Impact of Dynamic Wheel Loading on Flexible Pavement Responses for Non-Free Rolling Conditions

Abstract

Since 2000, pavement design methodologies have transitioned from empirical to mechanistic-empirical procedures. However, the complexities of loading patterns, contact stress distribution, material characterization, vehicle maneuvering, and dynamic load amplification are still not fully considered, despite their significant effect on pavement performance. In this study, state-of-the-art numerical models were used to investigate the changes in critical pavement responses derived from the combined effect of roughness-induced dynamic loading and vehicle maneuvering. A decoupled vehicle–tire–pavement interaction approach composed of a random process to generate artificial road roughness profiles, a mechanical full truck model, and three-dimensional finite element tire and flexible pavement models allow the prediction of the impact of road roughness on vehicle dynamics. In addition, the study quantifies the effect of dynamic loading on contact stresses, and consequently, on pavement critical responses. Because of the expected increase in axle weight from truck electrification, overweight scenarios were also considered. Changes in load history and distribution of strain fields were assessed for two typical pavement structures (thin and thick). The combined effect of roughness-induced dynamic loading and vehicle maneuvering greatly altered the critical pavement responses associated with bottom-up fatigue cracking, near-surface cracking, and rutting. Under the most adverse conditions, the critical tensile strains at the bottom of the asphalt concrete increased up to 125%, the shear strain increased up to 100%, and the compressive strain escalated to 120% when compared with the reference cases. Higher temperatures exacerbated the impact of dynamic wheel loading and vehicle maneuvering.

Keywords

infrastructure pavements design and rehabilitation of asphalt pavements asphalt mechanistic-empirical pavement design pavement design pavement modeling

The force applied by a wheel on a pavement surface varies with the dynamic response of the vehicle’s mechanical system, a phenomenon known as dynamic wheel loading (DWL). It arises from the oscillatory motions of the wheels as vehicles roll over irregularities on the surface. Rougher pavement surfaces could worsen the DWL effect, induce greater pavement responses, and accelerate pavement failure at locations where high dynamic loads are induced (1 –3).

Under free rolling conditions, DWL primarily affects pavement responses at depths beyond the near-surface region, specifically at the bottom of the asphalt concrete (AC) layer. This is because these depths are more sensitive to higher load magnitudes compared with near-surface responses. this does not hold true for non-free rolling conditions such as braking, acceleration, or cornering, however, because near-surface responses are considerably affected by tire–pavement surface tangential stresses ( 4 ). Therefore, the in-plane components of the contact stress distribution become of special interest.

Roughness-induced DWL leads to an increase in the static load value, primarily influenced by the pavement condition, traveling speed, and vehicle dynamic properties. In addition, the increased in-plane contact stresses generated by vehicle maneuvering results in higher longitudinal and shear components of the stress and strain fields in the regions under the tires. It has been shown that AC layers are more likely to experience distresses such as near-surface cracking and rutting under these circumstances ( 5 , 6 ). Indeed, Hajj et al. ( 7 ) found that braking induces greater rut depth in hot mix asphalt layers. Shear- and slippage-related distresses are typically observed in intersections, exits, ramps, and prolonged and steep segments of highways that experience frequent vehicle maneuvering, where modified design parameters might be necessary ( 8 , 9 ).

Consequently, quantifying the impact of incorporating vehicle maneuvering and DWL on pavement analysis is essential to update design guidelines to meet future demands caused by factors such as truck electrification ( 10 ), and deterioration of the highway system ( 11 ).

Objectives

The main objective of this study was to quantify the combined effect of roughness-induced DWL and non-free rolling conditions (full braking and full traction) on flexible pavement responses. To achieve this objective, the steps shown in Figure 1 were undertaken:

A target international roughness index (IRI) was defined to generate an artificial multi-track road roughness profile for a given pavement condition using band-limited white noise (BLWN).

Multi-track road roughness profiles were used as an input excitation to estimate the dynamic responses of a mechanical three-dimensional (3D) truck model and the changes in axle load were recorded through time for a given traveling speed (dynamic load profile).

Numerical simulations were conducted using representative discrete loads from the dynamic loading profile using a hyperelastic 3D finite element (FE) tire model to predict tire–pavement contact stresses.

The dynamic contact stresses were applied to a 3D FE flexible pavement model using the continuous moving load approach to record pavement responses.

In this framework, a decoupled vehicle–tire–pavement interaction approach is used, where the output of every step serves as the main input of the subsequent step until the final output is produced.

Numerical Modeling

State-of-the-art numerical models were considered to analyze the effect of various parameters on pavement responses, such as load configuration, axle load, temperature, pavement structure, and rolling condition.

Artificially Generated Road Roughness Profiles

The versatility of artificial road roughness profiles allows us to conduct thorough parametric studies. Standardized procedures (e.g., ISO 8608) have been developed to provide guidelines for roughness profile generation ( 12 ). However, overlooking inherent features of measured-on-site profiles, such as non-stationarity and local roughness variance, might affect the randomness of the generated profile. Liu and Al-Qadi ( 13 ) addressed this limitation, and their model was used in this study to artificially generate roughness profiles using BLWN.

A target IRI was defined to represent a pavement in poor condition (IRI > 172 in./mi), per the Federal Highway Administration (FHWA) classification. Multi-track road roughness profiles were artificially generated (left and right wheel paths) for an IRI of 220 in./mi, as shown in Figure 2. Because the IRI is a statistical measure of the amplitude and frequency of pavement surface unevenness, calculated from a measured longitudinal road profile by accumulating the output from a quarter-car model and dividing the result by the profile length ( 14 ), it is not influenced by the specific shape of the roughness profile. Consequently, completely different profiles may produce the same IRI as long as their amplitudes and frequencies are similar.

Road roughness profiles may be used as an input excitation to study the dynamic response of complex mechanical systems such as a Class-9 vehicle. Liu and Al-Qadi ( 15 ) developed and validated a 21-degree-of freedom (DOF) mechanical full truck model using the dynamic properties estimated by Kang et al. ( 16 ), Law et al. ( 17 ), and Abdelkareem et al. ( 18 ). By controlling some of the mechanical properties of the model, such as the axle weight, trailer mass, and tire stiffness and damping, various tire configurations and net weight scenarios were studied through numerical simulations. The 3D truck model used in this study assumed a non-deformable pavement surface, a solid axle system, and a linear viscoelastic suspension system. Other suspension types might require a fine-tuned model to be properly represented.

Three scenarios were selected, as shown in Figure 3. The average value of steering axle and drive axle weights, using analyzed weigh-in-motion data from across Illinois, were considered as references ( 3 ). Two possible distributions of extra weight derived from the incorporation of battery packs in electric trucks (E-truck) were assessed: (i) battery weight is fully carried by the steering axle, and (ii) battery weight is fully carried by the drive axle.

The maximum axle loads of the distribution were used for analysis. Therefore, an 11 kip and 19 kip steering axle, along with a 32 kip and 38 kip tandem axle, were selected. The dual-tire assembly (DTA), the most common configuration in the truck industry, was assumed for all of the axles, except for the steering axle, which was represented by considering one of the wheels from a DTA 275/80 R22.5.

Dynamic loading profiles were obtained after conducting simulations using the 21-DOF mechanical 3D truck model through a MATLAB/Simulink implementation. The simulations were performed at two traveling speeds: 35 mph and 70 mph, which corresponded to common speed limits for low-volume roads and interstate highways, respectively. Because the model was solved using the state–space representation in the time domain, the multi-track profiles needed to be converted into time-elevation data using a constant speed. The dynamic loading profiles for both the steering and the tandem axles are shown in Figure 4. The load carried by the tandem configuration was assumed to be evenly distributed among its axles. The stiffness of the tire configuration and the total weight carried by a given axle greatly influence the magnitude of the dynamic load amplification. In both cases, however, it is clearly shown that higher speeds induce higher load amplifications because the mechanical system is sensitive to the rate of the input excitation.

Discrete loads were randomly sampled from the dynamic loading profiles using a percentile approach, considering only the values within the fifth and 95th percentiles. Assuming an evenly distributed load, each wheel of the DTA configuration effectively carried only one-eighth of the total tandem axle load, while it carried one-half of the total load for the steering axle. As a result, a set of different wheel loads ranging from 3.50 to 10.50 kips were extracted and later used to conduct simulations using an hyperelastic 3D FE tire model for the two constant traveling speeds.

3D Tire Model

Contact stresses were predicted for a specific set of parameters using the 3D FE tire model of a DTA 275/80 R22.5 developed by Hernandez et al. ( 6 ). The model uses the Mooney-Rivlin hyperelastic constitutive model to characterize the stress–strain behavior of rubber materials. As a consequence, the temperature and frequency dependency were neglected. Moreover, the contact surface was assumed to be rigid because the rubber deformation is significantly greater than that experienced by the pavement.

Simulations were performed assuming a constant tire inflation pressure of 100 pounds per square inch and two traveling speeds (35 mph and 70 mph) for each of the loads included in the discrete loading profile. Because the free rolling condition occurs when the torque transmitted to the ground has a negligible value, the angular velocity $ω,$ at which this condition is met could be calculated through iteration. Once $ω$ was known, the braking and accelerating scenarios could be specified by defining the slip ratio. The slip ratio was set at 10%; beyond that value, no changes could be distinguished in the contact stress distribution. Although Liu et al. ( 19 ) developed a comprehensive database of contact stresses that quantified the influence of several input parameters, the slip ratio selected for this study was not covered and further numerical simulations needed to be conducted. Figure 5 illustrates the step-by-step process:

3D Pavement Model

Current design frameworks (e.g., AASHTOWare) use multilayer elastic analysis as the predominant means to calculate flexible pavement responses to loading. Under that approach, the AC, granular layer(s), and subgrade are assumed to be linear elastic materials. Furthermore, several assumptions are made, including that loads are static, which actually have a transient and dynamic nature. In addition, AC materials exhibit temperature and frequency dependency, while granular materials behave elastically only under low stress levels. Moreover, tire contact stresses are nonuniform. By employing FE methods and commercially available software, it is possible to conduct a thorough analysis that accurately represents field conditions. The latest version of the 3D FE pavement model developed by Al-Qadi et al. (20 –25) was utilized, and the most relevant considerations are discussed in the following sections.

Material Characterization

Elseifi et al. ( 22 ) demonstrated that linear elastic characterization of AC materials could grossly underestimate responses at intermediate and high temperatures and at low load frequencies. Linear viscoelastic constitutive models could be used instead. The time–temperature superposition principle could therefore be utilized to represent the time and frequency dependency of AC materials.

In mechanistic pavement design frameworks, the AC dynamic modulus ( $| E^{*} |$ ) is a key parameter to characterize material stiffness. If dynamic modulus test data are available, an AC mixture could be characterized by fitting a model to the test results. A $| E^{*} |$ master curve could be constructed by selecting a reference temperature and shifting the various test temperatures relative to the loading frequency. Although different mathematical models have been developed over the years, the sigmoidal function, applied by AASHTOWare, was employed.

Al-Qadi et al. ( 26 ) analyzed the database of the Long-Term Pavement Performance (LTPP) program and determined appropriate characterization to represent strong and weak sets for the AC surface course (WS), intermediate course (IM), and base course (BL). The master curves are shown in Figure 6.

On the other hand, the assumption of linear elasticity for granular layers might also lead to inaccuracies in the responses, especially for roads with thin surfaces. Nonlinear stress-dependent characterization offers a better means for characterization and has been found to result in better prediction of pavement responses ( 27 ). Kim et al. ( 28 ) characterized the stress-dependent behavior of subgrade and unbounded soils by incorporating a user material subroutine into ABAQUS. This implementation was based on the work of Tutumluer and Thompson ( 29 ). Their work presented a cross-anisotropic model based on extensive repeated-load triaxial tests to predict the nonlinear modulus of granular materials. Additionally, to characterize the resilient properties of the subgrade soil under high stress, a bilinear model proposed by Thompson and Robnett ( 30 ) was used to represent the stress-softening behavior.

Table 1 summarizes the characterization parameters used in the current study for the granular layers and the subgrade. It should be noted that for the subgrade, $E_{\max, \min}$ and $k_{3, 4}$ are expressed in MPa.

Table 1.

Stress-Dependency Parameters and Material Properties of Base Layer and Subgrade

Layer	Poisson’s ratio	Density (ton/mm³)	Parameter	Value
Base	0.35	2.0E-09	$K_{1}$	661.448
			$K_{2}$	0.932
			$K_{3}$	−0.569
			$K_{4}$	415.000
			$K_{5}$	0.906
			$K_{6}$	−1.240
			$K_{7}$	349.712
			$K_{8}$	0.969
			$K_{9}$	−0.672
Subgrade	0.40	1.50E-09	$E_{\max}$ (MPa)	69.20
			$E_{\min}$ (MPa)	24.60
			$k_{1}$	−635.00
			$k_{2}$	−201.00
			$k_{3}$ (MPa)	52.60
			$k_{4}$ (MPa)	0.047

Note: Conversion factors: 1 Ton/mm³ = 13,154.22 kip/in.³; 1 MPa = 145.04 pounds per square inch.

Under low stress levels, granular materials were assumed to be linear elastic (thick structures). The elastic modulus values used for the base and the subgrade under these conditions were 60.20 ksi (415 MPa) and 10.88 ksi (75 MPa), respectively. The value of the elastic modulus was selected based on seasonal conditions ( 31 ).

Temperature Profile

Wang et al. ( 27 ) derived a mechanistic approach to estimate the temperature field in a multilayered system. The temperature profiles shown in Figure 7 were constructed using this approach and assumed to be representative of winter (−10°C), fall (25°C), and summer (45°C) conditions.

Dynamic Transient Analysis

The frequency-dependence of AC materials renders static analysis unsuitable for predicting pavement responses. It underestimates in-plane strains at the bottom of the AC and the vertical strain on the subgrade ( 32 ). Although a quasi-static approach could incorporate a moving load in the analysis, it neglects inertial and damping effects. Thus, the dynamic transient approach was used. The analysis was performed within a time frame dependent on the traveling speed, which was later divided into time steps, where a given load could be applied at different and subsequent positions through increments at every step. This enabled smooth loading and unloading of surface elements in the wheel path. This smooth transition is key for the implementation of dynamic contact stresses because they change with the load position. In addition, sudden changes in load might induce convergence problems if not addressed properly. In this case, the 3D contact stresses transition from an initial to a final value throughout the time increments, each covering the same span of time. A constant speed for the moving load is assumed during the analysis. Given the space domain of the FE flexible pavement model, this was considered a reasonable assumption.

Contact Stresses

The contact stresses predicted using the FE tire model were shifted along the wheel path length in the direction of the traffic. The geometry of the wheel path is dependent on the tire imprint dimensions and the mesh grid chosen to represent it. Although a fine mesh could be used to capture most of the features in the contact stress distribution, it would be computationally inefficient. Moreover, the dimensions of the tire imprint change with load and tire inflation pressure. Consequently, after performing a sensitivity analysis to loading, an adequate tire imprint was defined based on the load range of the discrete loading profile.

To illustrate this concept, tandem axle modeling is described. A tire imprint 8.66 in. wide (220 mm) and 7.87 in. long (200 mm) was found to be sufficiently accurate in representing the contact area of a DTA for a load range between 7 and 11.5 kips. An element length of 1 in. was selected, and every rib was partitioned into two elements. The grooves were not included because their widths in the deformed configuration are negligible.

An element width of roughly 1 in. was selected for Ribs 1 and 5, while Ribs 2, 3, and 4 were assigned an element width of 0.8 in. Considering a tire spacing of 4 in. (100 mm), the final dimensions of the mesh grid were 21.26 in. × 7.87 in. (540 mm × 200 mm), as shown in Figure 8.

These considerations, along with an axle spacing of 41.34 in. (1,050 mm), resulted in a wheel path length of 169.29 in. (4,300 mm) to complete a single pass of the tandem axle. The number of time steps required to cover this distance depends on the number of elements shifted at every step. Al-Qadi et al. ( 26 ) concluded that a three-element span offered a good balance between accuracy and computational time. Therefore, the same number of time increments was defined to provide a smooth loading transition.

Based on the element length, the time increment value could be obtained for any constant speed (e.g., 0.8 ms at 70 mph and 1.6 ms at 35 mph). The consideration of a three-element span led to a 39 time-step process to represent a continuous moving load, translating 115.16 in. (2,925 mm) during the simulation. A similar process was conducted to determine the optimal dimensions for the steering axle.

Mesh Refinement

An adequate number of elements was selected to represent the model so a continuous stress transition throughout the domain could be achieved without jeopardizing the model accuracy. Mesh verification was performed for two traditional pavement structures representing a low-volume road and an interstate highway. The number of elements in the longitudinal and transverse directions (out of the wheel path) was selected to ensure the stresses and strains dissipate to nearly zero at the edges, which are represented by infinite elements. Although eight-node linear brick elements (C3D8) were mainly used, continuum isoparametric nonlinear elements (CIN3D8) were assigned to the elements located at the boundaries and at the bottom of the subgrade. The flexible pavement model was composed of three main regions: the wheel path, the transition zone, and the infinite elements, as depicted by Figure 9.

Table 2 presents in detail the mesh configuration and the overall dimensions (L, B, D, L1/B1, L2/B3, L3/B3, X and b) as shown in Figure 9 that were assumed for the thin and thick structures to represent a low-volume road and an interstate highway, respectively. To optimize the number of elements across the depth, a nonuniform distribution was selected by defining the ratio of coarsest element to finest element for every layer (bias). This means the element size increases with depth, instead of being constant.

Table 2.

Mesh Configuration and Geometric Dimensions of the Simulated Pavement Structures

Mesh configuration
Layer	Thin structure			Thick structure
Layer	Thickness (mm)	#Elem	Bias	Thickness (mm)	#Elem	Bias
Surface course	100	8	2.00	50	4	1.50
Intermediate course	–	–	–	75	4	1.50
Base course	–	–	–	150	6	1.20
Base layer	300	10	2.50	300	8	1.75
Subgrade	9 600	30	22.00	9 425	35	19.00
Geometric dimensions
Dimension	Magnitude (mm)		Dimension		Magnitude (mm)
Length (L)	15 800		Wheelpath length (X)		4 300
Width (B)	12 040		Wheelpath width (b)		540
Depth (D)	10 000		L2 and B2		500
L1 and B1	4 500		L3 and B3		750

Note: Conversion factor: 1 mm = 0.039 in.

Figure 10 illustrates an isometric view of a three-layer flexible highway pavement model, where the orientation in space can be observed.

Although the wheel path dimensions are small when compared with the overall size, the effects of the continuous moving load propagate throughout the model. In the absence of a transition zone, these effects can “bounce” near the edges, affecting the quality of the prediction and misrepresenting the physical behavior of the system.

In the subsequent sections, the various constituents of the theoretical framework are interconnected and deployed through a numerical matrix. Because each load had a corresponding contact stress distribution at a specific rolling condition, dynamic contact stresses were applied to two pavement structures (thin and thick) for different seasonal conditions.

Vehicle–Tire–Pavement Interaction System

A numerical matrix was constructed to assess the main responses of two different flexible pavement structures when subjected to DWL under non-free rolling conditions. The thin pavement structure comprises of a 4 in. (100 mm) AC layer, a 12 in. (300 mm) base layer, and a subgrade, while the thick pavement structure was composed of a 11 in. (275 mm) AC layer subdivided into a 2 in. (50 mm) surface course, a 3 in. (75 mm) intermediate course, and a 6 in. (150 mm) base course, along with a 12 in. (300 mm) base layer and a subgrade. Figure 11 illustrates the layer distribution and the material properties of the two pavement structures.

The factors included in the numerical matrix were the following:

• Two structures (thick and thin) representing high- and low-volume roads. Thick and thin pavements are simulated at a traveling speed of 70 mph and 35 mph, respectively. One pavement condition (poor, IRI > 170 in./mi).

• Two steering axle loads (11 and 19 kips) and two tandem loads (16 kips to represent a standard 32 kip tandem axle and 19 kips to represent an overweight 38 kip tandem axle).

• Two surface temperatures for the AC layer: 14°F (−10°C) to capture the material properties during winter and 113°F (45°C), for summer conditions. The subgrade elastic modulus was changed accordingly for the thick pavement structure.

• Two non-free rolling conditions: full braking and full acceleration, at 10% slip ratio.

• One constitutive model of the DTA (hyperelastic).

The numerical matrix therefore consisted of 32 models, based on a combination of the aforementioned parameters, as shown in Figure 12.

In addition to those cases, the following cases were defined for both the steering and tandem axle:

• Two baseline cases, one thick and one thin, both under static load, free rolling conditions, and a surface temperature of 77°F (25°C) to establish reference values.

• Two complementary cases, one thick and one thin, under dynamic loading using the overweight scenario, non-free rolling conditions but modifying the characterization of the AC layer to represent a weak set from the LTPP database so that the influence of the material strength could also be studied.

A total of 40 numerical simulations were conducted, including the baseline and additional cases.

Results and Discussion

Because of the similarities in the absolute values and distribution of contact stresses observed between the full braking and full acceleration scenarios, an analysis of the difference in the critical pavement responses caused by these cases was first conducted. An analysis of the variation in the main components of the strain field is also presented. The results were compared with the baseline cases to understand the influence of each parameter.

Load History

The continuous moving load approach and the viscoelastic characterization of the AC allowed for the analysis of the response at a fixed point throughout the simulation time. For instance, Figure 13 shows the results for a point located at the middle of the wheel path at a depth of 4 in. (100 mm). For the thin structure, this point is right at the interface between the AC and the base layer. Three cases were selected for comparison: the baseline case “FR” (free rolling, static load), a full braking case “OW_FB” (overweight, dynamic load), and a full acceleration case “OW_FT” (overweight, dynamic load) for the DTA tire (tandem axle)

The longitudinal tensile strain ( $ε_{11}$ ), transverse tensile strain ( $ε_{33}$ ), shear strain ( $ε_{23}$ ), and compressive strain ( $ε_{22}$ ) were examined and compared for different rolling conditions. The results showed that the free rolling case had considerably lower values than any of the non-free rolling cases for all the components of the analyzed strain fields. However, as the depth increased, the difference between the full braking and full acceleration scenarios became less significant. This could be because the in-plane components of the contact stress distribution mainly affect the near-surface responses.

Figure 14 presents the results at a depth of 2 in. (50 mm), which represents the interface between the surface course and the intermediate course in the thick pavement structure.

At shallow depths, the difference between the full braking and full acceleration scenarios became accentuated for all strain components. A shift in the distribution between the two cases could be observed, particularly in the $ε_{11}$ component. Nevertheless, the critical main responses were still quite similar. The free rolling scenario underpredicted all the responses because the effects of road roughness and vehicle maneuvering were overlooked. A snapshot at a location below the surface was analyzed and is presented in the following section.

Strain Field Distribution

Figures 13 and 14 illustrate the load history at a fixed point. Figure 15, however, illustrates the responses along a longitudinal profile drawn from the start to the end of the wheel path at a constant depth of 0.5 in. (10 mm), for the thick pavement structure. Full braking and full acceleration scenarios generated longitudinal contact stresses that were similar in magnitude but opposite in direction. While the tensile strain distribution was slightly shifted to the front at full braking, it was shifted to the back under full acceleration. Even though the location of the peaks of the distribution changed, the maximum values did not yield significant differences, even at a near-surface location. While the percentage difference was up to 16.91% for the transverse strain, the absolute value was too small to be considered significant. The longitudinal, shear, and transverse strains had differences of 3.23%, 8.42%, and 2.25%, respectively, which were negligible in their absolute values.

On the basis of these results, it was concluded that, as long as braking and accelerating scenarios are defined using the same slip ratio, the critical pavement responses remain in the same order of magnitude, despite the changes in the distribution.

Critical Pavement Responses

Because the full braking and full acceleration scenarios yielded similar maximum point responses despite the change in the location of the peaks and the overall distribution, the analysis in this section focused on the accelerating cases.

Responses Related to “Bottom-Up” Fatigue Cracking

Two components of the strain field are highly associated with bottom-up fatigue cracking: the longitudinal tensile strain ( $ε_{11}$ ) and the transverse tensile strain ( $ε_{33}$ ) at the bottom of the AC layer. Figures 16 and 17 illustrate the changes in the maximum values of these components for various AC surface temperatures and loading scenarios.

As expected, the thin structure yielded higher maximum responses than those of the thick pavement. Braking and roughness-induced DWL could greatly alter the predicted values of $ε_{11}$ and $ε_{33}$ , especially in the case of overweight scenarios. The results were approximately twice the value of the reference cases for the drive axle (95% increase). The difference was greater for the steering axle (125% increase), where one tire carries half the axle loading. During winter, the combined effect of the non-free rolling conditions and DWL could be minimized given the considerable increase in the AC dynamic modulus. The longitudinal and transverse strains were found to be comparable in magnitude for the drive and steering axles.

Responses Related to Near-Surface Cracking

The shear strain ( $ε_{23}$ ) within the AC layer is linked to near-surface cracking ( 33 ). For the thick pavement, the maximum shear strain was found within the first 3 in., while in the thin pavement, it was approximately 2 in. deep. Similar to the responses related to bottom-up fatigue cracking, the consideration of non-free rolling conditions and DWL severely altered the predicted maximum responses. For the drive axle, when overweight was ignored, the increase in the shear strain could be up to 70% for the thin pavement and 65% for the thick pavement during summer conditions. When overweight was considered, assuming the weight was entirely carried by the drive axle, the increase could be up to 100% for both the thin and thick pavements. The effects were less pronounced during winter conditions. When DWL and high temperature were combined, the effect was more pronounced (Figure 18).

Figure 19 shows that when considering overweight, DWL, high temperatures and non-free rolling conditions in the analysis, the increase in shear strain for the steering axle could be as high as 140% for both the thick and thin structures.

Responses Related to Rutting

The vertical compressive strain ( $ε_{22}$ ) within each pavement layer is the primary factor in permanent deformation. These values within the AC layer, and at the top of the base and the subgrade, are especially important because they constitute the main causes of rutting. Whatever the pavement structure and the axle configuration, the highest vertical compressive strain was found in the base layer for all cases. Figures 20 and 21 present the values of $ε_{22}$ for the thin structure for both the drive and steering axles.

Although the compressive strain was higher for the steering axle within the AC layer, this was not the case for the base layer and the subgrade. Even though a higher wheel load was associated with the steering axle, load history-related responses associated with the tandem configuration might explain the increased responses for the drive axle at deeper locations, where the effects of axle interaction could be observed.

For the thin structure, the combined effect of temperature, overweight, rolling condition, and DWL increased the compressive strain of the AC, base, and subgrade by up to 85%, 40%, and 45%, respectively, for the drive axle, and up to 120%, 85%, and 90% respectively, for the steering axle. Between the two axle configurations, an opposite trend in the variation of the compressive strain could be observed for the base and the subgrade. This could be because, for the thin structure, the base was subjected to stress hardening, while the subgrade underwent stress softening.

On the other hand, for the thick structure, the combined effect of temperature, overweight, rolling condition, and DWL increased the compressive strain of the AC, base, and subgrade up to 45%, 75%, and 50% respectively, for the drive axle, and up to 100%, 65%, and 105%, respectively, for the steering axle, as shown in Figures 22 and 23. Similar to the other responses, the values were generally lower for this structure. Extreme conditions were considered to illustrate the impact of DWL.

Summary

This study estimated the increase in flexible pavement responses induced by DWL and vehicle maneuvering for two different load configurations. The average axle weights were obtained from weigh-in-motion data. An increase in the axle loading caused by battery packs in electric trucks was also considered. A multi-track road roughness profile was generated to simulate a specific pavement condition based on a target IRI. This profile was then used as an input excitation to record the responses of a mechanical 3D model of a Class-9 truck.

Pavement structures were loaded with dynamic contact stresses to predict critical pavement responses. A DWL profile was extracted and discretized using a percentile approach to obtain a characteristic set of wheel loads. These wheel loads were translated into 3D contact stresses using a 3D FE model of a DTA for different rolling conditions. Full braking and full acceleration scenarios were selected to define a set of dynamic contact stresses. Typical flexible pavement structures (thick and thin) were simulated using tandem axle and single axle configurations to represent a drive axle and a steering axle, respectively. The main findings of this study are summarized below:

• The distribution and location of the peak values varied for the full braking and full accelerating scenarios. However, the critical main responses did not exhibit considerable differences. This could be because prediction of contact stresses did not establish significant changes for simulating braking and accelerating, apart from considering the slip ratio. Additionally, inertial effects or different acceleration rates were not considered for in the steady-state analysis of the FE tire model.

• The consideration of DWL, influenced by pavement roughness and maneuvering, had a substantial impact on in-plane stresses.

• The tensile strain at the bottom of the AC layer, associated with bottom-up fatigue cracking, increased by up to 125% and 133% for the thin and thick pavement structures, respectively.

• The shear strain, associated with near-surface cracking, showed an increase of approximately 100% for both structures under the most adverse conditions.

• The compressive strain, associated with rutting, increased by up to 45%, 75%, and 50% for the drive axle in the AC layer, base, and subgrade, respectively. The steering axle showed an increase of up to 100%, 65%, and 105% in the AC layer, base, and subgrade, respectively.

Footnotes

Acknowledgements

The computational infrastructure that facilitated the analysis carried out in this work was provided by the Illinois Campus Cluster.

Author Contributions

The authors confirm contribution to the paper as follows: study conception and design: J. J. Cárdenas Huamán, I. L. Al-Qadi; data collection: J. J. Cárdenas Huamán, I. L. Al-Qadi; analysis and interpretation of results: J. J. Cárdenas Huamán, I. L. Al-Qadi; draft manuscript preparation: J. J. Cárdenas Huamán, I. L. Al-Qadi. All authors reviewed the results and approved the final version of the manuscript.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Johann J. Cardenas

Imad L. Al-Qadi

The content of this paper reflects the views of the authors, who bear full responsibility for the factual accuracy and integrity of the data presented here. The content does not necessarily reflect the official views or policies of the Illinois Center for Transportation (ICT).

References

Wang

Asce

Zhao

Zhang

Flexible Pavement Response Analysis Under Dynamic Loading at Different Vehicle Speeds and Pavement Surface Roughness Conditions. Journal of Transportation Engineering, Part B: Pavements, Vol. 146, No. 3, 2020, p. 04020040. https://doi.org/10.1061/JPEODX.0000198.

Khavassefat

Jelagin

Birgisson

Dynamic Response of Flexible Pavements at Vehicle–Road Interaction. Road Materials and Pavement Design, Vol. 16, No. 2, 2015, pp. 256–276. https://doi.org/10.1080/14680629.2014.990402.

Said

Al-Qadi

Incorporation of Pavement Dynamic Loading in Mechanistic-Empirical Design Frameworks. Transportation Research Record: Journal of Transportation Research Board, 2023. 2677: 14–28. https://doi.org/10.1177/03611981231166943

Al-Qadi

I. L.

Yoo

P. J.

Effect of Surface Tangential Contact Stresses on Flexible Pavement Response (with Discussion). Journal of the Association of Asphalt Paving Technologists, Vol. 76, 2007, pp. 663–692.

Hammoum

Chabot

St-Laurent

Chollet

Vulturescu

Effects of Accelerating and Decelerating Tramway Loads on Bituminous Pavement. Materials and Structures, Vol. 43, 2010, pp. 1257–1269. https://doi.org/10.1617/s11527-009-9577-9.

Hernandez

J. A.

Gamez

Al-Qadi

I. L.

Effect of Wide-Base Tires on Nationwide Flexible Pavement Systems: Numerical Modeling. Transportation Research Record: Journal of the Transportation Research Board, 2016. 2590: 104–112.

Hajj

E. Y.

Thushanthan

Sebaaly

P. E.

Siddharthan

R. V.

Influence of Tire–Pavement Stress Distribution, Shape, and Braking on Performance Predictions for Asphalt Pavement. Transportation Research Record: Journal of the Transportation Research Board, 2012. 2306: 73–85.

Chen

Han

Liu

A Review on Solutions for Improving Rutting Resistance of Asphalt Pavement and Test Methods. Construction and Building Materials, Vol. 168, 2018, pp. 893–905. https://doi.org/10.1016/J.CONBUILDMAT.2018.02.151.

Gamez

Al-Qadi

I. L.

Turning Maneuver Effect on Near-Surface Airfield Pavement Responses. Transportation Research Record: Journal of the Transportation Research Board, 2019. 2673: 275–283.

10.

Harvey

Saboori

Miller

Kim

Jaller

Lea

Kendall

Saboori

Effects of Increased Weights of Alternative Fuel Trucks on Pavement and Bridges. Institute of Transportation Studies, Davis, CA, 2020. https://doi.org/10.7922/G27M066V.

11.

ASCE. 2021Report Card for America’s Infrastructure. American Society of Civil Engineers. https://infrastructurereportcard.org/. Accessed January 3, 2023.

12.

ISO. Mechanical Vibration — Road Surface Profiles — Reporting of Measured Data. ISO 8608:2016. ISO, Geneva, 2016.

13.

Liu

Al-Qadi

I. L.

Mechanistic Excess Fuel Consumption of a 3D Passenger Vehicle on Rough Pavements. Journal of Transportation Engineering, Part B: Pavements, Vol. 149, No. 1, 2023. https://doi.org/10.1061/JPEODX.PVENG-926.

14.

Sayers

M. W.

Gillespie

T. D.

Paterson

W. D. O.

Guidelines for Conducting and Calibrating Road Roughness Measurements. World Bank, Washington, D.C., 1986.

15.

Liu

Al-Qadi

I. L.

Development of a Simulated Three-Dimensional Truck Model to Predict Excess Fuel Consumption Resulting from Pavement Roughness. Transportation Research Record: Journal of the Transportation Research Board, 2021. 2675: 1444–1456.

16.

Kang

Ozer

Al-Qadi

I. L.

Spencer

B. F.

Stochastic Analysis of Rolling Resistance Energy Dissipation for a Tractor-Trailer Model. Transportation Research Record: Journal of the Transportation Research Board, 2019. 2673: 593–603.

17.

Law

E. H.

Janajreh

Frey

Vehicle Ride Response to New Widebase Tires and Conventional Dual Tires. SAE Technical Papers 2002-01-3114. Society of Automotive Engineers, Detroit, 2002. https://doi.org/10.4271/2002-01-3114.

18.

Abdelkareem

M. A. A.

Kaldas

M. M. S.

Kamal Ahmed Ali

Analysis of the Energy Harvesting Potential-Based Suspension for Truck Semi-Trailer. Proceedings of the Institution of Mechanical Engineers, Part D: Journal of Automobile Engineering, Vol. 233, No. 11, 2019, pp. 2955–2969. https://doi.org/10.1177/0954407018812276.

19.

Liu

Jayme

Al-Qadi

I. L.

ContactGAN Development – Prediction of Tire-Pavement Contact Stresses Using a Generative and Transfer Learning Model. International Journal of Pavement Engineering, Vol. 24, No. 2, 2022, p. 2138876. https://doi.org/10.1080/10298436.2022.2138876.

20.

Al-Qadi

I. L.

Wang

Yoo

P. J.

Dessouky

S. H.

Dynamic Analysis and In Situ Validation of Perpetual Pavement Response to Vehicular Loading. Transportation Research Record: Journal of the Transportation Research Board, 2008. 2087: 29–39.

21.

Al-Qadi

I. L.

Wang

Tutumluer

Al-Qadi

I. L.

Wang

Tutumluer

Dynamic Analysis of Thin Asphalt Pavements by Using Cross-Anisotropic Stress-Dependent Properties for Granular Layer. Transportation Research Record: Journal of the Transportation Research, 2010. 2154: 156–163.

22.

Elseifi

M. A.

Al-Qadi

I. L.

Yoo

P. J.

Viscoelastic Modeling and Field Validation of Flexible Pavements. Journal of Engineering Mechanics, Vol. 132, No. 2, 2006, pp. 172–178. https://doi.org/10.1061/(ASCE)0733-9399(2006)132:2(172).

23.

Gungor

O. E.

Al-Qadi

I. L.

Gamez

Hernandez

J. A.

In-Situ Validation of Three-Dimensional Pavement Finite Element Models. In The Roles of Accelerated Pavement Testing in Pavement Sustainability: Engineering, Environment, and Economics ( Aguiar-Moya

Vargas-Nordcbeck

Leiva-Villacorta

Loría-Salazar

, eds.), Springer, Cham, 2016, pp. 145–159. https://doi.org/10.1007/978-3-319-42797-3_10.

24.

Wang

Al-Qadi

I. L.

Stanciulescu

Effect of Surface Friction on Tire–Pavement Contact Stresses during Vehicle Maneuvering. Journal of Engineering Mechanics, Vol. 140, No. 4, 2014, p. 04014001. https://doi.org/10.1061/(ASCE)EM.1943-7889.0000691.

25.

Yoo

P. J.

Al-Qadi

I. L.

Elseifi

M. A.

Janajreh

Flexible Pavement Responses to Different Loading Amplitudes Considering Layer Interface Condition and Lateral Shear Forces. International Journal of Pavement Engineering, Vol. 7, No. 1, 2007, pp. 73–86. https://doi.org/10.1080/10298430500516074.

26.

Al-Qadi

Hernandez

Jayme

Ziyadi

Gungor

Kang

Harvey

, et al. The Impact of Wide-Base Tires on Pavement—A National Study. Illinois Center for Transportation, Rantoul, 2021.

27.

Wang

Roesler

J. R.

Asce

Guo

D.-Z.

Analytical Approach to Predicting Temperature Fields in Multilayered Pavement Systems. Journal of Engineering Mechanics, Vol. 135, No. 4, 2009, pp. 334–344. https://doi.org/10.1061/(ASCE)0733-9399(2009)135:4(334).

28.

Kim

Asce

A. M.

Tutumluer

Asce

Kwon

Nonlinear Pavement Foundation Modeling for Three-Dimensional Finite-Element Analysis of Flexible Pavements. International Journal of Geomechanics, Vol. 9, No. 5, 2009, pp. 195–208. https://doi.org/10.1061/(ASCE)1532-3641(2009)9:5(195).

29.

Tutumluer

Thompson

M. R.

Anisotropic Modeling of Granular Bases in Flexible Pavements. Transportation Research Record: Journal of the Transportation Research Board, 1997. 1577: 18–26.

30.

Thompson

M. R.

Robnett

Q. L.

Resilient Properties of Subgrade Soils. Transportation Engineering Journal of ASCE, Vol. 105, No. 1, 1979, pp. 71–89. https://doi.org/10.1061/TPEJAN.0000772.

31.

Mitchell

J. K.

Temperature Effects on the Engineering Properties and Behavior of Soils. Highway Research Board (HRB), Washington, D.C., 1969.

32.

Yoo

P. J.

Al-Qadi

I. L.

Effect of Transient Dynamic Loading on Flexible Pavements. Transportation Research Record: Journal of the Transportation Research Board, 2007. 1990: 129–140.

33.

Wang

Al-Qadi

I. L.

Wang

Near-Surface Pavement Failure Under Multiaxial Stress State in Thick Asphalt Pavement. Transportation Research Record: Journal of the Transportation Research, 2010. 2154: 91–99.