Farhad Salour Doctoral Thesis

particle interlock strength (Lekarp et al., 2000; Ekblad and Isacsson, 2006). In 
fine-grained materials and subgrade soils, in addition to the lubricating effect, pore 
water usually affects the effective stress of the soil structure due to a decrease in pore 
suction which will result in decreased reduction in material stiffness. In materials with 
low permeability that are exposed to frequent heavy traffic stress loads, pore pressure 
build-up may also take place which can result in considerable bearing capacity loss 
(Yang et al., 2008; Cary and Zapata, 2011; Nowamooz et al., 2011). All other factors 
remaining unchanged, an increase in moisture content results in larger resilient and 
plastic strains and accelerate the rate of distresses in pavement structures (ARA, 2004). 
Pavement moisture content and its influence on the behaviour of unbound materials 
has been the subject of many laboratory-based and field investigations. Some of the 
moisture related studies found in the literature are presented below. 
6.1.
 
Laboratory-based investigation and measurements 
Conventional repeated load triaxial testing
 
In pavement engineering, the mechanical response of the pavement materials are 
traditionally characterized using the total stress approach. In spite of the fact that 
pavement unbound layers are usually in partially saturated conditions, the effect of soil 
suctions is not generally considered. At laboratory scale, the material mechanical 
response is widely characterized using RLT tests. When using the conventional total 
stress approach, the effect of moisture content on the mechanical behaviour of 
unbound materials is generally taken into account later in the design process using 
adjusting models (described later). 
Ekblad and Isacsson (2006) investigated the influence of moisture on the resilient 
properties of coarse-grained granular materials with different grading using a large scale 

---

32 
triaxial cell. They monitored the moisture content of the specimen using two TDR 
probes buried in the specimen. They reported a significant loss in the resilient modulus 
with increase in moisture content, particularly in specimens with high fines content. The 
effect of moisture content on the resilient modulus was considerably less in specimens 
with coarser grading. 
Andrei (2003) conducted an extensive laboratory test program using the RLT method 
for both coarse-grained and fine-grained materials. The materials used in the study 
consisted of four base and four subgrade materials that are typically encountered in 
Arizona. The RLT tests were conducted at different moisture contents and compaction 
degrees to develop a resilient modulus predictive model that could estimate changes in 
the resilient modulus as a function of compaction degree, stress level and moisture 
content. Considerable changes in the resilient modulus were observed due to moisture 
content variations. Plastic subgrade materials were in particular very sensitive to 
moisture and the resilient modulus variations from 14 to more than 1350 MPa were 
observed due to changes in moisture content. However, for non-plastic soil and base 
material, the impact was much less. These materials exhibited up to three times higher 
resilient modulus values as the moisture content was reduced compared to the optimum 
moisture content. 
M
R
-Moisture adjustment models 
Several models have been developed to incorporate the effect of moisture content and 
its variations when predicting the resilient modulus of unbound pavement materials. 
Most of the developed models are based on laboratory testing of unbound materials at 
differing moisture contents. 
A simple approach that has gained popularity over the recent years is the predictive 
M
R
-Moisture model proposed in 
Mechanistic-Empirical Pavement Design Guide
(ARA, 2004). 
This one-dimensional model directly incorporates variations in the moisture content to 
the resilient modulus of unfrozen unbound materials using an adjustment factor given 
as the following: 
))
(
)
exp(ln(
1
log
opt
s
R
R
S
S
k
a
b
a
b
a
M
M
opt







[8] 
where 
opt
R
R
M
M
= resilient modulus ratio; 
R
M
is the resilient modulus at a given time 
and 
opt
R
M
is the resilient modulus at a reference condition. Further is 
a
= minimum of 
)
log(
opt
R
R
M
M
, 
b
= maximum of 
)
log(
opt
R
R
M
M
, 
s
k
= regression parameter and 
)
(
opt
S
S

= 
variation of degree of saturation expressed in decimal. 

---

33 
This empirical model and its regression parameters were developed through extensive 
laboratory triaxial tests conducted by Andrei (2003). Figure 16 shows the 
M
R
-Moisture 
model for both the fine-grained and coarse-grained materials proposed in MEPGD.