See discussions, stats, and author profiles for this publication at: [Link]

net/publication/337068917

Comparison of various rutting parameters and modelling of creep and

recovery behaviour of high modulus bituminous binders

Article in International Journal of Pavement Research and Technology · November 2019

DOI: 10.1007/s42947-019-0077-1

CITATIONS READS

17 592

4 authors, including:

Arunkumar Goli Muppireddy Amaranatha Reddy

Rowan University Indian Institute of Technology Kharagpur
26 PUBLICATIONS 79 CITATIONS 31 PUBLICATIONS 538 CITATIONS

SEE PROFILE SEE PROFILE

Braj Bhushan Pandey

Indian Institute of Technology,Kharagpur,India
12 PUBLICATIONS 64 CITATIONS

SEE PROFILE

All content following this page was uploaded by Arunkumar Goli on 10 November 2019.

The user has requested enhancement of the downloaded file.

 Chinese Society of Pavement Engineering
°Cd
International Journal of
Pavement Research and Technology
Journal homepage: [Link]/42947

Comparison of various rutting parameters and modelling of creep and recovery

behaviour of high modulus bituminous binders
Arunkumar Goli*, Anilkumar Baditha, Amaranatha Reddy Muppireddy, Braj Bhushan Pandey
Department of Civil Engineering, Indian Institute of Technology Kharagpur, 721302, India

Received 15 August 2019; received in revised form 15 September 2019; accepted 25 September 2019

Abstract

This paper evaluates the rut resistance of various high modulus bituminous binders based on different rutting parameters in cluding G*/sinδ, shenoys’
rutting parameter, zero shear viscosity (ZSV), viscous component of creep stiffness (Gv) determined using burger’s model and superpave plus rutting
parameter i.e. non-recoverable creep compliance (Jnr). The main objective of the study is to develop high modulus bituminous binders and evaluate their
performance in terms of rutting. Also, the creep and recovery curves obtained from the MSCR test were modelled to understand the viscoelastic creep
behavior of the binders. From the results, it was found that ranking provided by G*/sinδ, shenoys’ rutting parameter and ZSV approaches were identical
whereas Jnr and Gv provided similar rankings. The rutting resistance improvement ratio analyzed for the binders showed the trend which is identi cal for
G*/sinδ, shenoys’ rutting parameter and ZSV. However, the ratio is identical for Gv and Jnr. Burger’s model was unable to capture the nonlinear viscoelastic
behavior of modified high modulus bituminous binders under creep and recovery loading. Weibull model was used to simulate the creep and recovery
behavior. It was found to be fit well with the experimental curves. The parameters of the Weibull model was able to explain the viscoelastic behavior of
binders under creep and recovery loading. The composite modification of binder enhanced both the rutting resistance and perce nt recovery of high modulus
bituminous binders.

Keywords: High modulus bituminous binders; Rutting; Composite modified bituminous binders; Burger’s model; Weibull model.

1. Introduction design and construction. The key element in the production of

HiMA mixes is the usage of stiff bitumen of penetration 10/20 or
High modulus asphalt (HiMA) mixtures have become one of the 15/25 [5,6]. Stiffer binders for high modulus mixes can be
alternatives for sustainable pavements due to their reliable and produced by (i) adopting refining process to obtain hard grade
consistent performance. Such mixes were developed initially by bitumen from refineries without air blowing (ii) polymer
French engineers in early 1980’s and gained popularity all around modification, (iii) asphaltite modification, and (iv) air blowing the
the world [1,2]. Since last decade, it was observed that several softer grade bitumen. The mixes with above binders and additives
pavement research groups around the world were working on use improve the performance and also reduce the thickness of
of high modulus asphalt (HiMA) mixes for base course with a thin bituminous pavement for given traffic [3,8,9].
wearing course on it. Many countries such as South Africa, Bitumen plays a prominent role in the performance of the
Australia, a few European countries, USA, China experienced the pavement throughout the design life. Research studies claimed that
benefits of the HiMA technology [3-5]. HiMA mixes displayed blowing process of bitumen leads to premature cracking of
good performance in areas with slow moving traffic such as urban pavements [6]. Thus, research has been directed towards
intersections, areas with restricted crust thicknesses, heavy development of hard bitumen through modification using additives
trafficked areas, airfield pavements, sea ports etc., [5]. An [1,4,9,10]. Commonly used modifiers include Gilsonite, Ethylene
advantage of these HiMA mixes is the thickness reduction in the Vinyl Acetate (EVA), Styrene Butadiene Styrene (SBS), crumb
rubber and hydrated lime etc. Modifier selection plays a crucial
role to enhance the properties of binder. As mentioned earlier in
* Corresponding author the literature, modified bitumen performed better than the hard
E-mail addresses: goliarunkumar109@[Link] (A. Goli); grade bitumen. The asphaltite or gilsonite was proved to produce
anilkumarbadita@[Link] (A. Baditha); manreddy@[Link] HiMA mixes. However, these mixes become brittle at intermediate
(A. R. Muppireddy); bbpandey40@[Link] (B. B. Pandey). and lower temperatures. Hence the current study emphases on
Revised version of a paper presented at 15th World Conference on composite modified binders to improve rutting performance of
Transport Research (WCTR), Bombay, Mumbai, India, 26-31 May
HiMA mixes.
2019.

ISSN: 1996-6814 DOI: [Link]

Chinese Society of Pavement Engineering. Production and hosting by Springer Nature
 A. Goli et al. / International Journal of Pavement Research and Technology 12 (2019) 648-658 649

Permanent deformation is one of the major distress in flexible 3. Background of various rutting parameters
pavements and is observed as a continuous depression along the
wheel path [11]. The effect of rutting on the performance of 3.1. Superpave rutting parameter, G*/sinδ
pavement is significant at higher temperatures compared to service
temperatures. Stiffer binder enhances the rutting resistance of The energy dissipated per cycle of loading causes rutting [12].
bituminous mixes, especially mixes prepared with polymer Strain produced due to traffic loading is recovered partly by elastic
modified binders [12]. Several researchers proposed parameters component and partly dissipated by the viscous component. Using
that are capable of explaining anti-rutting potential of bituminous the dissipated energy concept, G*/sinδ was specified as a
binders including SHRP rutting parameter (G*/sinδ), zero shear superpave rutting parameter and is shown in Eq. (1).
viscosity (ZSV), shenoys’ rutting parameter (G*/(1-
2 1
(1/tanδ*sinδ))), viscous component of creep stiffness (Gv) and ∆𝑈 = 𝜋𝜏𝑚𝑎𝑥
(𝐺 ∗ /𝑠𝑖𝑛𝛿)
(1)
non-recoverable creep compliance (Jnr) [13-16]. This paper
compares the improvement in the rut resistance through different where, ∆U is energy dissipated per cycle, τ is the shear stress, G*
parameters for the composite modified binders that can be used for is the complex shear modulus and δ is the phase angle.
HiMA binders. The parameter, G*/sinδ is inversely proportional to ∆U. A
Laboratory testing and evaluation of binders is limited and is not higher G*/sinδ improves the rut resistance by reducing the energy
possible in all cases. Selection of binders for pavement dissipated from the material. Hence, the superpave research
construction requires well understanding on their behaviour. This program considered G*/sinδ as a specification parameter to
requires well established laboratory testing as field testing and evaluate rut resistance.
evaluation requires huge financial support. However, advanced
testing facilities are not well established in India. This brings the 3.2. Shenoy rutting parameter, G*/(1-(1/tanδ*sinδ))
utility of modelling binders and characterizing their viscoelastic
properties. Hence, many researchers studied on the mechanical Shenoy proposed a parameter G*/(1-(1/tanδ*sinδ)) to refine the
response of the bituminous binders through modelling [17-20]. existing superpave rutting parameter, G*/sinδ [25]. According to
There exists the mathematical models (Christensen-Anderson Shenoy, the non-recovered strain will be minimized if the term
model; Christensen-Anderson-Marasteanu model, master curves G*/(1-(1/tanδ*sinδ)) was maximized. This parameter includes the
etc.,), empirical models (relating the performance with the damping coefficient, tanδ which was shown in Eq. (2).
volumetrics and material properties) and analytical models 100𝜎 1
(Maxwell model, kelvin-voigt model, generalised models, %𝛾𝑢𝑛𝑟 = |𝐺 ∗ |
(1 − ) (2)
𝑡𝑎𝑛𝛿.𝑠𝑖𝑛𝛿
Burger’s model, Dibenedetto-Neifar (DBN) model etc.,) [21]. In
contrast to these mathematical and empirical models, these where, γunr is the non-recovered strain percentage, σ is the applied
analytical models could explain the viscoelastic behaviour of creep stress, G* is the complex shear modulus and δ is the phase
bituminous material physically using springs and dashpots [21]. angle.
The response of these springs and dashpots arranged in different At high temperatures, tanδ captures the rutting resistance of
forms subjected under loading and relaxation period were better bituminous binders better than G*/sinδ as sinδ becomes close to 1
explained. The creep and recovery loading approach and at higher temperatures. This approach was found most suitable for
modelling the behaviour of bitumen under the same was well modified binders too as it measures the non-recoverable properties
accepted around the world [17,20,22,23]. Among these analytical of binder through oscillation test.
models, Burger’s model was used mostly due to its simplicity and
the parameters has physical meaning [22]. 3.3. Zero shear viscosity, ZSV

The modification of bituminous binders introduce more elastic

2. Research motivation properties which could not be captured by G*/sinδ. It led to
development of zero shear viscosity. Zero shear viscosity
resembles the viscosity of bitumen at very lower frequencies and
The current study discusses about the HiMA binders which also was found to have a better correlation with rutting of modified
includes polymer modified bitumen. HiMA binders produced binders. Modified binders behave as pseudo plastic fluids i.e. with
using polymers comprises of higher polymer content which is increase in shear rate, the viscosity of these binders’ decreases.
uncommon. Their rheological behaviour has to be understood well Initially, complex viscosity, η* was determined at various
as they behave differently from neat binders and are complex in frequencies and the data was extrapolated to determine the
nature to model using Burger’s model. Modelling of these binders viscosity at lower frequencies at which the binder will be in steady
in non-linear viscoelastic domain may require more complicated state. The simplified cross model adopted by Biro [26] which is
set of Maxwell and kelvin-voigt models. Weibull distribution shown in Eq. (3) was used to estimate the ZSV.
function can be to represent the viscoelastic behaviour of 𝜂𝑜
polymeric materials [23]. It was found to be applicable for 𝜂∗ = (3)
1+𝐾𝜔𝑚
bituminous binders as it explained the creep and recovery
behaviour especially for modified binders [24]. Hence the main where, η* is the complex shear viscosity measured at frequency of
objective of the present study is to compare the rutting resistance ‘ω’. ηo is the zero shear viscosity. K and m are the coefficients.
of the composite modified binders and model their rheological
behaviour. Both Burger’s model and Weibull model were used to 3.4. Non-recoverable creep compliance, Jnr
model the viscoelastic behaviour of bituminous binder under and
creep and recovery loading. In this paper, the terms bitumen, Creep and recovery test aids in better understanding the
binders and asphalt binders were used mutually. rheological behaviour of bituminous binders both in linear and
650 A. Goli et al. / International Journal of Pavement Research and Technology 12 (2019) 648-658

non-linear domains. The delayed elasticity which has a prominent where, 𝜂𝑀 , 𝐸𝑀 , 𝜂𝐾 and 𝐸𝐾 are the model constants. 𝜂𝑀 represents
role in modified binders can be captured by creep and recovery the viscous flow behaviour of the bitumen. 𝜎 is the applied stress
tests. Multiple Stress Creep and Recovery (MSCR) test was and 𝜀(𝑡) is the strain response of bitumen at time t.
formulated to determine the bituminous binder’s resistance To overcome the uncertainty on the rutting parameter, G*/sinδ,
towards rutting [6,15]. Jnr is used for binder selection and rank the Bahia [14] introduced a parameter, viscous component of creep
binders based on rut resistance. The test is carried out at two stress stiffness (Gv). This parameter is based on the creep and recovery
levels, 0.1 kPa and 3.2 kPa with a creep loading of 1 s and rest test conducted by dynamic shear rheometer. The viscous
period of 9 s. At each stress level, 10 creep and recovery cycles component of creep stiffness is derived from the burger model and
will be performed and the average will be considered as given in is calculated using Eq. (8).
Eqs. (4) and (5). 𝜂𝑀
𝐺𝑣 = (8)
𝑆𝑈𝑀(𝐽𝑛𝑟 (𝜎,𝑁)) 𝑡
𝐽𝑛𝑟 = (4)
10 where, Gv = Viscous component of creep stiffness; ηM = Viscosity
where, coefficient of maxwell model and t = time at the end of creep
𝜀𝑁 period.
𝐽𝑛𝑟 (𝜎, 𝑁) = (5) The resistance to deformation depends upon the elastic and
𝜎
viscous components of binder. The viscous component should be
and Jnr is the Non-Recoverable Creep Compliance, σ is the applied minimum for reducing dissipation energy due to loading. This
stress, N =Number of cycles, 1-10, ε is the strain due to applied dissipation energy should be minimum and is inversely
stress. proportional to the viscosity coefficient of Maxwell model.
Therefore, the viscous component of creep stiffness should be
3.5. Burger’s model and viscous component of creep stiffness, G v maximum for better rut resistance.

The creep and recovery response of the binders were modelled 3.6. Weibull model
to further understand the rheological behavior. Many models were
used to explain the binder behavior under flow such as generalized Fancey [23] used Weibull distribution function to represent the
Maxwell model, generalized Kevin-voigt model, Burger’s model, rheological behavior of viscoelastic materials. As bitumen is a
the six element solid model and the power law model. However, viscoelastic material, this model was adopted to simulate the creep
the model should be able to analyze the behavior and its parameters and recovery response in MSCR test. This model does not have
should have a physical meaning [27]. Burger’s model has four
any theoretical basis and was a stretched exponential function [23].
parameters to capture the mechanical response of the binder under
However, model provides better understanding through various
the creep and recovery loading approach. The Burger’s model is a
parameters which is given in the Eq. (9) and (10) respectively. Eq.
combination of both Maxwell and kelvin-voigt model. The
(9) represents the total strain of the material under creep loading
Burger’s model consists of spring as well as dashpot both in
and Eq. (10) represents the strain after the release of the load.
parallel and series and has four elements (EM, ηM, EK, ηK). Each
𝑡 𝛽𝑐
element represents a parameter in the model as shown in the Fig. (−( ) )
𝜀1 (𝑡) = 𝜀𝑖 + 𝜀𝑐 (1 − 𝑒 𝜂𝑐 ) (9)
1. EM and EK are the spring elements representing elastic behavior
and ηM and ηK are the dashpots representing the viscous behavior 𝑡
(−( )𝛽𝑟 )
𝜀2 (𝑡) = 𝜀𝑟 (𝑒 𝜂𝑟 ) + 𝜀𝑓 (10)
induced in the Fig. 1. Burger’s model combination provides a
viscoelastic strain response under a stress load. The strain response
where, εi is the initial instantaneous creep strain, εc is the final creep
of the bituminous material depends upon the elastic and damping
strain, ηc is the characteristic life parameter, βc is the shape
coefficients.
parameter from the creep test; εr is the initial viscoelastic strain, εf
The strain during the creep loading and recovery periods are
is the strain from viscous flow, βr is the shape parameter, ηr is the
given in Eq. (6) and (7) respectively.
characteristic life parameter after the removal of creep load and t
−𝐸𝐾 𝑡
𝜎 𝜎 𝜎 is the time.
𝜀(𝑡) = + 𝑡+ (1 − 𝑒 𝜂𝐾 ) (6)
𝐸𝑀 𝜂𝑀 𝐸𝐾
𝐸𝐾 𝑡 𝐸𝐾 𝑡 4. Materials and methods
𝜎 𝜎
𝜀(𝑡) = 𝑡+ (𝑒 𝜂𝐾 − 1) . 𝑒 𝜂𝐾 (7)
𝜂𝑀 𝐸𝐾
As per the Indian Road Congress guidelines, IRC: 37 (2012)
[28], VG40 should be used for the construction of flexible
pavements at high temperature regions or/and with the traffic level
more than 30 msa. Therefore, VG40 was used for the study which
was supplied by the haldia refinery. A hard pitch binder produced
through propane de-asphaltene method was also used which was
supplied by the same. Gilsonite, Styrene Butadiene Styrene (SBS)
and Ethyl Vinyl Acetate (EVA) were used as additives in this
study. Gilsonite is a natural asphaltite used to stiffen the bitumen
whereas EVA and SBS are plastomer and thermoplastic elastomer
respectively.
SBS is a di-block copolymer and has 33% styrene bound to
butadiene. It forms a matrix of polystyrene spherical domains
Fig. 1. Components of Burger’s Model. attached to polybutadiene and has a two-phase morphology. It
 A. Goli et al. / International Journal of Pavement Research and Technology 12 (2019) 648-658 651

behaves as a rigid structure at high temperatures due to the were the complex modulus (G*), phase angle (δ), and the complex
presence of polystyrene and acts as an elastic material at viscosity (η*).
intermediate and low temperatures because of polybutadiene MSCR test was performed at 60°C at both linear and non-linear
content. EVA is a plastomer which forms a rigid complex network. viscoelastic range i.e., different stress levels, 0.1 kPa and 3.2 kPa
The material is formed by the copolymerization of ethylene and as per test protocol. In addition, to evaluate the rutting resistance
vinyl acetate and is characterized using Melt flow index and vinyl of binders at various stress levels, test was also performed at 0.8,
content. The EVA used in the present study has 28% of vinyl 1.6, 6.4, 12.8 and 25.6 kPa respectively. A temperature of 60°C
content. Higher the vinyl content in EVA, lower the crystalline was selected as it could represent the high service pavement
nature of the material due to amorphic nature of the vinyl acetate temperature in India. At each stress level, 100 cycles were
content. performed as preconditioning cycles and thereafter 10 cycles were
applied. The average strain of ten cycles creep and recovery data
4.1. Sample preparation obtained at the stress level of 3.2 kPa from the MSCR test was used
to predict the non-linear viscoelastic behavior.
A laboratory fabricated temperature-speed controlled stirrer was Table 1
deployed for binder modification and the stirrer is shown as Fig. 2. Nomenclature given to all set of binders.
Gilsonite was added at 4 % and 8% into the neat binder VG40
whereas EVA was doped at 1, 3, 5% percentages. But the addition Binder ID Nomenclature
of gilsonite may lead to the brittleness of binders. So, to escalate VG40 Viscosity grade 40
the elastic properties i.e. recovery properties of binders, 1 and 2% PEN23 Special hard binder with penetration of 23
of SBS was added to 4% gilsonite. Gilsonite and EVA stiffen the E1 VG40+EVA 1%
bitumen and SBS improves the elastic properties of the bitumen. E3 VG40+EVA 3%
To produce HiMA bitumen with good recoverable properties, it E5 VG40+EVA 5%
was proposed to blend Gilsonite and SBS at certain proportions G2 VG40+Gilsonite 2%
with VG40 the neat binder. G4 VG40+Gilsonite 4%
The above mentioned modifiers were added to the neat binder G8 VG40+Gilsonite 8%
when the temperature of the neat binder was about 150 - 160°C. G4S1 VG40+ Gilsonite 4% + SBS 1%
After the addition, binders were blended for 60 min at a speed of G4S2 VG40+ Gilsonite 4% + SBS 2%
2500 rpm maintaining the temperature at 180°C whereas the
temperature was selected appropriately based on previous research
studies [29-31]. Nomenclature provided for all the bituminous
binders was given in Table 1. The prepared binders were tested for
basic conventional tests and the results are presented in the Table
2. It was observed with increase in dosages of modifiers, a decrease
in penetration and increase in softening point.

5. Experimental program

5.1. Oscillation test and MSCR test

Two neat and eight modified binders were tested to understand

the rheological behavior at high temperatures. Dynamic shear
rheometer which is shown in Fig.3 was used to conduct the
frequency-sweep test and MSCR test. These tests were conducted
in accordance with ASTM D7175-05 [32] and ASTM D7405-10a
[33] respectively. A spindle of diameter 25 mm with a gap of 1mm
between the top spindle and bottom plates was used. Frequency
sweep test was performed at temperatures varying from 52°C to
88°C and frequencies 10 to 0.1 Hz. The test was conducted within
the linear viscoelastic range and in controlled strain mode with a
strain of 5%. The parameters acquired from the oscillation test Fig. 2. Blending setup used in the study.

Table 2
Physical properties of bituminous binders.
Tests Binder
VG40 PEN23 G2 G4 G8 E1 E3 E5 G4S1 G4S2
Penetration, 0.1 mm 42 23 35 28 22 39 35 30 24 22
Softening point, °C 54 63 59 64 67 56 59 63 66 68
Viscosity, Pa.s @60 °C 389.5 1791.3 563.3 861.7 1220 487.6 481.1 461.5 1180 1110
652 A. Goli et al. / International Journal of Pavement Research and Technology 12 (2019) 648-658

could be noticed that a dosage of 8% of gilsonite and composite

modified bitumen with 4% gilsonite and 2% SBS (G4S2) resulted
a Jnr similar to PEN23. EVA polymer did not improve the rutting
resistance which is contradictory with other studies. This might be
due to type of polymer (28% vinyl content) which was used in the
current study.
Since the anti-rutting potential of a binder is significantly
dependent upon the response of the binder under creep and
recovery after the removal of load. The percent recovery is also
important to consider for evaluation. It was observed from the
percent recovery data that VG40 showed poor recoverable
properties. As discussed earlier G8 and G4S2 has Jnr similar to
PEN23. However, G4S2 lowered the Jnr and increased the
recovery properties of base binder thereby ensuring superior rut
resistance. This is due to the dispersed SBS polymer in the base
binder. It is well known that SBS disperses in the bitumen and
behaves softer at lower temperatures and stiffer at higher
temperatures. Thus, it imparts elasticity. Higher recoverable
properties are attained for G4S2 bitumen followed by G4S1. EVA
modified binders improved neither Jnr nor recoverable properties
and hence it was recommended not to use EVA polymer with 28%
vinyl content. However, further investigations are needed for more
assurance. The Jnr difference represents sensitivity to stresses and
it should not be greater than 0.75 or 75% to ensure the binder is
resistant to stress variations. From the Table 3, it could be noticed
that none of the modified binders has a Jnr difference higher than
Fig. 3. Dynamic shear rheometer used in the study. 75%. As per AASHTO MP19 [34], VG40 could be graded as
standard grade (S), which is suitable a traffic of less than 3 million
6. Results and discussion standard axles. Modification changed the grades of binders from
standard to extremely heavy grade (E). Addition of EVA was
6.1. Evaluation rutting resistance of binders found to have improved the grade from S to heavy (H) whereas
other additives are able to improve the binder grades from S to E
grade.
To evaluate the rutting resistance of the composite modified
The incremental stress creep and recovery test results are also
bituminous binders, test results of the MSCR test were presented
presented in the Table 3 and Table 4. It was observed that Jnr
in the Table 3. Jnr and percent recovery at all the stress levels were
increased with stress levels. It can be seen that the Jnr of base
presented. MSCR test involves the creep and recovery curve which
binder, VG40 is higher than modified binders’ especially
could directly measure the permanent strain occurred in the binder
composite modified binders even at higher stress levels. The rate
through the Jnr parameter. Analysis was based upon the 3.2 kPa
of increase in Jnr has decreased for composite modified binders at
stress level as it is in the non-linear viscoelastic range. It can be
high stress levels. As expected, percent recovery of binders was
seen that Jnr was decreasing due to the incorporation of additives.
found to be decreasing with increase in stress levels. It has to be
VG40 was found to have poor rutting resistance with a Jnr of 2.5
noted that the percent recovery was more sensitive to stress levels
and as expected additives enhanced the anti-rutting potential of the
than the Jnr. Even at high stress levels, composite modified binders
base binder. It could be clearly observed that gilsonite improved
are able to recover satisfactorily. Thus composite modified binders
the Jnr exceptionally than EVA polymer. However, the brittle
are less sensitive to stress levels than the other binders including
nature of the binder increased with incremental dosages of
gilsonite and EVA modified binders.
gilsonite which could be noticed from the percent recovery. It
Table 3
Non-recoverable Creep Compliance (Jnr) at various stress levels.
Binder ID Jnr at Stress level, kPa Jnr Difference
0.1 0.8 3.2 6.4 12.8 25.6
VG40 2.33 2.43 2.50 2.65 2.79 3.10 7.3
PEN23 0.48 0.50 0.54 0.59 0.61 0.64 12.5
E1 2.03 2.15 2.19 2.37 2.66 2.83 8.3
E3 1.80 2.01 2.14 2.30 2.52 2.68 19.6
E5 1.7 1.92 2.12 2.14 2.48 2.53 24.9
G2 1.43 1.50 1.55 1.63 1.75 2.15 8.3
G4 0.97 0.98 1.06 1.16 1.25 1.29 9.5
G8 0.49 0.50 0.53 0.65 0.74 0.79 8.0
G4S1 0.54 0.58 0.62 0.65 0.68 0.72 14.1
G4S2 0.41 0.45 0.50 0.56 0.59 0.62 22.0
 A. Goli et al. / International Journal of Pavement Research and Technology 12 (2019) 648-658 653

In addition to the MSCR test, the oscillation test was performed that neither any modified binder reached a G*/sinδ value of PEN23
and various rutting parameters were determined as mentioned whereas Jnr and Gv showed the opposite trend. Ranking given using
earlier in this paper. G*/sinδ and the Shenoy rutting parameter ZSV parameter was found to have slightly change and better
were determined from the data obtained from the oscillation test. compared to other two as it distinguished the rutting potential
ZSV was predicted using the simplified cross model as mentioned between G8 and G4S2. The parameters Gv and Jnr resulted same
in section 2.3 and Gv was derived from burger’s model. It was rankings with G4S2 as best binder towards anti-rutting potential
observed from Table 5 that all the parameters indicated an followed by G8 and PEN23 binders. Through the observations, it
improvement in the rutting resistance of the bituminous binders. was found that Gv is also a promising parameter to assess the
However, the evaluation using various parameters lead to different permanent deformation characteristics of bituminous binders.
rankings. All the five parameters were considered for comparison In order to estimate and compare the five test methods,
and ranking was given as 1 to 10 as shown in Table 6. Rank 1 improvement of rutting resistance was considered as rutting
represents the best binder and rank 10 represents the worst binder resistance improvement ratio and plotted for each set of modifiers.
against rutting. Moreover, G*/sinδ, G*/(1-(1/tanδ*sinδ)) and ZSV Fig. 4 presents the rutting resistant improvement ratio for gilsonite
accord similar rankings. PEN23 binder was found to have highest modified binders evaluated using the five parameters. The rutting
G*/sinδ and G*/(1-(1/tanδ*sinδ)) next to G8 binder. It was shown resistance increased with larger content of modifiers
Table 4
Percent Recovery at various stress levels.
Binder ID Percent recovery at stress level, kPa
0.1 0.8 3.2 6.4 12.8 25.6
VG40 0.63 0.41 0.27 0.20 0.09 0.03
PEN23 8.89 7.13 5.39 4.17 2.88 0.95
E1 1.68 1.25 0.52 0.31 0.1 0.087
E3 6.42 2.14 0.99 0.73 0.49 0.12
E5 11.91 6.13 1.65 1.04 0.53 0.39
G2 2.66 1.47 0.81 0.63 0.49 0.33
G4 7.00 3.73 2.11 1.59 1.33 1.07
G8 8.7 7.98 6.57 5.14 3.91 2.32
G4S1 19.48 16.13 13.69 10.47 6.91 5.12
G4S2 28.53 25.91 24.88 22.14 18.37 15.45
Table 5
Various rutting parameters of all binders.
Binder ID Rutting parameters
G*/Sinδ, Pa G*/(1-(1/tanδ*sinδ)), Pa Gv, Pa ZSV , Pa-s Jnr, kPa-1
VG40 3883.30 4088.98 4002.50 398.83 2.50
PEN23 23517.62 26003.66 18610.15 2586.19 0.54
G2 8615.42 9421.83 6411.13 1166.28 1.55
G4 12246.79 14007.66 9506.82 1745.93 1.06
G8 13684.17 16211.13 18623.21 1924.13 0.53
E1 4614.94 5053.27 4560.4 497.95 2.19
E3 4803.61 5165.31 4619.7 634 2.16
E5 4875.30 5177.64 4721.3 664.22 2.12
G4S1 11050.00 12963.53 15513.4 1598.67 0.62
G4S2 11710.00 14494.40 20840.3 2280.155 0.50
Table 6
Ranking of rutting parameters.
Binder ID Ranking of rutting parameters
G*/Sinδ, Pa G*/(1-(1/tanδ*sinδ)), Pa Gv, Pa ZSV , Pa-s Jnr, Pa
VG40 10 10 10 10 10
PEN23 1 1 3 1 3
G2 6 6 6 6 6
G4 3 4 5 4 5
G8 2 2 2 3 2
E1 9 9 9 9 9
E3 8 8 8 8 8
E5 7 7 7 7 7
G4S1 5 5 4 5 4
G4S2 4 3 1 2 1
654 A. Goli et al. / International Journal of Pavement Research and Technology 12 (2019) 648-658

which was evident in all cases. The trend of the ratio with increase of recovery and hence higher percent of strain will be recovered
in dosages was found to be identical for G*/sinδ, G*/(1- rapidly leading to lower non recoverable strain. Table 8 clearly
(1/tanδ*sinδ)), and ZSV whereas Gv and Jnr followed different shows the decrease in these values which demonstrates enhanced
trend [35]. It can be also seen that for lower percentages of rutting potential of binders. βr has decreased from 2.019 (VG40) to
gilsonite G*/sinδ, G*/(1-(1/tanδ*sinδ)), and ZSV showed a 0.934 (G4S2) and ηr from 2.493 (VG40) to 2.212 (G4S2). εi is the
significant increase in rutting resistance whereas Jnr and Gv did not initial creep strain and it was observed to be higher for
demonstrated the same. It can be interpreted from Table 3 that
gilsonite do not improve recoverable strain compared to SBS
polymer. However, all the parameters are in the same region at
higher percentages which makes it inconclusive. This shows the
non-reliability towards the G*/sinδ, G*/(1-(1/tanδ*sinδ)), and
ZSV.
The similar trend was also observed for EVA modified binders
as shown in the Fig. 5. Yet in case of EVA modified bitumen, at
lower percentages, rutting resistance improvement ratio was found
to be in approximate region whereas at higher percentages only
ZSV was found to follow different trend implying higher viscosity.
ZSV was similar to that of Gv and Jnr for the composite modified
binders as observed in Fig. 6. This clearly indicates that the three
parameters (G*/sinδ, G*/(1-(1/tanδ*sinδ)), and ZSV) were not
able to judge the rutting resistance of modified bituminous binders.

6.2. Modelling the creep and recovery behavior

Fig. 4. Rutting resistance improvement ratio of gilsonite modified
binders based on various rutting parameters.
Creep and recovery curves of different binders measured at 60°C
and stress level of 3.2 kPa were modelled. Only few plots are
shown in this paper for representation. Burger’s model was found
to be well fit for base binders as well as the gilsonite and EVA
modified binders. In the case of composite modified binders, there
is deviation from the experimental curves which was shown in the
Fig. 7. Some part of the recovery portion of the curves were
deviated which can be clearly observed from the same Figure.
Burger’s model was unable to model the modified bitumen’s
viscoelastic behavior. Table 6 presents the burger’s model
parameters - 𝜂𝑀 , 𝐸𝑀 , 𝜂𝐾 and 𝐸𝐾 . It was found that ηM, the
viscosity of bitumen was increasing due to additives. Higher value
of ηM represents that the binder is has more recoverable strain
during the creep period. VG40 binder has the least value and
highest value was found for G4S2 binder as expected which proves
that composite modification improves rutting characteristics better
than other binders. G8 binder had shown same value of ηM along
with the PEN23 binder. EVA modified binders were found to Fig. 5. Rutting resistance improvement ratio of EVA modified
increase the rutting potential but was found to be very not as much binders based on various rutting parameters.
of composite bituminous binders in any approach. The parameter,
𝐸𝑀 represents the modulus of the Maxwell spring and implies the
instantaneous recoverable strain after the creep loading. A lower
value of 𝐸𝑀 represents that the binder has good recoverable
properties. It could be observed the modified binders has low 𝐸𝑀
indicating improved instantaneous recoverable properties. 𝐸𝐾
describes about stiffness of the material and 𝜂𝐾 is related to the
viscosity of kelvin unit in burger’s model. However, these
parameter’s values are inappropriate and unable to explain the
binders behaviour. Overall, the burger’s model was proved to be
not an appropriate modelling technique to model modified
bituminous binders. Hence weibull distribution function was
adopted in the study to simulate the creep and recovery behavior.
The creep and recovery curves were simulated and the Burger’s
model parameters were reported in the Table 7. Fig. 8 represents
the measured and predicted behavior of binders. It is noticed that
the modelled data fitted well with the experimental data. βr and ηr
Fig. 6. Rutting resistance improvement ratio of composite
represents the rate of recovery of bitumen after the removal of
modified binders based on various rutting parameters.
creep load. A lower values of βr and ηr represents the increased rate
 A. Goli et al. / International Journal of Pavement Research and Technology 12 (2019) 648-658 655

modified binder with higher dosages of additives as well as for viscoelastic properties and enhances the life of bitumen. Therefore
PEN23 binder. εc is related to the final creep strain and high values with the above results, it could be aforementioned that weibull
indicate high resistant to stress. As expected, the final creep strain model was able to explain the non-linear strain response of the
values decreased for modified binders with 8940 for VG40 and modified bituminous binders using the model parameters and also
73800 for G4S2 binder. ηc and βc represents the characteristic life was found to be in high agreement with the experimental data
and shape parameter during the creep loading period. εr represents curves.
the viscoelastic strain during recovery period and a lower value Overall, the composite modified binders were found to have
indicates that the binder is more viscous and less recoverable with lower creep strain due to higher stiffness and rapid recovery of
time. A higher value of εf indicates that the binder has high strain immediately after the removal of load. Also, as these
permanent strain at the end of recovery period. It could be composite binders were stiff enough to produce lesser strains, will
observed from the Table 8, that these parameters are in agreement immediately recovery whatever the strain accumulated due to
to that statement that the modified binder aided in improving the creep load.

(a) (b)

(c) (d)

(e) (f)

Fig. 7. Creep and recovery curves of binders (Burger Model fit); (a) VG 40 binder; (b) PEN23 binder; (c) E5 binder; (d) G4 binder; (e)
G4S1 binder; and (f) G4S2 binder.
656 A. Goli et al. / International Journal of Pavement Research and Technology 12 (2019) 648-658

(a) (b)

(c) (d)

(e) (f)

Fig. 8. Creep and recovery curves of binders (Weibull Model); (a) VG 40 binder; (b) PEN23 binder; (c) E5 binder; (d) G4 binder; (e) G4S1
binder; and (f) G4S2 binder.

Table 7
Burger Model Parameters.
Binder ID Model parameters
EM ηM EK ηK SSE
VG40 4.95E+17 4.00E+03 8.12E+05 1.01E+06 8.34E-04
PEN23 1.86E+07 1.86E+04 1.16E+05 2.68E+05 3.37E-05
E1 1.70E+14 4.56E+03 1.73E+03 2.24E+07 5.32E-04
E3 2.15E+13 4.62E+03 2.34E+05 2.87E+10 8.10E-04
E5 8.05E+11 4.72E+03 2.43E+05 3.73E+05 6.83E-04
G2 1.88E+13 6.41E+03 3.61E+04 2.72E+10 1.29E-03
G4 2.78E+09 9.51E+03 5.10E+04 8.18E+05 8.06E-04
G8 1.09E+06 1.86E+04 2.61E+05 2.51E+05 9.75E-05
G4S1 1.16E+06 1.55E+04 1.24E+05 8.97E+04 1.79E-03
G4S2 9.05E+05 2.08E+04 1.66E+05 3.06E+04 5.40E-03
 A. Goli et al. / International Journal of Pavement Research and Technology 12 (2019) 648-658 657

Table 8
Weibull Model Parameters.
Binder ID Model parameters
εi εc ηc βc εr εf ηr βr SSE
VG40 0 8940 1.81 1.62E-04 3.17E-03 0.800 2.493 2.019 0.000784
PEN23 0.00011 4150 1.76 7.73E-05 1.71E-02 0.174 5.314 3.963 0.000086
E1 0 51000 4.71 6.43E-05 8.10E-02 0.702 2.078 3.038 0.000369
E3 0 95100 10 7.29E-05 4.20E-02 0.690 1.356 2.132 0.000593
E5 0 86900 8.61 6.77E-05 7.66E-02 0.680 1.219 3.138 0.000747
G2 0 60500 9.00 7.33E-05 6.74E-03 0.495 4.800 2.164 0.000125
G4 0 63100 463 2.58E-03 1.29E-02 0.339 2.532 1.648 0.000093
G8 0.00043 46200 471 1.86E-03 1.46E-02 0.171 2.502 1.340 0.000028
G4S1 0.00155 44600 288 1.49E-03 3.85E-02 0.198 1.793 1.412 0.000175
G4S2 0.00303 73800 206 7.37E-01 9.52E-02 0.149 1.212 0.934 0.000407

7. Conclusions recoverable properties to ensure better rutting

performance.
The current study had focused on the improvement of rut 9. Burger’s model was found to be incapable to explain
resistance evaluated using various parameters for the modified the non-linear viscoelastic strain response of the
high modulus bituminous binders. These parameters include modified binders due to creep and recovery. These
superpave rutting parameter (G*/sinδ), shenoy rutting parameter Parameters for composite modified high modulus
(G*/1-(1/tanδ*sinδ)), zero shear viscosity (ZSV), viscous bituminous binders are inappropriate and were not able
component of creep stiffness (Gv) and Non-recoverable creep to explain their behavior.
compliance (Jnr). This study also tried to simulate the viscoelastic 10. Weibull model captured and modelled the strain
behavior of modified binder using both Burger’s model and response of bituminous binders with more appropriate
Weibull distribution function to understand the rut resistance in a fit with the experimental data and the parameters were
much appropriate manner. Various conclusions of the study were better in explaining the rut resistance.
as follows: 11. The model’s parameters are able to explain the
1. The superpave rutting parameter, shenoy rutting improvement in characteristic life, creep strain and
parameter and ZSV were found to follow same trend permanent strain of composite high modulus
whereas Gv and Jnr were following the same trend in bituminous binders.
evaluating the rut resistance. However, depending
upon the type of additive, rut resistance improvement References
ratio based on ZSV was different from G*/sinδ and
shenoy rutting parameter. [1] J. P. Serfass, A. Bauduin, J. F. Garnier, High modulus
2. Based on the G*/sinδ, G*/1-(1/tanδ*sinδ) and ZSV, asphalt mixes-laboratory evaluation, practical aspects and
PEN23 binder was found to be more rut resistant than structural design, 7th International Conference on Asphalt
the composite modified binder which was later found Pavements, Nottingham, United Kingdom, Vol. 1, 1992.
that composite modified binder has the highest rut [2] J. L. Delorme, C. De la Roche, L. Wendling, LPC bituminous
resistance based on Jnr and Gv. mixtures design guide, Laboratoire Central des Ponts et
3. Stress levels affected the non-recoverable creep Chaussées, Paris, France, 2007.
compliance (Jnr) and percent recovery of binders. The [3] M. E. Nunn, T. Smith, Road trials of high modulus base for
Jnr was found to be increasing with increase in stress heavily trafficked roads. Report number 231. Thomas
level whereas the percent recovery was decreasing Telford, Berkshire, UK,1997.
with increase in stress level. [4] J. J. Komba, B. M. J. Verhaeghe, J. S. O’Connell, J. K.
4. The stress sensitivity of composite modified binders is Anochie-Boateng, W. Nortje, Evaluation of the use of
less than the base binder as well as EVA modified polymer modified bitumen in the production of high modulus
binders in case of both Jnr and percent recovery. Stress asphalt for heavily-trafficked roads, 7th Africa
levels has more impact on percent recovery than the Transportation Technology Transfer Conference, Bulawayo,
Jnr. Zimbabwe, 2015.
5. In most of the cases, the resulting rutting resistance [5] L. Petho, A. Beecroft, J. Griffin, E. Denneman, High
improvement ratio found to be different for G*/sinδ, Modulus High Fatigue Resistance Asphalt (EME 2)
G*/1-(1/tanδ*sinδ) and ZSV while the ratio is same for Technology Transfer. Technical Report Number. AP-
Gv and Jnr. T283/14. Austroads, Australia, 2014.
6. The Non-recoverable creep compliance, Jnr was found [6] J. F. Corte, Development and uses of hard-grade asphalt and
to be effective in capturing the rut resistance of even of high-modulus asphalt mixes in France, Transp. Res.
high modulus bituminous binders. Circul. 503 (2001) 12-31.
7. Through all the rutting parameters, it was observed that [7] J. D'Angelo, R. Kluttz, R. N. Dongre, K. Stephens, L.
EVA with 28% vinyl content was not able to improve Zanzotto, Revision of the superpave high temperature binder
the rutting potential of the base binder. specification: the multiple stress creep recovery test (with
8. The composite modified high modulus bituminous
binders has a remarkable stiffness along with
658 A. Goli et al. / International Journal of Pavement Research and Technology 12 (2019) 648-658

discussion), J. Assoc. Asphalt Paving Technol. 76 (2007) [22] N. Saboo, P. Kumar, A study on creep and recovery behavior
123-162. of asphalt binders, Constr. Buil. Mater. 9 (2015) 632-640.
[8] Y. Brosseaud, F. Farcas, V. Mouillet, High modulus asphalt [23] K. S. Fancey, A Latch-Based Weibull Model for Polymeric
mixes with high rate of RA: what does it happen?, In Congrès Creep and Recovery, J. Polymer Eng. 21 (6) (2001) 489-510.
Eurobitume Eurasphalt, Turkey, 2012, pp. 1-13. [24] X. Liu, S. Zhang, X. Xu, Z. Zhang, L. Zhou, G. Zhang. Study
[9] M. I. G. Hernández, High Modulus Asphalt Concrete: a Long on the creep and recovery behaviors of UHMWPE/CNTs
Life Asphalt Pavement, J. Civ. Environ. Eng. 5 (5) (2015) 1. composite fiber, Fibers Polymers 14 (10) (2013) 1635-1640.
[10] G. Caroff, Investigation of rutting of asphalt surface layers: [25] A. Shenoy, Refinement of the Superpave specification
influence of binder and axle loading configuration, Transp. parameter for performance grading of asphalt, J. Transp.
Res. Rec. 1436 (1994) 28-37. Eng. 127 (5) (2001) 357–362.
[11] A. Golalipour, E. Jamshidi, Y. Niazi, Z. Afsharikia, M. [26] S. Biro, T. Gandhi, S. Amirkhanim, Determination of zero
Khadem, Effect of aggregate gradation on rutting of asphalt shear viscosity of warm asphalt binder, Constr. Buil. Mater.
pavements, Proc. Soci. Behav. Sci. 53 (2012) 440-449. 23 (2009) 2080-2086.
[12] H. Geng, C. S. Clopotel, H. U. Bahia, Effects of high [27] Y. Liu, Z. You, Determining burger's model parameters of
modulus asphalt binders on performance of typical asphalt asphalt materials using creep-recovery testing data, Pave.
pavement structures, Constr. Buil. Mater. 44 (2013) 207-213. Mater.: Modeling, testing, and performance, Minneapolis,
[13] D. A. Anderson, D. W. Christensen, H. U. Bahia, R. Dongre, Minnesota, USA, 2009, pp. 26-36.
M. G. Sharma, C. E. Antle, J. Button, Binder [28] Indian Roads Congress, Guidelines for the Design of
characterization and evaluation, volume 3: Physical Flexible Pavements, 3rd Revision. IRC: 37. New Delhi,
characterization. Strategic Highway Research Program, India, 2012.
National Research Council. Report No. SHRP-A-369. [29] B. V. Kök, M. Yilmaz, M. Guler, Evaluation of high
Washington DC, USA, 1994. temperature performance of SBS+ Gilsonite modified
[14] D. Anderson, Y. Le Hir, J. P. Planche, D. Martin, A. Shenoy, binder, Fuel 90 (10) (2011) 3093-3099.
Zero shear viscosity of asphalt binders, Transp. Res. Rec. [30] M. Liang, Y. Hu, X. J. Kong, W. Y. Fan, X. Xin, H. Luo,
1810 (2002) 54-62. Effects of SBS Configuration on Performance of High
[15] H. U. Bahia, H. Zhai, M. Zeng, Y. Hu, P. Turner, Modulus Bitumen Based on Dynamic Mechanical Analysis,
Development of binder specification parameters based on Kemija u industriji 65 (7-8) (2016) 379-384.
characterization of damage behavior, J. Assoc. Asphalt [31] J. Liu, P. Li, Experimental study on gilsonite-modified
Paving Technol. 70 (2001) 442–470. asphalt, Airfield and Highway Pavements: Efficient
[16] J. A. D'Angelo, The relationship of the MSCR test to rutting, Pavements Supporting Transportation's Future, Bellevue,
Road Mater. Pave. Des. 10 (sup1) (2009) 61-80. Washington, USA, 2008, pp. 222-228.
[17] C. Celauro, C. Fecarotti, A. Pirrotta, A. C. Collop, [32] American Society for Testing and Materials International,
Experimental validation of a fractional model for Standard Test Method for Determining the Rheological
creep/recovery testing of asphalt mixtures, Constr. Buil. Properties of Asphalt Binder Using a Dynamic Shear
Mater. 36 (2012) 458-466. Rheometer. ASTM D7175-05. West Conshohocken, PA,
[18] R. Delgadillo, H. U. Bahia, R. Lakes, A nonlinear USA, 2005.
constitutive relationship for asphalt binders, Mater. Struct. [33] American Society for Testing and Materials International,
45 (3) (2012) 457-473. Standard Test Method for Multiple Stress Creep and
[19] R. Delgadillo, H. U. Bahia, The relationship between Recovery (MSCR) of Asphalt Binder Using a Dynamic
nonlinearity of asphalt binders and asphalt mixture Shear Rheometer. ASTM D7405-10a. West Conshohocken,
permanent deformation, Road Mater. Pave. Des. 11 (3) PA, USA, 2011.
(2010) 653-680. [34] American Association of State Highway and Transportation
[20] F. Merusi, Delayed mechanical response in modified asphalt Officials, Standard specification for performance-graded
binders. Characteristics, modeling and engineering asphalt binder using Multiple Stress Creep Recovery
implications, Road Mater. Pave. Des. 13 (sup1) (2012) 321- (MSCR) Test. AASHTO MP 19. Washington, DC, USA,
345. 2010.
[21] L. Shan, Y. Tan, H. Zhang, Y. Xu, Analysis of linear [35] H. Pouria, R. Mohammad, F. M. Nejad, Comparing different
viscoelastic response function model for asphalt binders, J. rutting specification parameters using high temperature
Mater. Civ. Eng. 28 (6) (2016) 04016010. characteristics of rubber-modified asphalt binders, Road
Mater. Pave. Des. 16 (4) (2015) 751-766.

View publication stats