Effect of Vehicle-Pavement Interaction on Pavement Response

78 Paper No. 970578 TRANSPORTATION RESEARCH RECORD 1570 Effect of Vehicle-Pavement Interaction on Pavement Response MAGDY Y. MIKHAIL AND MICHAEL S. MAMLOUK The structural response of flexible pavements is studied under different dynamic loads and pavement roughness conditions. The factors affecting dynamic load variability are investigated with regard to pavementvehicle interaction. Furthermore, the study considers the viscoelastic nature of asphalt concrete and the nonlinearity and plasticity of granular and subgrade materials. The Florida COMPAS computer program was used to estimate the dynamic wheel force, and the ABAQUS threedimensional finite-element program was used to determine the pavement response. The effects of vehicle and pavement characteristics such as vehicle type, vehicle speed, suspension type, level of roughness, pavement stiffness, and layer thickness were studied and statistically analyzed. The walking-beam suspension causes more dynamic load variation than the air-bag and leaf-spring suspension. The dynamic load coefficient for the walking-beam suspension is approximately twice the other suspensions. Vehicle speed is an important factor; the 20 km/hr speed resulted in permanent displacement approximately 10 times the permanent displacement produced by the 130 km/hr speed. The pavement response varies with distance due to roughness. Pavement stiffness and thickness had some effect on pavement response, but truck type and truck suspension type did not have a large effect. Highways represent a major portion of the nation’s infrastructure affecting various activities of society on a day-to-day basis. Pavements are complicated physical structures responding in a complex way to the influence of many variables, such as load, materials, and environmental conditions. Compounding the problem is the fact that truck gross weights and tire pressures have grown in the last few decades, which has resulted in faster and more severe damage to pavements. Unlike other structural systems, pavements have relatively short service lives and deteriorate rapidly because of traffic loads and environmental effects. It is estimated that a large portion of the existing pavements in the United States need some kind of reconstruction or rehabilitation. The study of pavement response and performance is complex because it involves factors such as (a) dynamic loading, (b) different load configurations and magnitudes, (c) multilayer pavement systems consisting of different material types and layer thicknesses, (d) nonlinear material properties that are largely affected by environmental conditions such as temperature and moisture, (e) construction variabilities, ( f ) different types and intensities of failure, and (g) different cost-effective options for maintenance and rehabilitation (1). Most of the previous studies assume that pavements are loaded statically, whereas pavements are actually subjected to dynamic loads. The forces applied by vehicles vary instantaneously above and below the static weight because of the interactive effect between vehicles and pavement. Several factors contributing to this load M. Y. Mikhail, Department of Civil Engineering, University of Nevada, Reno, Nev. 89557-0152. M. S. Mamlouk, Department of Civil and Environmental Engineering, Arizona State University, Tempe, Ariz. 85281-5306. variation include road roughness, vehicle configuration, suspension type, tire type, and vehicle speed. Figure 1 shows typical patterns of dynamic wheel forces measured using a sensor mounted on the truck axle (2). Increasing dynamic forces increase the damage rate to pavements and shorten their service lives. An accurate study of pavement response should include the interaction between vehicles and pavement. Pavement roughness excites axle suspensions, causing vehicle dynamic forces, which in turn increase the roughness of the pavement surface. Consequently, more roughness of the pavement surface causes more vehicle dynamic forces. This is an accelerating process in which pavement and vehcle affect each other through time, a process that becomes more significant as the pavement deteriorates (3). Vehicle and pavement characteristics should be considered in a dynamic or time-dependent environment as well as with contributing factors, such as vehicle speed, in predicting the structural response of the pavement (4). A number of studies have addressed the subject of vehicle-pavement interaction (e.g., 5,6). However, current knowledge of the interaction between vehicles and pavement is as yet insufficient to provide the clear understanding needed to develop completely mechanistic pavement analysis and design methods. This study analyzes the structural response of flexible pavements under different dynamic loads and pavement roughness conditions. The factors affecting dynamic-load variability are investigated with regard to pavement-vehicle interaction. Furthermore, the study considers the viscoelastic nature of asphalt concrete and the nonlinearity and plasticity of granular and subgrade materials. Figure 2 shows a schematic diagram of the analysis performed. The road profile is input into a vehicle model from which the vehicle dynamic forces are determined. These dynamic forces are used in a pavement model to determine the dynamic pavement responses, such as displacements and critical strains. The study considers different vehicle and pavement variables. VEHICLE DYNAMICS Vehicle Dynamic Model When vehicles travel on the pavement, they develop different dynamic movements. Vehicle dynamic models have been developed to predict dynamic forces produced by different axle and wheel configurations at different locations on the pavement surface. Truck dynamic models were developed that represent the truck with a number of masses, springs, and dashpots. The Florida Comprehensive Pavement Analysis System (COMPAS) (7 ) is a research study conducted by the Texas Transportation Institute for the Florida Department of Transportation. The Florida COMPAS software includes a tire-force prediction model that Mikhail and Mamlouk Paper No. 970578 79 FIGURE 1 Typical instantaneous dynamic wheel force at 80 km/hr (50 mi/hr) speed for medium road roughness and walking-beam suspension (z). predicts dynamic tire loads for different types of vehicles traveling on a straight asphalt road at constant speeds. The model is capable of generating the dynamic force profile of a wheel load for different combinations of vehicle types, suspension types, vehicle speeds, and levels of pavement roughness. The COMPAS program simulates vehicles with combinations of masses, springs, and dashpots, as shown in Figure 3. The results of the Florida COMPAS were verified by comparing the output of the program with field measurements and results from other vehicle models (7,8). Vehicle dynamic-force measurements were obtained from the flexible pavement sections instrumented with piezoelectric film strips. Very close agreements were obtained between the Florida COMPAS results and both field measurements and other simulation models (8). Estimate of Vehicle Dynamic Forces The Florida COMPAS was used to determine the dynamic tire forces under various conditions. The following factors and levels were studied: • Trucks. Three types were used following the FHWA classification, two-axle tractor, two-axle trailer (2-S2); three-axle tractor, two-axle trailer, (3-S2); and five-axle multitrailer (2-S1-2). • Suspension. Air-bag, leaf-spring, and walking-beam suspensions were used. • Speed. Speeds of 20, 75, and 130 km/hr were used. • Roughness. Three levels with values of present serviceability under (PSI) 4.6, 3.5, and 2.4 were used. The combinations of the different truck types, suspension types, speeds, and pavement roughness levels produce different dynamic tire forces. These factors and levels resulted in 81 factor combinations. The gross vehicle weights for the three truck types were chosen so that the axle weights on rear axles were approximately equal to 80 kN for purposes of comparison. Numerical outputs were generated for all the above combinations. The results show the variation of dynamic tire forces with distance and time under various conditions. Figure 4 shows an example of the dynamic tire force for the second axle versus distance for the different types of suspension of a 2-S1-2 truck at a speed of 130 km/hr for an average PSI of 2.4. The data in Figure 4 show that the walking-beam suspension causes more dynamic load variation than the other suspension types. Figure 5 shows the dynamic tire force versus distance for three FIGURE 2 Schematic diagram of analysis performed. FIGURE 3 Schematic diagram of vehicle dynamic model used in Florida COMPAS. 80 TRANSPORTATION RESEARCH RECORD 1570 Paper No. 970578 FIGURE 4 Dynamic force versus distance for different suspension types (2S1-2 truck, 130 km/hr speed, PSI = 2.4). different speeds of the 2-S2 truck with an air-bag suspension and an average PSI of 3.5. The data show that at higher speeds more dynamic-force variability is expected. Figure 6 shows the dynamic force versus distance for different roughnesses for a 2-S1-2 truck with a 75 km/hr speed and a walking-beam suspension. The data show that as the pavement roughness increases, more dynamic-load variability is obtained. Dynamic-Load Coefficient The dynamic-load coefficient (DLC) is used to characterize the variability of the wheel force and is defined as (2) DLC = standard deviation of dynamic wheel force mean dynamic wheel force FIGURE 5 Dynamic force versus distance for different speeds (2-S2 truck, air-bag suspension, PSI = 3.5). (1) Mikhail and Mamlouk Paper No. 970578 81 FIGURE 6 Dynamic force versus distance for different roughness levels (2S1-2 truck, 75 km/hr speed, walking-beam suspension). Figures 7, 8, and 9 show the DLC of various suspension types, speeds, and levels of pavement roughness for the 2-S2 truck. The figures show that the air-bag suspension produces the least DLC, whereas the walking-beam suspension produced the highest DLC. Furthermore, increasing the truck speed or pavement roughness increases the DLC for all types of suspension. A full factorial experimental design was used for the above factors and levels. An analysis of variance (ANOVA) was performed on the results to test the significance of the different factors. All the main effects were found to be significant at a level of 0.05. ANALYSIS OF VEHICLE-PAVEMENT INTERACTION Pavement Dynamic Model The dynamic load profiles of the Florida COMPAS program were input to the ABAQUS program to evaluate the structural response of pavement. The ABAQUS program is a three-dimensional, dynamic finite-element program that has the capability to simulate actual vehicle loading conditions and estimate the structural re- FIGURE 7 Dynamic load coefficient for different serviceability indexes and speeds (2-S2 truck, air-bag suspension). FIGURE 8 Dynamic load coefficient for different serviceability indexes and speeds (2-S2 truck, leaf-spring suspension). FIGURE 9 Dynamic load coefficient for different serviceability indexes and speeds (2-S2 truck, walking-beam suspension). Mikhail and Mamlouk Paper No. 970578 sponse for flexible pavements (9,10). The program has been used previously to study the dynamic response of flexible pavements (11). ABAQUS solves the dynamic analysis of linear problems by using the eigenmodes of the system as a basis for calculating the response. In such cases, the necessary modes and frequencies must be extracted first. Other options are also available, such as time-history analysis; steady-state harmonic response analysis, including random-response analysis; and response-spectrum analysis. For the nonlinear dynamic analysis, direct integration of the system or subspace projection method is used. In the direct integration method, all the equations of motion of the system are integrated through time (10). The equations of motion of a multiple degree of freedom system with viscous damping present can be written in a matrix form as [ K ]{U} + [C ]{U˙ } + [ M ]{U˙˙} = {P} (2) where [K ] {U} [C ] [M ] { U̇} {Ü} {P} = = = = = = = stiffness matrix, displacement matrix, α[M ] 1 β[K], α and β are constants, mass matrix, velocity matrix, acceleration matrix, and force matrix. The nonlinear dynamic approach with the direct integration method was used here. FIGURE 10 Finite element mesh. 83 Features of the Finite-Element Model The three-dimensional finite-element model was developed to represent the pavement structure using brick elements, as shown in Figure 10. Because of symmetry, one-half of the wheelpath, together with one side of the surrounding region, was used. The finite-element mesh consists of a fine mesh close to the load and a coarse mesh far from the load. Mesh dimensions in the vertical direction was selected to match the pavement-layer thickness. A 12.2-m-long wheelpath with two 12.2-m unloaded ends on both sides was used to reduce the end effect. Both the asphalt concrete surface and base course were modeled as a single-element layer, whereas the subgrade was modeled as a multielement layer. The appropriate boundary conditions were used. Infinite elements were used to model the bottom layer of the subgrade. The mesh dimensions were constrained to satisfy the appropriate aspect ratio to have the loaded area required and to achieve the desired degree of detail. Paving materials were divided into asphalt mixtures, granular materials, and cohesive soils. The asphalt concrete was modeled as a viscoelastic material that is time- and temperature-dependent. The time-dependent properties were represented by the ratio of instantaneous shear modulus to the long-term shear modulus (G-ratio), which can be expressed as G-ratio = 1 − long- term shear modulus instantaneous shear modulus (3) 84 TRANSPORTATION RESEARCH RECORD 1570 Paper No. 970578 TABLE 1 Material Properties The Drucker-Prager model was used in the analysis to model base course and subgrade materials. This model is an elastic-plastic model in which materials are assumed to behave as elastic materials for low stress levels. When the stress level reaches a certain yield stress, the material will start to behave as an elastic-plastic material. The material characteristics required for this model include modulus of elasticity, Poisson’s ratio, damping coefficient, bulk density, angle of internal friction, and cohesion. The area of the elements in the wheelpath was taken as the tire-pavement contact area. To simulate truck loads moving at different speeds, the load cycle begins with a load magnitude equal to zero at time t0. After time t0, the load is increased linearly to a maximum value at time t1. The load magnitude remains constant between time t1 and time t2 after which the load is decreased linearly to zero at time t3. The time durations from t0 to t1, t1 to t2, and t2 to t3 are functions of the speed and the length of the contact area. This process is repeated for the different elements in the wheelpath at different times depending on the vehicle speed. In each case different load magnitudes were used to simulate the variability of the dynamic load that was obtained from the Florida COMPAS program. ABAQUS is installed on an HP9000 model J200 workstation with 128 Mbytes RAM, 9-Gbyte hard disk, and a 100-MHz PA-RISC 7200 CPU. Each computer run took an average of 3 hr. TABLE 2 The process of simulating the effect of moving loads and variable dynamic loading results in a huge amount of computation. Other Factors Considered In addition to the truck-type, suspension-type, vehicle-speed, and pavement-roughness variables used before, two more pavement variables were used. • Pavement stiffness. Three classes were used to represent stiff, medium, and weak sections. Table 1 shows the material properties used to represent these classes. • Layer thicknesses. Three sets of thicknesses were used to represent thick, medium, and thin pavements. Table 2 shows the values used to represent these sets. These factors and levels result in 729 factor combinations. A oneninth partial factorial design of experiment was used to reduce the number of combinations to 81. Three response parameters were calculated by the ABAQUS program for each factor combination. These parameters are the permanent displacement on the surface, total horizontal tensile strain at the bottom of the asphalt concrete layer, and the total vertical compressive strain on the top of the subgrade layer. The average of the Layer Thickness for Different Pavement Sections Mikhail and Mamlouk Paper No. 970578 85 FIGURE 11 Permanent displacement versus distance for 3-S2 truck for three different speeds (air-bag suspension, 2.4 PSI, stiff pavement, thick section). results of three points in the wheelpath was used, as shown in Figure 10. An ANOVA was performed on the results to test the significance of the different factors on the permanent surface displacement, horizontal tensile strain on the bottom of the asphalt concrete layer, and the vertical compressive strain on the top of the subgrade. The permanent displacement on the surface was selected because it is related to rutting and roughness development in the wheelpath. The horizontal tensile strain on the bottom of the asphalt layer and the compressive strain on top of the subgrade were selected because they are related to fatigue and rutting, respectively. For the permanent displacement parameter, the speed, pavement stiffness, and the interaction of layer thickness and truck type were found to be significant at the 0.05 level. That the pavement roughness and suspension type were not significant may be due to the fact that only one load repetition was considered. The effects of roughness and suspension type could have been more pronounced if several load repetitions were used. For the horizontal tensile strain parameter, the speed, thickness, and stiffness were found significant at the 0.05 significance level. For the vertical compressive strain parameter, the same factors as horizontal tensile strain were found significant. ANALYSIS OF RESULTS Vehicle speed was found to be an important factor that affects the pavement response. Figures 11, 12, and 13 show the permanent sur- FIGURE 12 Horizontal tensile strain versus distance for 3-S2 truck for three different speeds (air-bag suspension, 2.4 PSI, stiff pavement, thick section). 86 Paper No. 970578 TRANSPORTATION RESEARCH RECORD 1570 FIGURE 13 Vertical compressive strain versus distance for 3-S2 truck for three different speeds (air-bag suspension, 2.4 PSI, stiff pavement, thick section). face displacement, horizontal tensile strain at the bottom of the asphalt concrete layer, and vertical compressive strain on the top of subgrade versus distance for the three different speeds for the 3-S2 truck. It is clear that slow speeds produce higher displacements and strains in the pavement structure and can cause larger damage to the pavement as compared with high speeds. The 20 km/hr speed resulted in permanent displacement approximately 10 times the displacement produced by the 130 km/hr speed. This is because the time duration of loading is more for 20 km/hr than 130 km/hr. These figures show the variation in pavement response with distance (direction of travel) due to roughness. There was no big difference in pavement response between the different levels of pavement roughness. This may be attributed to considering one load repetition of the moving load. Figure 14 shows the horizontal tensile strain versus distance for three different PSI levels for the 3-S2 truck. The figure shows the variation of pavement response with distance due to roughness. The horizontal tensile strain for the 2.4 PSI level is higher than the 3.5 and 4.6 PSI levels. Pavement stiffness also affects pavement response. Figures 15 and 16 show the permanent displacement and vertical compressive strain versus distance for the three different pavement stiffness levels for the 2-S2 truck. These two figures show that weak pavement sections produce higher displacements and strains in the pavement structure and may deteriorate faster than stiff sections. Pavement thickness also affects the pavement response. Thin pavement sections produce higher displacements and strains in the pavement structure and may deteriorate faster than thick pavement sections. Based on the statistical analyses performed, the effect of vehicle type was not significant because the gross vehicle weights for the three truck types were chosen so that the axle weights on their rear axles were approximately equal for the purpose of comparison. The effects of suspension and roughness were not statistically significant, although these two factors were significant for the DLC, which may be due to the fact that only one load repetition was considered. FIGURE 14 Horizontal tensile strain versus distance for 3-S2 truck for three different PSI levels (air-bag suspension, 20 km/hr speed, stiff pavement, thick section). Mikhail and Mamlouk Paper No. 970578 87 FIGURE 15 Permanent displacement versus distance for 2-S2 truck for three different stiffnesses (air-bag suspension, 4.6 PSI, 75 km/hr speed, thick section). SUMMARY AND CONCLUSIONS The interaction between vehicles and pavement was investigated by using the Florida COMPAS computer program to estimate the dynamic wheel force and the ABAQUS three-dimensional finiteelement program to determine the pavement response. The effects of vehicle and pavement characteristics such as vehicle type, vehicle speed, suspension type, level of roughness, pavement stiffness, and layer thickness were studied and statistically analyzed. The walking-beam suspension causes more dynamic-load variation than the other two suspension types. Higher speeds produce more dynamic-force variability, and an increase in pavement roughness produces more dynamic-load variability. Vehicle speed is an important factor that affects pavement response. The 20 km/hr speed resulted in permanent displacement approximately 10 times the displacement produced by the 130 km/hr speed. Pavement response varies with distance (direction of travel) due to roughness. The stiffness of the pavement section affects the response of pavement, with weak pavement sections producing higher displacements and strains in the pavement structure. These weak pavement sections may deteriorate faster than stiff sections. The thickness of the pavement section also affects pavement response, with thin pavement sections producing higher displacements and strains in the pavement structure. The truck type and the suspension type did not have a large effect on pavement response. The vehicle-pavement interaction can be used to estimate the pavement response. ACKNOWLEDGMENT The authors would like to thank the Department of Civil and Environmental Engineering at Arizona State University for the financial support and for making its facilities available for the study. FIGURE 16 Vertical compressive strain versus distance for 2-S2 truck for three different stiffnesses (air-bag suspension, 4.6 PSI, 75 km/hr speed, thick section). 88 TRANSPORTATION RESEARCH RECORD 1570 Paper No. 970578 REFERENCES 1. Mamlouk, M. S. Effect of Vehicle-Pavement Interaction on Weigh-inMotion Equipment Design. International Journal of Vehicle Design on Vehicle-Road and Vehicle Bridge Interaction, 1994. 2. Sweatman, P. F. A Study of Dynamic Wheel Forces in Axle Group Suspensions of Heavy Vehicles. Special Report 27, Australian Road Research Board, Australia, June 1983. 3. Brademeyer, B. D. Mechanistic Vehicle-Pavement Interactions. Presented at the FHWA Load Equivalency Workshop, McLean, Va., Sept. 1988. 4. Cebon, D., and C. B. Winkler. A Study of Road Damage Due to Dynamic Wheel Loads Using a Load Measuring Mat. Report SHRP-ID/UFR-91518, Strategic Highway Research Program, Washington, D.C., 1991. 5. Morris, J. R. Effect of Heavy Vehicle Characteristics on Pavement Response and Performance—Phase I. National Cooperative Highway Research Program, TRB, National Research Council, Washington, D.C., Dec. 1987. 6. Gillespie, T. D., S. M. Karamihas, M. W. Sayers, M. A. Nasim, W. Hansen, and N. Ehsan. NCHRP Report 353: Effects of Heavy 7. 8. 9. 10. 11. Vehicle Characteristics on Pavement Response and Performance. TRB. National Research Council, Washington, D.C., 1993. Fernando, E. G., R. L. Lytton, W. L. Mcfarland, J. L. Memmott, F. Helin, and A. N. Jamy. The Florida Comprehensive Pavement Analysis System (COMPAS). Vol. 1, Texas Transportation Institute, Texas A&M University, 1991. Fernando, E. G., and Lytton, R. L. A system for Evaluating the impact of Truck Characteristics and Use on Flexible Pavement Performance and Life Cycle Costs. Proc., 7th International Conference on Asphalt Pavements, Nottingham, U.K., 1992. ABAQUS, Finite Element Computer Program (Version 5.5). Hibbitt, Karlsson and Sorensen, Inc., 1995. ABAQUS, Finite Element Computer Program, Theory Manual (Version 5.5). Hibbitt, Karlsson and Sorensen, Inc., 1995. Zaghloul, S., and T. D. White. Use of a Three Dimensional, Dynamic Finite Element Program for Analysis of Flexible Pavements. In Transportation Research Record 1388, TRB, National Research Council, Washington, D.C., 1993. Publication of this paper sponsored by Surface Properties—Vehicle Interaction.