FACTORS AFFECTING RESILIENT MODULUS

Saleh M & S J JI
University of Canterbury, Christchurch, New Zealand

n
s io
is
rm
ABSTRACT: Resilient modulus is an important property for asphalt concrete design and for

pe
mechanistic analysis of pavement response under traffic loading. This study investigates the
different factors affecting the resilient modulus of hot mix asphalt. A fractional factorial design

r's
of experiment was carried out to investigate seven factors each factor was studied at two levels.
ho
These factors are: the maximum nominal aggregate size, specimen diameter and thickness, the
load pulse form and duration, preset strain level and the compaction method. Two types of hot
ut
mix asphalts with different maximum aggregate sizes (10mm and 14mm) were studied. Gyratory
tA

and Marshall compaction methods were used to prepare the specimens. All mix specimens were
compacted to the same air voids content (5.0±0.5%). Sinusoidal and triangular load pulse forms
ou

were used in the measurement of the resilient modulus. This study attempts to examine how the
ith

different factors interrelate to affect the resilient modulus. It was found that the most significant
factor affecting the resilient modulus value is the maximum nominal aggregate size followed by
w

the load duration period, specimen thickness, specimen diameter, compaction method and strain
e

level, and then the interaction between specimen diameter and thickness, the interaction between
us

aggregate size and thickness, aggregate size and compaction method. The effect of wave pulse
form whether triangular or sinusoidal was found that it has no significant difference of the
re

resilient modulus value.

o
N

KEY WORDS: Resilient, strain, modulus, factors

1. BACKGROUND

The resilient modulus is an important parameter that is used in the mechanistic pavement design
as it is being used as an input to the multilayer layer elastic theories or finite elements models to
compute pavement response under traffic loading. These responses can be used through transfer
functions to calculate the optimum thickness design for new pavement or to estimate the

n
remaining life of an existing pavement. This makes the resilient modulus one of the most

io
important parameters in pavement design and analysis.
Due to the simplicity and ease of application to test laboratory compacted specimens and field

s
cores, the indirect tensile test is the most common repeated load test to measure the resilient

is
modulus of bituminous mixture. This involves preparing a compacted cylindrical asphalt mixture

rm
subjected to diametrical repeated loading. This test is standardised as the Australian Standard AS
2891.13.1-1995.
There are two parts to the resilient modulus test. The first is the preconditioning and test

pe
setting determination to find the load required to reach certain level of strain of 50 micro-strain as
required by the Australian standard. After that the resilient modulus is determined using
equation 1.

r's
P (υ + 0.27)
ho
E= Equation 1
Hhc
ut
Where E = resilient modulus (MPa)
tA

P = peak load (N)

υ = Poisson ratio (assumed as 0.4)
H = recovered horizontal deformation of specimen (mm)
ou

hc = height of specimen (mm)

ith

Although the resilient modulus test is used widely worldwide (Huang, 1993), there are a lot of
factors affecting resilient modulus of asphalt subjected to indirect tensile test. These include the
w

geometric factors of the test specimens (specimen thickness and diameter), maximum nominal
aggregate size, the load waveforms and pulse durations applied to the test specimens, the preset
e

strain level that is to be met during the test, and the type of compaction of the test specimen. This
us

research is to study the effects of these factors, their interactions, and their significance on the
resilient modulus through a fractional factorial design of experiment.
re

2 FACTORIAL ANALYSIS
o
N

There are a lot of factors that can be considered in the determination of resilient modulus.
However, for the purpose of this research, seven factors were considered. These are classified in
terms of geometric factors, experimental factors and compaction method. Table 1 shows the
factors that have been considered in the half fractional factorial analysis and the high and low
level of the numerical factors and the different levels for the categorical factors. The symbols
used are shown in Table 1 for each factor in the factorial analysis.
Table 1: Factors and their level that considered in the factorial analysis.

Factors Levels Units Symbol

Specimen Diameter 100 150 mm A

n
Specimen Thickness 40 70 mm B

s io
Max. Agg. Size 10 14 mm C

is
Compaction Method Marshall Gyratory Categorical D

rm
Load Wave Form Triangular Sinusoidal Categorical E

pe
Load Duration 100 200 ms F

Strain Level 20 60 µε G

r's
ho
For seven factors each at two level and with three replicates per test, the total number of
tests/specimens required is 3*27=384 runs. Therefore, it was decided to use half fractional
ut
factorial which reduces the number of runs to 3*27-1= 192 runs.
tA

3 RESULTS AND ANALYSIS

ou

Figure 1 is a half probability chart of the effects generated by Design Expert Software (2004).
Table 2 shows the analysis of variance for the different effects. The higher the F value or the
ith

lower the P value the higher the significance of the factor. Effects which lie on the straight line
are the insignificant effects, whereas the significant effects are far from the line (Douglas,2001).
w

It is clear that the aggregate maximum nominal size is the most important factor affecting the
resilient modulus, followed by the load duration, then the specimen geometry represented by the
thickness and diameter then the interactions between the different factors.
e

Figure 2 shows the effect of the aggregate nominal maximum size on the resilient modulus.
us

The coarser the aggregate gradation the higher the stiffness of the mix. This may be explained by
the higher partical to partical contact in the coarser aggregate structure which results in a higher
re

resilient modulus. The analysis of variance shown in Table 2 shows that this factor has the
highest F value which reflects the importance of the aggregate gradation on the resilient modulus
value. This results agrees with Figure 1. These results agreed with findings of Lim and Tan
o

(1995) and those of Brown and Bassett (1990).

N
 DESIGN-EXPERT Plot
Resilient Modulus
Half Normal plot
A: Diameter
B: Thickness C
C: Aggregate Size

n
D: Compaction 99
F

io
E: Load Form

Half Normal % probability

F: Duration B

s
97 A
G: Strain Lev el AC
CD

is
95
AB
ABC
AD
BC

rm
90 ABDD
BCD
BF
BCF
85 G
CF
80

pe
70
60

r's
40

20
ho
0
ut
tA

0.00 220.23 440.45 660.68 880.91

ou

|Effect|
ith

Figure 1: Half normal probability plot of the effects.

w
e
us
re
o
N
Table 2: Analysis of Variance of the significant factors.

Sum of Mean F P-Value

Source Squares DF Square Value Prob > F
Model 61571729 18 3420652 162.6082 < 0.0001

n
A 4345352 1 4345352 206.5659 < 0.0001

io
B 6589358 1 6589358 313.2397 < 0.0001
C 24831866 1 24831866 1180.438 < 0.0001

s
D 1095940 1 1095940 52.09794 < 0.0001

is
F 10507674 1 10507674 499.5055 < 0.0001

rm
G 192355 1 192355 9.144022 0.0031
AB 2108945 1 2108945 100.2533 < 0.0001
AC 3140645 1 3140645 149.2975 < 0.0001

pe
AD 1443301 1 1443301 68.61048 < 0.0001
BC 1171215 1 1171215 55.6763 < 0.0001
BF 311655.1 1 311655.1 14.81522 0.0002

r's
CD 2690620 1 2690620 127.9046 < 0.0001
CF 188498 1 ho 188498 8.96067 0.0034
ABC 1922761 1 1922761 91.40268 < 0.0001
ABD 456729 1 456729 21.71163 < 0.0001
BCD 334971.1 1 334971.1 15.92359 0.0001
ut
BCF 206885.3 1 206885.3 9.83475 0.0022
tA
ou
ith

The effect of the load duration on the resilient modulus is depicted in Figure 3. The shorter the
time duration the higher the resilient modulus (Huang, 1993). This effect can be attributed to the
w

viscoelastic nature of the bituminous materials which make the properties of these mixes load
rate dependent. It is well known that slow traffic has the most damaging effect on the asphalt
e

pavement causing severe rutting and distorations in the pavement structure. Therefore, in
us

measuring the resilient modulus in the laboratory an appropriate load duration should be selected
in order to measure a representative reslient modulus for the insitu conditions.
re
o
N
 4666

Resilient Modulus (MPa)

3855

n
io
3044

s
is
2233

rm
pe
1422

10.00 11.00 12.00 13.00 14.00

r's
ho
C: Aggregate Size
ut

Figure 2: Effect of maximum nominal aggregate size on the resilient modulus

tA

Figures 4 and 5 reveal the effect of the specimen geometry on the resilient modulus. It is clear
ou

that smaller diameter and thinner specimens yield higher resilient modulus than larger diameter
and thicker specimens. This finding is in agreement with Kandhal and Brown (1990) as they
found that the tensile strength of the 150mm diameter specimens were always lower than the
ith

100mm diameter specimen. This effect may be explained by the higher confinement of the
aggregate particles in the smaller diameter and thinner specimen. In addition for the larger
w

diameter and thicker specimens the probability of having higher percentages of micro-cracks is
higher than that in the smaller specimens, therefore, the rate of energy release in the larger
e

specimens is higher that that in the smaller specimens. A similar effect of that is noticed in
us

Portland cement specimens as the smaller cylinders always yield a higher strength than that of the
larger specimens. Therefore, a representative geometry should be selected in order to have a
resilient modulus that matches the actual field conditions.
re

The interaction between the aggregate size and the specimen diameter is shown in Figure 6.
The effect of the coarse gradation is very pronounced in the small size diameter specimen while
o

this effect is much less in the larger diameter specimen. This means using smaller molds, 100
N

mm diameter, in the laboratory is very sensitive to the aggregate gradation compared to the 150
mm diameter specimens. Again this is probably because of the degree of confinement in the
smaller mold is much higher than that in the larger diameter mold.
 n
4666

io
Resilient Modulus (MPa)

s
3855

is
rm
3044

pe
2233

r's
1422
ho
100.00 125.00 150.00 175.00 200.00
ut
F: Duration (ms)
tA

Figure 3: Effect of maximum load duration on the resilient modulus

ou

4666
ith Resilient Modulus (MPa)
w

3855
e
us

3044
re

2233
o
N

1422

40.00 47.50 55.00 62.50 70.00

B: Thickness(mm)
Figure 4: Effect of specimen thickness on the resilient modulus

4666

Resilient Modulus (MPa)

3855

n
io
3044

s
is
rm
2233

pe
1422

100.00 112.50 125.00 137.50 150.00

r's
A: Diameter (mm) ho
Figure 5: Effect of specimen diameter on the resilient modulus
ut
4666 C: Aggregate Size
tA
Resilient Modulus(MPa)

ou

3855
ith

AC
3044 14
w

AC10
2233
e
us

1422
re

100.00 112.50 125.00 137.50 150.00

o

A: Diameter (mm)
N

Figure 6: Effect of interaction between aggregate size and specimen diameter on the resilient
modulus.

Figure 7 shows the interaction between the specimen thickness and diameter. It is obvious that
the smaller diameter specimens are very sensitive to the specimen thickness as the effect of
thickness is quite pronounced, however, the large diameter specimens are not sensitive to the
thickness of the specimen as both the 40 mm and 70mm thick specimens produced a relatively
similar resilient modulus for the 150 mm diameter specimen. This will add another advantage for
the 150 mm diameter specimen compared to the 100 mm diameter.

4666 B: Thickness

n
io
Resilient Modulus(MPa)
3855

s
is
Thi
ckn
3044 ess=

rm
40
m m
Thickness=
70 m m

pe
2233

r's
1422
ho
100.00 112.50 125.00 137.50 150.00

A: Diameter (mm)
ut
Figure 7: Effect of interaction between thickness and diameter of the specimen on the resilient
tA

modulus.
ou

Figure 8 depicts the interaction between the compaction method and the specimen diameter. It is
clear that Marshall compaction is providing higher resilient modulus than gyratory compaction
ith

for smaller diameter specimens however for the larger diameter there is no effect between the two
compaction methods.
The interaction between the aggregate gradation and the thickness of the specimen is shown in
w

Figure 9. The coarser gradation is providing slightly higher effect on the resilient modulus for
the thin specimens (40 mm) and this effect is reduced for the thick specimens (70 mm).
e
us
re
o
N
 4666 D: Compaction

3855

Resilient Modulus

n
3044
Mars

io
hall

s
Gyratory

is
2233

rm
1422

pe
100.00 112.50 125.00 137.50 150.00
A: Diameter

r's
Figure 8: Effect of interaction between compaction method and diameter of the specimen on the
resilient modulus ho
C: Aggregate Size
ut
4666
tA
Resilient Modulus(MPa)

3855
ou

3044 AC14
ith

AC10
w

2233
e
us

1422
re

40.00 47.50 55.00 62.50 70.00

B: Thickness (mm)
o

Figure 9: Effect of interaction between aggregate maximum nominal size and specimen thickness
N

on the resilient modulus

4 CONCLUSION

Resilient modulus is affected by many factors. The aggregate gradation is the most significant
factor as the coarser gradation provides significantly higher resilient modulus compared to the
fine gradation. The time duration is the second most significant factor affecting the resilient
modulus due the visco-elastic nature of the bituminous materials. The shorter the loading time
(the faster the speed) the higher the resilient modulus and vice versa. The specimen geometry

n
(the thickness and diameter) are quite important factor affecting resilient modulus. The smaller

io
size specimens tend to have a higher resilient modulus than the larger size specimens. The
interactions between specimen geometry and aggregate size and the compaction method are also

s
important factors. The large size specimens (150mm diameter and 70 mm thick) tend to be less

is
sensitive to the compaction method and the aggregate gradation compared to the smaller size

rm
specimens.

ACKNOWLEDGEMENT

pe
The authors would like to thank Fulton Hogan contractor for the financial support which makes
this research possible. The authors also would like to express their greatest gratitude to the

r's
Department of Civil Engineering, University of Canterbury, for its support.
ho
REFERENCES
ut
AS 2891.13.1-1995 Methods of sampling and testing asphalt. Method 13.1: Determination of the
tA

resilient modulus of asphalt – indirect tensile method

Brown. E. R. and C. E. Bassett ,1990. Effects of maximum aggregate size on rutting potential and
other properties of asphalt-aggregate mixtures. Transportation Research Record (1259): 107-
ou

119.
Design Expert version 6, State-Ease Inc.2021 East Hennpin Ave., Suite 480, Minneapolis, MN
ith

55413
w

Douglas C. M., Design and Analysis of Experiments, John Wiley & Sons, Inc, 2001

Huang, Y.H. 1993. Pavement Analysis and Design, Prentice Hall Inc.
e
us

Kandhal, P.S and E.R. Brown 1990. Comparative evaluation of 4-inch and 6-inch diameter
specimens for testing large stone asphalt mixes. Serviceability and Durability of Construction
re

Materials – Proceedings of the First Materials Engineering Congress Part 1, Aug 13-15 1990.
o

Lim, C.T. and Tan, S.A 1995. Specimen size effects on the diametrical mechanical testing of
cylindrical asphalt mixes. Journal of Testing and Evaluation, Volume 23: 436-441.
N