See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/233072296

On Forecasting the Resilient Modulus from the CBR Value of Granular Bases

Article  in  Road Materials and Pavement Design · November 2007

DOI: 10.1080/14680629.2007.9690099

CITATIONS READS

7 51

1 author:

Sigurdur Erlingsson
University of Iceland and Swedish National Road and Transport Research Institute
82 PUBLICATIONS   556 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

International Conference on Earthquake engineering and Structural Dynamics, 12-14 June 2017, Reykjavik, Iceland View project

Mechanistic Empirical Pavement Performance Predictions View project

All content following this page was uploaded by Sigurdur Erlingsson on 25 May 2018.

The user has requested enhancement of the downloaded file.

This article was downloaded by: [Swedish Road and Transport Research Inst]
On: 05 March 2014, At: 23:43
Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer
House, 37-41 Mortimer Street, London W1T 3JH, UK

Road Materials and Pavement Design

Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/trmp20

On Forecasting the Resilient Modulus from the CBR

Value of Granular Bases
a b
Sigurdur Erlingsson
a
Dept. of Civil and Environmental Engineering , University of Iceland , 107, Reykjavik,
Iceland E-mail:
b
Dept. of Highway Engineering , Swedish National Road and Transport Research
Institute, VTI , 581 95, Linköping, Sweden E-mail:
Published online: 19 Sep 2011.

To cite this article: Sigurdur Erlingsson (2007) On Forecasting the Resilient Modulus from the CBR Value of Granular Bases,
Road Materials and Pavement Design, 8:4, 783-797

To link to this article: http://dx.doi.org/10.1080/14680629.2007.9690099

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of
the Content. Any opinions and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied
upon and should be independently verified with primary sources of information. Taylor and Francis shall
not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other
liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or
arising out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
 On Forecasting the Resilient Modulus
from the CBR Value of Granular Bases

Sigurdur Erlingsson
Downloaded by [Swedish Road and Transport Research Inst] at 23:43 05 March 2014

Dept. of Civil and Environmental Engineering

University of Iceland
107 Reykjavik, Iceland
sigger@hi.is
and
Dept. of Highway Engineering
Swedish National Road and Transport Research Institute, VTI
581 95 Linköping, Sweden
sigurdur.erlingsson@vti.se

ABSTRACT. The unbound granular materials (UGM), base and subbase layers, play an
essential role in the overall structural performance of thin pavement structures. They show
complex stress dependent elasto-plastic behaviour under external loading. Therefore the
UGM are commonly tested using the Repeated Load Triaxial (RLT) testing method to
estimate the stiffness of the material by applying haversine loading pulses. The RLT testing
method represents the actual stress situation quite adequately and gives satisfactorily
estimates of the stiffness characteristics of UGM. A simple test that has been used for a long
time in structural design of flexible pavements is the CBR (California Bearing Ratio) test. In
the CBR test, the load-deformation curve is acquired while a plunger is penetrated into the
material at a constant rate. In the literature one can find a number of relationships for UGM
where the CBR value is used to predict the stiffness. These connections usually do not take
into account that stiffness of UGM is both stress and moisture dependent. To investigate if a
relationship between the two tests exists, twenty materials have been tested with both methods
and the test results compared. The materials were of varying quality and were tested at four
different moisture contents. The results indicate that a simple power law can be used to
forecast the stiffness if the CBR-value is known.
KEYWORDS: Granular Materials, Mechanical Properties, Resilient Modulus, Triaxial Testing,
CBR Value.

DOI:10.3166/RMPD.8.783-797 © 2007 Lavoisier, Paris

RMPD – 8/2007. Water in Pavements, pages 783 to 797

 784 RMPD – 8/2007. Water in Pavements

1. Introduction

New flexible pavement design methods are under development where the aim is
to predict functional and structural conditions of roads over time. To do this the
response of the pavement structure due to vehicle loading is calculated based on a
mechanistic approach and thereafter a distress prediction (rutting, fatigue and
thermal cracking) is carried out. The key parameters for the response analysis are the
stress-strain relationships of the different layers of the pavement structure. Figure 1
illustrates the general stress regime experienced by an unbound base course element
Downloaded by [Swedish Road and Transport Research Inst] at 23:43 05 March 2014

in a pavement structure as the result of a moving wheel load within the plane of the
wheel track. Due to the wheel load, pulses of vertical and horizontal stress,
accompanied by a double pulse of shear stress with a sign reversal, affect the
element (Brown, 1996). A method, which represents this actual stress situation quite
adequately, is the Repeated Load Triaxial (RLT) testing method, where the stiffness
of the material is estimated by applying a number of haversine load pulses. However
the RLT testing method is a rather time consuming and therefore expensive to use.

Figure 1. a) Typical thin pavement structure and stresses, b) induced stresses in a

pavement element due to moving wheel load

The CBR (California Bearing Ratio) value has been used for a long time in the
structural design procedure of flexible pavements, even though it is known that it
poorly represents the actual stress situation pavements experiences during traffic
loading. One of the advantages of the CBR method is that it is a simple and cheap
test method. In the CBR test, a 50 mm cylindrical diameter plunger is advanced once
into a cylindrical mould of recompacted unbound granular material (UGM) at a
constant rate. The load required to cause 2.54 and 5.08mm insertion is recorded and
expressed as a percentage of the load required for the same penetration in a certain
standard material. The ratio of the insertion over the diameter of the plunger is
therefore 5 versus 10% respectively. These penetrations can cause high stress
concentrations locally in the material around the tip of the plunger with permanent
deformations as a result comprising many repeated light loading cycles.
In the literature one can find a number of relationships for UGM where the CBR
value is used to predict the stiffness. To investigate if a connection exist between the
 On Forecasting the Resilient Modulus 785

CBR value and stiffness, twenty materials have been tested with both the RLT test
and CBR test equipment and the test results compared. The major advantages if a
such relationship would exists is much cheaper and quicker estimation of the
stiffness as the RLT is quite an elaborate test and rather expensive but the CBR test
is a simple and cheap test.

2. Stiffness characteristics of UGM

Downloaded by [Swedish Road and Transport Research Inst] at 23:43 05 March 2014

The stress regime in pavements due to vehicle loading can be expressed by

introducing two stress invariants: the mean stress level p and the deviatoric stress q,
in which, for the case where σ2 = σ3, becomes
1
p= (σ 1 + 2σ 3 ) and q = σ1 − σ 3 [1]
3
In a similar way strain invariants can be introduced. The volumetric strain εv and
deviatoric strain εq, are defined as:
2
ε v = ε 1 + 2ε 3 and εq = (ε 1 − ε 3 ) [2]
3
in which ε1 and ε3 are the resilient axial strain and radial strain respectively, where it
has been assumed in a comparable way as for the stresses that ε2 = ε3.
The stresses and strains are now interconnected for isotropic elastic material
through the material properties and can be expressed in a diagonal matrix:

ε v  1 3(1 − 2ν ) 0   p
ε  =  [3]
 q Mr  0 2 1 + ν )  q 
(

where Mr and v are the material stiffness modulus (resilient modulus) and Poisson’s
ratio respectively, defined as:
q ε3
Mr = and ν =− [4]
ε1 ε1
It is well known that the stiffness modulus of UGM is stress dependent but the
Poisson’s ratio is not, or at least is to a much less extent and can usually be treated
as a constant. A number of relationships exist to describe the stress dependency of
the stiffness moduli. One of the most common and also one of the simplest is the so
called k - θ expression (Yoder and Witczak, 1975; Brown and Pappin, 1981; Correia
et al., 1999; Schwartz, 2002):

k2
 3p 
M r = k1θ k2 = k1   [5]
 pa 
 786 RMPD – 8/2007. Water in Pavements

where k1 and k2 are experimentally determined constants and pa is a reference

pressure, pa = 100 kPa.
By introducing Equation [5] into [3] it is obvious that the stresses and the strains
are interconnected in a nonlinear relationship, which can be written in matrix form
as
[ε] = [C( p )][σ] [6]

in which [ε]T = [εv, εq]T and [σ]T = [p, q]T and [C(p)] is the compliance matrix.
Downloaded by [Swedish Road and Transport Research Inst] at 23:43 05 March 2014

To determine the stress strain relationship experimentally, and therefore the k1, k2
and v, a number of measurements is needed to cover the actual range of mean
stresses p and deviatoric stresses q caused by different weights of the axle loads of
the traffic. The RLT test method can be used for this estimation where different
stress paths are applied (Correia et al., 1999; Erlingsson, 2000).
Many attempts have been made to connect values from CBR testing with
stiffness modulus (Witczak et al., 1995; Hoff, 1999). In Table 1 four equations are
shown where different organizations have made this attempt.

Table 1. Four equations, which have been used for the assessments of stiffness from
results from CBR testing. Stiffness is in MPa, and CBR-values in %. (Witczak et al.,
1995)

Equation Organization Equation No.

M r = 10.35 ⋅ CBR Shell Oil [7]

M r = 37.3 ⋅ CBR 0.711 U.S. Army corps of Engineers (USAGE) [8]

South African Council on Scientific and

M r = 20.7 ⋅ CBR 0.65 [9]
Industrial Research (CSIR)
Transport and Road Research Laboratory
M r = 17.25 ⋅ CBR 0.64 [10]
(TRRL)

These four equations give very different results as one can see in Figure 2 where
the predicted stiffness value according to the equations in Table 1 are given as
function of the CBR value.
It is well known that the stiffness of UGM is both stress and moisture dependent.
However, this fact is not expressed in Equations [7] – [10]. Therefore it is important
to give the stress state as well as for what moisture contents these equations are valid
to guide the user.
 On Forecasting the Resilient Modulus 787

1000
Eq 7
Resilient modulus [MPa]

Eq 8
800
Eq 9
Eq 10
600

400
Downloaded by [Swedish Road and Transport Research Inst] at 23:43 05 March 2014

200

0
0 20 40 60 80 100
CBR value [%]
Figure 2. Predicted stiffness values as a function of CBR values according to the
equations in Table 1

3. Test methods

3.1. CBR testing

The CBR test is frequently used in the structural design procedure of flexible
pavements. In the CBR test a specimen is compacted in a standard mould, with
compaction corresponding to Standard Proctor compaction. The CBR value is
obtained by measuring the relationship between force and penetration of a metal
plunge into the material, at a given constant rate, see Figure 3. The load required to
cause 2.54 and 5.08 mm insertion is recorded and expressed as a percentage of the
load required for the same penetration in a certain standard material.
It is commonly accepted that the induced stresses involved during the CBR test
poorly represents the actual stress situation that pavements experience during traffic
loading. The depth of insertion is more than sufficient to cause locally complex high
stress states in the material with possible permanent deformation as a result, which
comprises many repeated light loading cycles. Therefore it has been concluded that
the CBR value is not suitable as an input parameter for mechanistic design methods.
However as the CBR value is recorded as a stress at a given penetration (2.54 or
5.08 mm) one might argue that it represents the average slope of the stress
deformation curve and should therefore represent the stiffness over that interval of
deformation. For well graded compacted materials, where the aggregates are strong,
the largest part due to plunger penetration at the deformation 2.54 mm is due to
elastic response of the material and only up to a small extent due to plastic
deformation. Therefore can CBR-value give some indications of the actual stiffness
of the material.
 788 RMPD – 8/2007. Water in Pavements

Load

LVDT 20
Standard curve
15

Stress [MPa]
Result from test
Downloaded by [Swedish Road and Transport Research Inst] at 23:43 05 March 2014

10

5
Material
0
0 2 2,54 4 5,08 6 8 10

Penetration [mm]

Figure 3. CBR test set-up and a schematic view of the evaluation of the CBR value

In this project, the CBR test were performed at four different moisture contents
for each material, giving both the dry density as well as the CBR-value at these four
moisture contents. The highest CBR value after 2.54 mm penetration is thereafter
used in the comparison (Erlingsson and Magnusdottir, 2004).

3.2. Repeated Load Triaxial testing

In the RLT test performed here, 150 mm diameter specimen with a height of 300
mm were used. The materials were compacted according to the Proctor compaction
method (Erlingsson and Magnusdottir, 2002). In the testing procedure, first 20,000
load pulses are applied at the frequency 5 Hz for conditioning of the specimen,
thereafter sixteen stress paths are applied and during each stress path 100 symmetric
haversine load cycles are applied with a rise time of 50 ms (total length of pulse 0.1
sec) followed by a 0.9 sec rest time. During the last ten load cycle’s data from the
transducers as well as the axial load were collected to evaluate the specimen
response. The stress paths used are given in Table 2.
Typical results are shown in Figure 4 where measurements of the volumetric
strain εv and deviatoric strain εq are plotted for one material for all the sixteen stress
paths as a function of the mean stress level p. The data were further used to estimate
k1 and k2 from Equation [5] with the aid of the least square method and these results
are given as well in the figure.
 On Forecasting the Resilient Modulus 789

Table 2. Overview of the applied load pulses during stiffness testing

Confining Cyclic deviatoric Mean normal Number of

Sequence stress stress stress applied pulses
No σ3 ∆q p N
[kPa] [kPa] [kPa] [-]

0 50 120 90 20,000
1 30 30 40 100
Downloaded by [Swedish Road and Transport Research Inst] at 23:43 05 March 2014

2 30 45 45 100
3 30 60 50 100
4 30 75 55 100
5 50 45 65 100
6 50 60 70 100
7 50 75 75 100
8 50 90 80 100
9 90 75 115 100
10 90 90 120 100
11 90 120 130 100
12 90 150 140 100
13 150 120 190 100
14 150 180 210 100
15 150 240 230 100
16 150 300 250 100

1,4 1,2

1,2 1,0
1,0
0,8
ε q (10 )
ε v (10 )

-4
-4

0,8
0,6
0,6
0,4
0,4
K-Theta model 0,2 K-Theta model
0,2
Measurements Measurements
0,0 0,0
0 50 100 150 200 250 300 0 50 100 150 200 250 300
Mean stress level, p [kPa] Mean stress level, p [kPa]

Figure 4. Typical results from RLT testing of one material. Results from the
measurements of all the sixteen stress paths are shown along with the calculated
values based on the least square solution. a) Volumetric strain and b) deviatoric
strain versus the mean applied stress level
 790 RMPD – 8/2007. Water in Pavements

4. The materials

Totally twenty different granular materials were used in this study. They are all
typical base and subbase materials of different qualities, and somewhat different
petrology, however most of them were of basaltic origin. The maximum grain size
of all the materials was 22.4mm. Some parameters describing the materials can be
found in Table 3.
Downloaded by [Swedish Road and Transport Research Inst] at 23:43 05 March 2014

Table 3. Percent fines, uniformity coefficient, coefficient of gradation, unified soil

classification, Los Angeles abrasion value (LA), flakiness index (FI), specific gravity
(GS) and the dry density (ρdry) for the twenty materials used in this research

LA Density
D < 65µm Cu Cc FI Gs
Material USCS value ρdry
[%] [-] [-] [%] [-]
[%] [kg/m3]
Bakkasel 3.4 7.90 2.3 GW 16.5 19.1 2.98 1974
Bjorgun 5.0 36.0 7.1 GP-GM 16.4 1.3 2.93 2102
Brjanslaekur 2.3 10.0 0.7 GP 17.1 9.8 2.96 2084
Glera 4.0 18.0 1.4 GW 17.3 12.3 2.76 1958
Haukadalsa 4.6 23.3 1.5 GW 19.2 7.4 2.95 2186
Haumelar 1.2 5.30 1.7 GW 17.6 8.4 2.91 1969
Holabru 3.4 40.9 3.9 GP 15.7 6.3 2.98 2305
Holmkelsa 2.9 14.0 1.4 GW 21.7 6.3 2.74 1967
Hraunaos 5.2 44.0 4.9 GP-GM 15.6 13.6 2.99 2262
Jökulsa a Dal 2.2 25.0 53.0 GP 17.6 8.6 2.93 2172
Jökulsa a Fjö. 2.4 13.3 2.3 GW 22.2 4.8 2.78 1010
Krossanes 1.2 7.5 1.6 GW 16.7 34.5 2.99 1896
Larkot 3.9 26.8 1.9 GW 16.1 10.5 2.89 2205
Laekjarbotnar 4.3 20.0 2.2 GW 30.0 3.0 2.75 1821
Markarfljot 1.4 23.8 1.3 GW 18.5 7.0 2.81 2032
Nordfjardara 1.8 10.8 1.6 GW 19.2 14.0 2.93 2031
Raudamelur 5.1 48.0 12.0 GP-GM 23.5 3.0 2.60 1873
Stora-Fells. 3.9 44.0 5.8 GP 25.5 10.2 2.90 2178
Vallholt 3.3 30.0 1.2 GW 17.4 8.2 2.87 2165
Vatnsskard 3.7 30.0 4.8 GP 36.8 2.8 2.61 1822

From Table 3 it can be seen that all the tested materials are rather low in fine
contents, or D < 65 µm between 1.2 - 5.2%. Twelve of the materials are classified as
well graded gravel (GW) and only five as poorly graded gravel (GP) according to
the Unified Soil Classification System (USCS). Further have fourteen of the
materials Los Angeles, LA, abrasion values < 20 and would therefore be classified as
material with good resistance to fragmentations according to CEN norm 1342.
 On Forecasting the Resilient Modulus 791

Thirteen of the materials have flakiness index, FI, lower than 10 and are therefore
not considered flaky. However, two materials are rather flaky, the Bakkasel material
and Krossanes material with FI of 19.1 and 34.5 respectively. Further was the
percentage of crust particles (not shown in Table 3) for the material between 23 and
100%.
The grain size distribution curves of all the materials are given in Figure 5. Even
though it is difficult to separate the different curves on the figure one can get some
general overview of the shapes of the different grain size distribution curves.
Downloaded by [Swedish Road and Transport Research Inst] at 23:43 05 March 2014

The materials were tested at four different moisture contents values. Their target
moisture content values were: wopt – 2%, wopt – 1%, wopt and wopt + 2% where wopt is
the optimum moisture content. The compaction curve of the materials turned out to
show little dependency of moisture meaning that the dry density was fairly
independent of the moisture content. Therefore it was sometimes difficult to
estimate the optimum moisture content. This is probably due to the low fine content
of the materials. Even though the grain size curves of the materials includes in all
cases grains between 0 - 22.4 mm, due to the lack of fines, compaction with the
Standard Proctor method did not cause any difference in dry densities in the range of
used moisture contents. The resilient modulus and the CBR value showed on the
other hand much more dependency on moisture. In particular the CBR value, where
in some cases the difference between the lowest and the highest value was more than
100%. For the resilient modulus most material showed a variation less that 35% and
only occasionally the difference was about 50%. It was therefore decided to use in
this study only the highest measured CBR value as well as the highest resilient
modulus value for each material, although they where not always measured at the
same moisture content.

100

80
Percent fines [%]

60

40

20

0
0,01 0,1 1 10 100
Grain size, D [mm]
Figure 5. Overview of the grain size distribution of the twenty materials used in this
study
 792 RMPD – 8/2007. Water in Pavements

5. The results

Figure 6 shows the stiffness and the CBR values plotted as a function of the dry
density of the materials. There seems to be some relationship between CBR values
and dry density of the materials. That is as the density increases the CBR-value of
the materials increases. For the resilient modulus a relationship does hardly exists at
all.
800 200
Downloaded by [Swedish Road and Transport Research Inst] at 23:43 05 March 2014

Stiffness, M r [MPa]

CBR-value [%]
600 150

400 100

200 50

0 0
1600 1800 2000 2200 2400 1600 1800 2000 2200 2400
3 3
Dry density, ρ dry [kg/m ] Dry density, ρ dry [kg/m ]

Figure 6. Measured maximum stiffness at mean stress level p = 250 kPa and CBR
value as a function of dry density for all the twenty materials

To make a comparison between the resilient modulus and the CBR-value one
must decide what stress level should be used as the resilient modulus is stress
dependent. Here are mainly base course materials considered which are commonly
used under thin surface dressings. Under such conditions it is not uncommon that
induced vertical stress due to a 12 ton axle load with a tire pressure of 800 kPa is of
the order 450 kPa at 10-15 cm depth under the surface and the horizontal stresses are
approximately one third of that value (Erlingsson, 2002; Erlingsson and Ingason,
2004). The mean stress level p becomes therefore approximately 250 kPa. This
value is used here as a reference mean stress level. The predicted stiffness values
based on the CBR values from Table 1 are given again in Figure 7 along with the
actual stiffness measurements.
One can see from Figure 7 that the last two equations in Table 1 (Equations [9]
and [10]) seem to give a much better agreement with the measurements in particular
for CBR values higher than approximately 60%. Further gives Equation [9] better
agreement with the data than Equation [10]. Here one must bear in mind that it has
been decided to use the mean stress level as p = 250 kPa when plotting the stiffness
values. For higher stress levels it is possible that better agreement between the other
two equations could be established. The two materials showing the largest deviation
from Equation [9] in Figure 7 are the Krossanes and the Bakkasel material. They are
both very flaky, FI = 34.5 and 19.1 respectively, which is probably affecting their
behaviour.
 On Forecasting the Resilient Modulus 793

1000
Measurements
Resilient modulus [MPa]

Eq 7
750 Eq 8
Eq 9
Eq 10
500
Downloaded by [Swedish Road and Transport Research Inst] at 23:43 05 March 2014

250

0
0 20 40 60 80 100 120 140
CBR value [%]

Figure 7. Predicted resilient modulus based on the CBR value from the four
equations in Table 1 as well as the measured stiffness from the RLT test
measurements at the mean stress level p = 250 kPa

Table 4 gives then the k1 and k2 values according to Equation [5] for all the
twenty materials. One can see in Table 4 that the k1 value ranges from 137 to 250
and that the k2 value ranges from 0.272 to 0.630. Further it can be seen that the k2
value lies for most of the materials in the range 0.33 – 0.45 and the average value is
0.37. If it is assumed that k2 = 0.4 is sufficiently close to reality for the different
materials one can back-calculate the k1 value for the materials based on the
prediction equations relating CBR value to the resilient modulus here using p = 250
kPa, thus for Equation [9] in Table 1 this gives:

Mr 20.7 ⋅ CBR 0.65

kˆ1 = k2
= 0.4
= 9.25 ⋅ CBR 0.65 [11]
 3p   3 ⋅ 250 
   
 pa   100 

The last column in Table 4 gives the differences of the resilient modulus by
using k1 and k2 in Equation [5] and k̂1 and k2 = 0.4 respectively at the mean stress
level p = 250 kPa. One can see that thirteen of the materials have CBR values higher
than approximately 60% and of these, nine materials give ∆ lower that ±20% using
k̂1 and k2 = 0.4 instead of k1 and k2 for predicting the stiffness. Two of the other four
materials have rather high Flakiness Index, FI = 13.6 and 10.5 respectively which
might influence the result.
The Krossanes material shows the highest deviation of all the materials in the
differences in the resilient modulus at the mean stress level p = 250 kPa by the two
 794 RMPD – 8/2007. Water in Pavements

methods. That material has a CBR value of 30 and is very flaky or FI = 34.5 which
is probably influencing the results.

Table 4. CBR value and the average of k1 and k2 values from the RLT testing, the
calculated k1 value if k2 = 0.4 and the % difference between the stiffness at p = 250
kPa for all the twenty materials

CBR-
Downloaded by [Swedish Road and Transport Research Inst] at 23:43 05 March 2014

value
k1 k2 k̂1 ∆
Material
[MPa] [-] [MPa] [%]
[%]
Bakkasel 52 228.6 0.272 120.65 31.7
Björgun 91 152.3 0.630 173.59 24.0
Brjanslaekur 44 151.2 0.370 108.24 28.3
Glera 59 160.9 0.351 130.98 10.2
Haukadalsa 94 151.7 0.344 177.29 -30.8
Haumelar 40 175.5 0.394 101.74 41.3
Holabru 140 214.6 0.430 229.68 -0.7
Holmkelsa 68 159.6 0.374 143.64 5.2
Hraunaos 119 141.9 0.400 206.66 -45.6
Jökulsa a Dal 76 137.7 0.442 154.41 -3.0
Jökulsa a Fjöllum 49 151.6 0.296 116.08 5.6
Krossanes 30 464.2 0.153 84.39 70.1
Larkot 74 249.4 0.288 151.75 23.7
Laekjarbotnar 37 159.7 0.414 96.71 41.1
Markarfljot 81 157.1 0.419 160.94 1.4
Nordfjardara 42 153.3 0.383 105.02 29.1
Raudamelur 89 164.7 0.351 171.10 -14.7
Stora-Fellsöxl 95 177.9 0.339 178.51 -13.5
Vallholt 130 176.0 0.462 218.88 -9.8
Vatnsskard 68 144.9 0.340 143.64 -11.9

The stress dependency of the resilient modulus can now be estimated for the
materials by combining Equations [11] and [5] as:
0.4
 3p 
M r = 9.25 ⋅ CBR 0.65
⋅   [12]
 pa 
where the value of k2 has been assumed as 0.4. The resilient modulus versus mean
stress level for two of the materials are given in Figure 8 where both the actual
 On Forecasting the Resilient Modulus 795

measurements using k1 and k2 are given as well as the predicted stiffness based on
k̂1 and k2 = 0.4.
As one can see the difference is not very significant. For the Raudamelur
material the predicted curve deviates from the measurements as the mean stress level
increases. This is due to than the actual k2 value 0.351 is lower that 0.4 which is used
to create the predicted curve. For the other material, Jökulsa a Dal, the k2 value
0.442 is higher than 0.4 so the two curves get closer as the mean stress level
increases.
Downloaded by [Swedish Road and Transport Research Inst] at 23:43 05 March 2014

500 500
Stiffness, Mr [MPa]

Stiffness, Mr [MPa]
400 400
300 300
200 200
Measured
100 100 Measured
Predicted
Predicted
0 0
0 100 200 300 0 100 200 300
Mean stress level, p [kPa] Mean stress level, p [kPa]

Figure 8. Resilient modulus versus mean stress level for two materials based on
actual k1 and k2 values and k1 and k2 = 0.4 respectively. a) the Raudamelur
material with ∆ = -14.7% (at p = 250 kPa) and b) the Jökulsa a Dal material with
∆ = -3.0% (at p = 250kPa)

6. Conclusions

Here twenty unbound granular base course and subbase materials have been
tested with a CBR test to obtain the CBR value and RLT test to estimate their
resilient modulus. The grain size distribution of the materials is between 0 and
22.4mm. They are all rather low in fines and twelve of the twenty materials used in
the study are classified as well graded gravel according to the USCS. The grains of
the materials were of different quality in many aspects, as indicated by a large
variety of LA value, FI as well as the percentage of crust particles of the materials.
The tests were carried out at four moisture contents. However, the compaction curve
of the materials showed little dependency of moisture. This is interpreted here as due
to the low fine content of the materials (D < 65µm between 1.2 - 5.2%).
Based on the test results it is concluded that the CBR value of the unbound
granular materials show some dependency of the maximum dry unit weight of the
material. This is true at least for well graded gravel with a low fines content. It is
further concluded that for materials with higher CBR values than approximately
60%, a simple forecasting equation can be used to estimate their resilient modulus
based on the CBR value. For a mean stress level p = 250kPa the prediction equation
 796 RMPD – 8/2007. Water in Pavements

from the South African Council on Scientific and Industrial Research (Equation [9])
gives the best results of those equations presented here. It is further concluded that if
the k - θ expression is used for describing the stress dependency of the stiffness
modulus and using an average value of k2 = 0.4, then the k1 value can be predicted
by a simple power law based on the CBR value. This seems to give somewhat
satisfactory estimation of the stiffness behaviour for the range of mean stress level
between 0 - 300 kPa. As the materials had rather low fine contents their compaction
curves were rather independent of the moisture content. It was therefore not possible
to include moisture content dependency in the forecasting equation given in the
Downloaded by [Swedish Road and Transport Research Inst] at 23:43 05 March 2014

paper.
This is a limited analysis including only twenty materials with large variation in
terms of diverse properties although all have been used as either base course or
subbase materials in thin pavement structures. Study on other materials, with other
characteristics, may give different results. The reader should therefore consider
carefully if the materials presented here, match the ones he or she is interested in,
before applying the equations given in the paper.

Acknowledgements
The work described in this paper was sponsored by the Icelandic Road
Administration (ICERA) as well the Swedish National Road and Transport Research
Institute (VTI).

7. Bibliography

Brown S.F., “Soil mechanics in pavement engineering”, Geotechnique, Vol. 46, No. 3, 1996,
p. 382-426.
Brown S.F. and Pappin J.W., “Analysis of Pavements with Granular Bases”, Transportation
Research Record, No. 810, transportation Research Board, Washington, D.C., 1981.
Correia G., Hornych, P. and Akou, Y., “Review of models and modeling of unbound granular
materials”, Workshop on modeling and advanced testing for unbound granular materials,
Lisbon, Balkema, 1999, p. 3-15.
Erlingsson S., “Dynamic triaxial testing of unbound base course materials”, Proceedings of
the Nordic Geotechnical Conference - NGM 2000, Finish Geotechnical Society, Helsinki,
June 2000, p. 69-76.
Erlingsson S., “3-D FE Analyses of Test Road Structures – Comparison with Measurements”,
Proceedings from the 6th International Conference on the Bearing Capacity of Roads,
Railways and Airfields, Lisbon, 2002, p. 145-157.
Erlingsson S. and Magnusdottir B., “Dynamic Triaxial Testing of Unbound Granular Base
Course Materials”, Proceedings from the 6th International Conference on the Bearing
Capacity of Roads, Railways and Airfields, Lisbon, 2002, p. 989-1000.
 On Forecasting the Resilient Modulus 797

Erlingsson S. and Magnusdottir B., “Comparison between stiffness values from RLTT and
CBR-values for unbound granular materials”, Proceedings of the Nordic Geotechnical
Conference - NGM 2004, Swedish Geotechnical Society SGF Report 3, Vol. I, 2004,
p. B39-B47.
Erlingsson S. and Ingason Th., “Performance of two thin pavement structures during
Accelerated Pavement Testing under a Heavy Vehicle Simulator”, Proccedings from the
2nd International Conference on Accelerated Pavement Testing, Minnesota, 2004,
CD-Rom.
Downloaded by [Swedish Road and Transport Research Inst] at 23:43 05 March 2014

Hoff I., Material Properties of Unbound Aggregates for Pavement Structures, Ph.D. Thesis,
Institutt for veg- og jernbanebygging, NTNU, Trondheim, 1999.
Yoder E.J. and Witczak M.W., Principles of Pavement Engineering, 2nd Ed. John Wiley &
Sons, Inc, 1975.
Schwartz C.W., “Effect of Stress-Dependent Base Layer on the Superposition of Flexible
Pavement Solutions”, International Journal of Geomechanics, Vol. 2, No. 3, 2002,
p. 331-352.
Witczak M.W., Qi X. and Mirza M.W., “Use of Nonlinear Subgrade Modulus in AASHTO
Design Procedure”, Journal of Transportation Engineering, Vol. 121, 1995, p. 273-282.

Received: 16 October 2006

Accepted: 16 July 2007

View publication stats