You are on page 1of 11

Viscoelastic Nonlinear Multilayered Model

for Asphalt Pavements


Sudhir Varma 1 and M. Emin Kutay 2
Downloaded from ascelibrary.org by Indian Institute of Technology, Tirupati on 03/01/23. Copyright ASCE. For personal use only; all rights reserved.

Abstract: Flexible pavements are multilayer structures, typically with a viscoelastic asphalt layer followed by nonlinear unbound/bound
layers. Conventionally, multilayered elastic analysis is performed to obtain the response of flexible pavements for design and inverse analyses;
however, assuming asphalt pavement to be a linear elastic material is an oversimplification of its actual behavior. It is well known that the
responses of asphalt pavements are both rate and temperature dependent. In the present work, a computationally efficient model has been
developed to analyze flexible pavements, considering the top layer of linear viscoelastic asphalt concrete (AC), followed by a stress-
dependent (nonlinear) base layer, and an elastic subgrade. Constitutive equations are formulated for layered viscoelastic–nonlinear axisym-
metric systems. It is shown that the developed model can be used to simulate pavement response under stationary or transient loading. A
comparison between the model responses and results obtained using a finite-element (FE) model shows that the model could be used to
simulate both deflections and stress responses in multilayer viscoelastic nonlinear structures. The primary advantage of the model, as opposed
to FE models, is its computational efficiency. This makes it viable for use in backcalculation algorithms. DOI: 10.1061/(ASCE)EM.1943-
7889.0001095. © 2016 American Society of Civil Engineers.
Author keywords: Granular base; Pavement response; Finite-element method; Falling weight deflectometer; Asphalt pavement.

Introduction (jE j) at a certain frequency and temperature as a so-called equiv-


alent elastic modulus.
Flexible pavements are multilayer structures, typically with asphalt It is also well known that unbound granular materials exhibit non-
concrete (AC) as the top layer, followed by unbound/bound granu- linearity, i.e., stress-dependent modulus (Hicks and Monismith 1971;
lar layers. Deformation in flexible pavements under vehicular Witczak and Uzan 1988; Ooi et al. 2004). The modeling response of
loading depends on the properties of the constituting materials flexible pavements with stress-dependent granular layers requires iter-
as well as the geometric properties of each individual layer. Hence ative analysis, which makes the solution computationally intensive.
the mechanistic model used to analyze flexible pavements should To avoid this complexity, most of the standards (Shell International
utilize the constitutive material properties assumed for each indi- Petroleum Company 1978; AI 1999; Theyse et al. 1996; IRC 2001;
vidual layer. Traditionally, flexible pavements are analyzed using Austroads 2004) assume the granular layer to be elastic.
analytic multilayered elastic models [KENLAYER (Huang 2004); Wang and Al-Qadi (2013) developed a finite element (FE)–based
BISAR (De Jong et al. 1979); CHEVRONX (Warren and model for asphalt pavements in Abaqus and considered both the vis-
Dieckmann 1963)], which are based on Burmister’s elastic solution coelasticity of asphalt and the nonlinearity of granular layers. They
of multilayer structures (Burmister 1943, 1945). These models as- found that both the viscoelastic and nonlinear material properties need
sume that the material in each pavement layer is linearly elastic. to be considered in order to accurately predict flexible pavement re-
However, it is well known that AC (typically the top layer) is linear sponse. They also showed that the nonlinear response of the granular
viscoelastic at relatively low strain levels (Approximately equal to layer is influenced by the viscoelastic stress caused by the AC layer.
or less than 100 microstrain). Like any viscoelastic material, it Although FE-based modeling is very useful, is computationally ex-
shows properties dependent on time (or loading frequency) as well pensive and can be significantly affected by boundary conditions. Fur-
as temperature. Most of the present mechanistic design standards thermore, FE-based models are inefficient when used as an inverse
(Shell International Petroleum Company 1978; AI 1999; Theyse analysis tool, e.g., in back calculation algorithms. A computationally
et al. 1996; IRC 2001; Austroads 2004) assume that this layer efficient semianalytical model is inevitable in such algorithms.
is elastic during multilayer pavement analysis. Although the The formulation presented in this paper is inspired by quasi-
Mechanistic Empirical Pavement Design Guide (MEPDG 2004) linear-viscoelastic (QLV) constitutive modeling (Leaderman
recognizes the AC layer as being viscoelastic, it essentially per- 1943; Schapery 1969; Fung 1996; Masad et al. 2008; Yong et al.
forms a multilayer linear elastic analysis using a dynamic modulus 2010; Nekouzadeh and Genin 2013), which is often used in ana-
lyzing nonlinear viscoelastic (NLV) materials. In the literature, the
1
Research Assistant, Dept. of Civil and Environmental Engineering, various forms of the model are also called Fung’s model/Schapery’s
Michigan State Univ., East Lansing, MI 48824-1226. E-mail: nonlinearity model/modified Boltzmann’s superposition. In the
varmasud@msu.edu present study, quasi-elastic theory is combined with generalized
2
Assistant Professor, Dept. of Civil and Environmental Engineering, QLV theory, to develop an approximate model for predicting re-
Michigan State Univ., East Lansing, MI 48824-1226 (corresponding
sponse of multilayer viscoelastic nonlinear flexible pavement struc-
author). E-mail: kutay@msu.edu
Note. This manuscript was submitted on February 14, 2014; approved tures. An overview of the linear viscoelastic multilayer solution is
on July 24, 2014; published online on March 22, 2016. Discussion period presented in the next section, with special emphasis on quasi-elastic
open until August 22, 2016; separate discussions must be submitted for models. Before introducing the generalized QLV model, a brief
individual papers. This paper is part of the Journal of Engineering overview of granular nonlinear pavement models is also presented.
Mechanics, © ASCE, ISSN 0733-9399. The development of a generalized QLV model for a multilayer

© ASCE 04016044-1 J. Eng. Mech.

J. Eng. Mech., 2016, 142(7): 04016044


system is followed by numerical validation, in which the response mixture), if the response Rve ¼ εðtÞ, i.e., strain, then Rve H ¼ DðtÞ
of flexible pavements under stationary transient loading is ana- (i.e., creep compliance), and Iðτ Þ ¼ σðtÞ (i.e., stress). Using Schap-
lyzed. The model is implemented in general-purpose FE-based ery’s quasi-elastic theory, the viscoelastic response at time t to a
software, Abaqus, to validate the model. unit input function can be approximated by the elastic response
obtained using the relaxation modulus at time t as follows:

H ðx; y; z; tÞ ≅ RH ½x; y; z; EðtÞ


Rve ð2Þ
e
Theoretical Background
where ReH ½x; y; z; EðtÞ = unit elastic response at elastic modulus
Linear Viscoelastic Multilayer Solutions equal to the relaxed modulus [EðtÞ]. Characteristic viscoelastic
Various models have been developed by researchers to analyze properties, such as the relaxation modulus, creep compliance,
and dynamic modulus, are often expressed at a specific reference
Downloaded from ascelibrary.org by Indian Institute of Technology, Tirupati on 03/01/23. Copyright ASCE. For personal use only; all rights reserved.

multilayer pavement structures that are composed of linear visco-


elastic and linear elastic layers. Analytical solutions in these models temperature, in terms of a master curve. For thermorheologically
are typically derived based on the viscoelastic correspondence prin- simple materials, these characteristic properties can be generated
ciple (Huang 1973; Hopman 1996; Chen et al. 2009). Viscoelastic at any time and temperature using the time–temperature superpo-
correspondence methods are based on the fact that the integral sition principle. Flexible pavements are exposed to different
transform (Fourier or Laplace transform) of viscoelastic constitu- temperatures over time, which in turn affect their response. This
tive equations corresponds to linear elastic equations. The response variation in response can be predicted by extending Eqs. (1) and
obtained from elastic equations derived from the corresponding vis- (2) as follows:
coelastic structure is then inverted to get the final viscoelastic sol- Z t
R dIðτ Þ
ution. The complex process of deriving an elastic solution of the Rve ðx; y; z; tÞ ¼ ReH ½x; y; z; EðtR − τ Þ dτ ð3Þ
corresponding viscoelastic problem and subsequent inverse trans- τ ¼0 dτ
formation makes the methods cumbersome and inefficient. where tR ¼ t=aT ðTÞ; log½aT ðTÞ ¼ a1 ðT 2 − T 2ref Þ þ a2 ðT − T ref Þ
Kutay et al. (2011) developed a layered viscoelastic algorithm = shift factor at temperature T; T ref = reference temperature;
called LAVA to model the response of flexible pavement based on and a1 and a2 ; shift factor’s polynomial coefficients. Using Eq. (3),
quasi-elastic theory as introduced by Schapery (1965, 1974). The the formulation for predicting the vertical deflection of a linear vis-
time-domain solution uses a relaxation modulus EðtÞ master curve coelastic asphalt pavement system because of an axisymmetric
of the asphalt layer, which can be mathematically computed from loading can be expressed as
the dynamic modulus jE j master curve via Prony series–based in-
Z t
terconversion procedures (Park and Schapery 1999). The LAVA R dσðτ Þ
model differs from the conventional collocation method in the uve
z ðx; y; z; tÞ ¼ ueH−z ½x; y; z; EðtR − τ Þ dτ ð4Þ
τ ¼0 dτ
way the unit response functions are used. Unlike the conventional
collocation method, the unit response function of the flexible pave- where uvez ðx; y; z; tÞ = viscoelastic response of viscoelastic multilayer
ment obtained from a Heaviside step function is not approximated structure at time t and coordinates ðx; y; zÞ; ueH−z ½x; y; z; EðtR − τ Þ =
to exponential functions. Instead, the entire unit response is utilized elastic unit response of pavement system at reduced time tR because
as a direct numerical input, which is used in the convolution of the unit (Heaviside step) load; and σðτ Þ is the applied stress at
integral via linear interpolation. The LAVA viscoelastic model the pavement surface. ueH−z ½x; y; z; EðtR − τ Þ can be computed us-
was verified using frequency-based viscoelastic pavement analysis ing a layered elastic solution and inputting an asphalt modulus
software LAMDA and SAPSI (Al-Khoury et al. 2001a, b; Chen equal to EðtR − τ Þ for each ðtR − τ Þ.
1987; Kutay et al. 2011). Varma et al. (2013a, b) used the temper-
ature and time effect of viscoelasticity in the model to develop a
back calculation algorithm called BACKLAVA to derive the in situ Nonlinear Multilayer Elastic Solutions
properties (i.e., jE j master curve of AC) of existing pavements us- It is well known that, under constant cyclic loading, granular un-
ing falling weight deflectometer (FWD) time history data. In addi- bound materials exhibit plastic deformation during the initial
tion to these analytical methods, finite element–based models have cycles. As the number of load cycles increases, plastic deformation
also been used to analyze the viscoelastic response of flexible pave- ceases to occur and the response becomes elastic in further load
ments (Elseifi et al. 2006; Wang and Al-Qadi 2013). cycles (a phenomenon known as shakedown). Often this elastic re-
sponse is defined by the resilient modulus (M R ) at that load level,
Quasi-Elastic Multilayer Viscoelastic Solution which is expressed as
σd
Assuming there is full bonding between the asphalt layer and the MR ¼ ð5Þ
underlying base and subgrade layers, the overall response of the εr
entire pavement system becomes linear viscoelastic. Therefore,
where σd ¼ ðσ1 − σ3 Þ is the deviatoric stress in a triaxial test and εr
its response under arbitrary loading can be obtained using Boltz-
is recoverable strain. If the granular layer reaches this steady-state
mann’s superposition principle (i.e., the convolution integral)
condition under repeated vehicular loading, then the subsequent
(Levenberg 2013; Kutay et al. 2011; Levenberg 2008),
response can be considered recoverable and Eq. (5) can be used
Z t
dIðτ Þ to characterize the material. However, the MR value shown in
Rve ðx; y; z; tÞ ¼ H ½x; y; z; ðt − τ Þ
Rve dτ ð1Þ Eq. (5) is affected by the state of stress (or load level). Typically,
τ ¼0 dτ
unbound granular materials exhibit stress hardening (Yau and Von
where Rve ðx; y; z; tÞ = linear viscoelastic response at coordinates Quintus 2002; Taylor and Timm 2009), i.e., M R increases with in-
ðx; y; zÞ and time t; RveH ðx; y; z; tÞ = (unit) viscoelastic response creasing stress. Hicks and Monismith (1971) related hydrostatic
of pavement system to a Heaviside step function input HðtÞ; stress and the resilient modulus obtained in Eq. (5) to characterize
and dIðτ Þ is the change in input at time τ . It is worth mentioning the stress dependency of the material. The model suggested by
that for a uniaxial viscoelastic system (e.g., a cylindrical asphalt Uzan (1985) and Witczak and Uzan (1988) uses deviatoric stress

© ASCE 04016044-2 J. Eng. Mech.

J. Eng. Mech., 2016, 142(7): 04016044


as well as octahedral stress to incorporate the distortional shear ef- In the present study, the elastic nonlinearity is solved iteratively
fect into the model. The model has been further modified by vari- assuming an initial set of elastic moduli. The evaluated stresses ob-
ous researchers. Yau and Von Quintus (2002) analyzed long-term tained using the initial values of the modulus are used to evaluate a
pavement performance (LTPP) M R test data using the generalized new set of moduli using Eq. (7). The iteration is continued until the
form of Uzan’s (1985) model expressed as computed modulus from the stresses predicted by the layered sol-
 k  k ution and input modulus used in the layered solution converge.
θ − 3k6 2 τ oct 3 Note that the appropriate stress adjustments were made owing to
M R ¼ k1 þ k7 ð6Þ
pa pa the fact that unbound granular material cannot take tension. This
means that in such a case, either residual stress is generated such
where θ = hydrostatic stress; pa = atmospheric pressure; τ oct = that the stress state obeys a yield criterion or the tensile stresses are
octahedral shear stress; k1 ; k2 ; k3 ; k6 , and k7 = regression coeffi- replaced with zero.
Downloaded from ascelibrary.org by Indian Institute of Technology, Tirupati on 03/01/23. Copyright ASCE. For personal use only; all rights reserved.

cients. They found that parameter k6 in the equation regressed


to zero for more than half the tests, and hence the coefficient
was set to zero for the subsequent analysis. The subsequently modi- Proposed Constitutive Model for Multilayer
fied equation is shown as Eq. (7), which was used in the present Pavement Model under Uniaxial Loading: Combining
study to define the resilient modulus of an unbound granular layer: Linear Viscoelastic AC and Nonlinear Base
 k  k
θ 2 τ oct 3 The mechanical behavior of elastic and linear viscoelastic materials
M R ¼ k1 þ1 ð7Þ are is defined in the literature using springs and dashpots. In the
pa pa
present derivation the standard mechanical response using springs
where θ ¼ σ1 þ σ2 þ σ3 þ γzð1 þ 2K o Þ; τ oct = octahedral shear and dashpots is implied and has not been stressed for the sake of
stress; k1 , k2 , k3 = regression constants; K o = coefficient of the brevity. As shown in Fig. 1, for the uniaxial formulation, a two-
earth’s pressure at rest; and pa is atmospheric pressure. layer system comprised of a linear viscoelastic layer that is in a
Although the resilient modulus, M R , is not the Young’s modulus perfect bond with the underlying nonlinear unbound layer is con-
(E) (Hjelmstad and Taciroglu 2000), while formulating granular sidered. A third layer of elastic material is avoided because it can be
material constitutive equations, it is often used to replace E in easily lumped with the viscoelastic properties in the system. Under
the following equation: uniaxial loading, the formulation for the vertical surface response is
presented. The vertical deflection response of the two-layer system
vE E can be obtained as additional responses of individual layers. That
σij ¼ ε δ þ ε ð8Þ is, the vertical surface deflection at A in Fig. 1 can be expressed as
ð1 þ vÞð1 − 2vÞ kk ij ð1 þ vÞ ij
uz ðt; σÞ ¼ un ðσÞ þ uve ðtÞ ð9Þ
where E = Young’s modulus; v = Poisson’s ratio; σij = stress ten-
sor; εij = strain tensor; εkk ¼ ε11 þ ε22 þ ε33 ; and δ ij = Kroenecker
where uz ðtÞ = total vertical deflection at point A; un ðσÞ = vertical
delta. Nonlinear stress-dependent M R has been implemented in
deformation in the nonlinear layer; and uve ðtÞ = vertical deforma-
many FE-based models. These include GTPAVE (Tutumluer
tion in the linear viscoelastic AC layer. Unlike a system comprised
1995), ILLIPAVE (Raad and Figueroa 1980), and MICHPAVE
of only linear materials, a nonlinear system is expected to have dif-
(Harichandran et al. 1989). Typically FE-based nonlinear pavement
ferent unit response functions at different loads. In this study, the
analysis is performed by defining a user-defined material (UMAT)
unit response function of the system at any load level is defined as a
in FE-based softwares such as Abaqus and ADINA (Hjelmstad and
secant property (like a secant modulus) such that at any load level
Taciroglu 2000; Kim et al. 2009; ADINA 2012; Schwartz 2002).
ðσi Þ the unit response function uH ðt; σi Þ can have a definition sim-
Although FE-based solutions are promising, apart from being com-
ilar to that used for linear viscoelastic materials. Eq. (10) show
putationally expensive, they may exhibit a significant influence of
expressions for unit response functions at different load levels,
boundary conditions if the semi-infinite geometry of a problem is
not implemented adequately in the model.
An approximate nonlinear analysis of pavement can also be
performed using Burmister’s multilayered elasticity-based solution
(Burmister 1945). However, since the multilayer elastic theory as- A
sumes that each individual layer is vertically as well as horizontally
homogeneous, it can be used to illustrate nonlinearity only through
approximation. To incorporate variations in the modulus with Viscoelastic
depth, Huang (2004) suggested dividing the nonlinear layer into AC
multiple sublayers. Furthermore, he suggested choosing a specific
location in the nonlinear layers to evaluate the modulus based on
the stress state of the point. He showed that when the midpoint
under the load is selected to calculate modulus values, the predicted
Nonlinear base
response near the load is close to the actual response. However, the
material
difference between actual and predicted responses increases at
points away from the loading. Zhou (2000) studied the stress
dependency of the base layer modulus obtained from a base layer
mid-depth stress state. Zhou analyzed FWD testing at multiple load
levels on two different pavement structures. The study showed
that reasonable nonlinearity parameters could be obtained through
Fig. 1. Cross section of multilayer viscoelastic nonlinear system used
the regression of a back-calculated modulus with the stress state at
in uniaxial analysis
the mid-depth of the base layer.

© ASCE 04016044-3 J. Eng. Mech.

J. Eng. Mech., 2016, 142(7): 04016044


σi , according to the definition given. It is worth noting that in an
uniaxial case, the loading stresses are the same as the stress state for
calculating a nonlinear modulus in a nonlinear layer:

uiz ðtÞ uin þ uive ðtÞ


uiH−z ðtÞ ¼ uH−z ðt; σiÞ ¼ ¼
σi σi
u i
u ðtÞ
i
1
¼ ni þ ve i ¼ H−z ðtÞ
þ uve ð10Þ
σ σ K s ðσi Þ

where uiH−z ðtÞ ¼ uH−z ðt; σi Þ = nonlinear unit response of system at


stress equal to σi ; uin = deformation in nonlinear layer at stress
Downloaded from ascelibrary.org by Indian Institute of Technology, Tirupati on 03/01/23. Copyright ASCE. For personal use only; all rights reserved.

equal to σi ; uive ðtÞ = deformation in AC layer at stress equal to


σi ; K s ðσi Þ = secant stiffness of nonlinear layer at stress equal to
σi ; and uve H−z ðtÞ = unit response function of linear viscoelastic
AC material. It is clear from Eq. (10) that, for a constant load, a
constant modulus exists for the nonlinear layer, which can be used
in estimating the response of the combined viscoelastic–nonlinear Fig. 2. Secant response in nonlinear layer
system under any loading through Boltzmann superposition.
However, since the unit response functions are a function of
stress, the modified convolution is
Z t Although the uniaxial response derivation for a multilayer vis-
dσðτ Þ coelastic nonlinear system reduces to a simpler form, it cannot be
uz ðtÞ ¼ uH−z ðt − τ ; σÞ dτ ð11Þ
τ ¼0 dτ directly generalized for 2D or 3D (axisymmetric) conditions, for
the following reasons: (1) loading is typically concentrated over
where uz ðtÞ = vertical deflection of system at time t; uH−z ðt − τ ; σÞ = a specific loading area and (2) because of the relaxation of the
unit nonlinear response function at stress equal to σðτ Þ; and σðtÞ = AC layer, the stress state in the nonlinear layer can change over
arbitrary stress function used as the succession of infinitesimal steps. time. However, the impact of these issues can be minimal if the
The modified convolution is discussed in further detail in the next variation in the nonlinear modulus value in the radial direction
section. Similar to the numerical discretization of the convolution is not significant, which has been shown by researchers to be an
integral in linear viscoelasticity, the preceding equation can be adequate assumption (Huang 2004) in multilayer nonlinear elastic
numerically expressed as analysis. In the next section, this assumption is used in deriving a
response for a viscoelastic nonlinear system, and then error in the
uz ðtÞ ¼ uH−z ðt; σ0 ÞΔσ0 þ uH−z ðt; σ1 ÞΔσ1 analysis is discussed.
þ uH−z ðt − τ 1 ÞΔσ2 þ · · · þuH−z ðt − τ k−1 ; σk ÞΔσk ð12Þ

    Proposed Generalized Model for Multilayer


1 0 1
uz ðtÞ ¼ uve
H−z ðtÞ þ Δσ þ u ve
ðt − τ 1 Þ þ Δσ1 Pavement Model: Combining Linear Viscoelastic AC
K s ðσ0 Þ H−z
K s ðσ1 Þ and Nonlinear Base
1
þ ðuve
H−z ðt − τ 2 Þ þ ÞΔσ2 þ · · ·
K s ðσ2 Þ Applicability of Existing Theories of Nonlinear
 
1 Viscoelasticity
H−z ðt − τ k Þ þ
þ uve Δσk ð13Þ
K s ðσk Þ Mechanistic solutions for NLV materials exhibit variation depend-
ing on the type of nonlinearity present. Typical NLV equations
where Δσi = infinitesimal stress increment at t ¼ τ i . As shown in involve convolution integrals, based on an integration kernel, that
Eqs. (14)–(16), Eq. (13) can be further expressed as an additional are a function of stress or strain. Eqs. (17) and (18) show typical
independent response from linear viscoelastic and nonlinear forms of such expressions:
elastic constituents of the system. The first summation in the equa- Z t
tions represents the well-known convolution integration for linear dεðτ Þ
viscoelastic materials, whereas the second term represents the se- σðtÞ ¼ Eðt − τ ; εÞ dτ ð17Þ
τ ¼0 dτ
cant response from the nonlinear constituent. Fig. 2 further illus-
trates the secant computation of the response in the uniaxial Z t dσðτ Þ
analysis εðtÞ ¼ Dðt − τ ; σÞ dτ ð18Þ
τ ¼0 dτ
X
k X
k
1
uz ðtÞ ¼ where ε = strain; σ = stress; Eðt; εÞ = strain-dependent relaxation
H−z ðt − τ i ÞΔσ þ
uve Δσi ð14Þ
i

i¼0 i¼0
K s ðσi Þ modulus; and Dðt; σÞ = stress-dependent creep compliance. Typi-
cally, in many nonlinear materials, the shape of the relaxation
modulus of the material is preserved, even though the material
X
k X
k
uz ðtÞ ¼ exhibits stress or strain dependency (Shames and Cozzarelli 1997;
H−z ðt − τ i ÞΔσ þ
uve Δuin ð15Þ
i

i¼1 i¼0 Nekouzadeh and Genin 2013). Such NLV problems are solved by
assuming that time dependence and stress (or strain) dependence
Z Xk
can be decomposed into two functions:
t dσ
uz ðtÞ ¼ H−z ðt − τ Þ
uve dτ þ Δuin ð16Þ Dðt; σÞ ¼ hðσÞDt ðtÞ ð19Þ
τ ¼0 dτ i¼0

© ASCE 04016044-4 J. Eng. Mech.

J. Eng. Mech., 2016, 142(7): 04016044


Eðt; εÞ ¼ fðεÞEt ðtÞ ð20Þ

where hðσÞ = a function of stress; Dt ðtÞ = the (only) time=


dependent creep compliance; fðεÞ = a function of strain; and
Et ðtÞ = the (only) time-dependent relaxation modulus. This multi-
plicative decomposition facilitates an easy application of the super-
position principle. For such materials the following expression is
commonly used in NLV formulations to develop the convolution
integral (Nekouzadeh and Genin 2013):
Z t
R df½εðτ Þ dεðτ Þ
σðtÞ ¼ Et ðtR − τ Þ dτ ð21Þ
dε dτ
Downloaded from ascelibrary.org by Indian Institute of Technology, Tirupati on 03/01/23. Copyright ASCE. For personal use only; all rights reserved.

τ ¼0

where Et = relaxation function that remains unchanged at any strain


level; and f½εðτ Þ = a function of strain such that df½εðτ Þ=dε
represents the elastic tangent stiffness at different strain levels. Fig. 3. Flexible pavement cross section
These models are designated as Fung’s NLV material, which was
first proposed by Leaderman (1943). A generalized form of this
nonlinearity model was presented by Schapery (1969) using
thermodynamic principles. Yong et al. (2010) used the model to 1
describe the nonlinear viscoelastic-viscoplastic behavior of asphalt
0.95
sand, whereas Masad et al. (2008) used the model to describe the
nonlinear viscoelastic creep behavior of binders. The model sug- 0.9
gests that the nonlinear relaxation function can be expressed as
a product of function of time, Et ðtR − τ Þ, and a function of strain, 0.85

df½εðτ Þ=dε. In Eq. (23), nonlinearity is introduced by the elastic

g(σ )
0.8
component, dfðεðτ ÞÞ=dε, and the viscoelasticity derives from Et . 34 kPa
A direct extension of the concepts of the QLV model to develop 0.75 69 kPa
103 kPa
formulations for a viscoelastic nonlinear multilayer system, where 0.7 138 kPa
172 kPa
the unbound layer is nonlinear and the AC layer is linear viscoelas- 241 kPa
0.65 345 kPa
tic, leads to 483 kPa
Z t 0.6
965 kPa
R dfðεðτ ÞÞ dεðτ Þ 3 4 5 6 7 8
σðtÞ ¼ Et ðx; y; z; tR − τ Þ dτ ð22Þ 1 x 10 1 x 10 1 x 10 1 x 10 1 x 10 1 x 10
τ ¼0 dε dτ Relaxation modulus E(t) kPa

where Et ðx; y; z; tR Þ = relaxation function at location ðx; y; zÞ; and Fig. 4. Variation of gðσÞ with EðtÞ of AC layer at various stress load-
fðεðτ ÞÞ = a function of strain, εðτ Þ. Alternatively, to obtain vertical ings, illustrating dependency of gðσÞ on both loading stress and time
surface deflection in pavements, Eq. (22) can be expressed in terms
of a vertical deflection response to Heaviside step loading as
follows:
Z t Fig. 4 also illustrates that gðσÞ varies with different values of EðtÞ.
R This means that gðσÞ is not based solely on the stress, and as a
uve
z ðtÞ ¼ ueH−t ðtR − τ ; σ ¼ 1ÞgðσÞdσðτ Þ ð23Þ
τ ¼0 result, Fung’s model cannot be used in a layered pavement struc-
ture. This is meaningful since the change in the stress distribution
where uvez ðtÞ = surface (NLV) displacement; uH−t ðt; σ ¼ 1Þ = unit
e
within the pavement layers due to the viscoelastic effect [as EðtÞ
nonlinear elastic response due to a unit stress; and gðσÞ = a function varies] will impose changes in the behavior of the stress-dependent
of stress, which can be expressed as granular layer.
ueH ðtR ; σÞ
gðσÞ ¼ ð24Þ
uH−t ðtR ; σ ¼
e
1Þ Proposed Model: LAVAN
Boltzmann’s superposition integrals [e.g., Eq. (1)] assume that the
where ueH ðtR ; σÞ = nonlinear elastic unit displacement due to a
system is linear, i.e., it is not stress-dependent. This assumption
given stress ðσÞ. The LAVA algorithm was modified to implement
may be violated to approximate the nonlinearity of the system if
an iterative nonlinear solution for the granular base, which was as-
stress-dependent relaxation functions (i.e., the stress-dependent
sumed to follow Eq. (7). The ueH ðtR ; σÞ in Eq. (24) was calculated in
unit response functions) of the multilayered structure under many
a range of stress values from 0.7 to 965.3 kPa (0.1 to 140 psi) and
stress levels are precomputed and used in the modified superposi-
using EðtÞ values (for AC) in a range of times, ranging from 10−8 to
tion integral as follows:
108 s. Then, ueH−t ðtR ; σ ¼ 1Þ was calculated for unit stress. Sub-
Z t
sequently, using Eq. (24), gðσÞ values were computed for different R dIðτ Þ
stress levels and E(t) values. For Fung’s theory [i.e., Eqs. (22)–(24)] Rve ðx; y; z; tÞ ¼ ReH ½x; y; z; Iðτ Þ; tR − τ  dτ ð25Þ
τ ¼0 dτ
to hold, gðσÞ must be purely a function of stress and not a function
of EðtÞ. The gðσÞ values were computed for the pavement section where Rve ðx; y; z; tÞ = NLV response of layered pavement struc-
shown in Fig. 3 and plotted in Fig. 4. Fig. 4 shows the variation in ture; ReH ½x; y; z; Iðτ Þ; tR − τ  = unit response function that is a func-
gðσÞ, where the gðσÞ values decrease with increasing stress ðσÞ. tion of both time and input, Iðτ Þ, which in this study is the stress
This is expected behavior for a nonlinear material since, as the applied at the surface of the pavement. The reader is cautioned that
stress increases, the unbound layer moduli will increase. However, this approach is an approximation and usually can be accomplished

© ASCE 04016044-5 J. Eng. Mech.

J. Eng. Mech., 2016, 142(7): 04016044


8
10
through numerical (discrete) evaluation of Eq. (25). It should be Control 70−22
noted that in this formulation, unlike Fung’s QLV model, time CRTB

Relaxation modulus, E(t) (kPa)


dependence and stress (or strain) dependence are not separated. 7
10
Eq. (25) can be rewritten in terms of vertical surface deflection
under axisymmetric surface loading as follows:
Z t
R dσðτ Þ 6
10
uz ðr; z; tÞ ¼
ve
ueH−z ½r; z; σðτ Þ; tR − τ  dτ ð26Þ
τ ¼0 dτ

where uvez ðr; z; tÞ = vertical deflection at time t and location ðr; zÞ;
5
10
and ueH−z ðr; z; σðτ Þ; tR − τ Þ ¼ uez ½r; z; σðτ Þ; tR − τ =σðτ Þ, where
uez is the nonlinear response of the pavement at a loading stress
Downloaded from ascelibrary.org by Indian Institute of Technology, Tirupati on 03/01/23. Copyright ASCE. For personal use only; all rights reserved.

level of σ. The model in Eq. (26) can be expressed in a discretized 4


10
1.0E−10 1.0E−7 1.0E−4 1.0E0 1.0E4 1.0E7 1.0E10
formulation as follows:
Reduced time, at 19oC (s)
X
N
z ðr; z; ti Þ ¼ ueH−z ½r; z; σðτ j Þ; tRi − τ j Δσðτ j Þ ð27Þ
uve Fig. 5. Relaxation modulus master curve at 190°C reference
j¼1 temperature

where τ 1 ¼ 0, τ N ¼ t. The ueH−z ½σðτ j Þ; tRi − τ j  values are com-


puted via interpolation using the 2D matrix precomputed for
ueH−z ðσ; tÞ [which was computed in a range of stress values and Both the HMA mixes were analyzed using the granular nonlinear
EðtÞ values]. In this paper the developed model is referred to as model as shown in Eq. (7). The detailed geometric and material prop-
LAVAN (short for LAVA-Nonlinear). The step-by-step procedure erties for each layer for each test case are listed in Fig. 3.
to numerically compute a response is given as follows:
1. Define a discrete set of surface stress values: σk ¼ 0.7–965.3
ð0.1–140 psiÞσk . Comparison of LAVAN with FEM Software Abaqus
2. Calculate nonlinear elastic response ue ðσk ; tRi Þ in a range of tRi In Abaqus, the viscoelastic properties of the HMAs were input in
values using EAC ¼ EðtRi Þ for each tRi value. the form of normalized bulk modulus ðKÞ and normalized shear
3 .Recursively compute Ebase until the stress in the middle of the modulus ðGÞ (Abaqus). For the unbound nonlinear layer, a UMAT
base layer results in the same Ebase as that used in the layered was written. Abaqus requires that any UMAT should have at least
elastic analysis. In the present study recursive layered elastic two main components: (1) updating the stiffness Jacobian matrix
analysis is performed using the multilayer elastic program and (2) a stress increment. Eqs. (29) and (30) show the mathemati-
CHEVRONX (Warren and Dieckmann 1963). At this step, cal expressions for these two operations implemented in the
Eq. (7) is used in the nonlinear formulation for the base. UMAT:
4 .Calculate the nonlinear unit elastic response, ueH−z ðr; z; σk ; ti Þ,
as uez ðr; z; σk ; ti Þ=σk . ∂σij
Jijkl ¼ ð29Þ
5 .Perform the convolution shown in Eq. (27) to calculate the NLV ∂εkl
response.
σnþ1
ij ¼ σnij þ J ijkl ∂εnþ1
kl ð30Þ
Validation of LAVAN where J ijkl is the Jacobian matrix; and σnþ1 is the updated stress.
ij
For nonlinear analysis using LAVAN, the unbound modulus was
To validate the LAVAN algorithm, a flexible pavement is modeled as
calculated using the stress state at the midpoint of the unbound base
a three-layer structure, with a viscoelastic AC top layer, followed by a
layer (vertically). Since LAVAN cannot incorporate nonlinearity
stress-dependent (nonlinear) granular base layer on an elastic half-
along the horizontal direction, for comparison purposes, modulus
space (subgrade). Fig. 3 shows the geometric properties of the pave-
values were calculated using the stress state at r ¼ 3.5a (i.e., r in
ment structure used in the validation, where hAC ¼ 15 cm (5.9 in.)
Fig. 3), where a is the radius of loading. Estimating the base modu-
and hbase ¼ 25 cm (9.84 in.). The viscoelastic properties of two HMA
lus using the stress state along the centerline of the loading would
mixes, CRTB and Control [two materials from FHWA’s ALF 2002
result in a stiffer base (Kim et al. 2009); hence it was expected that
experiment—(Gibson et al. 2012)], were used for the AC layer in the
more accurate results would be obtained using the stress state at
analysis as Cases 1 and 2. The relaxation modulus master curves of
some location radially away from the centerline.
the two mixes are shown in Fig. 5. These curves were computed from
In Abaqus, a solution is sensitive to the boundary condition used
their |E*| master curves by following the Prony series–based intercon-
in the analysis. Kim et al. (2009) found that for multilayer nonlinear
version procedure suggested by Park and Schapery (1999). From the
axisymmetric problems with the vertical boundaries supported us-
theory of viscoelasticity, the creep compliance, relaxation modulus,
ing roller supports and the bottom fixed (finite boundaries), a do-
and dynamic modulus are interconvertible. As a result, the developed
main size of 140a in the vertical direction and 20a in the horizontal
viscoelastic nonlinear multilayer model can take any of three visco-
direction produce results comparable to an analytic solution. Alter-
elastic properties. The relaxation modulus EðtÞ can be approximated
natively, researchers have used infinite elements as the boundary to
with a sigmoid function as follows:
imitate semi-infinite geometry. In the present study, both boundary
c2 conditions were analyzed. It was observed that, when all the ver-
logðEðtÞÞ ¼ c1 þ ð28Þ
1 þ exp½−c3 − c4 logðtR Þ tical boundaries were supported using roller supports and the bot-
tom was fixed, a domain size of 133a in the vertical direction and
where tR = reduced time defined as the ratio of physical time t and 53a in the horizontal direction was found to produce a stable sur-
time–temperature shift factor aT ðTÞ; and ci = sigmoid coefficients. face deflection (with less than 1% error at the center). The same

© ASCE 04016044-6 J. Eng. Mech.

J. Eng. Mech., 2016, 142(7): 04016044


Table 1. Peak Surface Deflections (μm) for Different Boundary Conditions
Control mix CRTB mix
Sensor radial distance Abaqus (FB) LAVAN Abaqus (IB) Abaqus (FB) LAVAN Abaqus (IB)
0 cm (0 in.) 810.9 794.7 747.3 1,068.9 1,079.5 1,005.1
20.32 cm (8 in.) 757.4 734.6 693.6 966.3 955.0 902.2
30.48 cm (12 in.) 712.1 685.0 648.4 883.9 859.1 819.8
45.72 cm (18 in.) 639.5 605.5 575.7 758.6 715.5 694.5
60.96 cm (24 in.) 565.9 526.9 501.9 639.4 585.0 575.3
91.44 cm (36 in.) 433.9 388.4 369.6 445.5 384.3 381.3
121.92 cm (48 in.) 329.1 284.0 264.4 311.6 257.7 246.9
152.4 cm (60 in.) 252.5 210.4 187.0 227.2 183.4 161.9
Downloaded from ascelibrary.org by Indian Institute of Technology, Tirupati on 03/01/23. Copyright ASCE. For personal use only; all rights reserved.

Note: FB = finite boundary; IB = infinite boundary. Deflections are in micrometers (i.e., 10−6 m).

domain size is subsequently analyzed with all the vertical as well deflections predicted by LAVAN were found to lie between the
as bottom boundaries supported using infinite elements. For the FE results predicted by the two boundary conditions.
selected domain size, a FE mesh refinement of 10 mm in the Satisfactory performance of the model would require predicting
AC layer and 25 mm in the base layer is used. The model response a comparable time response by the model. The surface deflection
under transient loading is compared with results obtained from the history computed by LAVAN and Abaqus (analysis with finite
general-purpose FE software Abaqus. For this purpose, haversine boundaries) for the Control and CRTB mixes are plotted in Figs. 6
loading in an axisymmetric setup is used. Abaqus consumed ap- and 7, respectively. In the figures, the curves labeled AS show re-
proximately 17 min in analyzing a haversine loading of sults obtained using Abaqus, whereas the curves labeled LS show
951.5 kPa (138 psi) and 35 ms, whereas LAVAN could generate results obtained using LAVAN. A comparable response is visible in
the results in 3.6 min on the same desktop computer. the figures, which has been further quantified using Eqs. (31) and
The surface deflection obtained using FE for the two boundary (32). As expected, under the same geometric and loading condi-
conditions, (1) finite boundaries: roller support on the vertical tions, the stiffer Control mix generated lower deflections compared
boundaries and fixed support on the bottom and (2) infinite ele- to the softer CRTB mix. Fig. 6(a) shows the results when stress at
ments at boundaries, are compared with LAVAN in Table 1. It r ¼ 0 is used in LAVAN for nonlinearity computation, and, for
should be noted that both boundary conditions do not strictly re- comparison purposes, Fig. 6(b) shows the results when stress at r ¼
present the semi-infinite geometry of the problem. In Abaqus the 3.5a is used in LAVAN. Note that S1, S2, S3, S4, S5, S6, S7, and
solution in the infinite element is considered to be linear, which is S8 in the figures correspond to surface deflection at Sensor-1
assumed to match the material properties of the adjacent finite through Sensor-8 spaced at 20.3, 30.5, 45.7, 61.0, 91.4, 121.9,
element. Hence the infinite elements provide stiffness to the boun- and 152.4 cm (0, 8, 12, 18, 24, 36, 48, and 60 in.) from the
dary assuming the deflection at r ¼ ∞ to be zero. It can be seen centerline of the load. In general a better match for the deflection
from Table 1 that, for both the Control mix and the CRTB mix, the basin between the FE and LAVAN results can be found when a
surface deflection predicted by the FE method using finite boun- stress state at r ¼ 3.5a is used while incorporating nonlinearity.
daries was higher compared to results when infinite elements It is worth noting that, for the structure in Fig. 3, Huang’s
were used at the boundaries. Further, for both mixes the surface (2004) procedure for the location of stress used in the

1000 1000
AS1* AS1*
AS2 AS2
AS3 AS3
AS4 AS4
800 AS5 800 AS5
AS6 AS6
AS7 AS7
Surface deflection (μm)

Surface deflection (μm)

AS8 AS8
LS1* LS1*
600 LS2 600 LS2
LS3 LS3
LS4 LS4
LS5 LS5
LS6 LS6
400 LS7 400 LS7
LS8 LS8

200 200

0 0
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
(a) Time (ms) (b) Time (ms)

Fig. 6. Surface deflection comparison of Abaqus and LAVAN for Control mix (*AS1= Abaqus Sensor 1; *LS1 = LAVAN Sensor 1)

© ASCE 04016044-7 J. Eng. Mech.

J. Eng. Mech., 2016, 142(7): 04016044


1200 1200
AS1* AS1*
AS2 AS2
AS3 AS3
1000 AS4 1000 AS4
AS5 AS5
AS6 AS6
AS7 AS7
Surface deflection (μm)

Surface deflection (μm)


800 AS8 800 AS8
LS1* LS1*
LS2 LS2
LS3 LS3
LS4 LS4
600 LS5 600 LS5
LS6 LS6
LS7 LS7
Downloaded from ascelibrary.org by Indian Institute of Technology, Tirupati on 03/01/23. Copyright ASCE. For personal use only; all rights reserved.

LS8 LS8
400 400

200 200

0 0
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
(a) Time (ms) (b) Time (ms)

Fig. 7. Surface deflection comparison of Abaqus and LAVAN for CRTB mix (*AS1 = Abaqus Sensor 1; *LS1 = LAVAN Sensor 1)

nonlinear elastic analysis leads to r ¼ 2.8a, when a trapezoidal Next, the stress states ðσzz ; σrr ; σzr Þ obtained from LAVAN were
stress distribution with a 0.5 horizontal slope and 1 vertical slope compared to those computed by Abaqus. Figs. 8–10 show a com-
is assumed. It can be observed from Figs. 6 and 7 that the proposed parison between the flexible pavement stress state obtained for the
modified Boltzmann superposition works well close to the loading. CRTB mix using LAVAN and Abaqus at t ¼ 0.0175 s (for 35 ms
However, response prediction by the model gets less accurate as we haversine loading). It can be seen from the figures that the two sets
move away from the load. of contours closely follow each other. Furthermore, the difference
The difference between the FE results and LAVAN was quan- between the modulus predicted by the two models in the middle of
tified using the following two variables: the base layer under the center of the loading is less than 3.5%. It
can also be seen that the results obtained from LAVAN were more
jupeak peak
ABAQUS − uLAVAN j continuous (fewer breaks in the curves) compared to the results
PEpeak ¼ 100 ð31Þ
upeak
ABAQUS
obtained using Abaqus. This is due to the fact that the FE-based
solution poses difficulties at locations where there is an abrupt

1X juABAQUS ðti Þ=upeak peak


ABAQUS − uLAVAN ðti Þ=uLAVAN j
N
PEavg ¼ 100
N i¼1 uABAQUS ðti Þ=upeak
ABAQUS
ð32Þ Table 2. Percentage Error (PEpeak ) Calculated Using Peaks of Deflections
and Average Percentage Error (PEavg ) Calculated Using Entire Time
History
where PEpeak = percentage error in the peaks; upeak
ABAQUS = peak
Control mix CRTB mix
deflection predicted by Abaqus; upeak LAVAN = peak deflection
predicted by LAVAN; PEavg = average percentage error; r¼0 r ¼ 3.5a r¼0 r ¼ 3.5a
uABAQUS ðti Þ = deflection predicted by Abaqus at time ðti Þ; Peak deflection error, PEpeak (%)
uLAVAN ðti Þ = deflection predicted by LAVAN at time ðti Þ; and 0.8 1.0 2.9 6.3
N = total number of time intervals in the deflection time history. 1.8 0.1 0.8 4.3
Since the model integrates both viscoelastic and nonlinear material 2.6 0.7 0.9 2.7
properties, both peak deflection and relaxation of the deflection 4.2 2.2 3.8 0.2
time history should be predicted with accuracy. Therefore, the nor- 5.8 3.8 6.8 3.2
malized average error along the entire deflection history PEavg was 9.5 7.5 12.4 9.3
13.0 11.3 16.5 14.2
used to examine the model performance in creeping. As shown in
16.2 14.8 19.0 17.7
Table 2, the PEpeak and PEavg values for the Control mix showed a Average deflection history error, PEavg (%)
slight improvement in the results at r ¼ 3.5a. However, it can be 1.6 2.1 1.5 2.1
seen from the table that the CRTB mix shows more sensitivity to 1.5 1.9 1.3 1.8
the location of the stress state compared to the Control mix. As 1.4 1.7 1.3 1.5
expected, owing to the limitations of LAVAN in incorporating 1.3 1.5 1.6 1.3
horizontal nonlinearity, both mixes showed increasing differences 1.5 1.4 2.0 1.4
compared to the FE solutions at sensors away from the loading. 1.8 1.5 2.6 2.0
In both cases, however, errors in the first four to five sensors were 2.2 1.9 2.5 2.3
2.3 2.1 1.4 1.8
within 6%, indicating a good accuracy of LAVAN.

© ASCE 04016044-8 J. Eng. Mech.

J. Eng. Mech., 2016, 142(7): 04016044


0 0

−690

−6
−897 −897

90
−483 −6.9 −483 −6.9
10 10
−138 −138
−55
.2 −48.3
20 20
−4
4.
−4 85
8.3
30 30
−4 −4
4.8 1.4

z (cm)
5
z (cm)

−4
40 1.4 40
−3
4.5

−13.8

−13.8
− 20
50 −3 50
4.5

−20
.7
−2

.7
7.
−2

6
60 60 7.
6
Downloaded from ascelibrary.org by Indian Institute of Technology, Tirupati on 03/01/23. Copyright ASCE. For personal use only; all rights reserved.

70 70

80 80
0 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 90
(a) r (cm) (b) r (cm)

Fig. 8. Vertical stress distribution (σzz , kPa) in pavement: (a) LAVAN; (b) Abaqus

0 −2760 −10 −34 0 −276


−207−1
0 35 −690 5 −13 −6 −20070 −1380−103 −69
5 0
−345 −13
380 −86−93
−3

.9

.9
8 9
−6 4.5
4.

−6
5

10 −34 10 −34
−6 .5 − .5 −69
−138
9 1 38
−6.9 −13.8
20 20

30 30
−13

z (cm)
z (cm)

.8

40 40 −13.8

50 50
−6.
9
60 −6.9 60

70 70

80 80
0 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 90
(a) r (cm) (b) r (cm)

Fig. 9. Radial stress distribution (σrr , kPa) in pavement: (a) LAVAN; (b) Abaqus

0 0
7
3

20
.6

.7

276
48
276

20
27

483 34 34 69 20
10 5 138 69 34.5 10 5 138 27.6 .7
13.8 207 34.5
5.52 13.8
5.52
20 20

30 30
5.52
5.52
z (cm)

z (cm)

40 40 5.52
5.52

50 50

60 60 52
2

5.
5.5

70 70

80 80
0 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 90
(a) r (cm) (b) r (cm)

Fig. 10. Shear stress distribution (σrz , kPa) in pavement: (a) LAVAN; (b) Abaqus

© ASCE 04016044-9 J. Eng. Mech.

J. Eng. Mech., 2016, 142(7): 04016044


discontinuity in the material properties, which can be observed at and applications to the design of airport runways.” Proc., 23rd Annual
the layer interfaces. Meeting, Vol. 23, Highway Research Board, Washington, DC, 126–148.
Chen, S. S. (1987). “The response of multilayered systems to dynamic sur-
face loads.” Ph.D. dissertation, Univ. of California, Berkeley, CA.
Conclusions Chen, Y. G., Pan, E., and Green, R. (2009). “Surface loading of a multi-
layered viscoelastic pavement: Semianalytical solution.” J. Eng. Mech.,
In this study, a computationally inexpensive layered viscoelastic– 10.1061/(ASCE)0733-9399(2009)135:6(517), 517–528.
nonlinear model (called LAVAN) was developed. The model De Jong, D. L., Peutz, M. G. F., and Korswagen, A. R. (1979). “Computer
program BISAR, layered systems under normal and tangential surface
can consider both the linear viscoelasticity of an asphalt concrete
loads.” Koninklijke-Shell Laboratorium, Amsterdam, Netherlands.
(AC) layer and the nonlinear (stress-dependent) elastic moduli of un- Elseifi, M. A., Al-Qadi, I. L., and Yoo, P. J. (2006). “Viscoelastic modeling
bound layers. LAVAN combines quasi-elastic and quasi-linear vis- and field validation of flexible pavement.” J. Eng. Mech., 10.1061/
Downloaded from ascelibrary.org by Indian Institute of Technology, Tirupati on 03/01/23. Copyright ASCE. For personal use only; all rights reserved.

coelasticity (QLV) theories to simulate multilayer structures with (ASCE)0733-9399(2006)132:2(172), 172–178.


viscoelastic, nonlinear elastic, and elastic layers. Although the Fung, Y. C. (1996). Biomechanics mechanical properties of living tissues,
present form of the model considered a single point in estimating 2nd Ed., Springer, New York.
nonlinearity, comparison with a FE solution showed that the model Gibson, N., et al. (2012). “Performance testing for superpave and structural
was quite capable of predicting an accurate response to transient load- validation.” FHWA-HRT-11-045, Federal Highway Administration
ing, which can be used to simulate responses under moving load or Publication, Washington, DC.
falling weight deflectometer (FWD) testing. The accuracy of the pre- Harichandran, R. S., Baladi, G. Y., and Yeh, M. S. (1989). “MICH-PAVE
dicted response under loading was good when a stress state below user’s manual.” FHWA-MI-RD-89-023, Dept. of Civil and Environmen-
tal Engineering, Michigan State Univ., East Lansing, MI.
loading was selected (r ¼ 0) to incorporate nonlinearity. In general,
Hicks, R. G., and Monismith, C. L. (1971). “Factors influencing the resil-
the accuracy of the predicted response basin was improved when a ient properties of granular materials.” Highway Res. Rec., 345, 15–31.
stress state away from the loading was selected (r ¼ 3.5a). Further, Hjelmstad, K. D., and Taciroglu, E. (2000). “Analysis and implementation
the stress distribution predicted by LAVAN and the FE method were of resilient modulus models for granular solids.” J. Eng. Mech., 10
very close to each other. The computational speed of LAVAN was .1061/(ASCE)0733-9399(2000)126:8(821), 821–830.
about five times faster than that of Abaqus. While the study ignored Hopman, P. C. (1996). “VEROAD: A viscoelastic multilayer computer
the nonlinearity of the subgrade, one can easily incorporate the sub- program.” Transp. Res. Rec., 1539, 72–80.
grade nonlinearity by inserting multiple elastic layers with a modulus Huang, Y. H. (1973). “Stresses and strains in viscoelastic multilayer
that increases with depth. This is because the induced stress of surface systems subjected to moving loads.” Highway Res. Rec., 457, 60–71.
loading diminishes very quickly with depth, having minimal impact Huang, Y. H. (2004). Pavement analysis and design, 2nd Ed., Prentice-
on the modulus of the subgrade. The modulus of the subgrade is in- Hall, Englewood Cliffs, NJ.
fluenced much more by hydrostatic stresses than the induced stress IRC (Indian Roads Congress). (2001). “Guidelines for the design of flexible
pavements.” IRC:37-2001, New Delhi, India.
caused by the surface load.
Kim, M., Tutumluer, E., and Know, J. (2009). “Nonlinear pavement foun-
dation modeling for three-dimensional finite-element analysis of flex-
ible pavements.” Int. J. Geomech., 10.1061/(ASCE)1532-3641(2009)
Acknowledgments 9:5(195), 195–208.
Kutay, M. E., Chatti, K., and Lei, L. (2011). “Backcalculation of dynamic
The funding for this study was provided by the Federal Highway modulus from FWD deflection data.” Transp. Res. Rec., 2227, 87–96.
Administration (FHWA) Grant DTFH61-11-C-00026. This support Leaderman, H. (1943). Elastic and creep properties of filamentous materi-
is greatly appreciated. The authors also express their gratitude to als and other higher polymers, Textile Foundation, Washington, DC.
Dr. Nadarajah Sivanesvaran (Siva) from the FHWA for his valuable Levenberg, E. (2008). “Validation of NCAT structural test track experiment
comments. The authors also acknowledge the comments of the using INDOT APT facility.” Final Rep., North Central Superpave
FHWA project team members Prof. Karim Chatti, Dr. Imen Zabaar, Center, Joint Transportation Research Program, Purdue Univ., West
and Prof. Nizar Lajnef. Lafayette, IN.
Levenberg, E. (2013). “Inverse analysis of viscoelastic pavement properties
using data from embedded instrumentation.” Int. J. Numer. Anal. Meth.
References Geomech., 37, 1016–1033.
Masad, E., Huang, C., Airey, G., and Muliana, A. (2008). “Nonlinear
Abaqus version 6.11 [Computer software]. Dassault Systèmes, Waltham, viscoelastic analysis of unaged and aged asphalt binders.” Constr.
MA. Build. Mater., 22, 2170–2179.
ADINA. (2012). “ADINA theory and modeling guides.” Rep. No. ARD 12- MEPDG (Mechanistic Empirical Pavement Design Guide). (2004). “Guide
8, ADINA R&D, Watertown, MA. for mechanistic-empirical design of new and rehabilitated pavement
AI (Asphalt Institute). (1999). Thickness design-Asphalt pavements for structures.” National Cooperative Highway Research Program, Trans-
highways and streets, 9th Ed., Lexington, KY. portation Research Board, National Research Council, Washington,
Al-Khoury, R., Scarpas, A., Kasbergen, C., and Blaauwendraad, J. (2001a). DC.
“Spectral element technique for efficient parameter identification of lay- Nekouzadeh, A., and Genin, G. M. (2013). “Adaptive quasi-linear visco-
ered media. Part I: Forward calculation.” Int. J. Solids Struct., 38(9), elastic modeling.” Studies in mechanobiology, tissue engineering and
1605–1623. biomaterials, Vol. 10, Springer, Berlin, Germany, 47–83.
Al-Khoury, R., Scarpas, A., Kasbergen, C., and Blaauwendraad, J. (2001b). Ooi, P. S., Archilla, A. A., and Sandefur, K. G. (2004). “Resilient modulus
“Spectral element technique for efficient parameter identification of lay- models for compacted cohesive soils.” Transp. Res. Rec., 1874,
ered media. Part II: Inverse calculation.” Int. J. Solids Struct., 115–124.
38(48–49), 8753–8772. Park, S. W., and Schapery, R. A. (1999). “Methods of interconversion be-
Austroads. (2004). “Pavement design.” Australian Dept. of Transportation, tween linear viscoelastic material functions. Part I—A numerical
Sydney, Australia. method based on Prony series.” Int. J. Solids Struct., 36, 1653–1675.
Burmister, D. M. (1945). “The general theory of stress and displacements in Raad, L., and Figueroa, J. L. (1980). “Load response of transportation
layered soil systems. I.” J. Appl. Phys., 16(2), 89–94. support systems.” Transp. Eng. J., 106(1), 111–128.
Burmister, D. M., Palmer, L. A., Barber, E. S., and Middlebrooks, T. A. Schapery, R. A. (1965). “Method of viscoelastic stress analysis using
(1943). “The theory of stress and displacements in layered systems elastic solutions.” J. Franklin Inst., 279(4), 268–289.

© ASCE 04016044-10 J. Eng. Mech.

J. Eng. Mech., 2016, 142(7): 04016044


Schapery, R. A. (1969). “On the characterization of nonlinear viscoelastic dynamic modulus master curve of asphalt pavements.” Proc., 2013
materials.” Polym. Eng. Sci., 9(4), 295–310. Airfield & Highway Pavement Conf., ASCE, Reston, VA.
Schapery, R. A. (1974). “Viscoelastic behavior and analysis of composite Varma, S., Kutay, M. E., and Levenberg, E. (2013b). “Aviscoelastic genetic
materials.” Mechanics of composite materials, Academic Press, algorithm for inverse analysis of asphalt layer properties from falling
New York, 85–168. weight deflections.” Transp. Res. Rec., 4(2013), 38–46.
Schwartz, C. W. (2002). “Effect of stress-dependent base layer on the Wang, H., and Al-Qadi, I. L. (2013). “Importance of nonlinear anisotropic
superposition of flexible pavement solutions.” Int. J. Geomech., 10 modeling of granular base for predicting maximum viscoelastic pave-
.1061/(ASCE)1532-3641(2002)2:3(331), 331–352. ment responses under moving vehicular loading.” J. Eng. Mech., 139,
Shames, I. H., and Cozzarelli, F. A. (1997). Elastic and inelastic stress 29–38.
analysis, Taylor and Francis, Boca Raton, FL. Warren, H., and Dieckmann, W. L. (1963). “Numerical computation of
Shell International Petroleum Company. (1978). Shell pavement design stresses and strains in a multiple-layered asphalt pavement system.”
California Research Corporation, Richmond, CA.
manual-Asphalt pavement and overlays for road traffic, London.
Witczak, M. W., and Uzan, J. (1988). “The universal airport pavement
Downloaded from ascelibrary.org by Indian Institute of Technology, Tirupati on 03/01/23. Copyright ASCE. For personal use only; all rights reserved.

Taylor, A. J., and Timm, D. H. (2009). “Mechanistic characterization of


design system.” Rep. I of IV: Granular material characterization, Univ.
resilient moduli for unbound pavement layer materials.” NCAT Rep.,
of Maryland, College Park, MD.
Auburn Univ., Auburn, AL, 9–6. Yau, A., and Von Quintus, H. L. (2002). “Study of laboratory resilient modu-
Theyse, H. L., Beer, M., and Rust, F. C. (1996). “Overview of the South lus test data and response characteristics.” Publication FHWA-RD-02-
African mechanistic pavement design analysis method.” Transp. Res. 051, Final Rep., Federal Highway Administration, Washington, DC.
Rec., 1539, 6–17. Yong, Y., Yang, X., and Chen, C. (2010). “Modified Schapery’s model
Tutumluer, E. (1995). “Predicting behavior of flexible pavements with for asphalt sand.” J. Eng. Mech., 10.1061/(ASCE)EM.1943-7889
granular bases.” Ph.D. dissertation, School of Civil and Environmental .0000092, 448–454.
Engineering, Georgia Institute of Technology, Atlanta. Zhou, H. (2000). “Comparison of backcalculated and laboratory measured
Uzan, J. (1985). “Characterization of granular material.” Transp. Res. Rec., moduli on AC and granular base layer materials.” Nondestructive
1022, 52–59. testing of pavements and backcalculation of moduli, Vol. 3, ASTM
Varma, S., Kutay, M. E., and Chatti, K. (2013a). “Data requirements from STP 1375, S. D. Tayabji and E. O. Lukamen, eds., ASTM, West
falling weight deflectometer tests for accurate backcalculation of Conshohocken, PA.

© ASCE 04016044-11 J. Eng. Mech.

J. Eng. Mech., 2016, 142(7): 04016044

You might also like