Effects of truck load position on longitudinal joint deterioration

Transcription

1 Indian Journal of Engineering & Materials Sciences Vol. 15, February 2008, pp Effects of truck load position on longitudinal joint deterioration Muhammet Vefa Akpinar* İnsaat Muhendisligi Bolumu, Karadeniz Teknik Üniversitesi, Trabzon, Turkey Received 12 April 2006; accepted 21 January 2008 Longitudinal joint cracks between asphalt mats are common problems in asphalt pavements and often deteriorate faster than other areas. The cracks resulting from the deteriorated longitudinal joints are very important for determining the performance life of the asphalt pavements. Cracks in the longitudinal joints allow the ingress of water into the pavement leading to further disintegration. In the urban areas asphalt paving is often done in stages to facilitate traffic control. In most widening projects, the lane marking does not coincide with the edge of the pavement. This results in traffic loads traveling at variable distance from the edge although the most common assumption in modern flexible pavement evaluation and design is that the wheel path is located about cm from the edge of the pavement. It is suspected that the loading near the edge is one of the primary causes of longitudinal joint deterioration. In this study, mechanistic analysis with the finite element (FE) technique has been conducted to predict pavement response at the longitudinal joint under the truck load at different lateral positions relative to the joint. The FE analysis results showed that the position of the truck significantly affects the strains at the longitudinal. Tremendous increase in tensile strain was observed as the truck position nears the longitudinal joint. Maximum critical strains occur when the truck is positioned right at the longitudinal joint. The lane marking should be done in such a way that the right wheel path is located about 53 cm from the longitudinal joint. Key words: Asphalt pavement, Longitudinal joint, FEM The longitudinal joints between asphalt mats often deteriorate faster than other areas. The cracks resulting from the deteriorated longitudinal joints are very important for determining the performance life of the asphalt pavements. The cracks are the major zones for pavement distresses 1-9. Substantial difference in densities of the two asphalt mixes of the joint is the primary cause of the longitudinal joint crack 1-6. Cracks in the longitudinal joints allow the ingress of water into the pavement leading to further disintegration. When the first lane is paved, the asphalt concrete (AC) mixture shoves laterally under the roller, leading to a low density of the mix at the edge 1,7. This is primarily due to the fact that the edge of the cold lane is unconfined. The subsequent lane however has a confined edge and therefore, it generally has a higher density. In the urban areas asphalt paving is often done in stages to facilitate traffic control. In most widening projects, the lane marking does not coincide with the edge of the pavement. This results in traffic loads traveling at variable distance from the edge although the most common assumption in modern flexible pavement evaluation and design is that the wheel path is located about m from the edge of the pavement. It is suspected that the loading near the * mvakpinar@yahoo.com edge is one of the primary causes of longitudinal joint deterioration. Other causes for longitudinal cracks are known to be the temperature and environmental factors. Once the tensile stress due to temperature change or other environmental factors are higher than the tensile strength of the compacted AC mixture, the longitudinal construction joint splits apart 6. Figure 1 shows pavement with deteriorated longitudinal joints and associated cracks on urban highway in Manhattan, Kansas. The purpose of 2002 AASHTO was to advance to mechanistic related design procedure. Longitudinal joint techniques for asphalt pavement materials are empirical. It is important that a mechanistic procedure be developed which will reasonably predict the vertical deformation under roller loading and compare the longitudinal joint techniques. Finite element (FE) analyses provide significant basis for the development of mechanistic analysis. Available finite element programs are powerful tools for studying stress-strain analysis in pavement structure. Lane marking is directly related to the design of the highway and is used for traffic control. The lane stripes on Portland cement concrete pavement roads often coincide with the longitudinal joint. This is done primarily not to confuse the motorist since during inclement weather or when marks are faded motorists tend to follow the longitudinal joints to steer along the

2 AKPINAR : LONGITUDINAL JOINT DETERIORATION 37 Fig. 1 Deteriorated longitudinal joints (a) North bound K113 Highway and (b) US 24 West Highway, Manhattan, KS wheel path. These stripes also ensure that the traffic loads do not travel close to the joint since edge loading is considered most critical in concrete pavements. However, this is not the case in asphallt pavements. In most asphalt pavement projects in urban areas, the lane markings do not coincide with the longitudinal joints since the pavement is assumed to be homogenous and continuous. Project Selection and Data Collection Two sections on K-113, an urban arterial in Manhattan, Kansas, were selected as test sections in this study. The roadway was reconstructed to widen it from 2 lanes to 4 lanes. The pavement section consists of 11 inches of asphalt concrete on 6 inches of limetreated subgrade. In 1999, the AADT varied from 2,390 to 7,545 and the number of Equivalent 18-kip ESAL s per day varied from 55 to 61. The International Roughness Index (IRI) measured in March 18, 1999 was 138 cm/km. Ten percent low severity fatigue cracking was reported in some part of the roadway during the Pavement Management System (PMS) survey. During test section selection, longitudinal cracks were observed along the longitudinal joint line. Because of stage construction, pavement marking on the lane edge does not coincide with the pavement edge, and most of the cracking was observed on the sections where the wheel paths cross the longitudinal joints. No appreciable rutting was found on the test sections. Falling weight deflectometer (FWD) testing In this study, Falling Weight Deflectometer (FWD) tests were conducted to evaluate a multi-lane urban pavement that has shown premature deterioration of longitudinal joints. FWD tests were done at approximately 50 ft intervals in about 15 locations on the Northbound direction and 15 locations on the Southbound direction. Close proximity of the on and off-ramps prohibited more testing due to traffic control problems. At each location, three load drops ranging from 6,500 to 16,060 pounds were used. Target load levels were 7, 9 and 15 kips. The pavement surface deflections were measured by sensors located at 0, 8, 12, 18, 24, 36, and 60 inches from the center of the loading plate. Air temperature during FWD tests ranged from 29 to 74 F. Deflection data analysis Layer moduli were backcalculated with the EVERCALC 4.0 program. A three-layer system, consisting of the AC layer, a combined lime-treated and natural soil subgrade, and a stiff layer, was modeled in order to have better convergence. The root-mean-square (RMS) values were restricted to about 3-5%. The basins with higher RMS were excluded from the analysis. This resulted in exclusion of most of the data points in the north bound (NB) direction. The spatial variation was significant and the values varied from 300 ksi to 550 ksi. In general, the points near the longitudinal joints (less than 1 ft) had somewhat lower moduli values 5. The highest modulus was obtained when the test point was farthest from the longitudinal joint crack (about 2 ft). Higher variabilities were observed in the subgrade moduli. The results appear to be inconclusive with respect to the location of the FWD test points. This may indicate that the effectiveness of the lime

3 38 INDIAN J. ENG. MATER. SCI., FEBRUARY 2008 treatment was highly variable since the test locations are all on a fill section. Falling weight deflectometer tests at or near the longitudinal joints and cracks do not show any definite trends in the backcalculated layer moduli. The variability in subgrade modulus is higher than the asphalt concrete layer modulus. General pavement programs such as Elsysm and Michpave do not consider the elastic modul variations through the cross-section when computing the stress and strains. Rather, these programs assign one mudulus value for each layer which is not the case in the field according to FWD tests results. In this study, this was possible by deviding the AC layer into several zones and inputing the backcalculated modulus data to these zones. Finite element analysis The finite element analysis was done to compute the critical responses at the longitudinal joint location due to a tandem axle placed at different lateral locations from the joint. The critical responses are (i) tensile strain at the bottom of the asphalt layer and (ii) vertical compressive strain at the top of the subgrade. ANSYS Version 5.5 software was used to model the layers in three dimensions. The Solid 92 (10 node) element was used for both AC and subgrade layers. The effect of temperature on the AC layer was not considered. An asphalt section 114 cm long, 152 cm wide and 28 cm thick was defined. The subgrade thickness was 226 cm over a stiff layer. Since the AC layer moduli somewhat varied with respect to the location of the longitudinal joint, different elasticity modulus values of the AC layer were used. For the section away from the joint (about a foot), the AC modulus was Pa as shown in Fig. 2. The section of AC closer to the joint was assumed to have a lower modulus, Pa. The subgrade modulus was assumed to be the same throughout the section and was 10 8 Pa. Different types of element, such as, quad 4 node, 8 node, triangle 6 node, Brick 8 node, 20 node, and tet 10 node, were tried in modeling the layers in the pavement structure. Among them, Solid 92 3-D 10 node tetrahedral structural solid element resulted in the best model. Triangle-shaped and hexahedralshaped elements were better when used in 3-D meshing. Solid 92 has a quadratic displacement behaviour and is well suited for modelling irregular meshes, such as, those produced from various CAD/CAM systems 7. Ten nodes having three degrees of freedom at each node defined the Solid 92 element: translations in the nodal x, y and z directions. The element had plasticity, creep, swelling, stress stiffening, large deflections, and large strain capabilities. Temperatures could be input as element body loads at the nodes. Mesh dimensions were kept small enough to allow detailed analysis of the layers. However, as the mesh was made finer the number of elements increased resulting in increased memory and computational time. The mesh is finer for the AC layer in order to obtain more precise information. Time and memory requirements dictated that only the area under one half of a typical tandem axle load be considered. Tire loads were represented with rectangular areas of cm 2 resulting from a tire load of kn/m 2 and pressure of kn/m 2. The length of the contact area along the traffic direction was 10 inches (25.4 cm). Spacing between the axles in the tandem was 122 cm. Figure 3 shows the schematic diagram of the load positions used in the analysis. cases 1, 2, and 3 represent the truck locations when it was positioned far away from the longitudinal joint, at the joint, and close to the joint on the right lane, respectively. Figure 4 illustrates the load location for the load case 1 in ANSYS FEM. Areas A1, A2, A3 and A4 are the locations where the tires made contact with the pavement structure for this case. Fig. 2 Plan and profile view of the structure

4 AKPINAR : LONGITUDINAL JOINT DETERIORATION 39 Fig. 3 ing positions In all cases, the contact between the layers was modelled as full contact (full friction). As a result, translations of the nodes at the bottom of the AC layer were constrained, which means that there was no sliding along the contact surfaces. Edges of the AC and subgrade layers were also constrained in the horizontal direction as shown in Fig. 4. Results and Discussion The behaviour of the structural model was reasonable. Maximum vertical deformation occurred under the load. The vertical deformation for the load case 1 is shown in Fig. 4. Tensile strains at the bottom of the AC layer were used to compare the load cases for fatigue behaviour. Maximum tensile strain and compressive strain occurred at the bottom of the AC layer and at the top of the subgrade layer, respectively. The maximum values were obtained under the center of the loads. Table 1 shows the critical strains at the longitudinal joint while the truck was positioned at the center of the lane (load case 1), close to the joint on the left (load case 2) and right at the joint (load case 3). The position of the truck significantly affects the critical strains at the longitudinal joint location for load cases 2 and 3. However, the tensile strains increased by about 130% and 196% for the load cases 2 and 3 respectively. Thus, it is apparent that traffic loading close to the longitudinal joint will contribute to the premature failure of the joint. The computed tensile strain values may be underestimated since the dimensions of the FEM structure are small due to memory limitations. Fig. 4 (a) Constraints of the structure, load case 1 (with red mesh). Areas on the top represent the tire pressure and (b) case 1 vertical deformation Table 1 Maximum critical strains at the longitudinal joint location due to varying Truck position Critical strains At the longitudinal joint location Tensile at the bottom of the AC layer (10 6 ) Compressive strain at the top of the subgrade (10-6 ) case 1 case 2 case

5 40 INDIAN J. ENG. MATER. SCI., FEBRUARY 2008 The FWD tests at or near the longitudinal joints and cracks did not show any definite trends in the backcalculated layer modulus. High variability in backcalculated subgrade modulus was found. The FE analysis results showed that the position of the truck load significantly effect the strains at the longitudinal joint. Very high increase in tensile strain in asphalt concrete was observed as the truck position nears the longitudinal joint. Maximum critical strains occur when the truck is positioned right at the longitudinal joint. The sensitivity analysis results indicate that the lane marking should be done so that the wheel path is located about 53 cm from the longitudinal joint of the pavement. The most desirable load position would be the case where there is no tensile strain at the joint. case 1 was chosen as the best case. Further analysis shows that if the load case 1 comes 15 cm close to the joint, still there will not be any horizontal tensile strains at the bottom of the asphalt concrete layer. The distance between the second tire (on the right) in load case 1 and the joint is 38 cm. Thus, the lane marking should be done in such a way that the right wheel path is located about 53 cm from the longitudinal joint. Conclusions Based on the results of this study, the following conclusions may be drawn: (i) Falling weight deflectometer tests at or near the longitudinal joints and cracks do not show any definite trends in the backcalculated layer moduli. The variability in subgrade modulus is higher than the asphalt concrete layer modulus. Such as Elsysm and Michpave programs assignes one elastıc moduli value for each layer. The variations of the layer modulus values should be considered when designing pavement with general accepted FEM studies. (ii) Numerical analysis using the finite element method tends to indicate that the loading near the longitudinal joint is one of the primary causes of joint deterioration. Once the tensile stress due to traffic loads are higher than the tensile strength of the compacted asphalt concrete mixture, the longitudinal construction joint splits open. (iii) Position of the truck does not significantly affect the strains at the longitudinal joints as long as the truck is about 30 cm away from the joint. Maximum critical strains occur when the truck is positioned right at the longitudinal joint. (iv) The lane marking should be done in such a way that the right wheel path is located about 53 cm from the longitudinal joint. Acknowledgements The author gratefully acknowledges the testing support provided by the FWD crew of the Kansas Department of Transportation. References 1 Foster C R, Hudson S B & Nelson R S, J Transport Res Board, 51 (1964). 2 Kandhal P S & Mallick R B, J Transport Res Board, 106 (1996). 3 Kandhal P S & Rao S S, Evaluation of Longitudinal Joint Construction Techniques for Asphalt Pavements, (NCAT, Auburn), NAPA Longitudinal Joints: Problems and Solutions (NAPA Quality Improvement Series), Livneh M, Site and Laboratory Testing in Order to Determine the Bonding Method in Construction Joints of Asphalt Strips, AAPT Proc, Kandhal P S, Ramirez T L & Ingram P M, Evaluation of Eight Longitudinal Joint Construction Techniques for Asphalt Pavements in Pennsylvania, (NCAT, Auburn), Sherocman J A, Construction of Durable Longitudinal Joints, Proc Canadian Technical Asphalt Association, Sebaaly P E, Barrantes J C, Fernandez G & Loria L, Development of a Joint Density Specification Phase II: Evaluation of 2004 and 2005 Test Sections, Nevada DOT Report RDT , Scott J, Asphalt Pavement Joint Densities, presented at the 2005 Northwest FAA Regional Annual Conf, Denver, Colorado, ANSYS Elements Reference, ANSYS 5.5, 10 th ed, (ANSYS Inc., New York), 2006.