Materials and Structures

DOI 10.1617/s11527-013-0237-8

ORIGINAL ARTICLE

Numerical and experimental evaluation of the influence

of the filler–bitumen interface in mastics
Ebrahim Hesami • Bjorn Birgisson • Niki Kringos

Received: 18 August 2013 / Accepted: 11 December 2013

 RILEM 2013

Abstract The successful use of additives in modified of the filler–bitumen interface, considering shape and
asphalt mixtures, such as warm mix asphalt, depends size. The conclusion of this paper can thus be useful for
largely on the effect such modification has on the mastic. the effective development of modified asphalt mixtures
Previous research indicated that such modifiers do not and gives strong indications towards future research
simply change the bitumen properties, but can also directions.
change the interaction between the filler and the bitumen
matrix. Understanding the effect of the properties of the Keywords Mastic  Filler–bitumen interaction 
fillers, the bitumen and their interaction is thus important Viscosity  Interface properties  Asphalt
for future asphalt mix design. In order to investigate this modification
and to define the dominant relationships, this paper
combines a numerical and experimental approach. In
the experiments, the viscosities of modified and unmod- 1 Introduction
ified mastics with different filler concentrations and
types were systematically investigated utilizing a novel Viscoelastic behaviour of asphalt mixtures is mostly
testing protocol. In the numerical analyses, the Finite defined by the properties of its bituminous mastics.
Element Method was utilized for a micro-mechanical Traditionally, asphalt mix design is largely based on
analysis, in which the shape and size of the filler bitumen properties, such as viscosity. In the case of
particles were varied in the bitumen matrix. Combining modified asphalt mixtures, additives are added with a
the experimental and numerical results allowed for a specific design purpose in mind. For example, lower-
detailed investigation of the effect of the interface ing the mixing and compaction temperatures in the
properties, with and without modifiers. The research case of warm mix asphalt (WMA) or stretching the
further indicated that the effect of the shape and size of visco-elastic mix behaviour to lower temperatures, in
the fillers varied, depending on the interface properties. the case of polymer modification. Such modification
From the research relationships were established may, however, affect more properties than the
between the overall mastic viscosity and the influence intended design property only and thus affect the
commonly used design relationships. To some extend
one could account for these by empirically adjusting
E. Hesami (&)  B. Birgisson  N. Kringos these relationships. Such an approach, though, will
Division of Highway and Railway Engineering,
always be limited, does not lead to generic mix design,
Department of Transport Science, KTH Royal Institute of
Technology, Brinellvagen 23, 100 44 Stockholm, Sweden nor assist in successful predictions of in-time behav-
e-mail: hesami@kth.se iour of modified pavements.
 Materials and Structures

For this reason, a more thorough approach is

needed in which the actual effects of the modifications
are investigated by focusing on the fundamental
mechanisms. From such investigations, fundamental
relationships and dominant parameters can be identi-
fied that can assist in the development of successful
design protocols, material selection procedures and
pavement construction conditions.
This paper is investigating the effect of a WMA
additive on the mastic behaviour. To do so, experi-
mental and numerical analyses were developed in
which a focus was placed on the effect that the
modification has on the bitumen and the bitumen–filler
Fig. 1 The interaction model between filler and bitumen [9]
interface. The effect of the filler type, size and shape
were hereby important parameters that were consid-
It can be hypothesized that the interaction between
ered. To limit the number of variables and focus the
the filler and bitumen changes the properties of the
research, the temperature and bitumen source were
bitumen around the filler grains (Fig. 1) [9]. The
kept constant.
properties of this fraction of bitumen are influenced by
In the following sections the experimental and
fillers, and its characteristics, such as thickness and
numerical analyses are described in details and the
stiffness, can be altered by changing the type of filler.
various considerations are explained. Following these,
The bitumen in the vicinity of filler grains will thus
the found relationships and noted connectivity
have the highest stiffness, which decreases with
between the considered parameters are discussed.
distance from the filler grain until it reaches the
stiffness of the free bitumen. If a filler particle rotates
or displaces, its surrounding network will therefore
2 Research scope also change. The resistance to this movement can be
different if the thickness of the influenced layer is
The mastic phase has become an important point of changed or if slippage on the interface occurs. As such,
focus for many research groups in the past years e.g. this influenced layer of bitumen defines the interfacial
[1–7]. One of the challenging aspects in investigating properties of filler and bitumen, and can have a
mastic is to single out certain parameters without significant effect on the rheological behaviour of
affecting others. From earlier work [8], it was found asphalt mastics. A change of the influenced layer can
that filler shape, size distribution and filler chemistry occur by changes in the bitumen flow profile around
all had significant effects on the viscosity of mastics, the individual filler grain or when changes in the
and slightly changes in these parameters changed the adhesion of the filler and bitumen occur.
resulting mastic viscosity significantly. In this Considering the size and variability of this area,
research, three different fillers (M600, M10 and fly investigating the influenced layers directly is rather
ash) and one type of bitumen (penetration 70/100) complex. For this reason, in this research experimental
were investigated to produce several mastics with and numerical techniques are employed to investigate
different filler concentrations. From this, it was shown the influence of filler–bitumen interface in mastics.
that the relative viscosity of the mastics increases by To do so, in this study the viscosity of modified and
increasing the filler concentration, but that this incre- unmodified mastic samples with different filler types
ment is different for different filler types. at different filler concentrations were measured.
The differences in viscosity behaviour of mastics Utilizing the finite element approach, parameter-
partially could be explained from the different shapes ized meshes were developed to micro-mechanically
and sizes of fillers, but it was concluded that the filler– analyse the effect of shape and size of the filler
bitumen interaction also has an important effect on the particles in the bitumen matrix while varying the
mastic rheological behaviour. interface properties.
Materials and Structures

3 The hydraulic volume of fillers

The stiffening effect of fillers on bitumen has been

known for quite some time [3–6, 10] and can be
divided into three different mechanisms: (1) reinforce-
ment of bitumen due to filler particles themselves, (2)
(a) (b)
physiochemical interaction between filler and bitu-
men, and (3) particle–particle interaction. Physio- Fig. 3 Equal filler hydraulic volume concentration for two
chemical reinforcement refers to interfacial effects different circumstances, a coarser filler with lower surface
between filler and bitumen which forms adsorbed activity and b finer filler with higher surface activity
bitumen layer and a partially influenced bitumen layer
[7]. Particle–particle interaction defines the frictional
forces between particles which will be dominant at
higher concentrations, given more contact will occur.
The thickness of the adsorbed layer can be calcu-
lated from the concept of physiochemical reinforce-
ment. The size of this rigidly adsorbed layer was
previously suggested between 2 and 10 Å [11]. The (a) (b)
thickness of the adsorbed layer may, in most cases be
Fig. 4 Equal filler concentration in the unit volume of bitumen:
negligible, considering the size differences unless the a only fillers and b filler hydraulic volume
filler is fine enough that the adsorbed layer becomes
significant, for example in the case of lime [5]. In
addition to the adsorbed layer the influenced layer, concentration are kept the same, however due to the
described earlier, is acting around the particles. larger size of the hydraulic filler volume, the unit
Combination of the adsorbed layer, influenced layer volume of mastic needs fewer but coarser filler grain.
and the particle itself are the hydraulic volume of the As the viscosity of the mastic is very dependent on the
filler (Fig. 2) (named ‘micelle’ in earlier publications size and size distribution of the filler, the adsorbed and
[7, 11, 12] which can be used for calculating the effect influenced layer around the filler can change the
of filler on the viscosity of mastics [13]. viscosity results dramatically. Mastic with coarser
Figure 3 illustrates how the mastics with lower filler has a lower viscosity than the mastics with finer
filler concentration but higher filler surface activity filler at the same concentrations [14].
can have the same filler hydraulic concentration as a In addition to the size effect of the layer around the
mastic with a higher concentration but a lower filler filler, this layer can also have a smoothening effect.
surface activity. Depending on its properties, it can change the mobility
Since the hydraulic filler volume is always larger of the filler within the matrix as the interfacial
than the nominal filler volume, the viscosity of mastics properties between the filler and bitumen change.
will increase when the filler–bitumen interaction is For instance the hydraulic volume of filler can rotate
stronger [7]. Figure 4 shows the same concentration due to shear much easier than the filler with rough
for nominal filler volume and hydraulic filler volume texture and consequently the less resistance to flow. It
for the particular filler size. In both cases the is clear that there is not any sharp boundary between
layers and other part of bitumen, instead all work
together. If assuming the average velocity of bitumen
A and the velocity of the filler B, the rigid adsorbed
layer have the velocity of B and by increasing the
distance from filler the velocity of the influenced layer
increases up to A (Fig. 5) [13].
The hypothesis as explained above may be domi-
Fig. 2 Scheme of adsorbed layer, influenced layer and hydrau- nant in modified mastics, especially for adhesion
lic volume of filler promoter additives which act at the surface of the filler.
 Materials and Structures

Fig. 5 Velocity gradient in the surrounding layer

Fig. 8 Filler size distribution for M10, M600 and fly ash

mineral components, M600 is much finer than M10

(Fig. 8). This means comparing the results of mastics
with M600 and M10 allow for studying the size effect
Fig. 6 Typical scheme of surfactant molecular
of fillers on the behaviour of asphalt mastics.
Scanning electron microscopy pictures (Fig. 9)
show the angularity of M600 and M10, and round
shape of fly ash grains. So the effect of the shape of the
filler can be studied by comparing the viscosity of
mastics made with these fillers.
According to the test results from the Brunauer,
Emmett and Teller theory (BET) the specific surface
area of 0.93, 4.0 and 1.5 m2/g were obtained for M10,
M600 and fly ash respectively. The density of fillers
also measured by Helium Pycnometer and 2.79, 2.75
Fig. 7 Adhesion of the surfactant molecules to the solid polar
and 2.41 kg/cm3 were obtained as densities of M10,
surface
M600 and fly ash respectively (Table 1).
In order to evaluating the effect of filler properties in
A surfactant molecule is typically made of a modified and unmodified mastics, one type of bitumen
hydrocarbon chain plus a polar group (Fig. 6). If the (70/100 from Nynas AB) was used for all mastics.
filler has polar or polarizable components, the polar To study the effect of the filler–bitumen interface,
part of the surfactant molecule will make the polar AkzoNobel Rediset was used for producing the mod-
bond with surface of aggregate, and bitumen will ified mastics. According to the information released by
make a Van der Waals forces with the hydrocarbon the manufacturer, Rediset is designed to produce WMA
chain of the surfactant molecule (Fig. 7). This effect and it can decrease the mixing and compaction temper-
can increase the thickness of the adsorb layer. The ature by 30 C. Furthermore, Rediset actually improves
effects of surfactant molecules on the adsorbed layer adhesion between the binder and the aggregates by
may be more visible at higher temperatures when the amine surfactant [15]. In this research Rediset was
hydraulic volume of the polymer chain increases. added to the mastics by 1.5 % of weight of bitumen.
Each type of mastic was produced at two different
filler concentrations 20 and 40 % by volume. All
4 Experimental investigations samples were produced in two Rediset modified and
unmodified conditions.
4.1 Testing materials
4.2 Testing methods
In this study, three types of filler (M600, M10 and fly
ash) were chosen so as to allow investigation on certain Due to the sensitivity of the viscosity to the various
properties. While M600 and M10 have the same parameters, reliable and repeatable measurement of
Materials and Structures

Fig. 9 Filler’s scanning electron microscopy pictures

Table 1 Fillers physical properties Newtonian materials. In this standard, the converting
M600 M10 Fly ash equations are based on the assumptions of dealing with
a Newtonian material. As such, the shear stress
Specific surface area (m2/g) 4.0 0.93 1.5 gradient in the gap between inner and outer cylinder
3
Density (kg/cm ) 2.75 2.79 2.41 is considered linear, there is no plug zone and no
slippage on the wall of inner and outer cylinder. From
these assumptions, the limitations of using the cylin-
the viscosity of the mastics is a challenging task. For drical rotor are considerable.
this reason a protocol was established which is able to Bitumen is a Newtonian material, however when
measure the viscosity of mastics from low to high filler increasing the filler concentration, mastic starts to
concentrations at elevated temperature [16]. behave more non-Newtonian. For this reason, a vane
All mastic samples were mixed with high shear shaped rotor was chosen for the higher filler concen-
mixer at 140 C. As high temperature increases the tration. The range of filler concentration for using the
risk of sedimentation of filler and low temperature two rotors can vary from one type of mastic to the
changes the phase of mastic, 100 C was chosen as the other and it is highly depended on the filler size. From
optimum testing temperature. In this temperature the work has been done by Hesami et al. [16], most of
bitumen and mastics are liquid, so the suitable the mastics showed that around 20 % filler concen-
geometry is co-axial concentric cylinder. tration can be a suitable boundary for shifting from
Due to varying flow behaviour of the tested mastics cylindrical rotor to a vane shaped rotor.
with the different filler concentrations, it was consid-
ered that accurate measurement should be possible 4.3 Experimental results
when mastics display a different type of behaviour
such as Bingham flow, shear thinning or thixotropy. Figures 10, 11, 12 and 13 show the flow curves of
To do so a viscometer was chosen, able to use different modified and unmodified bitumen 70/100, mastic
geometries and procedures. The DIN standard M600, mastic M10 and mastic fly ash. From these
(3219:1993(E)) gives the geometry of inner rotor for can be seen that bitumen and mastics with 20 % filler
 Materials and Structures

Fig. 10 Flow curves of unmodified and modified bitumen Fig. 13 Flow curves of unmodified and modified mastics with
70/100 at 100 C 20 and 40 % of fly ash at 100 C

unmodified samples. While the unmodified bitumen

gives the viscosity of 2.5 Pa s, modified bitumen gives
the viscosity of 2.25 Pa s which is 10 % lower than the
viscosity of unmodified bitumen. The ratio of
decreases in the viscosity is not the same for all
samples. With the same amount of the modifier, the
viscosity of mastic with 40 % M600 is reduced by
63 %. This is the highest reduction of viscosity
observed among the modified mastics. The second
highest reduction in viscosity is for mastic 40 % M10
and then followed by 40 % fly ash, 20 % M600, 20 %
Fig. 11 Flow curves of unmodified and modified mastics with M10 and 20 % fly ash.
20 and 40 % of M600 at 100 C The differences in the reduction in the viscosity of
modified bitumen and mastics shows that the additive
not only change the viscosity of bitumen, but also
change the filler–bitumen interfacial properties. That
is why the M600, which has the highest surface area
among the other fillers, shows the highest reduction
compare to the other fillers. Also as mastics with 40 %
filler provide more surface area compared to mastics
with 20 % filler, the reduction in the viscosity of the
mastic with 40 % filler is more than mastic with 20 %
filler.
In addition to the surface area, shape of the filler
Fig. 12 Flow curves of unmodified and modified mastics with grains also has significant effect on the reduction of
20 and 40 % of M10 at 100 C viscosity. As the results show, the modified mastics
with fly ash show lower viscosity reduction than the
have the linear flow curve, while the mastics with 40 % mastic with M10, while fly ash filler has higher surface
filler have nonlinear flow curve. So to be able to area than M10. By comparison the results this can be
compare the viscosity of different samples, an arbitrary concluded that, due to the geometry of the angular
shear rate (6 1/s) was chosen to compare the viscosity particles they have more difficulty for movement and/
of bitumen and mastics at that shear rate (Table 2). or rotation, this means generally angular particles
As it is expected modified samples for all cases produce higher viscosity compared to the round shape
showed significantly lower viscosity compare to particles. Hence, softening the structured layer around
Materials and Structures

Table 2 Viscosity of binders and mastics at shear rate 6 (1/s) and 100 C
Viscosity (Pa s)
Bitumen Mastic M600 Mastic M10 Mastic fly ash

Filler concentration (%) 20 40 20 40 20 40

Unmodified 2.5 11.4 193.6 6.7 35.4 6.1 19.4
Modified 2.3 8.4 71.0 5.6 16.8 5.2 12.6
Reduction (%) 10 26 63 16 52 14 35

Table 3 Discrete relaxation spectrum for generalized Max-

well model
Number of Shear modulus, Gi Relaxation time, ki
branches (Pa) (s)

1 4.44E?06 1.09E-13
2 7.16E?06 2.21E-12
3 1.09E?07 2.74E-11
4 1.57E?07 2.50E-10
5 2.17E?07 1.84E-09
Fig. 14 Master curve at 25 C 6 2.84E?07 1.15E-08
7 3.53E?07 6.33E-08
the angular particles gives more freedom to them for
8 4.12E?07 3.15E-07
movement and/or rotation, while the round particles
9 4.47E?07 1.44E-06
have already had this freedom. Therefor the reduction
10 4.46E?07 6.13E-06
in the viscosity of the mastics with M10 is more than
11 4.02E?07 2.48E-05
the mastics with fly ash.
12 3.22E?07 9.59E-05
13 2.26E?07 3.60E-04
5 Micro-mechanical finite element model 14 1.37E?07 1.33E-03
15 7.13E?06 4.91E-03
5.1 Bitumen system 16 3.14E?06 1.82E-02
17 1.17E?06 6.93E-02
The generalized Maxwell model is employed to 18 3.68E?05 2.73E-01
simulate the viscoelastic behaviour of bitumen [17]. 19 9.81E?04 1.13E?00
To determine the discrete Maxwell model, the bitumen 20 2.22E?04 4.98E?00
70/100 was tested with the dynamic shear rheometer 21 4.24E?03 2.38E?01
(DSR) at 25, 30, 35, 40 and 45 C. Frequency sweeps, 22 6.81E?02 1.26E?02
from 0.1 to 100 rad/s, were performed at a shear strain 23 9.07E?01 7.51E?02
value within the linear viscoelastic range. The time– 24 9.74E?00 5.25E?03
temperature superposition principle was then used to 25 8.17E-01 4.61E?04
obtain master curve for the reference temperature
25 C. The complex modulus as a function of frequency
for bitumen 70/100 is shown in Fig. 14.
A nonlinear least square regression algorithm has  
X
n
kt
been used to obtain the Prony series coefficients for GðtÞ ¼ Gi e i
; ð1Þ
viscoelastic model, represented with generalized i¼1
Maxwell model (Table 3). A discrete generalized Xn
Gi x2 k2i
Maxwell model approximates moduli by using n terms G0 ðxÞ ¼ ; ð2Þ
2 2
as [18]: i¼1 1 þ x ki
 Materials and Structures

the boundaries of particles. The TEL contains a spring

and a dashpot in series to enable the simulation the
simple viscoelastic behaviour. Both spring and dash-
pot are available in normal and tangential directions
(Fig. 18). The TEL computes stress and strain at the
interface by the following equation.
 
oU
r  n ¼ K ðU Þ  D ; ð4Þ
ot
where K is the spring constant, D is the viscous force
and U is the relative displacement of two domains
Fig. 15 Complex modulus fitted by Prony series which the TEL is assigned on the boundary in
between.
Xn To study the effect of the properties of the interface,
Gi xki
G00 ðxÞ ¼ ; ð3Þ different parameters were assigned for spring constant
2 2
i¼1 1 þ x ki and dashpot’s viscous property. To do so the below
where GðtÞ, G0 ðxÞ and G00 ðxÞ are shear modulus, equations were used to calculate the spring constant
storage modulus and loss modulus, respectively. Gi and viscose force of interface based on the properties
and ki are shear modulus and relaxation time of each of viscoelastic binder around the particles. Then TEL
Maxwell branch. The fitted complex modulus is properties were assigned in three stages weaker, same
shown in Fig. 15. and stronger than what calculated. The spring con-
stants and viscose forces were used in the computation
5.2 Finite element model are given in Table 5.
E ¼ 2Gð1 þ tÞ; ð5Þ
In order to study the effects of filler–bitumen interface
Eð1  tÞ
a finite element analysis utilizing COMSOL Multi- K¼ : ð6Þ
physics 4.3b was performed. In Fig. 16 four different tð1 þ tÞð1  2tÞ
geometries which were used for FE modeling are In above equations E is an elastic modulus, G is a
depicted. To evaluate the effect of filler–bitumen shear modulus, K is a spring constant, t is an interface
interface with different particle shape and size on the thickness and t is a Poisson’s ratio. The instantaneous
behaviour of mastics, two different particle shapes and shear stress is the sum of the shear modulus of all
sizes were used. The areas of all angular and round, big Maxwell branches that is equal to 3.75e8 (Pa). In the
and small particles are almost the same; this means all low frequency domain bitumen is an almost incom-
samples have almost the same filler concentration. pressible material; therefore the Poisson’s ratio was
However surface area (primate) of angular particle is assigned 0.499 for bitumen. Utilizing Eq. 5, the elastic
higher than the round particle in both big and small modulus of the bitumen is equal to 11.24e8 (Pa). By
cases (Table 4). assuming 1 lm as a thickness of the interface and
The overall size of the geometry is 100 lm 9 100 lm using Eq. 6, the spring constant was obtained for
and the maximum and minimum size of triangular mesh 19e15 (N/m). In Table 5, cases 1–4 represent weaker,
elements are 2 and 0.0075 lm respectively. case 5 and 6 represent same and case 7 and 8 represent
Simulation has been done as a stress controlled test stronger interfacial properties than what calculated.
and the tensile stress was increased from zero to the The overall summation of the viscosity of the
maximum stress of 5,000 N/m2 during 0.005 s, then bitumen is used as the reference of the dashpot
stress was kept constant for 0.095 s. Afterward sample viscosity of the interface layer. Sum of the viscosity of
was unloaded by 0.005 s and finally for 0.9 s the the all Maxwell branches is equal to 87e4 (Pa s),
sample was relaxed (Fig. 17). however to see the significant effect of the dashpot
To define the interface layer between particles and viscosity two extreme values (compare to overall
bitumen the thin elastic layer (TEL) was assigned on viscosity) are chosen, 87 and 87e6 (Pa s).
Materials and Structures

Fig. 16 Four different FE

model’s geometries with
their meshes

Table 4 Physical dimension of particles

Big Small Big Small
round round angular angular
(BR) (SR) (BA) (SA)

Total perimeter (lm) 422 1,333 504 1,592

Total area (lm2) 2,829 2,829 2,847 2,846

5.3 FEM results

Fig. 17 Loading system
Figure 19 demonstrates examples of the deformed
meshes at the end of loading (t = 0.095 s). From these
Overall displacement ratio ðODRÞ
it can be seen that the levels of debonding are different Displacement with interface element; lm
for varying geometries. For comparing the effect of ¼
Displacement without interface element; lm
the interfacial properties on different filler sizes and
shapes, overall mesh displacement of a stress con- ð7Þ
trolled test of each geometry was divided by the Figure 20 illustrates the overall displacement ratio
displacement of the same mesh with rigid bonding (i.e. of all four geometries for weak spring and dashpot
no interface element) (Eq. 7). constants (case 1). As it can be seen in the figure, the
 Materials and Structures

overall displacement ratio for angular particles is

higher than the displacement for round particles. This
means the effect of weak interface in angular filler is
higher compared to the round filler. At weak interfa-
cial properties the shape of the filler is dominant and
can be explained through the high stress concentration
at interface of angular fillers.
Keeping spring constant the same but increasing the
Fig. 18 Schematic of thin elastic layer (TEL) dashpot constant, changes the relation of the overall

Table 5 Spring constant and viscous force of the interface

Thin elastic layer (TEL) properties

Case number 1 2 3 4 5 6 7 8
Spring constant (MN/m) 19 19 19e3 19e3 19e9 19e9 19e11 19e11
Dashpot viscosity (Pa s) 87 87e6 87 87e6 87 87e6 87 87e6

Fig. 19 Example of
deformed mesh and colored
map for Y component of
stress tensor at t = 0.095 s
with scale factor of 50 for
deformation
Materials and Structures

Fig. 20 Overall displacement ratio in Y direction at the top line Fig. 22 Overall displacement ratio in Y direction at the top line
for case 1 for case 3

Fig. 21 Overall displacement ratio in Y direction at the top line Fig. 23 Overall displacement ratio in Y direction at the top line
for case 2 for case 4

Note that in the case 3 (Fig. 22), the overall

displacement ratio and time as Fig. 21. In this new displacement ratio at the end of unloading
interfacial condition both shape and size of filler have (t = 0.1 s) is affected mostly by the shape of the
an influence on the overall displacement under the particles; and two small and big angular particles show
given loading system. Small particles provide higher higher overall displacement ratio compare to the round
surface area, hence show more interface effect upon particles samples. However at the end of the recovery
displacement. Due to higher stress concentration instead of the shape, surface area of fillers takes over
around the interface of angular particles, angular as the dominant effect on the displacement.
fillers (big and small) show more influence of the Figure 24 illustrates the displacement of all sam-
interface compared with round fillers. ples without interface layers between fillers and
Increasing the spring constant of the interface, the bitumen. For the models without interface or strong
effect of the surface area becomes dominant. As can be interface properties, the big round particle shows the
seen in Figs. 22 and 23, small angular and small round highest displacement followed by big angular, small
particle show significantly higher permanent defor- round and small angular. From this can be concluded
mation at the end of the recovering time (at time that for no interface system or stronger interfacial
1 s).This effect is even more visible when increasing condition, bigger filler particles will result in higher
the value of the dashpot constant. Small angular overall deformation. Within the bigger filler size the
particles showed the highest overall displacement round one gives more displacement compare to the
ratio both at the end of unloading and end of recovery, angular one.
followed by the small round, big angular and big round Figure 25 shows an example of displacement–time
meshes. chart for one geometry (big angular) and different
 Materials and Structures

is more visible in mastics with higher filler

concentrations. This indicates that the additive is
not only changing the viscosity of bitumen but also
has significant influence on the surface of the filler.
From both the experimental and the numerical
analyses, it was found that this influence is varying
for different filler shapes, sizes and filler
concentrations.
• Mastics with angular fillers are affected more than
mastics with round filler.
• Mastics with smaller filler size showed a signifi-
Fig. 24 Overall displacement ratio in Y direction at the top line cant higher influence compare to the coarser one.
without interfacial element • Mastic with higher filler concentrations showed
the highest level of influence by the changing the
interfacial properties.
In addition to this, the micro-mechanical finite
element model showed the influence of varying
interfacial properties on the resulting behaviour:
• For weak interfacial properties the effect of the
shape of the particles was dominant,
• For intermediate spring constant and a strong
dashpot viscosity the effect of the surface area is
dominant at the end of relaxation period while at
the end of the loading the effect of the filler shape
Fig. 25 Displacement of big angular sample for different is more dominant.
interfacial properties
The above conclusions are certainly dependent on
the test conditions chosen, limitations of the material
interfacial properties. Weak spring and dashpot con- selection and the FE parameter and mesh choices.
stant (case 1) gives the highest displacement at both From the research described in this paper, however, it
end of unloading and recovery. Intermediate spring has become very clear that mastic modification has to
and weak dashpot constant (case 3) gives the second take into careful consideration the influence of the
highest value of displacement at end of unloading, filler–bitumen interaction, especially with the pre-
however it gives the same permanent deformation at sence of modifiers. Accounting for an influenced layer
the end of recovery as same as intermediate spring around the fillers and being aware of the effect of
constant and strong dashpot constant (case 4). shape and size of the fillers and its influence on the
overall response can thus lead to more precise mastic
design.
6 Conclusions and recommendations
Acknowledgments Authors will specially like to thank Dr.
Denis Jelagin and Mr. Måns Collin for their ideas and help in
In this research the influence of filler–bitumen inter- development of this work. This research was financially
action on the rheology of asphalt mastic was inves- supported by The AkzoNobel and Swedish road administration.
tigated, placing special focus on shape and size of
filler. For this, both experimental and numerical
methods were utilized. The following conclusions References
were drawn from the investigation:
1. Wojciech G, Jaroslaw W (2008) The structure of mineral
• Measuring the viscosity of different asphalt mas- fillers and their stiffening properties in filler–bitumen
tics at 100 C showed that the effect of the additive mastics. Mater Struct 41:793–804
Materials and Structures

2. Ilan I, Joseph C (1977) Effect of the filler on aggregate- 10. Rigden PJ (1947) The use of fillers in bituminous road
bitumen adhesion properties in bituminous mixtures. Assoc surfacings. A study of filler-binder systems in relation to
Asph Paving Technol 46:228–258 filler characteristics. J Soc Chem Ind 66:299–309
3. Anderson UD, Bahia HU, Dongre R (1992) Rheological 11. Buttlar WG, Bozkurt D, Al-Khateeb GG, Waldhoff AS
properties of mineral filler–asphalt mastics and its impor- (1999) Understanding asphalt mastic behavior through mi-
tance to pavment performance. In: Meininger RC (ed) cromechanics. J Transp Res Rec 1681:157–169
Aggregates and mineral filler on asphalt mixtures perfor- 12. Shashidhar N, Shenoy A (2002) On using micromechanical
mance. ASTM, Ann Arbor models to describe dynamic mechanical behavior of asphalt
4. Anderson DA, Goetz WH (1973) Mechanical behavior and mastics. Mech Mater 34:657–669
reinforcement of mineral filler-asphalt mixtures. Technical paper, 13. Frankel NA, Acrivos A (1967) On the viscosity of a con-
publication FHWA/IN/JHRP-73/05. Joint Highway Research centrated suspension of solid spheres. Chem Eng Sci
Project Indiana Department of Transportation and Purdue Uni- 22:847–853
versity, West Lafayette. doi: 10.5703/1288284313845 14. Coussot P (2005) Rheometry of pastes, suspensions, and
5. Lesueur D (2009) The colloidal structure of bitumen: con- granular materials. Wiley, New Jersy
sequences on the rheology and on the mechanisms of bitu- 15. Rediset WMX (2013) When improved performance and
men modification. Adv Colloid Interface Sci 145:42–82 environment go hand in hand. AkzoNobel. http://www.
6. Faheem AF, Bahia HU (2010) Modelling of asphalt mastic akzonobel.com/innovation/our_innovations/more_innovations/
in terms of filler–bitumen interaction. Road Mater Pave- rediset/. Accessed June 2013
ment Des 11:281–303 16. Hesami E, Birgisson B, Kringos N (2014) A new protocol
7. Hesami E, Jelagin D, Kringos N, Birgisson B (2012) An for measuring bituminous mastic viscosity as a function of
empirical framework for determining asphalt mastic vis- its filler concentration. J Road Mater Pavement Des. doi:10.
cosity as a function of mineral filler concentration. J Constr 1080/14680629.2013.876926
Build Mater 35:23–29 17. Dai Q, You Z (2008) Micromechanical finite element
8. Hesami B, Bidewell N, Birgisson B, Kringos N (2013) framework for predicting viscoelastic properties of asphalt
Evaluation of environmental susceptibility of bituminous mixtures. Mater Struct 41:1025–1037
mastic viscosity as a function of mineral and biomass fillers. 18. Badami JV, Greenfield ML (2011) Maxwell model analysis
Transp Res Rec J Transp Res Board 2371:23–31 of bitumen rheological data. J Mater Civ Eng 23:1387–1395
9. Tan Y, Guo M (2013) Interfacial thickness and interaction
between asphalt and mineral fillers. Mater Struct 3:1–10