Motta, L.L.B. Bernucci, E. Moura & J. Pires

Polytechnic School of Sao Paulo, Brazil

L. Oliveira

Vale Company, Brazil

ABSTRACT: Subgrade is a global foundation upon which railway structure is constructed on a track.

It has paramount importance due to its influence on track general behavior in service. In order to present

a good performance, subgrade must have appropriate stiffness and enough bearing capacity to resist the

stresses and avoid excessive plastic deformations that influence the superstructure components deteriora-

tion. Special attention must be paid to a subgrade of heavy haul tracks, which are normally employed for

commodities transport and subjected to different operational conditions in terms of speed, loads, wagon

number, etc. On these particular conditions, it is reasonable to consider an increase in the track compo-

nents degradation rate. Aiming to evaluate the structural condition of a subgrade, laboratory and in situ

tests can be performed. This study was based on tests to evaluate the mechanical condition of a typical

tropical soil, addressing a subgrade rehabilitation process in a heavy haul track. This study included physi-

cal characterization and resilient modulus tests in laboratory, as well as two different in situ tests, namely

Dynamic Cone Penetrometer (DCP) and Light Weight Deflectometer (LWD). In general, in situ tests

results showed that the materials in both sections have high stiffness confirmed by the resilience modulus

laboratory tests. However, due to the used methods, it is necessary to evaluate a larger number of sections

to then, determine a correlation between the elastic modulus from the LWD test and the California Bear-

ing Ratio (CBR) from DCP test. Finally, among the three different soils classification methods verified,

the MCT methodology is the one that best suit to the evaluated material.

reported that, in the early 1940s, studies conducted

in Brazil with tropical soils showed that they had

different properties and behavior when classified

through traditional TRB and USCS soil classifi-

cations, normally used in soil mechanics. In addi-

tion, it was only from the 1970s that the proper use

of fine-grained lateritic soils began to be used for

evaluating materials for base layer of pavements.

In this context, two aspects of tropical soils need

to be considered: i) properties and behavior; and

ii) environment in which they are used. To evalu-

ate them, those authors developed a methodology

called MCT (Compacted Miniature for Tropical

Soils), which differs from traditional soils classifi-

cations. They observed that CBR (California Bear-

ing Ratio) values obtained with lateritic fine soils

are generally higher than predicted through tradi-

tional grain size and Atterberg limits parameters.

Tests in an experimental section of pavement with a

soil classified by traditional specifications as group

A-7-6 (liquid limit = 44%, plastic limit = 16%,

and more than 90% of material passing through

the 0.075mm sieve) were carried out. Traditional

1 INTRODUCTION

The subgrade is a component of the railway sub-

structure that has great importance, since it gives

support for all components and substantially con-

tributes to deflection of the rail under wheel load-

ing. Moreover, it can influence the deterioration

of ballast, sleepers and rails (Selig and Waters,

1994). Additionally, due to its distinct pedological

and geological units, the subgrade can present soil

composition, resistance and performance that vary

along the railway (Hay, 1982).

In this sense, because of its importance, it is

imperative that classification and geotechnical

properties of the soil present a biunivocal rela-

tionship between each other. On the other hand,

some of the most commonly used soil classifica-

tion methods in transportation engineering, such

as the Transportation Research Board (TRB) and

the Unified Soil Classification System (USCS), in

which the latter one is recommended by AREMA

(2013) to classify soils for railways, may not be suit-

able for tropical soils. Nogami and Villibor (1991)

1978

classifications indicated that soil should not be

used as subbase layer of a pavement, however after

15 years under more than 5×106 repetitions of a

standard axis (8,160kg), there were no differences

in performance between the sections in which the

subbase was composed with lateritic clay soil and

sandy soil.

Indraratna and Nutalaya (1991) evaluated a

typical lateritic soil from the Saraburi province

(Thailand) through compaction, CBR and shear

strength values. In general, the authors concluded

that the soil presented high resistance. Moreover,

they highlighted that although having low clay

content, lateritic soils are capable of sustaining

adequate cohesion, probably in part due to oxida-

tion and internal cementation, what can be advan-

tageous in landfill construction. The authors also

emphasize that, because of the complex and diversi-

fied behavior associated to tropical soils, there is an

urgent need for fundamental studies about the soil

genesis process and the influence of morphology

and physicochemical properties, as well as the verifi-

cation of tropical regions aspects (e.g. soil type and

climate) in the long-term performance of landfills.

Considering the complex behavior of lateritic

tropical soils, there was a lack of a method that

could quickly identify them in the field. In order to

solve this problem, in 1985, Nogami and Cozzolino

developed an expedited test procedure, which was

later successfully improved in Brazil by different

researchers, such as Fortes and Merighi (2003).

Although adopting one of the methods previ-

ously mentioned in the classification of tropical

soils, it is necessary to determine resilient charac-

teristics of the materials used in the railway track.

Studies from Li and Selig (1994) related to resilient

behavior of fine graded soils have shown that they

are significantly dependent on: i) loading condi-

tion or stress state, which include the deviation

and confinement stresses magnitude and number

of load cycles; ii) soil type and initial structure and

for compacted soils, the compaction method and

energy; and iii) soil physical state, which can be

defined by moisture content and dry density values.

Considering that fine graded soils are dependent

on these variables, Medina et al. (2006) evaluated

the resilient modulus of lateritic gravel finding val-

ues ranging from 300 to 600MPa.

It is known that soil stiffness and strength are

properties directly related one to another in the

sense that soils with low resistance also tend to

present low stiffness (Selig and Waters, 1994). So,

in order to analyze these properties in the field, the

use of Light Weight Deflectometer (LWD) and

Dynamic Cone Penetrometer (DCP) may be useful.

Fortunato (2009) reports that, in the process of

renewal of an old railway track, it is of extreme

importance to characterize the stiffness of the

existing platform, so tests such as DCP were used to

evaluate railways in Portugal and they were found

to be efficient in in situ characterizations. Also, the

results of studies from Abu-Farsakh et al. (2005)

showed that DCP can be used to evaluate stiffness

characteristics of pavements and subgrade layers.

Additionally, Chen et al. (1999) analyzed the cor-

relation between DCP and CBR and observed that

correlations were not adequate for high values of

DCP or low values of CBR.

On the other hand, LWD can be used to deter-

mine the deformation modulus and, according to

Stamp and Mooney (2013), the measured peak

deflection can be used directly as a measure of

soil stiffness or degree of compaction, along with

the peak force can be used to estimate the dynamic

modulus. Field and laboratory tests results con-

ducted by Nazzal et al. (2007) using LWD, DCP

and FWD (Falling Weight Deflectometer) showed

that the repeatability of the LWD values depends

on the stiffness of the material tested. In addition,

it was observed that subgrades with low bearing

capacity have poor repeatability. However, rigid and

well compacted layers presented significantly better

repeatability. Moreover, laboratory tests showed the

influence of depth on LWD results, which ranged

from 270 to 280mm on that study, depending on

the stiffness of the materials tested. In general, they

affirm that there is a good correlation between the

results obtained with LWD and DCP.

Concerning railways, sometimes characteristics

of substructure materials (e.g. ballast, subballast

and subgrade), such as strength, may not be taken

into account, although subballast and subgrade

are strongly susceptible to moisture variation.

These components have fundamental importance

in the track, in order to maintain adequate elastic

properties to the traffic. In this context, evaluat-

ing behavior of railway substructure material over

time may save costs, since material replacement

in a heavy haul railway may represent significant

parcel of the maintenance operation costs (Grabe

and Clayton, 2009). Besides, interruptions for

maintenance also have impact on costs in terms of

operation, since volume of materials transported

through a freight railway can be reduced.

2 OBJECTIVE

The objective of the present work is to evaluate the

structural condition of subballast/subgrade mate-

rial of Carajás Railway (EFC) in Brazil in two sec-

tions composed of recycled (cleaned) and fouled

(clogged) ballast by means of in situ tests using

DCP and LWD. In addition, repeated triaxial load-

ing tests were also conducted in laboratory, in order

to determine the resilient modulus. Characteristics

1979

of the tested materials were then compared through

different soils classification methods.

3 FIELD AND LABORATORY TESTS

Carajás railroad (EFC) is 892 km long and inter-

connects the Carajás mine to the terminal Ponta da

Madeira, in the state of Maranhao, northeastern

Brazil. The railroad is composed of metric gauge,

TR-68 rails, sleepers spacing of 0.61 m and bal-

last of crushed stone. In order to define the sec-

tions of the in situ tests, were selected places where

renewal activities had been scheduled between the

months of October and December of 2014, located

between the housings 24 and 25.

Two test sections on Carajás Railway were object

of this research. They were identified as “section I”

– km 407+485m (recently renewed track, recycled

ballast) and “section II” – km 409+931m (fouled/

contaminated ballast).

In situ evaluation using LWD and DCP was

carried out in two different periods of the year in

terms of rainfall magnitudes, 10 mm and 25 mm

(average).

Soil under the ballast/subballast layer was sam-

pled from the track, as illustrated by Figure 1

(between sections I and II), because according to

information provided by the maintenance depart-

ment of the railway, both sections were constructed

with similar material.

3.1 In situ tests with DCP and LWD

The configuration of the Light Weight Deflecto-

meter (LWD) used in the field tests has a 10.0kg

weight that slides over a 720 mm guide rod, in a

free fall towards to a 300mm plate. This latter has

an accelerometer coupled to it, in order to record

the deformation undergone by the layer, which is

used in the determination of the dynamic deforma-

tion modulus.

On the other hand, regarding to the Dynamic

Cone Penetrometer (DCP) apparatus, it consists

of a 8.0 kg weight (a 60° cone) that slides over a

guide rod, in a free fall from 575mm height into a

1,000mm shaft. The test was performed according

to ASTM D 6951/6951 M-09standard, in order

to determine the subgrade load capacity. The

penetration for each blow, at each depth of the

railway substructure, was used to estimate CBR

values through the US Army Corps of Engineer

equation (1).

CBR

DCP

=









292 1 12.

(1)

where DCP =Dynamic Cone Penetrometer (mm/

blow).

In situ tests with LWD and DCP were per-

formed in section I in November of 2014 (lower

rainfall – 10mm average) and in section II in April

of 2015 (higher rainfall – 25 mm average). These

two different periods were defined in order to ver-

ify if, in general, a change in the moisture content

condition could influence on the results. Figures2

and 3, respectively, illustrate LWD and DCP being

used in sectionI.

The results were obtained at each section in 3

different places and, in each one, 3measures were

performed nearby.

3.2 Laboratory tests

The following laboratory tests were carried out

with materials sampled in sections I and II: (i) par-

ticle size analysis; (ii) Atterberg limits; (iii) general

tests for soil classification by different methods;

and (iv) resilient modulus.

Figure1. Soil sample between sections I and II. Figure2. Use of LWD equipment in section I.

1980

Particle size analysis was performed in accord-

ance with standards ASTM 6913-04 and ABNT

NBR 7181-94 (Brazilian), which are similar.

Determination of Atterberg Liquidity (LL)

and Plasticity (LP) Limits, as well as the Plastic-

ity Index (PI) were performed according to ASTM

D4318-10standard, using material passed through

0.425mm sieve.

Classification through the Unified Soil Clas-

sification System (SUCS) and Transportation

Research Board (TRB) was applied as recom-

mended by ASTM D2487-11 and D3282-15. On

the other hand, MCT expedited classification

method for tropical soils was also performed,

which was originally developed to hierarchize

soils behavior for road geotechnical purposes. This

latter may indicate a more adequate classification,

due to the fact that the material is from a Brazilian

region with lateritic soil presence. In addition, it

is worth mentioning that MCT classification is a

quick and low cost procedure.

Due to the differentiation of this method, a brief

description is given as follows. In terms of grading,

the material is passed through the 2.00 mm and

0.42mm sieves. A soil paste is prepared using that

fraction, which is moistened and intensively mixed

until having a consistency determined by plasticity or

also fixed by a portable penetrometer (the optimum

water content is obtained). Subsequently, discs of

20mm in diameter by 5mm in height are molded,

and spheres are made. Afterwards, they are dried at

oven at 60ºC. Some characteristics are evaluated: i)

contraction, due to loss of moisture; ii) expansion

by water reabsorption; and iii) resistance to pen-

etration. Figure4shows the discs during one of the

MCT methodology test stages.

Concerning to resilient modulus, the test

was performed with the material at the optimum

Figure3. Use of DCP equipment in section I. Figure 4. Discs with soil samples used during one of

the MCT methodology test stages.

Figure5. Soil sample in the triaxial test chamber.

1981

moisture content (12.7%), as well as above that

content (14.6%) tested in 3 soil samples for each

moisture content. Compaction was performed

with normal Proctor energy.

Figure 5 shows the specimen (H = 200 mm,

D=100mm) in the triaxial test chamber.

Resilient modulus test was carried out accord-

ing to the recommendations of AASHTO T 307-99

and DNIT 134-10 ME (Brazilian standard), which

are similar. These standards are usually employed

for testing base and subbase materials for road

pavements, wherein the stress levels for determina-

tion of the resilient modulus are the same in both

standards.

4 RESULTS

4.1 In-situ tests with LWD and DCP

Results with LWD in-situ in sections I and II

are shown in Figure 6. It can be seen that there

were practically no differences between the val-

ues obtained, with an average of 58MPa (stand-

ard deviation of 1 MPa) and 61MPa (standard

deviation of 3MPa), respectively in sections I and

II. In the regions evaluated, the small difference

between rainfalls amount did not cause consider-

able changes in the general mechanical behavior of

the evaluated material.

Concerning DCP test, CBR results obtained

through Equation 1 over depth in both sections

(I and II) are shown in Figure7.

As expected, it was observed that both sub-

grades (sections I and II) presented an increase of

CBR over depth.

The analysis of the structural condition of the

materials that make up the subballast/subgrade

before (fouled) and after (recycled) the interven-

tion for maintenance is of paramount importance

to the rehabilitation process of the substructure

concerning to the optimization of the process, etc.

In relation to Figure7, the recycled ballast curve

did not reach the same depth as the fouled one,

because above 200 mm, the soil presented a very

high stiffness, not being possible this way, to con-

tinue with the penetration of the rod.

In terms of CBR, these results denote that both

sections presented good bearing capacity and may

not contribute significantly to the increase in the

general vertical displacements of the structure.

By means of the in situ test methods used, it was

possible to determine the stiffness condition of the

subballast/subgrade materials. However, it is neces-

sary to evaluate a larger number of sections and

points, to then determine a correlation between

ELWD (LWD) and CBR (DCP).

4.2 Laboratory tests

Figure8shows the results of grain size distribution

of the material collected between the two sections.

The results of grain size distribution showed

approximately 65% of sand. Besides, consist-

ency Atterberg limits obtained were: LL (Liquid

Limit) = 30%; LP (Plastic Limit) =18%; and IP

(Plasticity Index)=12%.

By grain size distribution and plasticity results

obtained, USCS classification indicated “group

Figure 6. Results of dynamic deformation modulus

with LWD.

Figure7. CBR results through DCP measurements.

Figure 8. Grain size distribution of the evaluated

material.

1982

SC” (clayey sand), whereas TRB classification

indicated “group A-6” (clayey soils), which would

represent a material with poor behavior.

However, expedited MCT classification indi-

cated “LA’- LG’” (Lateritic Sandy or Clayey).

Resilient modulus results of the sample com-

pacted in two different moisture contents are

shown in Figure 9. When comparing them, it

is observed that, considering a deviatoric stress

of 0.1 MPa, for example, resilient modulus val-

ues were approximately 600 MPa and 300MPa,

respectively for 12.7% and 14.6% of moisture con-

tent. Thus, it is noticed that an increase of 1.9% of

water entailed practically 50% in the decrease of

the resilient modulus value.

5 GENERAL CONCLUSIONS

In situ characterization of the subgrade with LWD

showed that there were no differences between

dynamic deformation modulus values for the

evaluated conditions (sections I and II), whereas,

through DCP it was noted that the material pre-

sented higher bearing capacity (higher CBR value).

Also, due to general low rainfall in the evaluated

region, a possible small change in moisture appar-

ently did not influenced on the mechanical behav-

ior in the field.

On the other hand, laboratory results of resil-

ient modulus as a function of the deviatoric stress

presented an increase on moisture content (1.9%)

in relation to the optimum, but reduced strength

values (50% less).

The test was performed with 2% of water con-

tent above the optimum water content in order to

verify how this value influence in the decrease of

the stiffness value. Besides, it was useful in order to

compare with the in situ tests values.

It is worth noting that despite the values of resil-

ience modulus obtained in laboratory show a loss

of almost 50% in stiffness, this decrease in terms of

resistance was not observed in the situ tests under

different rainfall conditions.

Soil classification methods USCS and TRB do

not adequately classified EFC sample material.

The latter classification indicated “poor material”,

while MCT expedited classification showed mate-

rial being sandy or clayey lateritic material. The

results using the MCT methodology and the resist-

ance tests (in situ and laboratory) indicated that it

is a material with good bearing capacity and suit-

able to be used as a subballast/ subgrade.

Therefore, all in situ and laboratory tests indi-

cated that the material analyzed presented an

adequate bearing capacity. On the other hand, if

USCS or TRB classifications were adopted, the

material would not be indicated to be used in the

subgrade. This fact shows that these classifications

may, in some cases, not adequately represent the

mechanical behavior of tropical soils.

From the results, it can be said that ven that ini-

tially, the soil evaluated was considered as a sub-

grade, according to the results obtained in terms of

in situ and laboratory tests results, the soil has most

of its characteristics (e.g. CBR and resilient modu-

lus) related to a subballast material. This fact was

due to the uncertainty about the existence of the

subballast component in that railway, constructed

approximately 35 years ago. The mentioned railway

is normally subjected to high loads (e.g. ≈37.5tons/

axle) and, in this scenario, is reasonable consider that

an excess of efforts acting into the track components

including subballast and subgrade, can provoke some

phenomenon that can cause for example the inter-

penetration of materials between the granular lay-

ers (e.g. subballast material into the subgrade one).

In addition, the high variation of the water content/

saturation degree in both subballast and subgrade

material can provoke a variation in the resistance and

deformability, which can collaborate for the interpen-

etration between the railway granular components.

REFERENCES

Abu-Farsakh, M.Y.; Nazzal, M.D.; Khalid, A.; Seyman,

E. 2005. Application of Dynamic Cone Penetro-

meter in Pavement Construction. Transportation and

Research Board Nº 1913. Washington, D.C., 53–61.

ABNT. NBR 7181. 1988. Solo—Análise granulométrica

- Método de ensaio. Associação Brasileira de Normas

Técnicas. Rio de Janeiro.

American Association of State Highway and Transpor-

tation Officials (AASHTO). 1999. T 307-99—Deter-

mining the resilient modulus of soils and aggregate

materials, USA. 37p.

American Railway Engineering and Maintenance of

Way Association. 2013. Manual for Railway Engineer-

ing, Vol. 1–4.