Professional Documents
Culture Documents
Responses
ABSTRACT
Aircraft induce high loads onto airfield pavement structures. High trafficking,
landing and traversing have been proven to accelerate pavement damage, with slippage
failure being one of the most significant distresses observed in airports, especially at
locations experiencing high shear stresses. Slippage is a bonding failure that happens
at the interface between layers. Tack and prime coats are the usual treatments for
bonding problems. This study focuses on modeling different layer interface properties
using the commercial element software ABAQUS. Properties thoroughly studied are
the shear stress and friction coefficient values. Those parameters were changed while
keeping the pavement structure and other parameters fixed to understand their effect
on pavement responses. The shear stress parameter is found to be only a limit value
that doesn’t affect model responses until it is surpassed, whereas it is observed that the
friction coefficient is more influential, especially at fully-bonded condition. However,
the effects of both parameters are found to be less important when compared to the effect
of loading. This investigation could eventually help in advancing pavement modeling
and especially layer interface properties making them closer to real conditions and real
asphalt concrete (AC) pavement behavior.
In the laboratory, test methods and protocols for measuring bond strength are gener-
ally grouped into three main load mechanisms; axial tension, torsional shear and direct
shear, as shown in Fig. 2 (White 2017). Axial tension tests measure the degree of
adhesion between the two layers. Torsional testing is less frequently reported. Direct
shear tests offer a more comprehensive assessment of the full interface strength likely
to be achieved in the field with adhesion, friction and interlock all contributing to the
resulting bond strength. Based on the test results, the interface bond can be measured
by its strength (peak stress), modulus/stiffness or work/energy.
It is well established that interface bonding characteristics are influenced by many
factors including: surface condition, preparation, temperature, tack coat material (emul-
sified asphalt, asphalt cutback or asphalt binder), tack coat curing time and tack coat
application rate (Tashman et al. 2008). According to Zhang et al., bonding shear strength
was found to decrease when the surface was not clean; had dust or moisture. Milled
surfaces always achieved higher bonding strength than non-milled, because of the added
friction. For temperature, increasing temperature results in decreased shear strength,
that is because at higher temperatures, asphalt viscosity is lowered and asphalt is more
likely to flow.
Additionally, AC mixtures designed with high air voids percentage or coarse aggregates
achieved lower shear strength (Zhang 2017). With regards to tack coat type, emulsified
tack coats are reported to have the highest shear strength at room temperatures. The
longer the curing time, the better the bond and finally for the application rate, there is
an optimum rate where adding more tack coat introduces a slippage plane (Du 2015).
At the interface between an AC layer and a granular layer, prime coats are used
instead of tack coats mainly to provide bonding and prevent migration of moisture.
Although emulsified asphalt prime coats can be used to help bond the top of the
granular base layer to the bottom of the AC layer (FAA 2018), little research has been
conducted on quantifying the required properties at this interface. The proper modeling
of the interface bonding conditions represents an important problem in understanding
the behavior of pavement structures and may lead to a reduction in maintenance and
rehabilitation costs (Romanoschi 1999).
This study focuses on a numerical approach for evaluating the effect of wheel load,
and the interaction properties (friction coefficient and shear stress limit) on pavement
responses, comparing the obtained maximum values for the stresses and strains of the
longitudinal, vertical and horizontal shear components. The simulations are performed
in the commercial software ABAQUS, and their inputs are based on the data obtained
from the major experimental efforts made at the National Airport Pavement Test Facility
(NAPTF). The NAPTF Construction Cycle 7 (CC7) was constructed on low-strength
DuPont Clay subgrade with a measured California Bearing Ratio (CBR) between 5%
and 6%, over two test areas, North and South, corresponding to perpetual pavement
testing and overload testing, respectively (FAA 2017). The pavement structure tested in
the South section is used in this study because its FEM model has been validated using
sensors data reported by FAA.
METHODOLOGY
A numerical model is generated using the commercial FEM software ABAQUS to
represent the chosen airfield pavement structure in order to enable a robust analysis of
the main responses under various combinations of wheel load (𝑃), friction coefficient
(𝜇), and shear stress limit (𝜇𝑚𝑎𝑥 ). All other parameters are kept constant for the purpose
of this study, using the considerations described in the following subsections. The
pavement structure corresponds to section tested as part of CC7 at the FAA’s National
Airport Pavement Testing Facility (NAPTF). A dynamic transient analysis is used to
For convenience, 4-8 Prony series terms are usually used. A 7-term Prony series
was found adequate for this study since the predicted values for 𝐸 ∗ reach convergence.
The least square method was used to optimize the initial shift factors and the fitting
parameters for the time-temperature superposition function. The shift factors were ob-
tained using the Williams-Landel-Ferry (WLF) equation as follows:
−𝐶1 (𝑇 − 𝑇𝑟 )
𝑙𝑜𝑔(𝛼𝑇 ) = (2)
𝐶2 (𝑇 − 𝑇𝑟 )
where:
𝑎𝑇 = AC time-temperature shift factor
𝑇 = temperature of interest
𝑇𝑟 = reference temperature
𝐶1 , 𝐶2 = regression coefficients
Finally, the obtained Prony series was normalized to the instantaneous modulus, to
be used as an input to ABAQUS to characterize the viscoelastic nature of the AC layer.
Granular Materials
The base and subbase materials are considered nonlinear cross-anisotropic and
stress-dependent to account for the thin 3-inch AC layer considered in the model. These
materials were characterized using a simplified procedure (Tutumluer and Thompson
1998), and implemented in ABAQUS through a user material subroutine (Wang and
Materials
Properties Base (P-209) Subbase (P-154) Subgrade (Dupont Clay
Material Model (1, Uzan Model) 1.00 1.00 1.00
𝐾1 1399.4 1526.0 1113.90
𝐾2 0.6075 0.7131 0.1369
𝐾3 -0.0806 -0.5467 -1.0290
𝜐12 0.35 0.35 0.40
𝜐31 0.35 0.35 0.40
Layer Interaction
The interaction properties between layers affect the pavement response to load
excitation. Therefore, their definition is critically important to understand the so-
called real conditions. An AC Coulomb friction model is usually implemented for
the interface between the AC and the granular base, the base and subbase, and the
subbase and subgrade. This model assumes that the resistance of movement is directly
proportional to the normal stress until the shear strength of the interface is reached.
Beyond this point, the layers slide relative to each other. The relationship between the
friction coefficient, the allowed maximum shear stress and normal stress is defined by
the following equation:
𝜏𝑐𝑟𝑖𝑡𝑖𝑐𝑎𝑙
𝜇= (3)
𝜎
where:
𝜏𝑐𝑟𝑖𝑡𝑖𝑐𝑎𝑙 = the allowed maximum shear stress prior to relative sliding of layers
𝜎 = the normal stress at the interface
𝜇 = friction coefficient
Those two parameters, namely the critical shear stress and the friction coefficient at
the interface are the focus of this study, and their effects on the pavement main responses
are investigated thoroughly.
Aircraft Tire Loading and Environmental Conditions
A tandem load configuration, a constant speed and tire inflation pressure are used in
this study. Tire inflation pressure was set as 200 psi and the tire moving speed was set to
2.5 mph representing taxiing speed at airports. Therefore, the estimated contact stress
distribution is on only dependent on the wheel load for the simulated airfield pavement.
The final mesh configuration (i.e., layer thickness, number of elements, and bias)
for the simulated airfield pavement structures is presented in Table 2:
Layers Element
Layer t # Elements Bias First Last
AC (P-401) 75 4 1.40 15.72 22.01
Base (P-209) 150 6 1.20 22.78 27.33
Uncrushed Stone (P-154) 500 15 1.10 31.77 34.95
Subgrade (P-154) 14275 40 35.00 36.49 1277.10
Pavement Responses
Stresses and strains in any direction can be extracted at any location within the
problem domain once a simulation is completed using the aforementioned 3D FEM
model. However, the targeted output values are located at the bottom of the AC layer, at
the interface with the granular base layer. Within this subdomain, the critical responses
include the following:
• Longitudinal stress (𝜎11 ), and longitudinal strain (𝜖11 ) for bottom-up cracking.
• Vertical compressive stress (𝜎22 ) and vertical compressive strain (𝜖22 ) for rutting.
• Horizontal shear stress (𝜎13 ) and horizontal shear strain (𝜖13 ) for slippage failure.
Fig. 5. Wheel load (P) vs stress (𝜎) for 𝜎11 , 𝜎22 and 𝜎13
Fig. 6 and Fig. 7 show the influence of the friction coefficient at the interface between
the AC and the granular base layer on the overall main responses. Fig. 6 presents the
longitudinal, vertical and shear strains whereas Fig. 7 shows the corresponding changes
in stress in those directions. The trend is clear, a greater bond strength (higher friction
coefficient) leads to lower responses. However, the effect is only considerable when
comparing a fully-bonded case to any other condition. A significant distortion is seen
when setting a fully-bonded interaction, with differences of 28%, 2% and 22% for the
longitudinal, vertical and horizontal shear stress, and up to 10%, 13% and 2% for the
longitudinal, vertical and horizontal shear strain. The differences observed in the main
responses when the friction coefficient is slightly modified are negligible.
Fig. 7. Friction Coefficient (𝜇) vs stress (𝜎) for 𝜎11 , 𝜎22 and 𝜎13 .
Fig. 8 and Fig. 9 show the influence of the shear stress limit at the interface between
the AC and the granular base layer on the main responses. Fig. 8 shows the longitudinal,
vertical and shear strains while Fig. 9 shows the corresponding changes in stress in those
directions. It can be seen that generally there is no effect on the overall responses when
changing this parameter from 1.5 to 0.6 MPa. Only when the value is set to as low as
0.2 MPa, a distortion is noticeable. However, at this point, the model failed to reach
convergence due to large nodal displacements between two consecutive time steps. Since
𝜏𝑐𝑟𝑖𝑡𝑖𝑐𝑎𝑙 establishes a limit, no effect is observed unless the value is surpassed. If the
maximum horizontal shear stress at the interface, which is a specific combination of the
other normal stresses, exceeds the established limit, the model becomes non-convergent.
Fig. 9. Shear Stress Limit (𝜏𝑚𝑎𝑥 ) vs stress (𝜎) for 𝜎11 , 𝜎22 and 𝜎13 .
As can be seen in Fig. 10 the model doesn’t reach convergence when the shear stress
limit value is set as 0.2 MPa, and fails suddenly after 16 time steps are performed. The
longitudinal stress seems to have a normal trend up to the failure point. However, a
sharp increase in the vertical stress occurs at step number 15 (notice that numbering
starts at zero). This might be explained by the fact that when slipping happens, the
area of contact between the tire and pavement decreases, which results in an increase
in vertical stress. Horizontal shear stress increases at failure as expected. The model
failure can be explained by the displacement thresholds set for the model. The model
started to slip, displacements exceeded the thresholds and the model failed.
The effects of changing the interaction properties when only vertical loading is
applied at every time step were not as noticeable as the effect of changing the wheel
load. The authors hypothesize that when operating conditions such as braking, cornering
and accelerating are considered, those properties will be more impactful on the pavement
overall responses.
RECOMMENDATIONS