You are on page 1of 37

This article was downloaded by: [Simon Fraser University]

On: 13 November 2014, At: 14:49


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

Influence of Water on Resilient Properties of Coarse


Granular Materials
a a
Jonas Ekblad & Ulf Isacsson
a
Royal Institute of Technology, Division of Highway Engineering , 10044, Stockholm,
Sweden E-mail:
Published online: 19 Sep 2011.

To cite this article: Jonas Ekblad & Ulf Isacsson (2006) Influence of Water on Resilient Properties of Coarse Granular
Materials, Road Materials and Pavement Design, 7:3, 369-404

To link to this article: http://dx.doi.org/10.1080/14680629.2006.9690043

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
Influence of Water on Resilient Properties
of Coarse Granular Materials

Jonas Ekblad — Ulf Isacsson

Royal Institute of Technology


Downloaded by [Simon Fraser University] at 14:49 13 November 2014

Division of Highway Engineering


10044 Stockholm, Sweden
ekblad@byv.kth.se
ulf.isacsson@byv.kth.se

ABSTRACT. The objective of this work was to investigate influence of water on resilient
properties of a coarse (maximum particle size 90 mm) granular material using various
gradings. Triaxial testing, using large-size samples (diameter 500 mm and height 1000 mm)
and constant confining pressures, was performed at varying water content up to practically
full saturation. Time Domain Reflectometry (TDR) was used to monitor transient in-sample
water contents. Evaluation comprised resilient modulus and Poisson ratio in terms of total
stresses, although the results were also discussed in the context of effective stresses. The
coarsest grading experienced only a small reduction in resilient modulus when brought close
to saturation, while specimens with an increased amount of fines and more even distribution
responded with a substantial loss of resilient modulus upon increased water content. It also
appeared that, as water content increased, the specimens became more dilative (increase in
Poisson ratio).
KEYWORDS: Granular Soils, Effective Stress, Resilient Modulus, Water Content, Time Domain
Reflectometry, Matric Suction.

Road Materials and Pavement Design. Volume 7 – No. 3/2006, pages 369 to 404
370 Road Materials and Pavement Design. Volume 7 – No. 3/2006

1. Introduction

From an engineering point of view, a road seems deceivingly simple. A more


careful look reveals the complexity of a layered structure of composites on a
supposedly infinite subgrade. The constituting materials are indeed complex,
unbound materials being non linear elasto-plastic and the bitumen bound layers
nonlinear viscoelastic-plastic. The structural integrity of a road construction relies
heavily on the unbound base and subbase layers; nevertheless mechanical and
hydraulic properties for coarse gradings are fairly unknown.
The influence of water and suctions on granular soils has been less investigated
than for fine grained and cohesive soils, the reason of which quite possibly can be
Downloaded by [Simon Fraser University] at 14:49 13 November 2014

lower suctions than in fine-grained soils. Furthermore, the suctions in granular soils
are expected to be considerably smaller than the stress levels imposed by self-weight
or construction for most engineering activities. Even though it might have a small or
negligible influence on mechanical behavior, soil suction is a crucial parameter
when modeling soil water movements.
The main objective of the research presented in this paper was to investigate
influence of water on resilient properties of coarse unbound granular materials in
saturated as well as unsaturated state. This study was limited to one type of
aggregate at four different gradings with maximum particle size 90 mm. Given the
size of the samples tested, it was deemed necessary to continuously monitor the
water content at various sample heights, which in turn necessitated evaluation and
use of instruments for water content determination. The general measures of
mechanical response were resilient modulus and Poisson ratio measured in triaxial
tests with constant confining pressure. Given the stress dependency of granular
materials in addition to externally applied stresses, also internal stresses, pore air and
water pressures were monitored.

2. Literature review

It is probably generally agreed that increased water content will have detrimental
effects on pavement response both in terms of resilient response and accumulation
of deformation, i.e. rutting. A more yielding resilient response may shorten life
length of the bituminous bound upper layers by increasing induced strain levels, thus
possibly increasing fatigue damage. It appears research on water influence on
overall pavement performance has mainly been focusing on properties of the
subgrade. Consequently, less information is available regarding influence of degree
of saturation on coarse crushed rock performance (base and subbase layers). In
many cases, testing has been performed close to optimum moisture content level,
slightly above or below. The reason is probably that this type of material is supposed
to be less susceptible to moisture changes due to coarser gradings and, furthermore,
also supposed to remain unsaturated during service. Pavements are constructionally
Influence of Water on Resilient Properties 371

designed and built to keep the pavement base unsaturated. Increased water content
might have an effect on several aspects of pavement behavior, e.g. resilient and
plastic deformation, permeability, pumping and channeling of fines and water, frost
heave etc. In this section, a short review of previous findings on the influence of
water on resilient properties of base and subbase materials (or similar) is given.
One of the earliest reports describing behavior of granular materials in
pavements and susceptibility of water was presented by Haynes and Yoder (1963).
They conducted triaxial laboratory testing on samples of material used in one of the
AASHO road test sections. One of their specific aims was to investigate how the
accumulative and resilient (rebound) response was affected by degree of saturation
up to complete saturation. The crushed material they used could retain water
Downloaded by [Simon Fraser University] at 14:49 13 November 2014

corresponding to roughly 80% saturation. In the tested interval between


approximately 60-80% saturation, the resilient deformation increased about one
third. They also performed tests on gravels which could be fully saturated under
laboratory conditions. At almost full saturation (about 98%), the resilient
deformation increased twofold compared to a saturation level of 70%. Morgan
(1966) performed, as part of an investigation on pavement behavior, triaxial tests on
dry and saturated granular materials. Morgan reported just a small tendency for
saturated samples to show larger permanent and resilient strains. Furthermore,
Morgan found that drained samples behave similar to dry. The grading of the
materials tested were sand-like i.e. fairly small maximum particle size and uniform
shape of the grading curve. Sweere (1990) performed triaxial tests on dry and wet
samples of granular base course materials, and observed a large reduction in resilient
modulus and for some materials a less nonlinear response of wet samples compared
to dry. A considerable reduction in characteristic resilient modulus by increased
water contents, in states drier than optimum, was observed by Dawson et al. (1996)
and Hornych et al. (1998) who observed reductions in characteristic modulus
(p = 250 kPa, q = 500 kPa) ranging from about 40% to well over 80% and in excess
of 50%, respectively. Measurements by Kolisoja et al. (2002) are also, basically, in
concurrence with previous results, i.e. loss of resilience upon wetting.
It is usually suggested that decrease in resilient modulus caused by increased
water content can be explained by the development of excess pore water pressures
and if effective stress or an extension of the effective stress concept into unsaturated
conditions is used, no reduction in resilient modulus is observed, e.g. Hicks (1970)
and Pappin et al. (1992). However, this statement is only in part supported by results
presented. Pappin et al. (1992) concluded that the results could be modeled provided
effective stresses and an equivalent pore pressure for the unsaturated state were
used. To derive the equivalent pore pressure, they compared unsaturated behavior
with dry and saturated behavior, thus adjusting what they called the true effective
stress. Unfortunately, these pore pressure corrections, used in the effective stress
calculations, did not appear to bear any direct relation to the applied suction values.
Thom (1988) showed excellent predictions of volumetric and shear strains on
saturated samples from equations derived from tests on dry samples by substituting
372 Road Materials and Pavement Design. Volume 7 – No. 3/2006

the total stress with effective stress. The evident conclusion was that the effective
stress principle is valid, even though caution is advised as this study was very
limited. Raad et al. (1992) noticed a small initial reduction of resilient modulus for
saturated compared to moist samples, when subjected to repeated loading. When
substantial pore pressures were induced, the modulus decreased (50-80% reduction).
Although the measurements were not analyzed in terms of effective stress, it seems
that presented data, at least to some extent, is supportive of the effective stress
principle for saturated samples. Tian et al. (1998) tested a coarse granular material at
moisture above and below optimum moisture content. At elevated moisture content,
a reduction of about 20% was observed. This reduction was attributed to decrease in
matric suction although no measurements of the actual level of matric suction are
Downloaded by [Simon Fraser University] at 14:49 13 November 2014

given. Based on this assumption, an increase in modulus could be expected for the
moisture content below optimum, but this increase was not recorded in the
measurements. In this context, a point made by Thom (1988) is noteworthy; he
discussed, based on measurements, the complexity of matric suction exerted in the
unsaturated state and Thom also mentions confounding mechanisms as cementation.
Estimation of the influence of water on resilient properties of granular soils can
also be done from field data. In an attempt to examine various factors influencing
resilient performance of a variety of granular materials, and possibly quantify them,
Rada and Witczak (1981) compiled a large amount of previously published data and
also performed some additional tests. Their overall conclusion in terms of moisture
effects was that resilience is lost by increased degrees of saturation, when saturated,
resilient modulus is reduced to one third of the value at low water contents. It should
be noted that there is a substantial amount of scatter in the compiled data, which is
to be expected considering the number of different materials and laboratories. On an
Icelandic test road section, Erlingsson et al. (2000) measured a seasonal change in
base layer modulus. From the weakest response during spring thawing, the modulus
almost doubled when reaching its maximum value in late summer. Ksaibati et al.
(2000) reports almost a total loss of resilience of the base layer in some cases, while
a more common reduction is around 30%. For all of these measurements, the actual
degree of saturation or level of matric suctions is unknown. It is probable that at
least some measurements were done close to saturation. Chandra et al. (1990)
concluded that effects on base modulus due to changes in (total) suction were too
small to be detected by backcalculating FWD-data (Falling Weight Deflectometer).
From the data used, the base moduli varied by less than 7 percent. The authors
pointed out that the range of moisture contents measured were fairly limited.
Approximately the same pattern was also reported by Ali and Lopez (1996). They
noticed no influence of base moduli by gravimetric water content from data
collected during one year within the Long-Term Pavement Performance Program
(LTPP). It is remarked that this site is located in a dry no-freeze zone. Further
analysis of LTPP data comprising 24 sites have been performed by Richter and
Schwartz (2003). Basically, they reached the same conclusion as Ali and Lopez, i.e.
no loss of resilient properties of the pavement layers by increased moisture content.
In this case the opposite reaction was the most common, that is increased modulus
Influence of Water on Resilient Properties 373

with increased moisture content. This apparent anomaly is discussed in some detail.
For all these investigations, it should be remembered that it is inherently difficult to
unambiguously isolate the influence of granular base behavior from backcalculated
falling weight data. Also, the scatter of different base materials and gradations is
considerable.
Fewer reports on the effect on Poisson ratio are available. Morgan (1966), as
previously mentioned, concluded that Poisson ratio remains essentially unaffected,
but a small increase at saturation was observed for some samples. Hicks and
Monismith (1971) found that in general, Poisson ratio decreased with degree of
saturation, a result also supported by Van Toan (1975), while Dawson et al. (1996)
generally observed an increase with water content (below saturation).
Downloaded by [Simon Fraser University] at 14:49 13 November 2014

To summarize, it seems that laboratory investigations commonly report a


substantial loss of resilience, when water content is increased, while experience
from field data is more uncertain. Concerning Poisson ratio, studied reports are
inconclusive but mainly showing only minor influence. Many studies used total
stress (generally normal stress) as state variable to which resilient modulus is
related. In some cases, effective stress in saturated state and extended effective stress
in unsaturated state were used successfully. However, these observations are more
ambiguous and scattered but commonly, the impression of the usefulness of the
effective stress approach is cautiously positive.

3. Experimental

3.1. Equipment

Testing of large maximum particle size materials requires a large sample size.
During the course of this study, tests have been performed on cylindrical specimens,
500 mm in diameter and 1000 mm in height. Furthermore, because of the test
schedule where individual samples were to be tested at increased water contents, it
was deemed necessary to be able to monitor water movements non-destructively in
real-time. Given the height and size of the sample, it would be very difficult to
estimate equilibrium conditions in terms of matric suction and water distribution. In-
sample measurement is achieved by burying TDR-probes (Time Domain
Reflectometry) at two heights surrounding the part of the triaxial sample where
displacements are measured.
The main components of the triaxial testing equipment are a large confining cell
and two separate hydraulic actuators, one controlling the deviatoric load and the
other confining pressure (Lekarp and Isacsson, 2000). Either air or silicon oil can be
used as confining medium. If silicon oil is used, it is possible to cycle the confining
pressure independently from the deviatoric load. During this work, testing was
carried out using air to provide constant confining pressures.
374 Road Materials and Pavement Design. Volume 7 – No. 3/2006

500
Pore pressure

TDR probe

Circumferential
extensometer
Downloaded by [Simon Fraser University] at 14:49 13 November 2014

Pore pressure
1000

Tensiometer
LVDT

TDR probe

In-sample instrumentation On-sample instrumentation

Figure 1. Specimen instrumentation (dimensions in mm)

To allow for water increase and drainage, separate tubing is connected to the top
and bottom plate, respectively. In the plates drainage is provided through three
coarse filters (diameter 80 mm) evenly distributed on the surface facing the sample.
In Figure 1, a schematic of instrumentation is shown. To the left, transducers
inside the sample are shown and to the right, on-sample transducers measuring axial
and circumferential length changes. Three LVDT’s, spaced 120°, apart were used to
measure axial deformation. The LVDT’s are attached to anchoring devices buried
during compaction. Circumferential measurement is provided by a strain gaged
extensometer attached between the ends of a roller chain, wrapped around the
specimen at mid height. The rollers on the chain extend outside the link plates. Steel
springs are used to connect the chain ends. In addition to transducers shown in
Figure 1, a load cell, a confining pressure sensor and a LVDT measuring top-plate
movements are attached.
To measure matric suction, a tensiometer, 2100F, manufactured by Soilmoisture
Equipment Corp. was used. The standard Bourdon type vacuum gauge was replaced
by a strain gaged pressure sensor. This replacement allowed for continuous data
logging and, maybe equally important, reduced response time by decreasing the
Influence of Water on Resilient Properties 375

amount of water that has to flow through the porous cup to reach stress equilibrium.
The porous ceramic cup has a nominal bubble pressure of approximately 100 kPa.
The plastic tube, hydraulically connecting the porous cup to the pressure sensor, has
to be fed through both the membranes and the outer pressure cell, thus providing an
uninterrupted connection. Furthermore, while passing between the sample surface
and the cell wall, the plastic tube has to be shielded from the confining pressure,
which would otherwise disturb the pressure measurements. This is achieved by
routing the plastic tube through a nylon tube used for compressed air, which in both
ends is attached to standard pneumatic push-in adaptors. At the sample-membrane
interface, the connection is sealed using an o-ring. This connection is shown in
Figure 2. Similar connections were also used for the cable feedthroughs (TDR
Downloaded by [Simon Fraser University] at 14:49 13 November 2014

probes and pressure sensor).

Membrane

Anchor
: O-ring seals

Pneumatic "Pus h-In-F itting"


8 mm adaptor Ceramic porous cup

8 mm nylon tube

Was her

50 mm

Figure 2. Tensiometer tube feedtrough

The anchoring device, shown in Figure 2, is buried during compaction.


Remaining parts are subsequentially attached to the anchor after the sealing
membrane has been added. The other end of the nylon tube is connected to a similar
push-in adaptor in the bottom plate. This seal has been working flawlessly for both
air and silicon oil. No leaks have been detected. Pressure readings from the
tensiometer are gauge pressure, i.e. the water pressure deficiency relative to ambient
atmospheric pressure, while the matric suction is the difference between the pore air
pressure and pore water pressure. The pressure sensor located in the top-plate is used
to measure pore air pressure and can correct for any differences in pressure.
To measure excess pore water pressure when the sample is saturated, a pressure
transducer is buried inside the sample as shown in Figure 1. In the unsaturated state,
this sensor will act as an additional air pressure sensor. As stated by Dempsey
(1982), pore pressure transducers with a porous filter, usually ceramic, cannot be
used, mainly because of slow response time. Furthermore, during repeated testing,
this type of transducers can become clogged with fines thus reducing the sensibility
even more. Nevertheless, the transducers need some kind of protection to reduce the
376 Road Materials and Pavement Design. Volume 7 – No. 3/2006

influence of effective soil pressure caused by the material weight and imposed loads.
Two types of pressure sensors have been used in this study, a miniature transducer
Entran EPX and a Druck PDCR similar to the sensor located at the top of the
specimen. Both are measuring absolute pressures. The miniature transducer used a
coarse titanium filter with pore sizes with low bubble pressure, thus permeable to
both atmospheric pressure air and water, whereas the head of the Druck PDCR
transducer was filled with glass wool, preventing fines intrusion, and capped with a
cone containing holes. In the unsaturated state, none of the transducers seemed to be
influenced by soil stress, when compared to the pore air transducer in the top plate.
Even though it is larger in size compared to the miniature sensor by Entran, the
Druck PDCR sensor is not excessively large compared to the largest particles. Since
Downloaded by [Simon Fraser University] at 14:49 13 November 2014

the sensor is placed within the region of dimensional measurements, both axial and
radial, its influence on mechanical properties of the sample should be negligible.
As previously mentioned the technique to measure water content was Time
Domain Reflectometry (TDR). It has been known for a long time (e.g. Smith-Rose,
1933) that measured dielectric properties of soils can be related to water content.
Early studies were usually performed in the frequency domain but in the late 1960s
and early 1970s reports of measurements in time domain were published by e.g.
Fellner-Feldegg (1969), Hoekstra and Delaney (1974) and Davis and Chudoboviak
(1975). A large part of published work concerns agricultural engineering and
consequently soils commonly studied differ from the crushed aggregate used in this
study. However, the main purpose of the measurements performed was to monitor
water dynamics when adding water in the unsaturated state and draining from
saturation. Consequently, no elaborate review of TDR-technology or actual
measurements are given in this paper. The basis of TDR measurement is to measure
the speed of an electromagnetic pulse, which is dependant of the relative
permittivity of the surrounding medium. The relative permittivity of free water is
much higher than the other soil constituents, why measured permittivity is mainly
influenced by water content. TDR-probes used in this work were manufactured by
Campbell Scientific (CS605). They are a 3-rod design with 300 mm length and 45
mm spacing between outer rods. Time domain measurements were performed using
a reflectometer, TDR100, from Campbell Scientific. It includes a step-pulse
generator, data acquisition and an instrument-implemented waveform analysis
algorithm.

3.2. Materials

3.2.1. Basic characteristics


Four different gradings of the same crushed aggregate were tested. The origin of
the material used was Skärlunda (Östergötland, Sweden) and is characterized as a
foliated medium grained granite with (based on point counting) quarts, K-feldspars
and plagioclase as main constituents. Phyllosilicates (muscovite, biotite and chlorite)
Influence of Water on Resilient Properties 377

are also present, comprising about 10% by point counts. The nominal particle size
distributions (gradings) were derived using the equation:

n
 d 
P =   [1]
 DMax 

where P is the percent smaller than d, Dmax is the maximum particle size and n is the
grading coefficient describing the shape of the curve. This equation (or similar in
formulation), describing a continuous particle size distribution, has been used by e.g.
Fuller (1905), Talbot and Richart (1923) and Andreasen and Andersen (1930).
Downloaded by [Simon Fraser University] at 14:49 13 November 2014

Fuller (1905) assumed that a grading coefficient of 0.5 will give “the best theoretical
mix”, while Andreasen and Andersen (1930) gave it some theoretical justification.
For crushed rock, in practice maximum density (minimum voids), is achieved at
lower n-values than 0.5 (e.g. Fuller and Thompson, 1907; Andreasen and Andersen,
1930). In this investigation, a maximum particle size of 90 mm and four different
distributions corresponding to gradings coefficient: 0.3, 0.4, 0.5 and 0.8,
respectively, were used. This gives a range spanning from a fairly coarse material to
a material containing a substantial amounts of fines. Nominal and actual particle size
distributions are visualized in Figure 3. Eight narrow fractions with known particle
size distributions were used to compose the required gradings, which allowed for a
close agreement to nominal distributions. This measure also reduced the risk of
aggregate separation during sample preparation, when each triaxial sample was
manufactured from sub samples equaling the amount required for compaction of one
layer (cf. Section 3.2.2).
All the distributions described, except n = 0.3, fit within the limits, for a subbase,
given by Swedish guidelines (ATB Väg, 2003). Grading 0.3 contains too much of
particles smaller than 1 mm with a maximum deviation from the upper limiting
curve of approximately 4%-units. A grading parameter of 0.5 is quite close to an
average curve drawn in the middle between the given limits.
Other parameters used to describe the gradings are given in Table 1. Uniformity,
(U), is defined as:

D60
U= [2]
D10

where D60 is the nominal size for which 60% of the material is passing, and D10 is
the corresponding size for 10%.
378 Road Materials and Pavement Design. Volume 7 – No. 3/2006

100
Nominal
90 Actual
80
Passing [%-by weight]

70

60

50

40 0.3

30 0.4
0.5
Downloaded by [Simon Fraser University] at 14:49 13 November 2014

20

10 0.8
0
0.063 0.125 0.25 0.5 1 2 4 8 16 31.5 63 90

Sieve size [mm]

Figure 3. Nominal and actual particle size distributions for the different gradings
used. Maximum particle size was 90 mm and grading coefficients: 0.3, 0.4 0.5 and
0.8

Table 1. Grading characteristics for Skärlunda (maximum particle size 90 mm)

Grading Effective size Fines <63 µm


D10 [mm] Uniformity [ ] [% by weight]

0.8 5.1 9.4 0.7

0.5 0.90 36 2.6

0.4 0.28 88 5.9

0.3 0.04 390 12.5

Densities for the different gradings are given in Table 2. Specific gravity was
measured according to standard methods concerning densities of aggregate
materials. These methods are based on submerging the aggregate in various fluids,
depending on particle sizes, measuring dry and submerged weights respectively.
Maximum density was determined according to ASTM D4253, using a vibratory
table on which a sample is contained in a cylinder and heavily vibrated under a static
pressure and a surplus of water. The specified validity of this method is restricted to
cohesionless, free-draining soils containing less then 15% by weight particles
smaller than 74 µm and with maximum particle size of 75 mm. Even though the
chosen gradings might violate one or more of these restrictions, it was decided not to
Influence of Water on Resilient Properties 379

alter the gradings by removing particles above 75 mm, thereby changing the
material. It was also decided not to perform testing for optimum moisture content.
The main reasons behind this decision were, limitations in the current methods
regarding coarse materials, and the fact that the compaction of the samples was to be
done at fairly low water contents.

Table 2. Densities for the different distributions designated by the grading


coefficient

Grading Specific gravity [kg/m³] Maximum density [kg/m³]


Downloaded by [Simon Fraser University] at 14:49 13 November 2014

0.8 2637 1863

0.5 2630 2124

0.4 2629 2273

0.3 2615 2353

While there is only a small variance (<1%) in specific gravity, the maximum
density is clearly influenced by grading, the coarser the grading the lower the
maximum bulk density. The decreasing density is almost linearly related to
increasing grading coefficient.

3.2.2. Specimen preparation


Since no reference to a more or less standardized compaction procedure could be
made, given the size of the triaxial specimens, a detailed description of the
procedure and equipment used is given.
Compaction was done manually in a three-section split mold. A nitril rubber
membrane (NBR, 1.1 mm thickness) is held inside the mold by vacuum during the
compaction. This membrane is necessary to allow demolding without sample failure
or partial loss of material. Due to the coarse and sharp nature of the aggregate and
the heavy compaction, it is punctured during compaction, why a second membrane
is used for sealing. An absolute seal is critical for further testing. Batches of
approximately 45 kg, enough for one layer, were prepared by mixing deionized
water and the initially dry aggregate in a 125 liter concrete mixer. The specimen is
compacted in 10 layers with equal mass, varying the height of each layer according
to the method suggested by Ladd (1978). Since the lower layers are subjected to
more compaction, their initial density should be lower than the layers above. The
degree of under-compaction of each layer is varied linearly from the bottom to the
top. For specimens tested, a degree of under-compaction of 5% was somewhat
arbitrarily chosen because of lack of data. During a previous study (Lekarp and
Isacsson, 2001), the actual density distribution was measured after compaction. No
380 Road Materials and Pavement Design. Volume 7 – No. 3/2006

indication of systematically varying densities was observed. The exception might be


the lowest layer, which shows the lowest degree of compaction. It was also the layer
subjected to least compaction according to the Ladd procedure.
Compaction was performed with a vibrating Kango hammer with a static weight
of approximately 50 kg and a vibrating frequency of 23 Hz. The foot diameter was
175 mm, i.e. smaller than the sample diameter, which allows for a shearing action.
Initial compaction was done according to the pattern shown in Figure 4. After initial
compaction, the vibrating hammer was moved in a circular manner with some
overlap and was continued until the desired height was reached. If the layer height
was not reached within one minute, the compaction was discontinued. Each layer
surface was scarified before the next layer was added to increase particle interlock
Downloaded by [Simon Fraser University] at 14:49 13 November 2014

and bulk continuity, thus decreasing compaction induced anisotropy.

3 5 4

2
Figure 4. Initial compaction pattern

The surface roughness or texture of the top layer was smoothed by using
intermediate size particles, 1-20 mm, of the same material. This was done to create a
smooth level surface on which the top plate was placed. A smooth level reduces the
risk of uneven stress distribution. Intermediate size particles were used to avoid
alteration of water retention and permeability characteristic of the sample. A low
retention capability and high permeability for gas and water were considered
essential to reduce the influence of this measure. On layers 2, 5 and 8 transducers
was placed on the scarified surface (cf. Figure 1). During compaction and repeated
load conditioning, the tensiometer porous cup was replaced by a metal dummy.
Insertion of the porous cup was done after the conditioning. This measure was found
necessary, because on a pre-test specimen it proved that the large initial aggregate
displacements during the first load cycles broke the fragile ceramic cup. Anchoring
arrangement for fitting of the three axial transducers was also embedded during
compaction. After compaction the sample was demolded, a second membrane was
Influence of Water on Resilient Properties 381

fitted and all transducers connected. Transducer feedthroughs and membrane-plate


interfaces were finally sealed with silicon.
Initial properties for the tested specimens are summarized in Table 3. The target
initial dry density level was chosen to be 95% of maximum dry density given in
Table 2. In some cases, this target had to be abandoned during compaction because
of the obvious difficulties reaching or maintaining the desired density levels. The
initial water contents were chosen so that the matric suction would be in the range
between 15-20 kPa, estimated from previously measured soil water characteristic
curves relating matric suction to volumetric water content.
Downloaded by [Simon Fraser University] at 14:49 13 November 2014

Table 3. Initial properties of the triaxial samples

Water content Dry density Density ratio


Grading
[% by weight] [kg/m³] (actual/max. dens.) [%]

0.8 0.5 1923 103

0.5 1.0 2112 99

0.4 1.5 2154 95

0.3 2.0 2134 91

Even though the density ratio indicates differences in compaction, it is believed


that they depend to a large extent on inaccuracies and difficulties in the method to
determine maximum density caused by the coarse nature of the aggregate. The
overall level of compaction is believed to be high.
The basic scheme was to test the specimens at an initially low water content,
after which water was added stepwise until fully soaked. For each step during
wetting, estimated amounts of water was added through the top plate and allowed to
percolate the entire sample. The stage before soaking corresponds to maximum
water retention capability on wetting, i.e. the point where the specimen cannot retain
more water without draining. Soaking, or saturation, was achieved by adding water
through the bottom plate until water flowed through the top plate valve. Since no
particular measures to fully saturate the sample could be undertaken, it is not
believed that full saturation was reached and, therefore, this state is termed soaked.
After testing at maximum water content the specimens were allowed to drain freely,
after which they were tested again. This stage represents water retentive capacity
when drying, and the water content is usually higher than for the corresponding
wetting sequence, due to the hysteresis of the soil water characteristics curve. In
Table 4, the different stages for each sample are summarized. For each stage, water
content and degree of saturation are given as specimen average values, and
382 Road Materials and Pavement Design. Volume 7 – No. 3/2006

measured matric suction represents value at midheight. Matric suction is the


difference between pore air pressure (ua) and pressure in the water phase (uw).

Table 4. Properties at different stages of testing for the triaxial specimens


designated by grading coefficient

Gravimetric Volumetric Matric


Degree of
Water state water water suction
Grading saturation, S
coding content, w content, θv (ua-uw)
[m³/m³]
[% weight] [% vol.] [kPa]
Downloaded by [Simon Fraser University] at 14:49 13 November 2014

0.5 (Initial state) 0.5 1.0 0.04 15


Retention limit 1.3 2.5 0.09 4
0.8
Soaked* 14.1 27.1 1.00 (1.00) -2
Drained* 2.3 4.5 0.17 5
1 (Initial state) 1.0 2.1 0.11 16
Retention limit 2.0 4.2 0.21 4
0.5
Soaked* 9.9 0.21 >1.00 (1.06) -3
Drained* 3.8 8.0 0.40 4
1.5 (Initial state) 1.5 3.2 0.18 13
2.5 2.5 5.4 0.30 6
0.4 Retention limit 4.3 9.4 0.52 4
Soaked* 8,8 19,0 >1.00 (1.05) -2
Drained* 5.1 11.0 0.61 5
2 (Initial state) 2.0 4.3 0.23 20
3 3.0 6.4 0.35 11
0.3 Retention limit 5.2 11.1 0.60 4
Soaked* 8.8 18.8 >1.00 (1.02) -4
Drained* 7.5 16.0 0.87 7
* Predicted values, except matric suction which are measured values ( )Values within
parenthesis are nominal predictions.

It should be noted that the negative suctions indicated at soaked conditions, are
actually pore pressures generated by the hydraulic head, i.e. the pressures are above
atmospheric. Given the height of the sample, it is fair to assume gradients in terms
of water content and matric suction. For the soaked and drained states, the water
content and saturation value are predicted from volumetric water content
measurements by TDR. There is a fair amount of uncertainty in the prediction of
water content why care must be observed when analyzing those measurements. It
Influence of Water on Resilient Properties 383

was considered that nominal estimations of water content from measuring actual
volume of water flowing in and out of the sample was even more uncertain. Because
of the height of the specimen, a considerable hydraulic pore pressure is exerted to
the membrane at soaked conditions. Even though the soaking process was performed
under elevated confining pressures, it seems as there is a ballooning effect on the
membrane, which cannot easily be accounted for.
Nominal water contents at retention limit and drained states were compared to
numerical predictions using the soil water characteristics curve. All predictions were
within less than 1 unit-% volumetric water content (standard error 0.3%-by weight)
compared with measured except for grading 0.4 at retention limit, where the
predicted value was exceeded with approximately 2%-by volume. The drained state
Downloaded by [Simon Fraser University] at 14:49 13 November 2014

was, in contrast, predicted more accurately. The higher water content after soaking
and draining the specimens, compared to the retention limit reached when increasing
water content, is caused by the hysteresis of the soil water characteristic curve.

3.2.3. Triaxial test procedure


In the cylindrically confined triaxial test, where only principal stresses can be
applied and two principal stresses are equal, usually σ2 and σ3, it is common to use
the following stress invariants:

σ 1 + 2σ 3
p= [3]
3

q = σ1 −σ 3 [4]

θ = σ 1 + 2σ 3 [5]

where p is the mean normal stress, q is called deviatoric stress and θ is total stress.
The mean normal stress, p, is equal to the mean stress in the volumetric and
deviatoric stress tensors, total stress, θ, is the first invariant, I1, of the principal stress
tensor and q is a function of the second invariant of the deviatoric tensor and the
octahedral shear stress. In the following, the material is treated as a continuum, and
stresses and strains are defined as so called engineering stresses and strains, i.e. the
force and length change are related to initial unstressed area and length, respectively.
Provisionary European standard prEN 13286-7 (CEN, 2000) formed the basis for
the testing schedule. Resilient behavior of granular materials is generally determined
during repeated load triaxial tests after steady-state, in terms of resilient response, is
reached. Before measuring the resilient properties, a conditioning is therefore
required to reach a practically resilient response with negligible further permanent
accumulation. The conditioning comprised 20 000 haversine load repetitions at a
384 Road Materials and Pavement Design. Volume 7 – No. 3/2006

frequency of 1 Hz. Confining pressure was held constant at 40 kPa, and deviator
stress cycled between 0 and 280 kPa.
Details of the resilient test schedule used are shown in Figure 5 as total stresses.
In addition to the stress paths outlined in prEN 13286-7 for constant confining
pressure test (10, 20, 30, 40 kPa, respectively), two further confining pressures, 70
and 100 kPa, were added. At soaked conditions, the triaxial test mode was
consolidated and undrained. Excessive pore pressure induced by increased confining
stress was allowed to dissipate, after which the drainage valve was closed during
repeated deviatoric loading. In practice, there is only a very small difference
between undrained and drained conditions. Measured pore pressures are of the same
magnitude; generally no difference could be discerned. This observation is probably
Downloaded by [Simon Fraser University] at 14:49 13 November 2014

influenced by several phenomena, e.g. the relative length and cross-sectional area of
tubing used compared to sample size, dilative behavior of the sample, low induced
pore pressures and the comparably short duration of the load (1s). The similarity
between drained and undrained tests is also true for the unsaturated states where
minute pore air pressures are induced by cyclic loading. The magnitude of these
pressures was independent of testing mode, drained or undrained.

300
40 kPa

250
Deviator stress, q [kPa]

30 kPa 70 kPa 100 kPa


200

150 20 kPa

100
10 kPa

50

0
0 50 100 150 200
Mean normal stress, p [kPa]

Figure 5. Stress conditions used in the constant confining pressure tests (total
stress). Also indicated are the actual confining pressures used. At soaked conditions
the tests were consolidated undrained

Each stress path was repeated for 50 cycles as a sinusoidally (haversine)


oscillating axial load at a frequency of 1 Hz. For some samples at higher water
Influence of Water on Resilient Properties 385

contents, some of the harshest most severe stress conditions were omitted to avoid
failure or excess accumulation of permanent deformation.
The stress values given are nominal values, which in practice rarely could be
strictly fulfilled. Initial stresses caused by sample weight, dead weights of loading
equipment, hydrostatic pressure by confining liquid and contact loads are present to
some extent, depending on the test setup used. In these test, air was used as
confining fluid, causing no significant contribution to confining pressure (σ3). Dead
weight of loading equipment (i.e. top plate and loading piston) and the applied
contact force cause an initial deviator stress (q) of approximately 2 kPa.
Two elastic parameters are calculated from the measured behavior, resilient
Downloaded by [Simon Fraser University] at 14:49 13 November 2014

modulus, Mr, and Poisson ratio, ν, defined by:

σ1 −σ 3 q
Mr = r
= r [6]
ε1 ε1
and

r
ε3
ν =− r
[7]
ε1
where ε3r and ε1r are the resilient radial and axial strains, respectively. Resilient
modulus and Poisson ratios are determined as secant values in terms of cycled
deviator stress. Even though these parameters can be questioned from a theoretical
point of view, they will be used in the analysis due to their frequent use and the fact
that they are easily interpreted. It is noted, that at this stress levels granular materials
are commonly inelastic. In addition, in an elastic framework Poisson ratio can not
exceed 0.5.
The complete set of data was also fitted to equations describing the nonlinear
dependence of resilient modulus to mean normal stress p. Probably the most well-
known and commonly used model is generally called the k-θ model. It assumes that
resilient modulus can be related to the sum of principal stresses θ according to:

M r = k1θ k 2 [8]

where k1 and k2 are regression parameters. The origin of the model dates back to at
least 1960s. In Brown and Pell (1967), the in-situ modulus was plotted versus the
first stress invariant (I1 or θ) in a log-log diagram. Seed et al. (1967) suggested a
power law relationship relating resilient modulus to sum of principal stresses.
Reference to preceding work by Biarez in 1962 was given by e.g. Hicks (1970).
Hicks and Monismith (1971) concluded that Equation [8] was verified in most of
their test results. Since, it has gained widespread use.
386 Road Materials and Pavement Design. Volume 7 – No. 3/2006

May and Witczak (1981) indicated, based on extensive analysis of field data, that
the resilient modulus is not solely dependent on sum of principal stresses but also on
shear strain level. For low levels of shear strain, the modulus is larger compared to
at higher levels of strain. They used a fitting parameter to normalize this behavior.
As a consequence, May and Witczak stated that using the sum of principal stresses
as the only stress variable might be justifiable at low strains. For higher shear strain
levels an adjustment is necessary. This shear-weakening effect was later recognized
by Uzan (1985), who suggested the following relationship:

M r = k1θ k 2 σ dk 4 [9]
Downloaded by [Simon Fraser University] at 14:49 13 November 2014

where k1, k2 and k4 are regression parameters and σd is the deviatoric stress invariant
q. Uzan also suggested to use octahedral shear stress when major, minor and
intermediate principal stresses are not equal, instead of deviatoric stress, which
applies for triaxial testing. A similar approach have also been suggested by Lade and
Nelson (1987), using the first invariant of the stress tensor and the second invariant
of the deviatoric stress tensor.

4. Results

In this section, the stresses used are total stresses, i.e. only externally applied
loads are taken into account. Internally acting stresses as matric suction in the
unsaturated states and pore pressure (hydrostatic) for the soaked/saturated state are
disregarded in this chapter but shortly discussed in Chapter 5. Furthermore, for
soaked/saturated conditions only very small load induced pore pressures were
observed, commonly below 1 kPa with a maximum measured value of 2.5 kPa for
all samples and stress paths used. As this value is far below the hydraulic gradient of
10 kPa, induced pore pressures were ignored in analysis. Assuming an effect of
internal stresses, they will cause a shift in mean normal stress.
In Figure 6, resilient modulus, as a function of mean normal stress (p), and
Poisson ratio, as a function of the ratio deviator stress to confining stress, at initial
states (i.e. the driest state with a matric suction of between 13-20 kPa) of the
different gradings are shown. In this and following figures, fitted lines are included
for ease of viewing: resilient modulus to a power law equation and Poisson ratio to a
second degree polynomial. As recognized by e.g. May and Witczak (1981) and
Uzan (1985), resilient modulus is not solely dependent on the mean normal stress (or
sum of principal stresses) but also on level of shear straining.
Influence of Water on Resilient Properties 387

700

0 .8
R esilien t M o d u lu s, M r [M P a ]

600
0 .4
0 .3
500 0 .5

400

300

200 n= 0 .8
Downloaded by [Simon Fraser University] at 14:49 13 November 2014

n= 0 .5
100 n= 0 .4
n= 0 .3
0
0 50 100 150 200
M e a n n o r m a l s tr e s s , p [k P a ]

0 .6

0 .4
0 .5 0 .3

0 .5
P o iss o n ra tio , ν

0 .4
0 .8
0 .3

0 .2
n= 0 .8
n= 0 .5
0 .1
n= 0 .4
n= 0 .3
0
0 2 4 6 8
S tre s s ra tio q / σ 3

Figure 6. Resilient modulus and Poisson ratio at initial state, (matric suction 13-20
kPa)
388 Road Materials and Pavement Design. Volume 7 – No. 3/2006

Unfortunately, when presenting resilient modulus as a function of mean normal


stress, this dependency is seen as increased scatter; the shear dependency is
projected on the mean normal stress axes as apparent scatter. This scatter is more
obvious at elevated water contents. A similar effect is also seen for Poisson ratio as a
function of the ratio deviator stress to confining pressure (q/σ3).
At initial state, all gradings are fairly equal in terms of resilient modulus but
diverge for Poisson ratio. The general tendency is that a higher grading parameter
yields lower Poisson ratio. The overall resilient response of gradings 0.3 and 0.4 is
similar both in terms of resilient modulus and Poisson ratio.
In Figure 7, the resilient response at retention limit during wetting is shown for
Downloaded by [Simon Fraser University] at 14:49 13 November 2014

the different gradings.


Comparing Figures 6 and 7, it is obvious that all gradings have lost some
resilience at retention limit but for gradings 0.8 and 0.5 the loss is smaller in
magnitude than for gradings 0.4 and 0.3. The finer the grading (lower grading
coefficient), the larger the reduction of resilient modulus seems to be. A similar
pattern can be seen for Poisson ratio, where the finer gradings were mostly affected
while gradings 0.8 and 0.5 remain essentially unaffected. Poisson ratio in excess of
0.5, as measured for gradings 0.3 and 0.4 at high shear stress ratios, is usually
referred to as dilation and is an indicative of volume increase when stressed in
compression. This behavior was termed dilatancy by Reynolds (1885) and is typical
of particulate materials under certain stress conditions and density states.
Figure 8 shows the influence on resilient modulus and Poisson ratio when water
content is increased to almost full saturation. It is probable that a complete saturation
was not achieved.
While grading 0.8 remains almost unaffected by increasing water content to
almost full saturation, all other gradings suffer a substantial loss of resilient
modulus. The largest decrease in modulus is seen for gradings 0.3 and 0.4 at
elevated stress levels (cf. Figures 6-8). It can also be seen that the influence of stress
level on resilient modulus is considerably smaller, compared to initial state modulus,
for all gradings except 0.8. This is most pronounced for gradings 0.3 and 0.4 which
again behave fairly equal. Poisson ratio is also influenced to a high degree by
saturation. All specimens showed an increase, but for grading 0.8 it can barely be
distinguished. All the other gradings experienced ratios clearly higher than 0.5,
gradings 0.4 and particularly 0.3 showing the highest. Comparing Figures 6-8 it is
evident that dilatancy is increased with increased water content except for grading
0.8 which throughout different water contents remains comparatively unaffected.
Influence of Water on Resilient Properties 389

700
M r [M P a]

600

500 0 .8

400
R esilient M od ulu s,

0 .5
300

0 .4
200 n= 0 .8
Downloaded by [Simon Fraser University] at 14:49 13 November 2014

0 .3
n= 0 .5
100 n= 0 .4
n= 0 .3
0
0 50 100 150 200
M e a n n o r m a l s tr e s s , p [k P a ]

700
M r [M P a]

600
0 .8
500
0 .5

400 0 .4
R esilient M od ulu s,

0 .3
300

200 n= 0 .8
n= 0 .5
100 n= 0 .4
n= 0 .3
0
0 50 100 150 200
M e a n n o r m a l s tr e s s , p [k P a ]

Figure 7. Resilient modulus and Poisson ratio at retention limit (wetting)


390 Road Materials and Pavement Design. Volume 7 – No. 3/2006

0 .6 0 .3

0 .4
0 .5
P o is s o n ra tio , ν

0 .4 0 .5

0 .3 0 .8

0 .2
Downloaded by [Simon Fraser University] at 14:49 13 November 2014

n= 0 .8
n= 0 .5
0 .1
n= 0 .4
n= 0 .3
0
0 2 4 6 8
S tre s s ra tio q / σ 3

1 .2
n= 0 .8
0 .3
n= 0 .5
1
n= 0 .4 0 .4
n= 0 .3
P oisson ratio , ν

0 .8
0 .5

0 .6

0 .4 0 .8

0 .2

0
0 2 4 6 8
S tre ss ratio q / σ 3

Figure 8. Resilient modulus and Poisson ratio for soaked/saturated samples

To emphasize how resilient behavior is affected by varying water contents and


matric suction, results from tests using a confining pressure of 40 kPa are shown for
Influence of Water on Resilient Properties 391

each grading in Figures 9-12. Corresponding water contents, degrees of saturation


and matric suctions are found in Table 4.

500
w = 0 .5 σ 3 = 40 kP a
450
[M P a ]

R e te ntio n lim it
400 S oaked
350 D ra ine d
r
R e s ilie n t M o d u lu s , M

300
250
Downloaded by [Simon Fraser University] at 14:49 13 November 2014

200
150
100
50
0
0 50 100 150 200
M e a n n o r m a l s tre s s , p [k P a ]

1
w = 0 .5 σ 3 = 40 kP a
0 .9
R e te ntio n lim it
0 .8 S oaked
D ra ine d
P o is s o n ra tio , ν

0 .7
0 .6
0 .5
0 .4
0 .3
0 .2
0 .1
0
0 2 4 6 8
S tr e s s ra tio q /σ3

Figure 9. Resilient modulus and Poisson ratio, grading 0.8

For grading 0.8, a continuous loss of resilient modulus is observed when water
content is increased. The average loss from dry conditions to soaked is
approximately 10-15% with retention limit state in between. After draining the
initial response is fully recovered. Poisson ratio is only influenced to a very small
392 Road Materials and Pavement Design. Volume 7 – No. 3/2006

degree. It cannot be distinguished whether the decreased resilient modulus is an


effect of reduced matric suction or a change in frictional properties of load carrying
contact points. From the soil water characteristics curve, it seems that matric suction
is mainly caused by surface adhesion forces rather than capillary air-water
interfaces, thus in effect not contributing to internal stresses.

500
w=1 σ 3 = 40 kP a
450
[M P a ]

R e te ntio n lim it
400 S oaked
350 D ra ine d
r
Downloaded by [Simon Fraser University] at 14:49 13 November 2014

R e s ilie n t M o d u lu s , M

300
250
200
150
100
50
0
0 50 100 150 200
M e a n n o rm a l s tr e s s , p [k P a ]

1
w=1 σ 3 = 40 kP a
0 .9
R e te ntio n lim it
0 .8 S oaked
D ra ine d
P o is s o n ra tio , ν

0 .7
0 .6
0 .5
0 .4
0 .3
0 .2
0 .1
0
0 2 4 6 8
S tre s s r a tio q /σ3

Figure 10. Resilient modulus and Poisson ratio, grading 0.5


Influence of Water on Resilient Properties 393

Grading 0.5 shows a clear effect at soaked conditions. The reduction is between
approximately 20-30% depending on mean stress. There seems to be no effect of the
decreased matric suction between 1% gravimetric water content and retention limit
(2% by weight) neither for resilient modulus nor Poisson ratio. The average matric
suction is reduced by 12 kPa (cf. Table 4). After draining, the sample stiffness is not
fully recovered. It can be seen that the water content after draining is slightly
elevated, indicating a hysteretic behavior of the soil water characteristic curve (cf.
Table 4). To further decrease the water content a means to subject the sample to
elevated matric suctions e.g. a semipermeable filter would be required. This could
not be accomplished with the triaxial cell used.
Downloaded by [Simon Fraser University] at 14:49 13 November 2014

500
w = 1 .5 σ 3 = 40 kP a
450
[M P a ]

w = 2 .5
400 R e te ntio n lim it
350 S oaked
r
R e s ilie n t M o d u lu s , M

D ra ine d
300
250
200
150
100
50
0
0 50 100 150 200
M e a n n o r m a l s tr e s s , p [k P a ]

1
w = 1 .5 σ 3 = 40 kP a
0 .9 w = 2 .5
0 .8 R e te ntio n lim it
S oaked
P o is s o n ra tio , ν

0 .7
D ra ine d
0 .6
0 .5
0 .4
0 .3
0 .2
0 .1
0
0 2 4 6 8
S tr e s s r a tio q /σ3

Figure 11. Resilient modulus and Poisson ratio, grading 0.4


394 Road Materials and Pavement Design. Volume 7 – No. 3/2006

500
w=2 σ 3 = 40 kP a
450
[M P a ]

w=3
400 R e te ntio n lim it
350 S oaked
r
R e s ilie n t M o d u lu s , M

D ra ine d
300
250
200
Downloaded by [Simon Fraser University] at 14:49 13 November 2014

150
100
50
0
0 50 100 150 200
M e a n n o r m a l s tr e s s , p [k P a ]

1
w=2 σ 3 = 40 kP a
0 .9
w=3
0 .8 R e te ntio n lim it
S oaked
P o is s o n ra tio , ν

0 .7
D ra ine d
0 .6
0 .5
0 .4
0 .3
0 .2
0 .1
0
0 2 4 6 8
S tr e s s r a tio q / σ3

Figure 12. Resilient modulus and Poisson ratio, grading 0.3

Grading 0.4 displays a similar but more pronounced pattern as grading 0.5, i.e.
weakening with increased water content. Comparing initial state modulus to
saturated conditions the average reduction is roughly 40% Both in terms of resilient
Influence of Water on Resilient Properties 395

modulus and Poisson ratio, the response seems unaffected by increasing water from
initial state gravimetric water content 1.5% to 2.5% even though the degree of
saturation is increased and matric suction is decreased (cf. Table 4). A clear effect
can be seen by further increasing water content to the wetting retention limit;
resilient modulus decreases and Poisson ratio increases. In this case, the initial
unsaturated properties are only partially regained after draining. A clear effect,
manifested as a decrease in resilient modulus and increase in Poisson ratio, remains.
It can be noted that the average water content is higher than at the wetting retention
limit, but the matric suction is equal.
Finally, grading 0.3 behaves similar to 0.4. It remains essentially unchanged by
the first water content increase (2 to 3%-by weight), followed by additional
Downloaded by [Simon Fraser University] at 14:49 13 November 2014

reduction, when matric suction is further reduced to the wetting retention limit
(w = 5.2%). At soaked conditions, the reduction of modulus is large and Poisson
ratio is notably increased. Hysteretic behavior in terms of water content is also
obvious after draining. Also, a considerable reduction in modulus and increase in
Poisson ratio remains compared to the retention limit state during wetting.
For a summarized comparison of tested gradings, the normalized resilient
modulus at mean normal stress (p) of 100 kPa and Poisson ratio at stress ratio q/σ3 =
4 was calculated and shown in Figure 13 as a function of degree of saturation. The
normalized resilient modulus is calculated as the ratio of current state modulus to
initial state modulus. Poisson ratios shown are actual values. Since the resilient
modulus at initial state was fairly equal for all gradings (cf. Figure 6), the
normalized values are mutually comparable. For the wetting test sequence,
measurements are connected with lines while the drained result is shown with a
similar marker in inverted colors (i.e. black is replaced with white and vice versa).
A notable result is the similarity between gradings 0.3 and 0.4 in terms of
resilient modulus. Although tested at different water contents, modulus as a function
of degree of saturation is similar. This similarity is also true for the drained states
which also falls close to the same line. For Poisson ratio, the similarity between
gradings 0.3 and 0.4 is less compelling at water contents closer to saturation. At
soaked/saturated conditions grading 0.3 is far more dilative, a difference that
remains for the drained state. For the other gradings, 0.8 is the least susceptible to
increased water contents both in terms of resilient modulus and Poisson ratio while
grading 0.5 falls somewhere between these extremes, grading 0.8 on one side and
0.3 and 0.4 on the other. Grading 0.5 shows a reduction in resilient modulus and
increased dilatancy for increased degrees of saturation.
The complete set of data was also fitted to equations describing the nonlinear
dependence of resilient modulus to mean normal stress p and deviator stress q (cf.
Section 3.3.3). In Table 5, fitted parameters for Equations [8] and [9], with sum of
stresses (θ) substituted by mean normal stress (p), are given. Units in the regression
were mean normal stress in kPa and modulus in MPa.
396 Road Materials and Pavement Design. Volume 7 – No. 3/2006

Strong agreement between k1 and k2 obtained using the k-θ and Uzan models,
respectively, is demonstrated. They are approximately of the same magnitude and
show the same pattern for different water levels. Overall, there exists a strong
correlation between k1 and k2 within each model; k2 increases as k1 decreases. This
pattern is shown for all the gradings investigated. Rada and Witczak (1981) used k2
as indicator of nonlinearity i.e. a larger k2 would imply greater degree of
nonlinearity.

1
Downloaded by [Simon Fraser University] at 14:49 13 November 2014

[P a /P a ]

0 .8

0 .6
r
N o rm a liz e d M

0 .4 W e ttin g D ra in e d sta te
n = 0 .8
0 .2 n = 0 .5
n = 0 .4
n = 0 .3 p = 100 kP a
0
0 0 .2 0 .4 0 .6 0 .8 1
S a tu r a tio n , S [m ³/m ³]

1 W e ttin g D ra in e d sta te
0 .9 n = 0 .8
0 .8 n = 0 .5
0 .7 n = 0 .4
P o is s o n ra tio ν

n = 0 .3
0 .6
0 .5
0 .4
0 .3
0 .2
0 .1
q /σ 3 = 4
0
0 0 .2 0 .4 0 .6 0 .8 1
S a tu ra tio n , S [m ³/m ³]

Figure 13. Normalized resilient modulus at p = 100 kPa and Poisson ratio at q/σ3 =
4 as a function of degree of saturation. Wetting test schedule is interconnected with
lines and drained state shown in similar markers with inverted colors
Influence of Water on Resilient Properties 397

Table 5. Model parameters for k-θ and Uzan models for resilient modulus (MPa) as
a function of mean normal stress p (kPa) and deviator stress q (kPa)

k-θ Uzan
Grading Water state
k1 [MPa] k2 k1 [MPa] K2 k4
0.5 10.2 0.79 13.3 1.00 -0.25
Ret. limit 17.7 0.65 19.2 0.84 -0.19
0.8
Soaked 16.7 0.65 20.1 0.87 -0.23
Downloaded by [Simon Fraser University] at 14:49 13 November 2014

Drained 16.6 0.69 19.9 0.85 -0.18


1 13.2 0.71 13.6 0.86 -0.14
Ret. limit 16.2 0.66 18.3 0.80 -0.15
0.5
Soaked 26.6 0.50 30.4 0.61 -0.14
Drained 27.2 0.53 29.9 0.71 -0.18
1.5 13.0 0.73 13.0 0.82 -0.09
2.5 16.9 0.67 17.1 0.69 -0.02
0.4 Ret. limit 31.5 0.49 32.3 0.44 0.04
Soaked 50.8 0.28 49.4 0.27 0.01
Drained 27.2 0.49 27.8 0.61 -0.11
2 12.8 0.72 11.3 0.85 -0.09
3 14.3 0.69 13.5 0.75 -0.04
0.3 Ret. limit 35.5 0.44 32.3 0.29 0.15
Soaked 53.8 0.24 60.6 0.40 -0.17
Drained 33.2 0.41 35.6 0.53 -0.13

It is obvious from Table 5 and Figures 9-12 that increasing water content reduces
stress dependency. For grading 0.3 and 0.4 at soaked conditions the results approach
a straight line indicating linear behavior i.e. stress level independence. The Uzan
parameter, k4, shows no correlation to either k1 or k2 and is usually negative
indicating, shear-weakening. In a few cases k4 is positive i.e. higher resilient
modulus at large strains. However, in, these cases the parameter is close to zero and
its effect on predicted modulus very small. The material response is basically
following the more simple single power law relationship given by the k-θ model.
398 Road Materials and Pavement Design. Volume 7 – No. 3/2006

5. Discussion and conclusions

The general question whether granular base and subbase layers reach saturation
or not during field conditions, is not treated in this study but obviously important.
Design procedures usually assume a weakened response during spring and autumn,
indicating, from experience, a large influence of water. Given the possible ways of
water enrichment in upper pavement layers, e.g. rising ground water table,
infiltration from surface and surrounding areas, accumulation caused by freezing
and thawing etc, it seems possible that granular base and subbase layers, at least at
certain periods, reach high degrees of saturation, thus reducing bearing capacity. In
this work, it has been shown that that some gradings experience a substantial loss of
Downloaded by [Simon Fraser University] at 14:49 13 November 2014

resilient capacity at elevated water contents. If saturated, an essential characteristic


of the materials is the response to draining, a topic not treated. The fairly low matric
suction measured seems to have only a minor influence on resilient properties but
could be of importance in draining saturated pavement structures.
An important observation is, it appears that resilient modulus of granular
frictional materials is not a simple function of stresses, applied and internal (matric
suction and pore air pressures). Based on the results presented in this paper it does
not seem possible to form a continuous curve in terms of resilient modulus for a
single grading for all tested states of water contents, expressed in terms of a single
state variable. In the case of saturated soils, the effective stress, introduced by
Terzaghi (1924), have proven to be a powerful state variable:

σ ′ = σ − uw [10]

where σ’ is effective normal stress, σ is total normal stress and uw is pore water
pressure. The original notation used by Terzaghi has been altered in Equation [10] to
agree with notation common in geomechanics. Experimental evidence have since
convincingly shown that the mechanical behavior, in the saturated state, can be
related to effective stress changes. A simple and straightforward way to comprise
unsaturated states is simply to extend the effective stress principle assuming it to be
valid provided negative pore pressure (matric suction) is used. Generally, a factor to
adjust for the unsaturated water stress levels is introduced. Some examples can be
given. Croney et al. (1958) suggested:

σ ′ = σ − β ′uw [11]

where β’ is an empirical measure of “the number of bonds of water under


tension”. A well-known relationship, which also introduces the air-phase, was given
by Bishop (1959):

σ ′ = (σ − u a ) − χ (u w − u a ) [12]
Influence of Water on Resilient Properties 399

where ua is pore-air pressure and χ is a parameter related to the degree of saturation


of the soil. Sometimes, χ is simply replaced by degree of saturation, S. At saturated
conditions (χ = 1), Equation [12] simply reduces to the Terzaghi effective stress
equation (Equation [10]). This approach might be an oversimplification for several
reasons. The most serious, from a theoretical point of view, is maybe that it
incorporates inherently empirical parameters in the state variables i.e. the variables
will depend on the material tested. The internally acting stress might also need to be
separated in to air and water pressure respectively. Accounting for matric suction or
internal capillary tension for these tests means adjusting the mean normal stress at
the stress levels tested. In the previous figures presented the total mean stress is
used. Assuming an internal tension caused by capillary forces (matric suction) this
Downloaded by [Simon Fraser University] at 14:49 13 November 2014

would shift the resilient modulus curve to the right, i.e. towards higher effective
mean normal stresses while at saturated conditions, the effective mean normal stress
would be lower following the Terzaghi effective stress principle. This extended
effective stress principle is falsified already by the observation that, for gradings 0.3
an 0.4, the increase in water content from the initial dry state to the next level to be
tested, caused a decrease in mean matric suction of 9 and 7 kPa, respectively. No
effect of this change in mean normal stress could be discerned from the measured
resilient response, while the smaller decrease in matric suction caused by the
increase of water content to maximum retentive capability (7 and 2 kPa,
respectively) caused a significant weaker and more dilative response. Compensating
for degree of saturation, as suggested by Equations [11] and [12] means that the
differences in effective mean normal stress become substantially smaller with
decreased normalizing power, i.e. the differences become so small that the
horizontal shift is hardly noticeable. Considerably higher matric suction reductions
are needed to explain the loss of resilience. In conclusion, there seems to be further
mechanisms involved. An obvious factor, in this connection, is altered frictional
properties of contact points. This phenomenon has previously been discussed by e.g.
Hveem and Carmany (1948) and Thom and Brown (1987). Another approach might
be to separately consider the effect of the externally applied stress tensor and the
matric suction. Generally, in this discussion it is probably important to remember
that compared to externally applied stresses the measured levels of internal stresses
are low. Also, given the height of the samples, the gradient is of a magnitude
comparable to measured matric suctions. It is also to be noted that the tests in this
study were conducted using constant confining pressures and modulus defined as
ratio of deviator stress (unaffected by change in mean normal stress) to axial
deformation. Under tests were more principal stresses are cycled, the results might
be different.
The magnitude of measured induced pore pressures (air or water), or rather lack
of elevated pressures, was rather surprising. Few reports on generated pore pressures
during resilient testing have been found. Elevated pore pressures are commonly
presented in studies of liquefaction in e.g. earth quake engineering but also in
highway engineering, when a large amount of traffic is simulated. These tests are
usually performed at higher stress levels than resilient testing and for a large number
400 Road Materials and Pavement Design. Volume 7 – No. 3/2006

of repeated load cycles, during which pore pressures are induced and accumulated.
Granular unbound materials used in pavements are not, in general, prone to liquefy.
Hicks and Monismith (1971) reported magnitudes of induced pore pressures of 5-
10% of applied deviator stresses, while Van Toan (1975) noted no pore pressures
generated by deviatoric stresses. The low levels of generated pore pressures might
be explained by insufficient saturation and dilatancy. Since no major efforts to reach
complete saturation could be performed, it is highly possible that pore air is still
present. Even small amounts of entrapped air will swamp any generated pressures at
the comparably small strain levels encountered. Furthermore, the dilative behavior
observed will counteract any generation of pore pressures because of the volume
increase. This effect is also discussed by Bonaquist and Witczak (1996) who also
Downloaded by [Simon Fraser University] at 14:49 13 November 2014

estimated that saturation levels must be greater than 98 percent to change the
effective stress in any significant way. There are also practical uncertainties that
have to be contemplated. Pore air pressures might be hard to measure at elevated
saturation levels because of a discontinuous air-phase. Air bubbles are dispersed in
the water and pressures might be very local. Even though, air pressures were
measured at two points, which showed equal levels of generated pressures, some
caution is advised, whether these recorded pressures are representative or not.
Transient changes in the matric suction caused by change in geometry of the
capillarity’s when the sample is loaded might occur, which are not possible to
measure with tensiometric technique. However, there are previous reports indicating
that this might be unlikely. Several authors have concluded that dry density has a
minor effect on matric suction properties (Box and Taylor, 1962; Olson and
Langfelder, 1965; Krahn and Fredlund, 1972; Rasiah and Aylmore, 1998). This
phenomenon may in part be explained by that densification mainly affects large
pores while the originally small pores do not change their size significantly. This
also implies that samples are fairly insensitive to different confining pressures used
during triaxial tests in terms of matric suction. Although, based on the results of
some of the investigations mentioned it is probable that the conclusion, regarding
insensitivity of confining pressures, is less solid in the low suction range. The acting
capillarity’s might be of a size that to some extent is affected by densification but no
indication of this could be discerned during testing. The matric suctions measured
remained unchanged during testing at varying confining pressures.
Based on the results in this study the following conclusions can be drawn:
– Increased water content causes reduction in resilient modulus and increase in
Poisson ratio. This behavior is more pronounced as the grading parameter is
decreased. The large increase in Poisson ratio is to some extent in contrast to
previous reports.
– In terms of resilient response, a difference between wetting and drying can be
seen for the finer gradings. This is probably caused by the hysteretic appearance of
the soil water characteristics curve.
– For simple modeling of nonlinear behavior, both the k-θ and Uzan models give
a fair estimate of resilient modulus. At high levels of shear stress, the Uzan model
Influence of Water on Resilient Properties 401

shows closer agreement with measurements. The assumption of a constant Poisson


ratio, assumed by both models is clearly invalid. Poisson ratio is strongly stress
dependent.
– Measuring volumetric water contents with Time Domain Reflectometry for
large maximum size granular materials was successful, at least to monitor relative
changes of even fairly small increments. The TDR probes can sustain heavy
compaction during burial without loosing accuracy and can be reused a large
number of times.

Acknowledgements
Financial support was given by:
Downloaded by [Simon Fraser University] at 14:49 13 November 2014

– The Swedish National Road Administration (Vägverket);


– Development Fund of the Swedish Construction Industry (SBUF);
– Swedish Aggregates Producers Association (SBMI).
The NCC quarry in Skärlunda, Norrköping provided the aggregates used during
testing. Appreciation is also due to Dr. Karl-Johan Loorents at the Swedish National
Road and Transport Research Institute (VTI) for mineralogical characterization of
the aggregate used.

6. Bibliography

Ali H.A., Lopez A., “Statistical analysis of temperature and moisture effects on pavement
structural properties based on seasonal monitoring data”, Transportation Research
Record, No. 1540, 1996, p. 48-55.
Andreasen A.H.M., Andersen J., “Om relationen mellem kornsammansætning og
mellemrum i produkter af løse korn med nogle eksperimenter”, Ingeniøren, Vol. 39,
No. 9, 1930, p. 99-107.
ASTM D4253, “Standard test methods for maximum index density and unit weight of soils
using a vibratory table”, 1996.
ATB Väg, “General Technical Construction Specifications for Roads”, Swedish, Swedish
National Road Administration, 2003.
Bishop A.W, “The principle of effective stress”, Teknisk Ukeblad, Vol. 106, No. 39, 1959,
p. 859-863.
Bonaquist R., Witczak M.W., “Plasticity modeling applied to the permanent deformation
response of granular materials in flexible pavement systems”, Transportation Research
Record, No. 1540, 1996, p. 7-14.
Box J.E., Taylor S.A., “Influence of soil bulk density on matric potential”, Proc. of the Soil
Sci. Soc. of America, Vol. 26, 1962, p. 119-122.
Brown S.F., Pell P.S., “An experimental investigation of the stresses, strains and deflections
in a layered pavement structure subjected to dynamic loads”, Proceedings, Second
402 Road Materials and Pavement Design. Volume 7 – No. 3/2006

International Conference Structural Design of Asphalt Pavements, Ann Arbor, USA,


1967, p. 487-504.
CEN, “Unbound and hydraulically bound mixtures. Part 7: Cyclic load triaxial test for
unbound mixtures”, European Standard Draft prEN 13286-7, 2000.
Chandra D., Chua K.M., Lytton R.L., “Effects of temperature and moisture on the load
response of granular base course material in thin pavements”, Transportation Research
Record, No. 1252, 1990, p. 33-41.
Croney D., Coleman J.D., Black W.P.M., “Movement and distribution of water in soil in
relation to highway design and performance”, Highway Research Board, Special report,
No. 40, 1958, p. 226-252.
Downloaded by [Simon Fraser University] at 14:49 13 November 2014

Davis J.L., Chudobiak W.J., “In situ meter for measuring relative permittivity of soils”,
Geological Survey of Canada, Part A, Paper 75-1, 1975, p. 75-79.
Dawson A.R., Paute J.L., Thom N.H., “Mechanical Characteristics of Unbound Granular
Materials as a Function of Condition”, Flexible Pavements, ed. Gomes Correia A., 1996,
p. 35-45.
Dempsey B.J., “Laboratory and field studies of channeling and pumping”, Transportation
Research Record, No. 849, 1982, p. 1-12.
Erlingsson S., Bjarnason G., Thorisson V., “Seasonal variation of moisture and bearing
capacity on Icelandic test road sections”, Proceedings of the 5th int. symposium on
unbound aggregates in roads-UNBAR 5, Nottingham June 2000, p. 317-324.
Fellner-Feldegg H.,. “The measurement of dielectrics in the time domain”, The Journal of
Physical Chemistry, Vol. 73, No. 3, 1969, p. 616-623.
Fuller W.B., "Proportioning concrete", A treatise on concrete plain and reinforced,
eds.:Taylor F.W. and Thompson S.E., 1905, p. 183-215.
Fuller W., Thompson S.E., “The laws of proportioning concrete”, Transactions of the
American Society of Civil Engineers, Paper No 1053, 1907, p. 67-143.
Haynes J.H., Yoder E.J., “Effects of repeated loading on gravel and crushed stone base course
materials used in the AASHO road test”, Highway Research Record, No. 39, 1963,
p. 82-96.
Hicks R.G., Factors influencing the resilient properties of granular materials, Ph.D. Thesis
University of California, Berkeley, 1970.
Hicks R.G., Monismith C.L., “Factors influencing the resilient response of granular
materials”, Transportation Research Record, No. 345, 1971, p. 15-31.
Hoekstra P., Delaney A., “Dielectric properties of soils at UHF and microwave frequencies”,
Journal of Geophysical Research, Vol. 79, No. 11, 1974, p. 1699-1708.
Hornych P., Hameury O., Paute J-L., “Influence de l’eau sur le comportement mecanique des
graves non traitees et des sols supports de chaussees”, International Symposium on
Subdrainage in Road Pavements and Subgrades, Granada, November 1998, p. 249-257.
Influence of Water on Resilient Properties 403

Hveem F.N., Carmany R.M., “The factors underlying the rational design of pavements”,
Highway Research Board, Proc of the 28th annual meeting, Washington. 1948,
p. 101-136.
Kolisoja P., Saarenketo T., Peltoniemi H., Vuorimies N., “Laboratory testing of suction and
deformation properties of base course aggregates”, Transportation Research Record,
No. 1787, 2002, p. 83-88.
Krahn J., Fredlund D.G., “On total, matric and osmotic suction”, Soil Science, Vol. 114,
No. 5, 1972, p. 339-348.
Ksaibati K., Armaghani J., Fisher J., “Effect of moisture on modulus values of base and
subgrade materials”, Transportation Research Record, No. 1716, 2000, p. 20-29.
Downloaded by [Simon Fraser University] at 14:49 13 November 2014

Ladd R.S., “Preparing test specimens using undercompaction”, Geotechnical Testing Journal,
GTJODJ, Vol. 1, No. 1, 1978, p. 16-23.
Lade P.L., Nelson R.B., “Modelling the elastic behaviour of granular materials”, Int. J.
Numer. Anal. Meth. Geomech, Vol. 11, 1987, p. 521-542.
Lekarp F., Isacsson U., “Development of a large-scale triaxial apparatus for characterization
of granular materials”, International Journal of Road Materials and Pavement Design,
Vol. 1, No. 2, 2000, p. 165-196.
Lekarp F., Isacsson U., “The Effects of Grading Scale on Repeated Load Triaxial Test
Results”, International Journal of Pavement Engineering, Vol. 2, No. 2, 2001, p. 85-101.
May R.W., Witczak M.W, “Effective granular modulus to model pavement responses”,
Transportation Research Record, No. 810, 1981, p. 1-9.
Morgan J.R., “The response of granular materials to repeated loading”, ARRB Proceedings
3rd Conf. Sydney 1966, 1966, p. 1178-1192.
Olson S.A., Langfelder L.J., “"Pore water pressures in saturated soils", Journal of the Soil
Mechanics and Foundation Division, ASCE 91SM4, 1965, p. 127-150.
Pappin J.W., Brown S.F., O’Reilly M.P., “Effective stress behavior of saturated and partially
saturated granular material subjected to repeated loading”, Géotechnique, Vol. 42, No. 3,
1992, p. 485-497.
Raad L., Minassian G.H., Gardin S., “Characterization of Saturated Granular Bases Under
Repeated Loads”, Transportation Research Record, No. 1369, 1992, p. 73-82.
Rada G., Witczak M.W., “Comprehensive evaluation of laboratory resilient moduli for
granular material”, Transportation Research Record, No. 810, 1981, p. 23-33.
Rasiah V., Aylmore L.A.G., “"Sensitivity of selected water retention functions to compaction
and inherent soil properties", Aust. J. Soil Res., Vol. 36, 1998, p. 317-326.
Reynolds O., “"On the dilatancy of media composed of rigid particles in contact", Phil. Mag.
S.5, Vol. 20, No. 127, 1885, p. 469-480.
Richter C.A., Schwartz C.W., “Modeling stress- and moisture-induced variations in pavement
layer moduli”, TRB 2003 Annual meeting, 2003.
404 Road Materials and Pavement Design. Volume 7 – No. 3/2006

Seed H.B., Mitry F.G., Monismith C.L. Chan C.K., “Prediction of flexible pavement
deflections from laboratory repeated load test”, NHCRP Rep. No. 35, National
Cooperative Highway Research Program, 1967.
Smith-Rose R.L., “The electrical properties of soils for alternating currents at radio
frequencies”, Proc. Royal Soc. London Ser. A. Mathematical and physical sciences,
Vol. 140, 1933, p. 359-377.
Sweere G.T.H., Unbound granular bases for roads, Ph.D. Thesis Delft Technical University,
1990.
Talbot A., Richart F.E., “The strength of concrete. Its relation to the cement aggregates and
water”, University of Illinois Bulletin, Bulletin No 137, Vol. 21, No. 7, 1923.
Downloaded by [Simon Fraser University] at 14:49 13 November 2014

Terzaghi K., “Die Theorie der hydrodynamishen spannungserscheinungen und ihr


erdbautechnishes anwendungsgebiet”, Proc. Int. Cong. App. Mech Delft, 1924,
p. 288-294.
Thom N., Design of road foundations, Ph.D. thesis, Department of Civil Engineering,
University of Nottingham, 1988.
Thom N.H., Brown S.F., “Effect of moisture on the structural performance of a crushed-
limestone road base”, Transportation Research Record, No. 1121, 1987, p. 50-56.
Tian P., Zaman M.M., Laguros J.G., “Gradation and Moisture Effects on Resilient Moduli of
Aggregate Bases”, Transportation Research Record, No. 1619, 1998, p. 75-84.
Uzan J., “Characterization of Granular Material”, Transportation Research Record, No. 1022,
1985, p. 52-59.
Van Toan D., Effects of basecourse saturation on flexible pavement performance, PhD Thesis
Auckland University, 1975.

You might also like