Professional Documents
Culture Documents
Transportation Engineering: Shahram Misaghi, Cesar Tirado, Soheil Nazarian, Cesar Carrasco
Transportation Engineering: Shahram Misaghi, Cesar Tirado, Soheil Nazarian, Cesar Carrasco
Transportation Engineering
journal homepage: www.elsevier.com/locate/treng
a r t i c l e i n f o a b s t r a c t
Keywords: Several mechanistic-empirical software packages have been developed in the last two decades to address the
Pavements impact of axle load and axle configuration on pavement responses and their performance. These software packages
Dynamic loads generally do not consider the vehicle pavement interaction. The interaction of truck suspension system with the
Damage
roughness of the road surface may exert additional forces to the pavement.
Road roughness
Truck suspension A process to quantify the impact of truck suspension systems and road surface condition on the damage exerted
Truck-pavement interaction to the pavement is presented in this paper. The International Roughness Index (IRI) was used to simulate the
road roughness. The truck-pavement interaction was then modeled to estimate the dynamic load applied to
the pavement. The process followed to develop the algorithms is discussed, followed by a parametric study to
demonstrate the interaction among the suspension properties, road roughness and vehicle speed. The impact of
dynamic load on the pavement distresses and the progress of deterioration is also discussed.
Introduction layered linear elastic theory program called JULEA, developed by Uzan
[24]. That program uses an algorithm for calculating IRI based on the
Trucking accounts for about 80% of freight transportation in the quarter-car model to predict the rehabilitation required for a pavement
United States. The impact of axle loads and axle configurations on the section [7]. Mulungye et al. [15] used viscoelastic response models from
US highway network has been a topic of continuous study by researchers ANSYS/ED finite element software to characterize the response of flex-
and highway agencies. Several software packages have been developed ible pavements under traffic load. They also analyzed the dynamic re-
to predict the pavement responses to traffic based on the axle load and sponse of flexible pavements in relation to tire pressure, axle load and
axle configuration. The responses obtained are then incorporated into axle configuration, considering non-linear viscoelastic characteristics of
models to predict pavement distresses, e.g. permanent deformation (rut- pavement material and cyclic wheel loads [16]. Shi and Cai [20] formu-
ting) and/or cracking, as a function of vehicular traffic. Several com- lated a three-dimensional (3D) vehicle-pavement coupled model using
puter programs have been released to calculate stresses (e.g. BISAR, EL- ANSYS to simulate the pavement dynamic loads induced by the vehicle-
SYM5, WESLEA) based on linear elasto-static layered theory. A natural pavement interaction considering road surface condition and vehicle
extension of layered elastic programs has been the viscoelastic-plastic parameters consisting of vehicle speed, weight, damping, and rigidity
programs such as VESYS [12,27]. Saliba [18] described a brief review to reflect the vehicle’s suspension system. Tirado et al. [23] developed
of the mathematical theory of visco-plasticity and the computational a process based on a mechanistic-empirical (ME) analysis to estimate
procedure used to model tire/soil interaction. the permit fees based on truck-axle loading and configuration and the
Finite element techniques provide the flexibility for incorporating associated predicted pavement deterioration using a finite element pro-
nonlinear and viscoelastic-plastic behavior of a system. Chen et al. gram called IntPave. The authors conducted a parametric study show-
[4] compared the results obtained by ABAQUS with the ones coming ing how the truck gross vehicle weight and axle configuration, as well
from layered-elastic programs under the same circumstances. Zaghloul as the pavement structure and the damage threshold to rehabilitation
and White [26] compared the AASHTO load equivalency factors with affect the permit fee. Xu et al. [25] used a quarter vehicle model to
responses from a three-dimensional (3-D) dynamic finite element (FE) calculate the dynamic tire force variation due to surface roughness in
model of an asphalt layer on top of a granular base as well as for full jointed plain concrete pavements (JPCPs) to predict long-term degrada-
depth asphalt pavements. The Pavement-ME software contains a multi- tion. The study also considered the deflection and flexural temperature
∗
Corresponding author.
E-mail addresses: misaghi.sh@gmail.com (S. Misaghi), ctirado@utep.edu (C. Tirado), nazarian@utep.edu (S. Nazarian), ccarras@utep.edu (C. Carrasco).
https://doi.org/10.1016/j.treng.2021.100045
Received 10 June 2020; Received in revised form 31 December 2020; Accepted 1 January 2021
2666-691X/© 2021 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/)
S. Misaghi, C. Tirado, S. Nazarian et al. Transportation Engineering 3 (2021) 100045
variations due to curling. A whole life pavement performance model was mean value, 𝐹̄ , that is usually assumed to be equal to the static load, and
proposed considering the dynamic interaction between vehicles and the a standard deviation, 𝜎. The dynamic load coefficient (DLC) is defined
JPCP pavement, yet the case study did not consider the feedback of as a dimensionless variable obtained by dividing the standard deviation
surface and structural degradation during the pavement’s service life. by the mean static load [9]. In an equation form, this relationship is
Elnashar et al. [6] evaluated the effects of suspension damping and vehi- shown as
cle speed on flexible pavements. In that study, the vehicle was simulated 𝜎
𝐷𝐿𝐶 = _ (1)
as a two-degree-of-freedom quarter-vehicle model, and the pavement- 𝐹
foundation system was described by a simply supported Euler-Bernoulli The DLC concept allows the probabilistic magnitude of the dynamic axle
beam on top of a Pasternak model foundation. The dynamic responses load at a certain vehicle speed and road roughness to be determined.
of the system were investigated using the Newmark-𝛽 method. The au- Theoretically, a truck passing over a smooth pavement should have a
thors observed the coupled system responses decreased with increasing DLC close to zero.
suspension damping, and its impact on the pavement dynamics is more Tire force time-history can be used to determine the statistical dis-
significant is stiffer and rougher pavements. tribution parameters. If N measurements of tire forces, fk , are available,
In summary, none of these studies consider the gradual increase in 𝜎 is derived from:
the dynamic loads exerted to the pavement due to the vehicle-road in- 1 ∑𝑁
𝜎2 = 𝑓2 (2)
teraction as the pavement surface roughness gradually increases. 𝑁 𝑘=1 𝑘
The dynamic impact factor, DI, which can be multiplied by the static
Objective load to determine the dynamic load due to vehicle-pavement interac-
tion, is estimated from:
The main objective of this paper is to present a method that incorpo- 𝐷𝐼 = 1 + 𝑍𝑟 × 𝐷𝐿𝐶 (3)
rates the effects of the road-vehicle interaction in the assessment of the
pavement performance. The truck-pavement interaction was modeled to where Zr is reliability index considered.
estimate the additional dynamic loads exerted to pavements due to the Cebon [2] experimentally measured the dynamic loads under rigid 3-
interaction of the truck suspension system and the road surface rough- and 4-axle trucks. Sprung mass, unsprung mass and dynamic tire forces
ness. This procedure can be easily integrated into a tool that makes use were measured during his experiments. The results explained several es-
of finite element analysis to predict pavement response due to truck traf- sential features of heavy vehicle vibration behaviors. Other researchers
fic loads and that is capable to predict pavement distresses. The method have conducted experiments to obtain the DLC using weigh-in-motion
described in the paper was utilized to periodically modify the amplitude (WIM) sensors mounted on the tire-pavement interface [8] or stress-in-
of load applied to the pavement as it becomes more distressed. motion (SIM) measuring mats [5]. Cebon [2] also attempted to validate
the 2- and 3-dimensional numerical models developed for the simula-
tion of those vehicles. The agreement between the behavior of the mod-
Dynamic load coefficient
els and measurements was acceptable. The author presented additional
guidelines to improve the accuracy from numerical simulations by point-
The road roughness may exert additional loads to each wheel. A
ing up that road surface roughness is one of the most important factors
typical dynamic axle load probability density distribution, as shown in
that cause discrepancy between the models and measurements [3].
Fig. 1, represents the likelihood that the wheel load would have a given
Sweatman [22] studied the influences of the vehicle speed and
magnitude as it moves along a pavement section. The distribution has a
roughness on DLC with different suspension systems. His study showed
2
S. Misaghi, C. Tirado, S. Nazarian et al. Transportation Engineering 3 (2021) 100045
Table 1 Methodology
Percentage usages of different suspension systems (from [14]).
Suspension Percent Usage Two different approaches for solving a model excited by a rough
pavement can be pursued. This problem can be either solved in the time-
Tractors Trailers
domain or the frequency-domain [3]. The time-domain simulations are
Leaf spring 55 - 77 > 80 particularly attractive for nonlinear dynamic systems. The main concern
Walking beam 15 - 25 <2
with this type of solution is the computation time because of the intense
Air spring 15 - 20 10 - 15
Other 2-4 Nil numerical integration algorithms necessary. For complex dynamic sys-
tems that can be approximated as linear systems, the frequency domain
solutions are preferred. In this study a frequency domain solution, based
on the frequency response model, was implemented.
that the DLC was linearly related to the vehicle speed. Sun [21] con- Frequency response model is an input/output linear model with the
ducted a numerical simulation of IRI by using the power spectral density applied force being the input and displacement being the output [3].
(PSD) obtained from Newmark sequential integration. Statistical analy- For a given frequency, 𝜔, the input and output are related according to:
sis of his simulations showed a linear relation between the IRI and the
standard deviation of relative sprung mass vertical velocity. That study 𝑈 ( 𝜔 ) = 𝐻 (𝜔 ) × 𝐹 ( 𝜔 ) (4)
showed that if the PSD of roughness was defined by a polynomial func-
tion, the IRI could be calculated simply by means of the square root of where U(𝜔) is a vector of displacements at frequency 𝜔, F(𝜔) is the vec-
the sum of the weighted regression coefficients of roughness PSD. tor of resulting forces at frequency 𝜔 and H(𝜔) is called the frequency
Park et al. [17] conducted numerical simulations using a two- response function (FRF) or transfer function. The FRF is a matrix whose
dimensional (2D) half-truck FE model using OpenSees. The model con- elements are related to the masses, spring constants and damping char-
sisted of four elastic beam elements for the vehicle frame, four elastic acteristic of the suspension system. Since the goal is to obtain the forces
springs and four dashpot zero-length elements for the suspension and from known displacements, Eq. (4) can be rearranged to:
tires. The study consisted on simulating the moving of the vehicle pass-
𝐹 (𝜔) = 𝐻 (𝜔)∗ × 𝑈 (𝜔) × 𝐻 (𝜔)𝑇 (5)
ing through a typical horizontal land rigid road pavement with different
IRI and travel velocities. Similar to other studies, the DLC increased as where symbols ∗ and T signify the complex conjugate and the transpose
road roughness and vehicle speed increased. of matrix H(𝜔). In terms of power spectral density, the equation can be
Kazemi et al. [11] investigated the impact of roughness-induced dy- rephrased to [3]:
namic truck axle loadings on the pavement responses of two different
[ ] [ ]
pavement structures using 3D-Move. The authors introduced a parame- 𝑆𝐹 (𝜔) = [𝐻 (𝜔)]∗ × 𝑆𝑢 (𝜔) × [𝐻 (𝜔)]𝑇 (6)
ter called the Dynamic Response Coefficient (DRC), based on the coef-
ficient of variation for a specific response at a specific point within the where SF (𝜔) is a matrix of the power spectral densities of tire forces,
pavement structure, to account for the variation in pavement responses Su (𝜔) is a matrix of road profile power spectral density of displacement.
under variation in dynamic loading. Their results showed that the DRC Element H(i,j) of the matrix H(𝜔) is the dynamic force generated by tire
differed from the DLC, and also found that DRC was highly sensitive to i due to a unit harmonic displacement input at tire j.
temperature and location within the pavement structure.
The studies presented hereby agree that, given the impact of the
Simulating road roughness profile
pavement-vehicle interaction on the applied load to the pavement, it is
desirable that software packages that predict pavement performance be
The serviceability of a pavement is traditionally determined by its
powered by a method that considers the road-vehicle interaction.
smoothness. The serviceability concept was first described by Carey and
Irick [1] at the AASHO road test. The World Bank developed the In-
ternational Roughness Index (IRI) as a quantitative expression of the
smoothness of a pavement [19]. The IRI (ASTM E 1926) is defined as
Suspension systems
the average rectified slope (ARS) which is a ratio of accumulated suspen-
sion motion to the distance traveled. A mathematical model of a quarter
Three types of suspensions are popular in the US and throughout the
car one-wheel vehicle passing a measured profile at 80 km/h is used for
world. As reflected in Table 1, the most common suspension types are
that purpose.
the leaf spring, the walking beam, and the air spring suspension. Spring
Since there is not a direct approach to simulate a road based on a
ride suspensions, i.e. leaf spring and walking beams, consist of several
given IRI, a road corresponding to an IRI is reproduced to assess the im-
strips (leaves) of steel with rectangular cross sections are formed into
pact of road roughness. Based on measurements on European roads, ISO
semi-elliptical arcs and tied together. The axle is secured to the center
[10] proposed Eq. (7) to relate the vertical profile (u) spectral density
of the arc, while the two ends are connected to the vehicle frame. The
to wavenumber:
stacked strips act as a spring to carry the payload of the truck elasti- { −𝑛
cally. These springs also absorb the energy by dry (‘Coulomb’) friction 𝑆𝑢 (𝜅0 )( 𝜅𝜅 ) 1 𝜅
𝜅0
≤1
among leaves and contact points. Parabolic leaf spring is the newest de- 𝑆𝑢 (𝜅) = 0
𝜅 −𝑛2 𝜅 (7)
𝑆 𝑢 ( 𝜅0 ) ( 𝜅 ) 𝜅
> 1,
0 0
sign that uses fewer leaves. In this design, the thickness of the leaves
changes from the center to the end following a parabolic curve. Bene- where 𝜅 = wavenumber, cycle/m, 𝜅 0 = datum wavenumber, cycle/m,
fits of this type of design are reducing weight and better flexibility. Leaf Su (𝜅) = displacement spectral density, m3 /cycle, and Su (𝜅 0 ) = spectral
spring suspensions are usually used for tandem, tridem or more than density at 𝜅 0 , m3 /cycle or initial spectral density.
three axles. Traditionally, the behavior of leaf springs is modeled using After several iterations, Cebon [3] recommended the following val-
simple beam theory. The walking beam suspension system consists of ues for other parameters in Eq. (7), n1 = 3, n2 = 2.25, k0 = 1/(2𝜋)
two sets of leaf springs rigidly supported by beams that transmit the cycles/m. The initial power spectral density is the power of spectral
load to a beam centrally pivoted. Upper and lower shock insulators are density at datum wavenumber (1/2𝜋). The initial power displacement
attached to the beam at each side of the walking beam, and a large cap spectral density Su (𝜅 0 ) is the parameter which generates different road
clamps the insulators to the axles. No shock absorbers are used in this conditions. Table 2 provides a relationship between Su (𝜅 0 ) and the sub-
system. jective road roughness.
3
S. Misaghi, C. Tirado, S. Nazarian et al. Transportation Engineering 3 (2021) 100045
Table 2
Relationship between Su (𝜅 0 ) and road roughness (from [3]).
∑𝑁−1 √ 2𝜋𝑘𝑟
𝑍𝑟 = 𝑘=0 𝑆𝑘 𝑒𝑖(𝜃𝑘 + 𝑁
)
𝑟 = 0, 1, 2, ..., 𝑁 − 1 (8) Fig. 3. Correlation between Su (𝜅 0 ) and IRI.
4
S. Misaghi, C. Tirado, S. Nazarian et al. Transportation Engineering 3 (2021) 100045
Table 3
Statistical information about IRI for 1000 simulations.
Descriptive Statistics Road Roughness Initial Power Displacement Spectral Density Su (𝜅 0 ) (10−6 m3 /cycle)
Table 4 different parameters impact with speed on the DI, the baseline parame-
Specifications assumed for single tandem leaf spring baseline. ters were multipled by 14 , 12 , 2 and 4 times to consider a wide range of
Symbol Parameter Value values.
Fig. 4 shows the impact of varying the first and second suspension
ms Sprung Mass 6700 kg
stiffnesses from 1 MN/m to 16 MN/m simultaneously. In this figure,
mu1 = mu2 Unsprung Mass 500 kg
Is Pitch Inertia 930 kg⋅m2 as well as the following ones, baseline stiffness properties shown in
cs1 = cs2 Suspension Damping 80 kN⋅s/m Table 4 are represented in solid black. This figure clearly shows that
ct1 = ct2 Tire Damping 4 kN⋅s/m DI increases with increasing vehicle speed and spring stiffness. For a
ks1 = ks2 Suspension Stiffness 4 MN/m
stiffness of 16 MN/m and a speed of 137 km/h, the load applied to
kt1 = kt2 Tire Stiffness 3.5 MN/m
IRI Road Roughness 2 mm/m
the pavement is about 1.88 times the static load. However, a decrease
in magnitude of DI will occur if only one suspension stiffness is var-
ied while the second one is maintained at 4 MN/m, as shown in Fig. 5.
In this case, the dynamic load is only 1.45 times the static load for a
geometry of the model, describes how much force is transferred to each
stiffness of 16 MN/m and a speed of 137 km/h.
wheel.
Likewise, the impact of the tire stiffness was evaluated by keeping all
Generally, the equation of motion for a linear vehicle model can be
other parameters constant. As shown in Fig. 6, the tire stiffness does not
written in matrix form as:
have much of an impact on DI. The impact of the suspension damping
[𝑀 ]𝑞̈ + [𝐶 ]𝑞̇ + [𝐾 ]𝑞 = 𝑄 (11) characteristics on DI is shown in Fig. 7. The suspension damping co-
efficient significantly impacts the DI, especially as the vehicular speed
where [M] is the mass matrix, [C] is the damping matrix, and [K] is the
increases.
stiffness matrix. Combining Eqs. (10) and (11), one obtains:
At a given speed, DI decreases with the increase in the suspen-
[𝐵 ]𝑞̄ = [𝐷][𝑅]𝑢̄ (12) sion damping coefficient, except for a very high damping coefficient
(320 kN⋅s/m) that DI slightly increases. Even though not shown here,
where the damping coefficient of the tire has a minimal effect on DI.
[𝐵 ] = −𝜔2 [𝑀 ] + 𝑖𝜔[𝐶 ] + [𝐾 ]. (13) The impact of the load applied to the suspension (from 20% under to
20% over legal limits) is shown in Fig. 8. Similar to the previous case,
The vector of dynamic tire amplitude 𝐹̄𝑡 is given by as the load increases, the DI slightly decreases. This decrease is more
𝐹̄𝑡 = [𝑅]([𝐷]𝑇 𝑞 − 𝑢̄ ). (14) pronounced at higher speeds.
Finally, the impact of the change in IRI on DI is shown in Fig. 9.
Substitution of Eqs. (12) and (13) into Eq. (14) one obtains: At low speeds the change in dynamic load is small as the IRI increases.
𝐹̄𝑡 = [𝐻(𝜔)]𝑢̄ (15) However, as the speed increases, the DI is significantly impacted by the
IRI.
where
Impact of the dynamic impact (DI) factor and pavement distresses on
[𝐻(𝜔)] = [𝑅]([𝐷]𝑇 [𝐵 ]−1 [𝐷][𝑅] − [𝐼 ]) (16)
progress of deterioration
For n tires, matrices [R] and [D] can be defined as:
The previous section showed that the truck suspension system char-
⎡𝑘𝑡 + 𝑖𝜔𝑐𝑡1 0 ⋯ 0 ⎤
⎢ 1 0 𝑘𝑡2 + 𝑖𝜔𝑐𝑡2 0 ⋮ ⎥ acteristics and the road roughness have an impact on the magnitude
𝑅𝑛×𝑛 =⎢ ⎥ (17) of the load exerted to a pavement which, consequently, also affect the
⎢ ⋮ 0 ⋱ 0 ⎥
⎢ pavement distresses. In the case of flexible pavements, this increase in
⎣ 0 ⋯ 0 𝑘𝑡𝑛 + 𝑖𝜔𝑐𝑡𝑛 ⎥⎦
load magnitude will speed up propagation of fatigue cracks and rutting.
Thus, this will eventually lead to faster deterioration, and consequently,
⎡0 0 0 ⋯ 0⎤
⎢0 0 0 ⋯ 0⎥
a reduction in pavement service life. As these distresses increase with the
⎢ ⎥ number of truck passes, the road roughness will change as well. After
⎢1 0 0 ⋯ 0⎥
an extensive literature search, the WesTrack model [13] was selected
𝐷𝑛+2×𝑛 = ⎢0 1 0 ⋯ 0⎥ (18)
⎢ ⎥ to evaluate the impact of the additional loads on the surface rough-
⎢0 0 1 ⋯ 0⎥
⎢0 ness. This model relates the IRI and the major distress types for flexible
0 0 ⋱ 0⎥
⎢ ⎥ pavements. The proposed relationship was developed using field mea-
⎣0 0 0 ⋯ 1⎦
surements obtained from testing conducted at the WesTrack accelerated
pavement testing facility which involved pavement sections with differ-
Single tandem leaf spring suspension impact on dynamic impact
ent asphalt pavements and similar subsurface structure, as well as same
(DI) factor
traffic loading and environmental conditions with no frost heaving. This
model was evaluated using LTPP data was observed to have an average
A parametric study was conducted to evaluate the impact of the sus-
difference of 13.9% from the actual measured values. The relationship
pension spring stiffness and the pavement roughness on the dynamic
is in the form of
impact. For this purpose, a single tandem, leaf spring suspension system
with properties shown in Table 4 was used as baseline. To evaluate the 𝐼 𝑅𝐼 = 0.597𝐼 𝑅𝐼𝑖𝑛𝑖𝑡 + 0.0094𝐹 𝑎𝑡𝑖𝑔𝑢𝑒% + 0.00847𝑅𝑢𝑡𝐷𝑒𝑝𝑡ℎ + 0.382 (19)
5
S. Misaghi, C. Tirado, S. Nazarian et al. Transportation Engineering 3 (2021) 100045
1.88
2.00
Spring Stiffnesses, MN/m tandem leaf spring.
1 2 4 8 16
1.68
1.80
1.61
Dynamic Impact Factor
1.53
1.60
1.47
1.37
1.36
1.36
1.29
1.40
1.25
1.25
1.22
1.21
1.20
1.17
1.15
1.15
1.15
1.13
1.11
1.10
1.09
1.20
1.06
1.04
1.03
1.00
8 40 72 105 137
Vehicle Speed (km/hr)
1.45
1.50
1.43
First Spring Stiffness, MN/m DI for single tandem leaf spring.
1.37
1 2 4 8 16
1.35
1.40
1.33
1.29
1.29
Dynamic Impact Factor
1.27
1.25
1.24
1.30
1.23
1.22
1.21
1.19
1.18
1.17
1.15
1.15
1.20
1.12
1.11
1.08
1.07
1.06
1.04
1.04
1.10
1.00
8 40 72 105 137
Vehicle Speed (km/hr)
1.50 Fig. 6. Impact of the tire stiffness on DI for single tandem leaf
Tire Stiffness, MN/m spring.
1.38
1.37
1.37
1.37
1.28
1.25
1.24
1.30
1.23
1.22
1.20
1.17
1.16
1.16
1.15
1.20
1.13
1.10
1.06
1.06
1.06
1.10
1.03
1.02
1.00
8 40 72 105 137
Vehicle Speed (km/hr)
where IRI is measured in mm/m, RutDepth is in millimeters, and Fa- written as:
tigue% is the percentage of fatigue cracking in total lane area. This re- 𝑁2
lationship allows for an iterative process where, as rutting and fatigue 𝑅 𝐷 (𝑁 ) = 𝑈 ⋅ 𝜇 ⋅ 𝑁 −𝛼 𝑑𝑁, (21)
∫𝑁1
cracking progress, the IRI and the DI are updated.
In the performance model, the progression of rutting with load rep- where RD is the rut depth at the Nth load repetition and U is the layer
etition, N, was calculated using the VESYS model [12,27] in the form elastic deflection.
To predict fatigue cracking the Asphalt Institute MS-1 model was im-
plemented. Such model is suggested by the MEPDG as well [7]. Fatigue
𝜇 cracking is assumed to be related to tensile strain, 𝜀t , occurring at the
𝜀𝑝 = ⋅ 𝜀 ⋅ 𝑁 1−𝛼 , (20)
1−𝛼 𝑟 bottom of the HMA layers (bottom-up cracking). The form of the load
repetitions to failure of the cracking model is as follows:
where 𝜀p is the accumulated permanent strain, 𝜀r is the resilient elastic −𝑘2 −𝑘
𝑁𝑓 = 𝑘1 𝜀𝑡 𝐸𝐴𝐶𝑃
3
(22)
strain and 𝛼 and 𝜇 are the rate of increase in permanent deformation
against the number of load applications and the permanent deformation, where Nf is the number of load applications to failure, k1 = 0.0796,
respectively. The total elastic strain within a pavement layer is simply k2 = 3.291, and k3 = 0.854 are regression parameters based on a thresh-
the total compression of that layer, given by the difference in deflections old value of 20% cracked area and a typical HMA mix. Fatigue cracking
of the top and bottom of the layer [27]. For any layer this difference is predictions are calculated based on Miner’s law for cumulative damage,
6
S. Misaghi, C. Tirado, S. Nazarian et al. Transportation Engineering 3 (2021) 100045
1.61
1.70 Damping Coefficients, kNs/m Suspension
20 40 80 160 320
1.60 characteristics
Dynamic Impact Factor
1.47
1.47
1.50
1.40
1.37
1.36
Road condition
1.35
1.35
1.32
1.40
1.29
1.27
1.27
1.25
1.24
1.22
1.30
1.21
1.19
1.19
DI calculation
1.15
1.15
1.20
1.09
1.07
1.07
1.06
1.06
1.10
Pavement static
1.00 response
8 40 72 105 137
Vehicle Speed (km/hr)
Pavement
IRI recalculation
distresses
Fig. 7. Impact of damping coefficients on DI for single tandem leaf spring.
1.50 Fig. 10. Progressive growth of IRI due to additional surface roughness devel-
Sprung Mass, kg
oped by growth of pavement distresses due to truck passes.
1.41
1.39
Dynamic Impact Factor
1.37
1.36
1.35
1.40
1.32
1.30
1.29
1.30
1.23
1.22
1.21
Base
1.17
1.16
1.15
1.15
1.14
1.20
1.07
1.06
1.06
1.05
1.05
1.00
8 40 72 105 137
Vehicle Speed (km/hr) Fig. 11. Control pavement section.
quires the recalculation of DI after each truck pass since additional pave-
1.60
1 1.5 2 2.5 3
1.50 Fig. 10. These additional dynamic loads result in a continuous increase
1.38
1.37
1.40
ration even further.
1.29
1.28
1.25
1.24
1.30 To evaluate the impact of a variable IRI to a constant DI, both ap-
1.22
1.20
1.20
1.18
1.15
1.14
1.20
1.10
1.10
1.09
1.07
1.06
1.04
1.03
7
S. Misaghi, C. Tirado, S. Nazarian et al. Transportation Engineering 3 (2021) 100045
10 40
Failure Criterion
5 20
0 0
0.0 0.5 1.0 1.5 2.0 0.0 0.5 1.0 1.5 2.0
Standard Truck Passes (millions) Standard Truck Passes (millions)
2.0
(c) Surface Roughness
1.8
IRI (mm/m)
1.6 Constant DI
1.4 Progressive DI
1.2
1.0
0.0 0.5 1.0 1.5 2.0
Standard Truck Passes (millions)
This described methodology was incorporated into an algorithm that [6] G. Elnashar, R.B. Bhat, R. Sedaghati, Modeling and dynamic analysis of a vehi-
made use of a finite element program and used to obtain the response of cle-flexible pavement coupled system subjected to road surface excitation, J. Mech.
Sci. Technol. 33 (7) (2019) 3115–3125.
a typical flexible pavement to a standard truck loading. The algorithm [7] ERES Consultants, IncGuide for Mechanistic-Empirical Design of New and Rehabil-
was coupled with mechanistic-empirical distress models to predict pave- itated Pavement Structures. ELEC Final Report – NCHRP Project 1-37A. National
ment damage to evaluate the impact the DI has on the rate of deteriora- Cooperative Highway Research Program, Transportation Research Board, National
Research Council, Washington, DC, 2004.
tion. This involved the comparison of two approaches, one considering [8] J. Gajda, P. Burnos, R. Sroka, Accuracy assessment of weigh-in-motion systems
a constant DI factor and another involving the continuous growth in DI, for vehicle’s direct enforcement, IEEE Intell. Transp. Syst. Mag. 10 (1) (2018)
which involved the additional dynamic loads due to the deterioration 88–94.
[9] T.D. Gillespie, S.M. Karamihas, M.W. Sayers, M.A. Nasim, W. Hansen, N. Ehsan,
of the pavement with truck traffic. The additional dynamic loads due to D. Cebon, Effects of Heavy Vehicle Characteristics on Pavement Response and Per-
vehicle-road interaction was found to have a significant impact on the formance, Transportation Research Board of the National Academies, Washington,
deterioration of pavement in terms of rutting and fatigue cracking and DC, 1993 NCHRP Report 353.
[10] International Organization for StandardizationMechanical Vibration – Road Surface
a smaller impact on the functional performance of the pavement.
Profiles – Reported of Measured Data. ISO 8068: 1995 (E), Technical Committee
ISO/TC, Mechanical Vibration, Shock. Subcommittee SC2 Measurement, Evaluation
of Mechanical Vibration, & Shock as Applied to Machines, International Organiza-
Declaration of Competing Interest
tion for Standardization, Geneva, Switzerland, 1995.
[11] S.F. Kazemi, P.E. Sebaaly, R.V. Siddharthan, E.Y. Hajj, A.J. Hand, M. Ahsanuzza-
The authors declare that they have no known competing financial man, Dynamic pavement response coefficient to estimate the impact of variation in
interests or personal relationships that could have appeared to influence dynamic vehicle load, in: Proceedings of the 10th International Conference on the
Bearing Capacity of Roads, Railways and Airfields, 2017, pp. 702–718.
the work reported in this paper. [12] W.J. Kenis, VESYS users manual: Predictive design procedures, Federal Highway
Administration, 1977, FHWA A-RD-77-154.
[13] J.A. Mactutis, S.H. Alavi, W.C. Ott, Investigation of relationship between roughness
Acknowledgments and pavement surface distress based on WesTrack project, in: Transportation Re-
search Record: Journal of the Transportation Research Board, No. 1699, Transporta-
The authors would like to thank the New York State Department of tion Research Board of the National Academies, Washington, DC, 2000, pp. 107–113.
[14] J.R. Morris, Effects of Heavy Vehicle Characteristics on Pavement Response -
Transportation (NYSDOT), Federal Highway Administration (FHWA), Phase 1. National Cooperative Highway Research Program, Transportation Research
the Texas Department of Transportation (TxDOT) and Ohio Department Board, Washington, DC, 1987.
of Transportation (ODOT) for funding this study and providing valu- [15] R.M. Mulungye, P.M.O. Owende, K. Mellon, Analysis of response of flexible pave-
ments using finite element methods, Inst. Technol. Blanch. J. (2005) 40–52.
able input. The contents of this paper reflect the authors’ opinions, not [16] R.M. Mulungye, P.M.O. Owende, K. Mellon, Finite element modelling of flexible
necessarily the policies and findings of the listed agencies. pavements on soft soil subgrades, Mater. Des. 28 (3) (2007) 739–756.
[17] D.-W. Park, A.T. Papagiannakis, I.T. Kim, Analysis of dynamic vehicle loads us-
ing vehicle pavement interaction model, KSCE J. Civ. Eng. 18 (7) (2014) 2085–
References
2092.
[18] J.E. Saliba, Elastic-viscoplastic finite-element program for modeling tire/soil inter-
[1] W.N. Carey, P.E. Irick, The pavement serviceability–performance concept, Highw. action, J. Aircr. 27 (1990) 350–357.
Res. Board Bull. (1960) 40–58. [19] M.W. Sayers, T.D. Gillespie, W.D.O. Paterson, Guidelines for Conducting and Cali-
[2] D. Cebon, Heavy vehicle vibration - a case study, in: Proceedings of the 9th IAVSD brating Road Roughness Measurements, The World Bank, Washington, D.C., 1986
Symposium on the Dynamic of Vehicles on Roads and Tracks, 1985. World Bank Technical Paper Number 46.
[3] D. Cebon, Handbook of Vehicle-Road Interaction, Swets and Zeitlinger B.V., Lisse, [20] X.M. Shi, C.S. Cai, Simulation of dynamic effects of vehicles on pavement using a
The Netherlands, 1999. 3D interaction model, J. Transp. Eng. 135 (10) (2009) 736–744.
[4] D.H. Chen, M. Zaman, J. Laguros, A. Soltani, Assessment of computer programs [21] L. Sun, Simulation of pavement roughness and iri based on power spectral density,
for analysis of flexible pavement structure, Transp. Res. Rec. 1482 (137) (1995) Math. Comput. Simul. 61 (2003) 77–88.
123–133. [22] P.F. Sweatman, A Study of Dynamic Wheel Forces in Axle Group Suspensions of
[5] M. De Beer, C. Fisher, Stress-in-motion (SIM) system for capturing tri-axial tyre–road Heavy Vehicles, Australian Road Research Board, 1983.
interaction in the contact patch, Measurement 46 (7) (2013) 2155–2173.
8
S. Misaghi, C. Tirado, S. Nazarian et al. Transportation Engineering 3 (2021) 100045
[23] C. Tirado, C. Carrasco, J.M. Mares, N. Gharaibeh, S. Nazarian, J. Bendaña, Process Symposium on Heavy Vehicle Transport Technology (HVTT13), San Luis, Argentina,
to estimate permit costs for movement of heavy trucks on flexible pavements, in: 2014.
Transportation Research Record: Journal of the Transportation Research Board, No. [26] S. Zaghloul, T.D. White, Load equivalency factors for asphalt pavements (with dis-
2154, Transportation Research Board of the National Academies, Washington, DC, cussion), J. Assoc. Asphalt Pav. Technol. 63 (1994).
2010, pp. 187–196. [27] Zhou, F. and T. Scullion. VESYS5 Rutting Model Calibrations with Local Accelerated
[24] J. Uzan, JULEA Linear Elastic Analysis Computer Program, US Army Waterways Pavement Test Data and Associated Implementation. TX Report No. 9-1502-01-2.
Experiment Station, Vicksburg, MS, 1989. FHWA, U.S. Department of Transportation, Texas Transportation Institute, College
[25] M.C. Xu, R. Roebuck, D. Cebon, Modelling the performance of jointed plain con- Station, 2002.
crete pavements to dynamic vehicle loads, in: Proceedings of the 13th International