materials

Article
Research of Method for Solving Relaxation Modulus
Based on Three-Point Bending Creep Test
Yazhen Sun 1, *, Zhangyi Gu 2 , Jinchang Wang 3 and Xuezhong Yuan 4, *
1 School of Transportation Engineering, Shenyang Jianzhu University, Shenyang 110168, China
2 College of Civil Engineering and Architecture, Zhejiang University, Hangzhou 310058, China;
guzhangyi315@163.com
3 Institute of Transportation Engineering, Zhejiang University, Hangzhou 310058, China; wjc501@zju.edu.cn
4 School of Science, Shenyang Jianzhu University, Shenyang 110168, China
* Correspondence: syz16888@126.com (Y.S.); yuanxuezhong1@163.com (X.Y.);
Tel.: +86-139-4002-1867 (Y.S.); +86-136-1404-0298 (X.Y.)




Received: 20 May 2019; Accepted: 20 June 2019; Published: 24 June 2019


Abstract: A method was developed for solving the relaxation modulus of high viscosity asphalt
sand (HVAS) based on the three-point bending creep test, and was verified by comparison with
experimental results. In this method, firstly, a transcendental equation was obtained by the convolution,
and then equations were obtained by Taylor’s formula, which were solved by Mathmatica to obtain
the relaxation modulus by Newton’s method. Subsequently, the laboratory investigations of the
viscoelastic parameters of the Burgers model for the HVAS by three-point bending creep tests were
carried out. In addition, the method was verified by comparing the relaxation moduli with the indoor
relaxation experiments. Results showed that the numerical calculation and the test data were in good
agreement, and the relaxation characteristics of the HVAS were reflected more accurately. The method
can be used to study the relaxation characteristics of the asphalt mixtures effectively. In addition, this
study provides a research basis for road crack prevention.

Keywords: HVAS; transcendental equation; relaxation modulus; creep compliance; convolution;

Newton’s method; Boltzmann superposition principle

1. Introduction
The pavement performance and durability of the asphalt mixture road will be seriously affected
after pavement cracking [1–3]. The reason is that the road spans a vast area, where the range of
temperature could be very large due to environmental changes, so the inconsistency of deformation
between the surface and the base causes thermal stress [4–6]. If the relaxation ability is strong,
the thermal stress is released so that it does not exceed the allowable value. Relaxation modulus is an
important parameter for analyzing and evaluating relaxation ability [7]. Therefore, simple and accurate
calculation of relaxation modulus is an important basis for the study of viscoelastic parameters of the
asphalt mixtures, which provides the research basis for road crack prevention.
The relaxation modulus could be directly obtained from relaxation test [8–10]. However, constant
strain should be applied to the specimen instantaneously and remains unchanged during the
relaxation test, which poses a great challenge to instrument selection, test design and operation [11].
Higher hardness of asphalt mixture at low temperature leads to longer relaxation time, which also
increases the difficulty of the relaxation test [12]. Therefore, direct measurement of the relaxation
modulus by the relaxation test often results in large errors, and it is difficult to operate. By contrast,
a creep test is easier to be carried out, so solving E(t) by J(t) was one of the focuses of the researches [13–15].
In such case, apart from the relaxation tests, the interconversion of the creep compliance was
recommended [16].

Materials 2019, 12, 2021; doi:10.3390/ma12122021 www.mdpi.com/journal/materials

Materials 2019, 12, 2021 2 of 14

It is to be noted that in a rheological application, the convolution of J(t) and E(t) for the LVEs is
expressed according to the Boltzmann superposition integral [17] as
Z t
J (t − τ)E(τ)dτ = t (1)
0

where J(t) is the creep compliance function, E(t) is the relaxation modulus function and t is the time.
In the early application of the integral, the relation of the approximate J(t) and E(t) were usually
taken as reciprocal, being only applicable to weakly viscous materials instead of viscoelastic materials.
Lately, with the introducing of several variables [18], it is possible to derive an inverse reciprocal
relation, and only when t = 0 or t = ∞, the reciprocal relation applies [19], i.e.,
(
t = 0 E(0) = 1/J (0)
(2)
t = ∞ E(∞) = 1/J (∞)
With the increasing need of interconversion between the LVE functions of the HVAS, approaches
were proposed to solve the interconversion equation between J(t) and E(t). For example, a numerical
integration method for determining of E(t) was proposed according to the measured J(t) based on the
accurate approximate transformation between the volume J(t) and the bulk relaxation modulus [20];
a numerical interconversion based on the convolution integral, which relates E(t) to J(t), was made
to discrete the time domain into a finite number of sub domains, and an iterative expression was
obtained [21]. However, in the abovementioned methods, E(t) depended on the accuracy of the
numerical calculation; this led to large differences in the calculations with the test data when the
computational accuracy was poor. As noted by some authors [13,22,23], J(t) and E(t) were substituted
into integral relationship simplified as a matrix form, and the Dirac Delta function was introduced
to solve the creep compliance. With respect to the integrals expressed in the Prony series, a poorly
conditioned matrix equation increases the complexity of the calculation.
Many scholars have studied and some achievements have been made in determining the
relaxation modulus of the asphalt mixture by using the complex modulus. Zhao determined the
main curve of the storage moduli by using the test results of complex moduli of the asphalt mixture,
and converted the storage moduli into the relaxation moduli of asphalt mixture by using the collocation
method and related viscoelastic theoretical formula [24]. On the basis of the analysis of previous
findings, Liu gave the relationship between the complex moduli and the relaxation moduli [14].
However, in determining the complex moduli of asphalt mixtures, tests need to be carried out at
several temperatures. Moreover, at each temperature, the complex moduli of the asphalt mixtures
with different angular frequencies need to be measured to obtain the dynamic moduli and phase angle
of asphalt mixtures.
In this paper, a new scheme is adopted to solve the convolution formula. Previous studies have
used the same model to characterize the relaxation and creep of asphalt at the same time [19,25,26].
Common models, such as the Burgers model, Maxwell model and Kelvin model, are good at creep and
relaxation alone, but it is very difficult to take both into account [27,28]. Therefore, the advantages of
the Burgers model and Generalized Maxwell model (GMM) are integrated in their respective fields to
better characterize the properties of the HVAS. A more general formula not confined to the form of the
Prony series is introduced and the solution process is also simplified. First, J(t) was obtained based on
the three-point bending creep tests. Then, by substituting the parameters for the Burgers model into
the convolution integral expanded by the Taylor series, E(t) was derived from Wolfram Mathematica
8 using Newton’s method. For verifying, the relaxation tests were carried out and the E(t) obtained
was compared with that determined by the suggested approach. Moreover, with the application of
the time-temperature equivalence principle, the master curve of E(t) was obtained within a larger
temperature range.
This method is not only straightforward and accurate in the interconversion (without the
application of Laplace transform) between the compliance and the modulus functions, but can also
Materials 2019, 12, x FOR PEER REVIEW 3 of 14
Materials 2019, 12, 2021 3 of 14

avoid the high requirements for experiment equipment. Therefore, the method is recommended to
study the
avoid the relaxation characteristics
high requirements of the HVAS.
for experiment equipment. Therefore, the method is recommended to
study the relaxation characteristics of the HVAS.
2. Interconversion of Creep Compliance and Relaxation Modulus
2. Interconversion of Creep Compliance and Relaxation Modulus
2.1. Interconversion Equation
2.1. Interconversion Equation
When a constant stress is applied to the asphalt mixture, the development of the corresponding
strainWhen
can be divided into
a constant stress three stages, to
is applied i.e.,
thetheasphalt
instantaneous
mixture,elastic stage, the viscous
the development flow stage and
of the corresponding
the delayed
strain can be elastic
divided deformation
into three stages, stage. i.e.,
Thethecreep behaviour of
instantaneous the HVAS
elastic canviscous
stage, the be characterized
flow stagewith
and
the delayed
the four-parameters ( E1 , E2 , stage.
elastic deformation 1 , 2 )The
  of the Burgers
creep model,
behaviour as shown
of the HVAS incanFigure 1.
be characterized with the
four-parameters (E01 , E02 , η01 , J(t)
The creep compliance η02 ) for theBurgers
of the Burgersmodel,
model asof shown
the HVAS is written
in Figure 1. as:
The creep compliance J(t) for the Burgers model of the HVAS E
is written as:
1 t 1  2t
J (t )    (1  e E0 2 ) (3)
1 E1 t 1 1E2 − 02 t
J (t) = 0 + 0 + 0 (1 − e η2 ) (3)
E1 η1 E2
where 1 and 2 are the viscosity coefficients in the Burgers model, E1 and E2 are the elastic
moduliη0inand
where theηBurgers
0 are themodel.
viscosity coefficients in the Burgers model, E01 and E02 are the elastic moduli in
1 2
the Burgers model.

Figure 1. Burgers model.

Figure 1. Burgers model.
A reliable performance prediction model is conducive to the development of the relaxation
A reliable
behaviour [29]. performance prediction
To this end, the model
Generalized is conducive
Maxwell model (GMM) to the development
assembled by of the relaxation
parallel Maxwell
behaviour [29]. To this end, the Generalized Maxwell model (GMM) assembled by
elements representing the relaxation module is employed, as shown in Figure 2. The overall parallel Maxwell
elements
relaxationrepresenting
modulus is the relaxation module is employed, as shown in Figure 2. The overall relaxation
modulus is
n n t
− τt 
E(tE) =
t   EE 
X
e
i i e i i (4)
i=1i 1

where ii is
where is the
the i-th
i-th spring-dashpot
spring-dashpot element,
element, ii is
is the
the relaxation
relaxation time
time (the
(the higher
higher its
its value,
value, the
the longer
longer itit
takes for
takes forthe
thestrain/stress
strain/stress
to to relax)
relax) related
related to dashpot
to the the dashpot viscosity,
viscosity, Eimodulus
Ei is the is the modulus ofMaxwell
of the i-th the i-th
Maxwellassociated
element element associated to stiffness,
to the spring the springn isstiffness, n is of
the number the number of spring-dashpot
spring-dashpot elements
elements accounted for
accounted for and t is the time.
and t is the time.

Figure 2. Generalized Maxwell model (GMM).

Figure 2. Generalized Maxwell model (GMM).
Materials 2019, 12, 2021 4 of 14

Substituting Equation (3) and (4) into the convolution integral of Equation (1), we obtain

n E0
 
t
t−τ − η02 (t−τ) 
Z
− ττ  1 1
X
[ Ei e i ] 0 + 0 + 0 (1 − e 2 )dτ = t

 (5)
0 E1 η1 E2
i=1

2.2. Derivation of Relaxation Modulus

The target number of spring-dashpot elements n is predefined before solving. After the expansion
of the integral, Equation (5) can be re-rewritten in terms of Ei , τi , etc., as:
 n
P  n
P
− τt n n
Ei Ei
1 1
Ei τi − 1 1
Ei τi e η01 τi t − η01 τi +
P P
E01 + E02 E01 + E02
i +
i=1 i=1 i=1 i=1
n n
E0
2 n
(6)
Ei − τt τi η0 − 0t − τt τi ·η02
i τ2 1
Ei η0 −τ2E0 e η2 1
P P P
η01 e i
− E02 + E02 Ei e i · η2 −τ2 E02
0 =t
2 i 2
i=1 i=1 i=1

where Ei (i = 1 to 6) is the elastic modulus in GMM, τi (i = 1 to 6) is the relaxation time in GMM,

η01 and η02 are the viscosity coefficients in the Burgers model, E01 and E02 are the elastic moduli in the
Burgers model. Actually, this is a transcendental equation. Although the commands Solve and NSolve
in Mathematica can be applied to solve exponential and trigonometric equations in a limited way,
they are not designed to solve complicated transcendental equations. Fortunately, Equation (6) can be
solved by Mathematica using Taylor’s formula.

2.2.1. Taylor’s Formula

To simplify the calculations, multinomials obtained by Taylor’s formula were used in the alternative
method. The application of the multinomials in the study of the HVAS is a progress, especially in the
computer program field, which can be done with high precision and is an applicable approximate
method for the calculations of the complex functions. Taylor’s formula with a surplus item Peano
meets n-order differentiable at the point x0 [30]. In other words, this type of function, with a simple
form and widely applicable conditions, is very convenient to deal with some qualitative problems.
Due to Equation (6), n-order derivative exists at 0, Taylor’s formula with a surplus item Peano is used
to expand at 0, namely, the Maclaurin formula. In our study, there are 12 unknown parameters of
the generalized Maxwell model, so the equation is expanded to 12 polynomials about t and a Peano
remainder, and the required accuracy of the calculations can be acquired. We have:
 
(E1 τ1 + E2 τ2 + E3 τ3 + E4 τ4 + E5 τ5 + E6 τ6 ) E10 + E10 − η10 + 1
1 2 1
η0
 
E1 τ1 E2 τ2 E3 τ3 E4 τ4 E5 τ5 E6 τ6
− E20 η0 −E 0τ + η0 −E0 τ + η0 −E0 τ + η0 −E0 τ + η0 −E0 τ + η0 −E0 τ
2 2 2 1 2 2 2 2 2 3 2 2 4 2 2 5 2 2 6 (7)
+(− E1 τ1 +E2 τ2 +E3 τ3η+0 E4 τ4 +E5 τ5 +E6 τ6 + E1 τ1
η2 −E02 τ1
0 + E2 τ2
η2 −E02 τ2
0 + E3 τ3
η2 −E02 τ3
0 + E4 τ4
η02 −E02 τ4
1
5 τ5 E6 τ6
+ η0E−E 0 τ + η0 −E0 τ + E1 + E2 + E3 + E4 + E5 + E6 + 1)t + ··· + O[t]
12
= 0
2 2 5 2 2 6

where Ei (i = 1 to 6) is the elastic modulus in GMM, τi (i = 1 to 6) is the relaxation time in GMM, η01 and
η02 are the viscosity coefficients in the Burgers model, E01 and E02 are the elastic moduli in the Burgers
model and the ellipsis “···” represents a total of 11 items.

2.2.2. Solutions of Transcendental Equation

For this kind of problem, the command FindRoot in Mathematica can be used. The kernel
algorithm of FindRoot is the iterative Newton’s method [31]. In the calculation, the first or the first two
points (the initial guess) should be specified, for the best results, the initial guess should be as close to
the expected root as possible.
Materials 2019, 12, x FOR PEER REVIEW 5 of 14
Materials 2019, 12, x FOR PEER REVIEW 5 of 14

two points (the initial guess) should be specified, for the best results, the initial guess should be as
two points (the initial guess) should be specified, for the best results, the initial guess should 5beof as
close to 2019,
Materials the expected
12, 2021 root as possible. 14
close to the expected root as possible.
By default, 15 iterations are performed before FindRoot is aborted. The number of iterations is
By default, 15 iterations are performed before FindRoot is aborted. The number of iterations is
controlled by the Max Iterations, which can be used to increase the number of iterations to obtain
controlled by the15Max
By default, Iterations,
iterations which canbefore
are performed be used to increase
FindRoot the number
is aborted. of iterations
The number to obtain
of iterations is
more accurate values and to prevent early termination of operations before the desired results are
more accurate values and to prevent early termination of operations before the desired results are
controlled by the Max Iterations, which can be used to increase the number of iterations to obtain more
obtained.
obtained.
accurate values and to prevent early termination of operations before the desired results are obtained.
For example, to improve the accuracy, the FindRoot command in Mathematica can be written as
For
For example,
example, to toimprove
improvethe theaccuracy,
accuracy,thetheFindRoot
FindRootcommandcommand inin
Mathematica
Mathematica can
canbebewritten as
written
[lhs = = rhs, {x, x0, xmin, xmax}, Max Iterations → 200]. Here, the equation solved is lhs = = rhs, 200
as [lhs = =
[lhs = = rhs, {x, x0, x0min, min
rhs, {x, x , x xmax},max
, x Max Iterations → 200]. Here, the equation solved is lhs = = rhs, 200
}, Max Iterations → 200]. Here, the equation solved is lhs = = rhs,
iterations will be carried out in the interval [xmin, xmax], and the roots will be found near x0. In order to
iterations willwill
200 iterations be carried
be carriedoutout
in the interval
in the [xmin
interval [x , xmax
, x],max
and the the
], and roots willwill
roots be found
be foundnear x0.xIn. order
near to
In order
solve the unknowns of Ei and i in Equation (7), min the bending creep tests were carried out. 0
solve the unknowns of Ei and
to solve the unknowns of E and τ
i i ini Equation (7), the bending creep tests were carried out.
in Equation (7), the bending creep tests were carried out.
3.
3. Creep
CreepTests
Tests and Calculations of Relaxation Moduli
3. Creep Tests and Calculations
and Calculations of
of Relaxation
Relaxation Moduli
Moduli
3.1.
3.1. Three-Point
Three-PointBending
Bending Creep
Creep Tests
Tests
3.1. Three-Point Bending Creep Tests
3.1.1.
3.1.1. Material
MaterialProperties
Properties
3.1.1. Material Properties
The
The bitumen
bitumen with
with 70 70 penetrations
penetrations is
is used
used asas the
the asphalt
asphalt binder
binder for
for the
the samples.
samples. Limestones
Limestones are
are
The bitumen with 70 penetrations is used as the asphalt binder for the samples. Limestones are
used
used asas the
the aggregates.
aggregates. The The optimum
optimum ratio
ratio of
of oil
oil to
to aggregate
aggregate is 8.1%. The
is 8.1%. The continuous
continuous aggregate
aggregate
used as the aggregates. The optimum ratio of oil to aggregate is 8.1%. The continuous aggregate
gradation
gradation has has aa nominal
nominal maximum
maximum sizesize of
of 10
10 mm.
mm. AnAnadditional
additional 0.7%
0.7% TCA (temperature controlling
TCA (temperature controlling
gradation has a nominal maximum size of 10 mm. An additional 0.7% TCA (temperature controlling
viscosity
viscosity acid)
acid) additive
additive of of the
the asphalt
asphalt mixture
mixture mass
mass and
and 1%
1% activated
activated rubber
rubber crumb
crumb and
and 0.7%
0.7% TCA
TCA
viscosity acid) additive of the asphalt mixture mass and 1% activated rubber crumb and 0.7% TCA
additive
additive of of the
the mass
mass ofof the
the asphalt
asphalt mixture were added during the blending process.
additive of the mass of the asphalt mixture were added during the blending process.
3.1.2. Sample
3.1.2. SamplePreparations
Preparations
3.1.2. Sample Preparations
The track
The track plate
plate samples
samples mademadeby bywheel
wheelrolling
rollingwhich
whichagree
agreewith
withthe
thestandard
standardtest
testmethod
method[32],
[32],
were The
cut trackbeams
into plate with
samples made byofwheel
dimensions 250 × rolling
30 × 35which
mm agree with
3 (Figure 3), theeffective
the standard testofmethod
span the [32],
beam is
were cut into beams with dimensions of 250 × 30 × 35 mm (Figure 3), the effective span of the beam
3
were
200 cut as
mm, into beams
shown in with dimensions
Figures 3 and 4. of 250 × 30 × 35 mm3 (Figure 3), the effective span of the beam
is 200 mm, as shown in Figures 3 and 4.
is 200 mm, as shown in Figures 3 and 4.

Figure 3. Specimen for bending creep test.

Figure 3. Specimen for bending creep test.
Figure 3.

P
P
3535

25 100 100 25 30
25 100 100 25 30

Figure
Figure 4.
4. Dimensions
Dimensionsof
ofthree-point
three-point bending
bending beam
beam (mm).
(mm).
Figure 4. Dimensions of three-point bending beam (mm).
3.1.3. Bending Creep Tests Procedure
3.1.3. Bending Creep Tests Procedure
3.1.3. Bending Creep Tests Procedure
The creep test is a test method for determining the viscoelastic parameters of materials, which
The creep test is a test method for determining the viscoelastic parameters of materials, which is
The creepused
is commonly test is
toaevaluate
test method for determining
the creep properties the viscoelasticThe
of materials. parameters
effects ofof materials, which
temperatures on theis
commonly used to evaluate the creep properties of materials. The effects of temperatures on the
commonly used to
creep compliance evaluate
were the creep
investigated. Theproperties of materials.
bending creep tests wereThe effects
carried out of temperatures
at three on the
temperatures of
creep compliance were investigated. The bending creep tests were carried out at three temperatures
creep compliance
0 ◦ C, −5 were investigated. The bending creep tests were carried out at three temperatures
◦ C and −15 ◦ C, respectively. After the specimens were put into the environmental chamber,
of 0 C, −5 C and −15 C, respectively. After the specimens were put into the environmental
of
the0temperature
C, −5 C and
was −15 C, respectively.
increased Aftervalue.
to the expected the specimens reachput
In order towere into the
thermal environmental
equilibrium in the
specimens, it was conditioned for over 4 h, after that, a stress of 10% of the failure load was applied
Materials 2019, 12, x FOR PEER REVIEW 6 of 14
Materials 2019, 12, 2021 6 of 14

chamber, the temperature was increased to the expected value. In order to reach thermal equilibrium
in the specimens, it was conditioned for over 4 h, after that, a stress of 10% of the failure load was
according to the standards. The creep time was designed as 4 h. The lever-loading device applied
applied according to the standards. The creep time was designed as 4 h. The lever-loading device
the load and the displacement meter was used to measure the center displacement of the beam,
applied the load and the displacement meter was used to measure the center displacement of the
as shown in Figure 5. The experiment was repeated three times, and the averages were determined by
beam, as shown in Figure 5. The experiment was repeated three times, and the averages were
Equations (8)–(10) given below.
determined by Equations (8)–(10) given below.

Figure 5.
Figure 5. Bending creep test.
Bending creep test.

The bending
The bending tension stress σ00 is
tension stress is calculated
calculated as:
as:
3LP
3LP
σ0=
0 
00
2 2
(8)
(8)
2bh
2bh

where P
where P00 is the concentrated
is the concentrated force applied at
force applied at the
the middle
middle ofof the
the specimen
specimen (in
(in N),
N), bb is
is the
the width
width ofof the
the
specimen section
specimen section (in
(in m),
m),hhisisthe
theheight
heightofofthe
thecross
crosssection
sectionofofthe
the specimen
specimen (in(in
m)m)
and and
L isL the
is the span
span of
of the
the specimen
specimen (in(in
m).m).
The bending
The bending tension
tension strain
strain is
is calculated
calculated as:
as:

6hd (t )
) 6hd(t2)
ε(t)(t= (9)
(9)
L2L
where d(t)
where d(t) is
is the
the midspan
midspan deflection
deflection which
which varies
varies with
with the
the time
timettduring
duringthe
theloading
loading(in
(inm).
m).
The bending creep compliance is calculated
The bending creep compliance is calculated as: as:

 (t )
J ( t ) ε(t) (10)
σ0 0
J (t) = (10)

3.1.4. Test Results and Calculation of Creep Compliance

3.1.4. Test Results and Calculation of Creep Compliance
The bending strains of the specimens changing with the time at different temperatures under the
The bending strains of the specimens changing with the time at different temperatures under
same stress level are shown in Figure 6.
the same stress level are shown in Figure 6.
 Materials 2019, 12, x FOR PEER REVIEW 7 of 14

2.5
Materials 2019, 12, x FOR PEER REVIEW 7 of 14
Materials 2019, 12, 2021 15℃ 7 of 14
10℃
2.0
0℃

2.5
1.5

Stran/10-3
15℃
10℃
2.0
1.0 0℃

1.5
0.5

Stran/10-3
1.0
0.0
0 2 4 6 8 10 12
Time/103s
0.5

Figure 6. Bending tension

0.0 strain compared to time at temperatures 0 C, 10 C and 15 C.
0 2 4 6 8 10 12
Time/103s
As shown in Figures 6 and 7, the creep process of the HVAS can◦be divided into three stages,
Figure 6. Bending tension strain compared to time at temperatures 0 C, 10 ◦ C and 15 ◦ C.
i.e., the initial creep stage, the stable creep stage and the accelerating creep stage. The creep rates
Figure 6. Bending tension strain compared to time at temperatures 0 C, 10 C and 15 C.
increase
As shownwith the temperature.
in Figures 6 and 7, the creep process of the HVAS can be divided into three stages, i.e.,
According to Equation (10), the stress and the
strain can be used to determineThethe creep compliance.
As shown in Figures
the initial creep stage, the 6 and
stable 7, the
creep creep
stage process
and of the HVAS
accelerating creepcan be divided
stage. into
creep three
rates stages,
increase
The creep compliances at different temperatures are shown in Figure 7.
i.e., the
with the temperature.
initial creep stage, the stable creep stage and the accelerating creep stage. The creep rates
increase with the temperature.
0.004
According to Equation (10), the stress and strain can be used to determine the creep compliance.
15℃
10℃
The creep compliances at different temperatures are shown in Figure 7.
-1

0℃ 0.003
Creep compliance/MPa

0.004
0.002 15℃
10℃
-1

0.003 0℃
Creep compliance/MPa

0.001

0.002
0.000
400 800 1200 1600 2000
0.001 Time/s

Figure 7. Creep compliances at temperatures 0 ◦ C, 10 ◦ C and 15 ◦ C.

Figure 7. Creep
0.000 compliances at temperatures 0 C, 10 C and 15 C.
400 800 1200 1600 2000
According to Equation (10), the stress and strain can be used to determine the creep compliance.
Time/s
3.2.creep
The Determination of Model
compliances Parameters
at different temperatures are shown in Figure 7.

Mechanical of
3.2. Determination models
Figuremay
Model be considered
7. Creep
Parameters compliancesasatdifferent combinations
temperatures of linear
0 C, 10 C and 15 C. spring(s) and linear
dashpot(s) in various series and/or parallel arrangements depending upon the complexity of
3.2. Mechanical
Determination
viscoelastic models
of Model
material may be considered
Parameters
behaviour. These basic as different
elementscombinations of linear spring(s)
and their combinations allowand thelinear
better
dashpot(s) in various series and/or parallel arrangements depending upon the complexity
modelling of the viscoelastic behaviour of the asphalt mixtures and the binders than the empirical of viscoelastic
materialMechanical
behaviour. models
These may beelements
considered andastheir
different combinations of linear spring(s) andoflinear
mathematical models. Thebasic
linear spring response combinations
is the same as allow
a linearthe better
elastic modelling
material, whilethe
the
dashpot(s)behaviour
viscoelastic in various of series
the and/or
asphalt parallel
mixtures and arrangements
the binders depending
than the empiricalupon the complexity
mathematical models.of
basic response of a linear dashpot is the same as that of a Newtonian fluid. Combining these two
viscoelastic
The linear material
spring response behaviour.
is the These basic
same arrangements elements
as a linear elastic and their
material, combinations
while the basic allow the
response better
of a
basics in various series and/or parallel produces the viscoelastic mechanical models,
modelling
linear dashpotof the
is viscoelastic
the same as behaviour
that of a of the asphalt
Newtonian mixtures
fluid. and the
Combining binders
these two than
basicsthein empirical
various
some of which, e.g., Maxwell model, Kelvin model, etc., are too simple to adequately model the
mathematical
series models. The linear spring response is the same as a linear elasticsome
material, whilee.g.,
the
actualand/or parallel
behaviour of arrangements
asphalt mixtures, produces
while the
someviscoelastic
other ones,mechanical models,
e.g., the Burgers modelofmay which,
properly
basic
Maxwell response of a linear dashpot is the same as that of a Newtonian fluid. Combining these two
capture model, Kelvin
the actual model, of
behaviour etc.,
theare too simple to adequately model the actual behaviour of asphalt
mixture.
basics inwhile
mixtures, various
someseries
otherand/or
ones, parallel
e.g., the arrangements
Burgers produces the viscoelastic mechanical models,
The creep compliances were obtained by model
fittingmay
the properly
test datacapture the actual
at different behaviour
temperatures; of
the
some
the of which, e.g., Maxwell model, Kelvin model,
mixture. of the four-component Burgers model are listed in Table 1. etc., are too simple to adequately model the
parameters
actual behaviour of asphalt mixtures, while some other ones, e.g., the Burgers
The creep compliances were obtained by fitting the test data at different temperatures; the model may properly
capture theofactual
parameters behaviour of the
the four-component mixture.
Burgers model are listed in Table 1.
The creep compliances were obtained by fitting the test data at different temperatures; the
parameters of the four-component Burgers model are listed in Table 1.
Materials 2019, 12, 2021 8 of 14

Table 1. Burgers model parameters at different temperatures.

Model Parameters 0 ◦C 10 ◦ C 15 ◦ C
E01 3771.931 1893.139 1105.924
E02 5996.615 1272.372 705.477
η01 2.4 × 107 4.2 × 106 2.0 × 106
η02 5 × 106 1.2 × 106 4.6 × 105
R2 0.998 0.999 0.999

3.3. Calculation of Relaxation Moduli

Substituting the parameters of the Burgers model into Equation (7), the recovery of the relaxation
moduli from the creep compliance at different temperatures is realized. The results are listed in Table 2.

Table 2. GMM parameters from calculations at 15 ◦ C and 0 ◦ C.

Model Parameters 15 ◦ C 0 ◦C
E1 /MPa 2032 431
E2 /MPa 2415.65 675.65
E3 /MPa 3742.5 1452.5
E4 /MPa 7464 431
E5 /MPa 5645 3713
E6 /MPa 2032 431
τ1 /s 0.01 0.0015
τ2 /s 1.54 0.00019
τ3 /s 0.011 0.00012
τ4 /s 0.00018 0.00089
τ5 /s 0.0000007 0.0000003
τ6 /s 0.0000005 0.0000006

Substituting the data in Table 3 into Equation (4), the relaxation moduli at temperatures 15 ◦ C and
0 ◦ C are obtained:

E(t) = 2 032e−0.01t + 2 415.65e−1.54t + 3 742.5e−0.011t + 7 464e−0.000 18t + 5 645e−0.000 007t + 2 032e−0.000 000 5 t (11)

E(t) = 431e−0.001 5t + 675.65e−0.000 19t + 1 452.5e−0.000 12t + 431e−0.000 89t + 3 713e−0.000 000 3t + 431e−0.000 000 6 t (12)

Table 3. GMM parameters at 15 ◦ C.

Model Parameters 0.3 mm 0.9 mm 1.5 mm 3.0 mm 6.0 mm

E1 /MPa 1 1.3 1200 12.21 65.442
E2 /MPa 1.6 1.33 25.56 1 114.98
E3 /MPa 159 106.4 182.06 95.30 54.45
E4 /MPa 341 181.3 94.33 567.09 31.396
E5 /MPa 341.6 181.3 89.3 567309 31.396
E6 /MPa 1 1.3 145.1 1 114.93
τ1 /s 1 1 1 1 1
τ2 /s 1 1 1 18.84 3.16
τ3 /s 12.248 7.63 11.43 18.84 19.5
τ4 /s 1 1 1 1 286.09
τ5 /s 1 1 1 1 296.14
τ6 /s 1 1 1 18.84 3.16
Materials 2019, 12, 2021 9 of 14

4. Verification of Calculated Results and Uniaxial Compression Tests

4.1. Uniaxial Compression Relaxation Tests

4.1.1. Determination of Constant Levels of Input Strains for Relaxation Tests

The input constant strain is an important parameter for the relaxation test. The input constant
strain was obtained for the characterization of the viscoplasticity of the HVAS below the undamaged
limit. It has been widely accepted that a sample was not damaged as long as the stress did not reach
the vertex of the stress-strain diagram. The property of the mixture changed from linear viscoelasticity
to viscoplasticity as the input strain increases during the experiment [33]. Under the condition of small
strain, the test specimen was considered not damaged (the strain was in the linear viscoelastic scope),
and the constant input strain was assigned conservatively 10% of the strain corresponding to the
maximum normal stress. Due to the relaxation nature of the HVAS, the magnitude of the deformation
is constant according to the preset program.

4.1.2. Uniaxial Compression Relaxation Tests Procedure

The stress relaxation test is an experiment method to determine the viscoelastic parameters of
materials, and it is commonly used to obtain the stress relaxation properties. Since the viscoelastic
materials have the memory effect, the stress responses of the materials depend on their loading histories.
To this end, the effects of loading histories on the relaxation moduli were investigated. The maximum
strain of breaking 20% [34] is recommended to reduce the coefficient of variation in compressive
strength tests. Therefore, the designation of the initial strain (0.3 mm) is conservative, that is, 10% of
the strain at strength (3 mm).
The direct compression relaxation modulus experiments, at constant input strains of 0.004285,
0.012857, 0.021428, 0.042857, 0.085714, respectively (at input displacements of 0.3 mm, 0.9 mm, 1.5 mm,
3 mm and 6 mm, respectively), and at temperatures of 15 ◦ C, −5 ◦ C and −15 ◦ C, respectively, were
carried out on the specimens. The temperature was increased to the predetermined value when the
specimens were put inside the environmental chamber. To reach thermal equilibrium in the specimens,
it was conditioned for over 4 h. Vaseline was applied on the surfaces of the specimens to reduce the
boundary effect and the friction. After the sample was placed between the base and the pressure
head of the WDW testing machine (Shanghai Xunrong Testing Equipment Co., Ltd., Shanghai, China),
the constant input strain was imposed.

4.1.3. Determination Parameters of GMM Model

The GMM seems to be the best phenomenologic model to represent the HVAS relaxation
behaviour [35]. Since the GMM with six arms has the best phenomenologic representation of the
viscoelastic behavior of the HVAS, the model was chosen to study the stress relaxation behavior of the
HVAS, and the model parameters are listed in Table 3.

4.1.4. Construction of Master Curves for Relaxation Modulus of Asphalt Mixture

The tests were conducted at several temperatures, so a master curve of the relaxation modulus
was constructed using the time-temperature superposition principle [36]. The relaxation modulus test
protocol is theoretically sound, but practically, the test machine may not be able to control the specimen
deformation at a desired constant level. Therefore, the relaxation modulus cannot be simply calculated
by dividing the relaxing stress by the strain. Based on the time-temperature equivalence principle,
the relaxation modulus curves (RMC) was constructed from short-time relaxation measurements with
relevant temperature-shift-factor rates. Williams, Landel and Ferry’s model [37] is used in the analysis:

C1 (T − T0 ) C ∆T
lgαT = − =− 1 (13)
C2 + (T − T0 ) C2 + ∆T
relaxation
relaxation measurements
measurements with
with relevant
relevant temperature-shift-factor
temperature-shift-factor rates.
rates. Williams,
Williams, Landel
Landel and
and Ferry’s
Ferry’s
model [37] is used in the analysis:
model [37] is used in the analysis:
CC1 ((TT TT0 )) C T
lgTT  C 
lg 1 0   C11 T (13)
 (13)
C22  TT
Materials 2019, 12, 2021 (T T ) C 10 of 14
C22  (T  T00 )
where
where CC11 and
and CC22 are
are material
material parameters
parameters and
and TT00 is
is the
the reference
reference temperature,
temperature, in in this
this study,
study, TT00 == 00 C.
C.
C1 and
whereShift C are material parameters and T is the reference temperature, in this study, T0 = 0 ◦ C.
Shift factors
factors were
were calculated
calculated byby the
the WLF
WLF (a formula
formula about
about time-temperature
time-temperature equivalence
2 0
(a equivalence
Shift factors were calculated by the WLF (a formula about time-temperature equivalence principle)
principle
principle ) to
to control
control the relative horizontal
horizontal displacement at
at different
different temperatures as
as listed in
in Table
to control )the relative the relative displacement
horizontal displacement
at different temperatures temperatures
as listed in Table listed
4 [38]. Table
44 [38].
[38].
Table 4. Temperature shift factors.
Table
Table 4.
4. Temperature
Temperature shift
shift factors.
factors.
Temperature/◦ C Temperature Fluctuation ∆T/◦ C Shift Factors lgαT
TT /C
Temperature/C
Temperature/C Temperature
Temperature Fluctuation
Fluctuation /C
Shift
Shift Factors lgTT
Factors lg
15 15 1.6636
15
15 −5 15
15 1.6636
1.6636
5 −0.5959
5
5 −15 55
15 0.5959
0.5959
−1.8569
15
15 15
15 1.8569
1.8569
According
According to
to the shift factors given in Table 4, the master curve of stress relaxation moduli was
According
obtained by to the
the shift
superposition
factors
shiftatfactors
0
given
given
◦ C, as
in
shownin Table
Table
in
4,
4, the
Figure the
8.
master
master curve
curve of
of stress
stress relaxation
relaxation moduli
moduli was
was
obtained by superposition at 0 C, as shown in Figure
obtained by superposition at 0 C, as shown in Figure 8. 8.
6000
6000
moudulus/MPa

5000
Relaxationmoudulus/MPa

5000
4000
4000
3000
3000
Relaxation

2000
2000
1000
1000
0
01 10 100 1000 10000
1 10 100 1000 10000
Time/s
Time/s

Figure
Figure 8.
Figure
8. Relaxation
Relaxation master
8. Relaxation
master curve
curve at
master curve at 00 ◦C.
C.
at 0 C.

Figure
Figure 88 shows
Figure
shows that
8 shows that
that the
the stress
stress relaxation
the stress
relaxation rate
rate decreases
relaxation rate
decreases significantly
significantly with
decreases significantly
with the
the decrease
with the
decrease of
decrease of
of
temperature.
temperature. In
temperature.
In addition,
In
addition, the
addition,
the stress
the
stress relaxation
stress
relaxation can
relaxation
can be
can
be roughly
be
roughly divided
roughly
divided into
divided
into an
into
an attenuation
an
attenuation stress
attenuation stress
stress
relaxation
relaxation stage
relaxation
stage and
and aa steady
stage and
steady stress
stress relaxation
a steady stress
relaxation stage.
relaxation stage.
stage.
4.2.
4.2. Verification of Calculated Results
4.2. Verification
Verification of
of Calculated
Calculated Results
Results
For
For comparisons,
comparisons, the
therelaxation
relaxation moduli
modulibyby
thethe
direct measurements
direct andand
measurements the the
calculations fromfrom
calculations the
tests For comparisons,
◦
at 15 C, are shown the relaxation
in Figure 9. moduli by the direct measurements and the calculations from
the
the tests
tests at
at 15 C, are
15 C, are shown
shown in in Figure
Figure 9.
9.
800
800
500
modulus/MPa
moudulus/MPa

500 test result

Relaxationmodulus/MPa

test results
Relaxationmoudulus/MPa

test results test result

400 caculation 600 calculation
400 caculation 600 calculation

300 400
300 400
200
Relaxation

200 200
Relaxation

200
100
100
0
0 0
0
0 20 40 60 80 100 120 0 50 100 150
0 20 40Time/s
60 80 100 120 0 50
Time/s 100 150
Time/s Time/s

(a)
(a) (b)
(b)
Figure 9. Cont.
 Materials 2019, 12, x FOR PEER REVIEW 4 of 14
Materials
Materials2019,
2019,12,
12,x2021
FOR PEER REVIEW 4 11
ofof1414

1000 5000
1000 5000 test result
test result

Relaxation modulus/MPa
Relaxation modulus/MPa
test result 4000 test result
calculation

Relaxation modulus/MPa
calculation
Relaxation modulus/MPa calculation 4000 calculation
3000
500 3000
500
2000
2000
1000
0 1000
0 0
0
0 50 100 150 200 0 100 200 300
0 50 100
Time/s 150 200 0 100 Time/s200 300
Time/s Time/s

(c) (d)
(c) (d)
400
400 test result
test result
Relaxation modulus/MPa

calculation
Relaxation modulus/MPa

300 calculation
300

200
200

100
100

0
0 0 50 100
0 50
Time/s 100
Time/s
(e)
(e)
Figure 9. Comparisons of relaxation test results and calculation results at 15C. (a) 0.3 mm; (b) 0.9
Figure
Figure9.9.Comparisons
Comparisons of relaxation test
of relaxation testresults
resultsand
andcalculation
calculation results
results at 15C.
at 15 ◦ C. (a)(a)
0.30.3
mm;mm;
(b) (b) 0.9
0.9 mm;
mm; (c) 1.5 mm; (d) 3.0 mm; (e) 6.0 mm.
mm; (c)mm;
(c) 1.5 1.5 mm; (d)mm;
(d) 3.0 3.0 mm; (e)mm.
(e) 6.0 6.0 mm.
Theresults
resultsshow
showthat that the relaxation moduli based oncreep
the creep teststhe
and the relaxation tests
The
The results show thatthetherelaxation
relaxationmodulimoduli based
based on onthe the creepteststests
and andrelaxation
the relaxation tests
tests accord
accord with the relaxation properties of the HVAS and have manyaspects,
similar which
aspects, which indicates
accord
with with
the the relaxation
relaxation propertiesproperties
of the HVASof the and
HVAShaveand havesimilar
many many similar aspects, which indicates
indicates that E(t)
that E(t) can be determined from J(t) (obtainable from three-point bending creep tests), based on the
that E(t) can be determined from J(t) (obtainable from three-point bending creep tests), based on the
can be determined from J(t) (obtainable from three-point bending creep tests), based on the expansion
expansion of the convolution.
expansion of the convolution.
of the convolution.
The curves
curves of
of relaxation
relaxation moduli
moduli obtained
obtained by
by the
the two
two methods
methods are are almost
almost overlapping,
overlapping, but but there
there
The curves of relaxation
The moduli obtained by the two methods are almost overlapping, but there
still exists a narrow margin at the inflection point. This is due to the error between the Prony series
still
stillexists
existsaanarrow
narrowmargin
marginatatthe theinflection
inflectionpoint.
point.This
Thisisisdueduetotothe
theerror
errorbetween
betweenthe theProny
Pronyseries
series
and Burgers model in the process of characterizing the viscoelastic properties.
and
andBurgers
Burgersmodel
modelin inthe
theprocess
processof ofcharacterizing
characterizingthe theviscoelastic
viscoelasticproperties.
properties.
Similarly,forfor comparisons, the relaxation moduli by the direct measurements and the
Similarly, for comparisons, the relaxation moduli by the direct measurements
Similarly, comparisons, the relaxation moduli by the direct measurements and the
and the calculations
calculations from the creep
◦ tests at 0 C are shown in Figure 10.
calculations
from the creep from theat
tests creep
0 Ctests at 0 Cin
are shown are shown
Figure 10.in Figure 10.

6000
6000 test result
Relaxation modulus/MPa

5000 test result

calculation
Relaxation modulus/MPa

5000
calculation
4000
4000
3000
3000
2000
2000
1000
1000
0
0 0 50 100 150 200
0 50 100
Time/s 150 200
Time/s
Figure 10. Comparisons of relaxation test results and calculation results at 0 ◦ C.
Figure 10. Comparisons of relaxation test results and calculation results at 0 C.
Figure 10. Comparisons of relaxation
◦ test results and calculation results at 0 C.
Compared with the results at 15 C, the relaxation modulus curves calculated from the two
Compared
methods displaywith thetrends
similar resultsatat0 15
◦ C.C, the relaxation
Moreover, modulus
with the decreasecurves calculated the
of temperature, from the two
relaxation
Compared with the results at 15 C, the relaxation modulus curves calculated from the two
methods increases
modulus display similar
sharply,trends at 0 C.
this proves theMoreover,
validity ofwith the decrease
the method again.ofHowever,
temperature,
there the
are relaxation
still some
methods display similar trends at 0 C. Moreover, with the decrease of temperature, the relaxation
modulus increases
deviations sharply,modulus
in the relaxation this proves theinflection
at the validity point,
of the especially
method again. However,
at 0 ◦ C. there
This is due are fact
to the still
modulus increases sharply, this proves the validity of the method again. However, there are still
Materials 2019, 12, 2021 12 of 14

that the experimental data of the relaxation modulus were calculated by the WLF formula, and the
characteristics of the rheological properties of materials were characterized by the Burgers model.

5. Conclusions
The three-point bending creep tests were carried out and the experimental data were fitted by the
four-element Burgers model to determine the creep compliance. The R-squared values of Burgers are
greater than 0.995, which indicates that the model accurately characterizes the viscoelasticity used for
the subsequent analysis.
A New Method for the Solving Convolution formula was proposed. To be more specific,
the advantages of the Burgers model and GMM are integrated in their respective fields to better
characterize the properties of the HVAS.
The creep test is easy to be carried out, which provides creep compliance as the basis for solution.
Additionally, the relaxation modulus can be obtained by the transformation relationship between the
creep compliance and relaxation modulus. This method avoids the error of the direct relaxation test
and reduces the requirement for equipment and operation level.
The method for the recovery of the relaxation modulus from the creep compliance is proposed for
solving the relaxation moduli of the HVAS based on the convolution between the creep compliance and
the relaxation modulus. In this method, a transcendental equation was obtained by the convolution of
the creep compliance and the relaxation modulus, and the polynomial functions were obtained by the
expanding of Taylor’s formula, which was solved to obtain the relaxation modulus by Mathmatica
using Newton’s method. The method was verified by the relaxation tests.
The results show that the method provides a good agreement between the experiment data and
the experimental curves. Therefore, the method can better reflect the relaxation characteristics of the
HVAS, and can be used for further study on the relaxation characteristics of the material.

Author Contributions: Y.S. organized the research; Z.G. and J.W. performed creep and relaxation tests; Z.G.
and Y.S. carried out theoretical calculation; Y.S. and Z.G. wrote the manuscript; Y.S., X.Y. and J.W. checked
the manuscript.
Funding: The research is funded by the National Natural Science Fund (51478276), the Natural Science Foundation
of Liaoning Province “Research on freezing-thawing damage and improvement of interfacial behaviours of asphalt
mixture basing on reliability theory”.
Acknowledgments: This research was conducted at the Shenyang Jianzhu University and Institute of
Transportation Engineering of Zhejiang University.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Park, H.M.; Choi, J.Y.; Lee, H.J.; Hwang, E.Y. Performance evaluation of a high durability asphalt binder
and a high durability asphalt mixture for bridge deck pavements. Constr. Build. Mater. 2009, 23, 219–225.
[CrossRef]
2. Doyle, J.D.; Mejias-Santiago, M.; Brown, E.; Howard, I.L. Performance of High RAP-WMA Surface Mixtures,
2011. The National Academies of Sciences, Engineering, and Medicine. Available online: https://trid.trb.org/
view/1135253 (accessed on 24 May 2018).
3. Saha, G.; Biligiri, K.P. Fracture properties of asphalt mixtures using semi-circular bending test: a state-of-the-art
review and future research. Constr. Build. Mater. 2016, 105, 103–112. [CrossRef]
4. Behnia, B.; Dave, E.V.; Ahmed, S.; Buttlar, W.G.; Reis, H. Effects of recycled asphalt pavement amounts on
low-temperature cracking performance of asphalt mixtures using acoustic emissions. Transport. Res. Rec.
2011, 2208, 64–71. [CrossRef]
5. Al-Khateeb, G.G.; Ghuzlan, K.A. The combined effect of loading frequency, temperature, and stress level on
the fatigue life of asphalt paving mixtures using the IDT test configuration. Int. J. Fatigue 2014, 59, 254–261.
[CrossRef]
Materials 2019, 12, 2021 13 of 14

6. Souza, F.V.; Castro, L.S. Effect of temperature on the mechanical response of thermo-viscoelastic asphalt
pavements. Constr. Build. Mater. 2012, 30, 574–582. [CrossRef]
7. Walubita, L.F.; Alvarez, A.E.; Simate, G.S. Evaluating and comparing different methods and models for
generating relaxation modulus master-curves for asphalt mixes. Constr. Build. Mater. 2011, 25, 2619–2626.
[CrossRef]
8. Fu, Q.; Xie, Y.J.; Long, G.C.; Meng, F.; Song, H. Temperature sensitivity and model of stress relaxation
properties of cement and asphalt mortar. Constr. Build. Mater. 2015, 84, 1–11. [CrossRef]
9. Tabatabaee, H.A.; Velasquez, R.; Bahia, H.U. Modeling thermal stress in asphalt mixtures undergoing glass
transition and physical hardening. Transport. Res. Rec. 2012, 2296, 106–114. [CrossRef]
10. Sun, Y.; Gu, Z.; Wang, J.; Fang, C.; Yuan, X. Study on Relaxation Damage Properties of High Viscosity Asphalt
Sand under Uniaxial Compression. Adv. Civil Eng. 2018, 2018, 1498480. [CrossRef]
11. Huang, W.K.; Zhang, L.J. Study on transformation from asphalt mixture creep compliance to relaxation
modulus. J. Transport Sci. Eng. 2015, 31, 7–12.
12. Zheng, J.L.; Tian, X.G.; Ying, R.H. A laboratory research on the thermo-viscoelastic constitutive model of
bituminous mixtures. J. Changsha Univ. Sci. Technol. 2014. Available online: http://en.cnki.com.cn/Article_en/
CJFDTOTAL-HNQG200401001.htm (accessed on 15 September 2018).
13. Park, S.W.; Schapery, R.A. Methods of interconversion between linear viscoelastic material functions. Part I:
a numerical method based on prony series. Int. J. Solids Struct. 1999, 36, 1653–1675. [CrossRef]
14. Liu, X.M.; Tang, Z.P.; Lin, X.Z. Discussion on transformation formulas between complex compliance
and creep compliance, complex modulus and relaxation modulus in linear viscoelasticity. J. Univ. Sci.
Technol. China 1989. Available online: http://en.cnki.com.cn/Article_en/CJFDTotal-ZKJD198904009.htm
(accessed on 22 December 2018).
15. Kim, J.; Scholar, G.A. Determination of accurate creep compliance and relaxation modulus at a single
temperature for viscoelastic solids. J. Mater. Civil Eng. 2008, 20, 147–156. [CrossRef]
16. Forough, S.A.; Nejad, F.M.; Khodaii, A. Comparing various fitting models to construct the tensile relaxation
modulus master curve of asphalt mixes. Int. J. Pavement Eng. 2016, 17, 314–330. [CrossRef]
17. Boltzmann, L. Zur Theorie der elastischen Nachwirkung. Ann. Phys.-Berlin 1878, 241, 430–432. [CrossRef]
18. Yan, M.; Wang, J. Study on Conversion Between Creep Compliance and Relaxation Modulus of Asphalt
Mixture. Technol. Highw. Transp. 2016. Available online: http://en.cnki.com.cn/Article_en/CJFDTOTAL-
GLJT201605008.htm (accessed on 10 June 2018).
19. Anderssen, R.S.; Davies, A.R.; de Hoog, F.R. The effect of kernel perturbations when solving the
interconversion convolution equation of linear viscoelasticity. Appl. Math. Lett. 2011, 24, 71–75. [CrossRef]
20. Zhao, B.H. Study on the Volume Relaxation Modulus and the Volume Creep Complianee. J. Solid
Rocket Technol. 1995. Available online: http://www.cnki.com.cn/Article/CJFDTOTAL-GTHJ501.012.htm
(accessed on 20 March 2018).
21. Hopkins, I.L.; Hamming, R.W. On creep and relaxation. J. Appl. Phys. 1957, 28, 906–909. [CrossRef]
22. Park, S.W.; Kim, Y.R.; Schapery, R.A. Aviscoelastic continuum damage model and its application to uniaxial
behavior of asphalt concrete. Mech. Mater. 1996, 24, 241–255. [CrossRef]
23. Park, S.W.; Kim, Y.R. Fitting Prony-series viscoelastic models with power-law presmoothing. J. Mater.
Civil Eng. 2001, 13, 26–32. [CrossRef]
24. Zhao, Y.Q.; Tang, J.M.; Bai, L. Determination of relaxation modulus using complex modulus of the asphalt
mixture. J. Build. Mater. 2012, 15, 498–502.
25. Findley, W.N.; Davis, F.A. Creep and Relaxation of Nonlinear Viscoelastic Materials, 1st ed.; Elsevier Publishing:
Amsterdam, The Netherlands, 2013.
26. Koeller, R.C. A theory relating creep and relaxation for linear materials with memory. J. Appl. Mech. 2010, 77,
031008. [CrossRef]
27. Zhang, J.; Xu, L.; Wang, B. Modification of creep model of asphalt mixture and parameters determination.
J. Wuhan Univ. Technol. (Transp. Sci. Eng.) 2010, 34, 699–702. Available online: http://www.en.cnki.com.cn/
Article_en/CJFDTotal-JTKJ201004016.htm (accessed on 4 June 2018).
28. Blab, R.; Harvey, J.T. Modeling measured 3D tire contact stresses in a viscoelastic FE pavement model.
Int. J. Geomech. 2002, 2, 271–290. [CrossRef]
29. Shan, L.; Xu, Y.; He, H.; Ren, N. Optimization criterion of viscoelastic response model for asphalt binders.
Constr. Build. Mater. 2016, 113, 553–560. [CrossRef]
Materials 2019, 12, 2021 14 of 14

30. Wang, Q. The Taylor’s formula with surplus item Peano and applications. J. Shenyang Archit. Univ. (Nat.
Sci. Ed.) 2005. Available online: http://www.cnki.com.cn/Article/CJFDTOTAL-SYJZ200506041.htm (accessed
on 17 May 2018).
31. Mangano, S. Mathematica Cookbook; O’Reilly Media, Inc.: Sebastopol, CA, USA, 2010.
32. Research Institute of Highway Ministry of Transport. JTG E20-2011 Standard Test Methods of Bitumen and
Bituminous Mixtures for Highway Engineering; China Communications Press: Beijing, China, 2011.
33. Lytton, R.L.; Gu, F.; Zhang, Y.; Luo, X. Characteristics of undamaged asphalt mixtures in tension and
compression. Int. J. Pavement Eng. 2018, 19, 192–204. [CrossRef]
34. Walubita, L.F. Comparison of Fatigue Analysis Approaches for Predicting Fatigue Lives of Hot Mix Asphalt
Concrete (HMAC) Mixtures. Ph.D. Thesis, Texas A&M University, College Station, TX, USA, August 2006.
35. Hong, Z. Mechanism Analysis of the Differences of Asphalt Fatigue Damage under Stress and Strain Control
Mode. Master’s Thesis, Harbin Institute of Technology, Harbin, China, July 2014.
36. Daniel, J.S.; Bisirri, W.; Kim, Y.R. Fatigue evaluation of asphalt mixtures using dissipated energy and
viscoelastic continuum damage approaches. J. Assoc. Asphalt Paving Technol. 2004, v73, 557–583.
37. Ferry, J.D. Viscoelastic Properties of Polymers; John Wiley & Sons: New York, NY, USA, 1980.
38. Zhu, X.L. Experimental study on aging performance of asphalt mixture. Jilin Univ. China 2011. Available online:
http://cdmd.cnki.com.cn/Article/CDMD-10183-1011099648.htm (accessed on 20 March 2018).

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).