Transportation Research Record: Journal of the Transportation Research Board

It was verified that finite-element modeling could be successfully used to analyze concrete pavements with partial-depth cracks. An existing finite-element program, ILLI-SLAB, was modified (ILSL97) to allow for partial-depth crack analysis. To model a partial-depth crack, a special line spring element was added to the finite-element code. The line spring elements mimic the behavior of a crack by acting as a rotational hinge between two continuous slabs. By using available fracture mechanics techniques, a relationship was derived between the amount of moment load transfer across a crack and the crack depth. This analytical solution was then used to formulate the element stiffness matrix for the line spring element. The deflections predicted by the new finite-element program are correct, but the stresses in the vicinity of the crack tip needed to be corrected to match the stress singularity zone in front of cracks. Several example problems were used to verify the proposed finite-element model, and an example of a typical highway loading condition was analyzed.

Application of fracture mechanics concepts developed in various branches of engineering to the pavement problem can address current limitations, thereby advancing considerably existing pavement design procedures. The state of the art in fracture mechanics applications to pavement engineering is summarized, and an in-depth discussion of one of the major concerns in such applications, the specimen-size effect, is provided. It is concluded that the fictitious crack model proposed by Hillerborg appears most promising for computerized application to pavements. The similitude concepts developed by Bache will be very useful in such efforts. Both the desirability and the scarcity of suitable candidates to replace Miner’s cumulative linear fatigue hypothesis in conventional pavement design are confirmed. Fracture mechanics is shown to be a very promising engineering discipline from which innovations could be transplanted to pavement activities. Nonetheless, it is pointed out that rather slow progress characterizes fracture mechanics developments in general. Pavement engineers clearly need to remain abreast of and involved in fracture mechanics activities.

An innovative approach to characterize the micro- and macromechanical damage phenomena resulting from fatigue cracking and relate them to the design and durability of pavements has been developed. In this approach, crack propagation and its associated damage are considered as irreversible processes, and hence the general framework of the thermodynamics of irreversible processes is used to model fatigue crack propagation behavior. A constitutive equation, the modified crack layer model, is used to extract material parameters characteristic of the pavement’s resistance to fatigue crack propagation. These parameters are the specific energy of damage, γ′, and a dissipative characteristic, β′, of the paving mixture. The general applicability of the proposed approach has been extended to examine fatigue crack propagation data of concrete reported in the literature. It is found that the proposed model describes fatigue crack propagation behavior of concrete pavements over the entire range of the crack driving force.

A structural evaluation and analysis of instrumented plain portland cement concrete pavement was performed in a rural primary highway subjected to heavy dynamic loads. Two full-scale slabs were constructed and tested to verify the dynamic analysis of the computer program UPR-PAVI2, which was developed at the University of Puerto Rico at Mayagüez using finite elements. It has the capability to analyze the effect of temperature differential, dead load, and dynamic and static trucks on rigid pavements. UPR-PAVI2 can consider different soil stiffness under the slab, the slab geometry, and multiple trucks and their longitudinal paths, velocities, stiffnesses, geometries, and load distributions. The instrumented slabs included both doweled and undoweled. Their material properties were obtained through laboratory and field tests. The subbase course consisted of a cut section in limestone bedrock for the doweled slab and borrowed material from limestone rock for the fill section used in the undoweled slab. The base course was composed of two layers of nonerodable black bituminous base. The slabs were constructed with concrete with a compressive strength of 34 MPa (5,000 psi) at 28 days. A loaded truck generated the established repetitions by moving over the instrumented slabs at different velocities, producing static and dynamic deflections that were measured with a data acquisition system. The replication and behavior of the response were analyzed, and the validated data were compared with the response obtained using UPR-PAVI2. The results obtained from this simulation were in good agreement with the field data.

The effect of coarse aggregate on pavement performance has been attributed to the volume of aggregate used in pavement construction. The different patterns of crack development for limestone (LS) and siliceous river gravel (SRG) are a typical example of aggregate-induced variable performance in continuously reinforced concrete pavement (CRCP). An attempt was made to find a reasonable solution for pavements with SRG. As a way to solve the performance problem observed from the SRG pavement, a blended aggregates mixture was suggested. Laboratory and field tests were performed to check the feasibility of their application in pavements. From the laboratory test, a 50:50 blending ratio was suggested after considering the effect on tensile strength and thermal coefficient of expansion. Field test sections were also constructed to verify previous performance observations for the two aggregates and to provide performance data for new variables such as blended aggregates and special curing methods. Unexpectedly, the blended mixture did not improve the performance of SRG pavement; rather it experienced worse cracking than SRG alone. A controlled experiment with additional field test sections is needed to verify or disprove this finding. The only definitive finding was that selection of aggregate in the concrete pavement is a vital consideration for the design of the pavement. The CRCP8 analytical program reasonably predicted crack spacing for both SRG and LS pavements, predicting mean crack spacing of 0.99 m (3.25 ft) for SRG and 1.98 m (6.41 ft) for the limestone. These values are somewhat below the actual spacing observed at 100 days. Data collected after the first winter period will be required to calibrate the program.

Elastic layered programs (ELPs) are currently being used in mechanistic-based pavement design procedures for the analysis of jointed portland cement concrete pavements. Corrections must be made to stresses obtained from ELP solutions to account for the effects of finite slab size, load location on the slab, and load transfer efficiency of the joints. A preliminary artificial neural network (ANN) model is trained and used as a tool to predict the results of a finite-element analysis program for a standard pavement section. Under identical loading conditions, the trained neural network produces stresses within 1.2 percent of those obtained from finite-element analyses. The trained ANN model is found to be very effective for correcting ELP stresses, practically in the blink of an eye, with no requirements of complicated finite-element inputs. The preliminary ANN algorithm is currently being expanded to handle more general input conditions covering a wide range of slab sizes, slab thicknesses, subgrade supports, and loading conditions. Design curves created from these ANN algorithms will eventually enable pavement engineers to easily incorporate sophisticated state-of-the-art technology into routine practical design.

The development of a mechanistically based model (CRACK) that predicts the spacing of cracks in roller-compacted concrete (RCC) pavements due to restraint of drying shrinkage and thermal contraction is presented. The model considers the RCC pavement thickness, paving lane length, base friction characteristics, climatic time-temperature regimes, as well as the RCC drying shrinkage, coefficient of thermal contraction, strength gain, and modulus of elasticity. The model also determines the tensile stresses and displacements of the RCC slab along its length, from which the crack width is predicted. The CRACK program was used to predict the crack spacing and width at four RCC pavement field sites, and the predicted results were compared with field measurements. The model tends to overpredict the observed crack spacings, in part because of the simplifying assumptions used in the model development and the fact that the model does not consider fatigue stresses from warping, curling, or traffic-related loads, which cause additional cracking. However, the predicted crack spacing provides a basis for selecting joint spacings, and the predicted crack widths may help in the design of joint sealant reservoirs.

Temperature and moisture gradients can lead to significant tensile stresses at the slab top and bottom. Current techniques for assessing the internal stresses due to such gradients are based on the assumption that temperature and moisture distributions through the slab thickness are linear. However, the actual distributions of such gradients have been found to be highly nonlinear. A new closed form solution technique for calculating the stresses in a pavement slab due to nonlinear gradients is introduced. The analysis is separated into two parts. In the first, an expression is presented for calculating the self-equilibrated stresses within a cross section due to internal restraint (i.e., satisfying equilibrium conditions and continuity of the strain field within the cross section). These stresses are independent of slab dimensions and boundary conditions. In the second, the stresses due to external restraint (i.e., self-weight and subgrade reaction) are calculated using an equivalent linear temperature gradient obtained from the first part and existing closed form solutions by Westergaard or Bradbury. The solution to this step includes slab length and boundary conditions. Total internal stresses due to nonlinear gradients are obtained by using the superposition principle. The methodology has been applied to field data from two studies in which the temperature profiles were recorded throughout a 24-hr period. Linear gradient solution methods cannot accurately predict the curling stresses in concrete pavements. This is especially pronounced during nighttime and early morning hours, during which nonlinear analysis predicts tensile stress in both the slab bottom and top before the application of any traffic loading.

A numerical analysis of a multilayered rigid pavement system is presented. To efficiently model the behavior of the subgrade soil layers in the horizontal direction, infinite elements are used to represent the decaying feature of the field variables (displacements and stresses) in the far field. Ordinary finite elements are used in a relatively small region around the applied loads to accurately capture the stress field. Compared with a conventional finite-element analysis of pavements, wherein a finite size of the computational domain must be chosen and artificial boundary conditions must be assumed at the truncated boundary, the current approach can significantly reduce the total number of elements, saving considerable computation time and thereby making feasible more extensive analysis.

A new stress analysis and thickness design procedure for jointed concrete pavements was developed. On the basis of Westergaard’s edge stress solution and several prediction models for stress adjustments for a variety of loading and environmental (i.e., thermal curling) conditions, a modified Portland Cement Association (PCA) equivalent stress analysis and thickness design procedure was proposed and implemented in a highly user-friendly, Windows-based program, TKUPAV, for practical trial applications. The proposed approach has been further verified by reproducing results very close to the PCA’s equivalent stresses and fatigue damages using a spreadsheet program and the TKUPAV program. The possible detrimental effect of loading plus daytime curling has been illustrated in a case study, which also indicated that the effect of thermal curling should be considered in the thickness design of concrete pavements.

The effectiveness of a concrete pavement construction specification depends on the degree of correlation the criteria in the specification have to the material properties and behavior characteristics that govern the performance of pavements. Since concrete strength is determined on the basis of small test specimens, it is important that the strength measured in these specimens be representative of the strength within the actual configuration of a pavement system. A primary deficiency in this regard is that strengths obtained from different test specimens can be quite different from each other and from the strength of the concrete in the pavement. This fact raises the question of which specimen should serve as a representative concrete strength standard that would allow a reliable measure of concrete quality and yet provide a parameter that is compatible with design stress calculations. This deficiency is encompassed within size and geometry effects that are present when estimating strength from small test specimens. Since most distresses in concrete pavements are due to tension-induced cracking, it is only logical that tension-type tests be used to represent concrete resistance to failure. A simplified tension test method, based on the fracture mechanics theory, provides a solution to the deficiency by taking into account size and geometry effects. The method considers the geometry of the specimen from which the strength is being measured (e.g., cylinder) and the geometry of the structure to which the measured strength is being applied (e.g., pavement slab). This method makes it possible to circumvent the problem of obtaining a representative strength value from a small laboratory specimen.

The primary source of water in pavements is from surface infiltration. Water from surface infiltration and other sources increases pavement maintenance costs and shortens pavement life. These effects can be mitigated by including a subdrainage system in the pavement design. A study of pavement subdrainage systems has been conducted that included a phase for laboratory tests for the hydraulic characteristics of materials used in three test sections built on I-469 at Fort Wayne, Indiana. The purpose of building these test sections was to evaluate the subdrainage performance and materials used in the sections. The sections differ in the permeable drainage layer as well as the filter (separation) layer. The sections were instrumented with different types of sensors. Instrumentation installed in the sections included time domain reflectometry (TDR) probes, moisture suction blocks, temperature probes, resistivity tree, and rainfall and outflow tipping buckets. Other features of the instrumentation site included an air temperature probe and access tube for a nuclear moisture content probe. The laboratory hydraulic characteristic test results on the materials are presented. Field moisture condition in the layers is also presented. Calibration equations for the TDR are developed for the field materials. A summary of three rainfall events and pavement outflow volumes recorded for the tests is presented. Recommendations are made for materials to be used in subdrainage systems.

Accurate in situ evaluation of the changes in material properties of an asphalt surface layer due to fatigue loading and rest periods is important for a better understanding of fatigue mechanisms of asphalt concrete in the field and therefore more realistic prediction of the fatigue life of asphalt pavements. A nondestructive technique based on the stress wave propagation method is presented as a means of measuring the changes in the “apparent” modulus of an asphalt surface layer due to fatigue damage growth and healing during rest periods. The method was applied to four asphalt pavements in the FHWA Turner Fairbank Highway Research Center that were loaded by the Accelerated Loading Facility to induce fatigue damage. The dispersion analysis is performed using the Short Kernel Method on the wave transients measured at different loading cycles and after rest periods. The results indicate that (a) the test and analysis technique used provides a sensitive means of evaluating the changes in asphalt surface layer properties during fatigue loads and rest periods, (b) the elastic modulus of the asphalt layer decreases as the number of loading cycles increases in an S-shape decaying curve, and (c) the effective modulus of the asphalt layer increases because of the introduction of rest between loading cycles.

Surface deflection measurements and backcalculation of layer moduli in flexible pavements are significantly affected by the temperature of the asphalt concrete (AC) layer. Correction of deflections or backcalculated moduli to a reference temperature requires determination of an effective temperature of the AC layer. For routine deflection testing and analysis in state highway agencies, it is preferable, from a practical point of view, to use a nondestructive prediction method for determining the effective AC layer temperature instead of measuring the temperature directly from a small hole drilled into the AC layer. A simplified procedure to predict asphalt pavement subsurface temperatures is presented. The procedure is based on fundamental principles of heat transfer and uses the surface temperature history since yesterday morning to predict the AC layer mid-depth temperature at the time of falling weight deflectometer (FWD) testing today. The surface temperature history is determined using yesterday’s maximum air temperature and cloud condition, the minimum air temperature of today’s morning, and surface temperatures measured during FWD tests. FWD tests and temperature measurements have been conducted on seven pavement sections with varying structural designs located in three different climatic regions of North Carolina. The field temperature records from these pavements have provided values of pavement thermal parameters and coefficients in temperature functions that are needed in the prediction procedure. A set of verification results are presented using examples with different climatic regions, changing AC layer thicknesses, and varying weather patterns in different seasons.

A design procedure was developed for layered elastic flexible airfield pavement systems in terms of probability and reliability. A user-friendly computer program, RELIFLEX, was prepared to carry out the computations. The Rosenblueth method is used to estimate the expected value and variance of the computed strains (dependent parameters) on the basis of the input mean values of independent parameters (i.e., aircraft load, layer thicknesses, and material moduli). RELIFLEX has the capability of considering the strain distribution to be normal, lognormal, or beta, and a capacity-demand analysis is included. Only normal distribution is presented. Environmental effects and the lack of fit of the prediction models are not considered in RELIFLEX. The relationships between the reliability level and the allowable strain repetition of the designed system, established using RELIFLEX, provide a decision-making tool for engineers to design pavement at the desired reliability level. For a desired reliability level of the design system, pavement thickness can be varied to agree with the designed coverage or the allowable coverage can be modified for a given pavement thickness.

The design and construction of pavement rehabilitation projects is one of the greatest challenges afforded today’s pavement engineers. All project information pertinent to the performance of a pavement section is needed to select, design, and successfully construct a rehabilitation strategy. Case studies of three rehabilitation projects constructed by the Pennsylvania Department of Transportation are provided. Each case study uses a different rehabilitation strategy. The process for selecting the appropriate requirements for each one is discussed. The pavement performance for each rehabilitated section is reviewed. The pavement performance information is tied back to the original criteria for project strategy selection and design to evaluate the success of the rehabilitation in addressing the issues that necessitated the rehabilitation and providing quality pavement performance.

The design and construction of flexible overlays has become a popular exercise. However, there is not a simple, straightforward, and yet reliable design procedure that the design engineer can implement on a routine basis. The data needed for overlay design are not easily accessible to the design engineer, and yet the accessible data are not fully reliable in most cases. The process by which the design engineers at the Nevada Department of Transportation handle overlay design is presented. The various steps followed and the obstacles that the design engineer encounters in the search for the necessary data and the final design process are described. Major assumptions must be made along the way that could significantly affect the final design. Three case studies are presented. Each project was designed using three different design methods, including the AASHTO nondestructive testing method, the AASHTO condition survey method, and the Nevada Department of Transportation (NDOT) mechanistic overlay design method. All three methods used the same data gathered by the design engineer for each project. The analysis indicates that there is a significant discrepancy between the two AASHTO methods, whereas the NDOT method and the AASHTO condition survey method agreed on one project.

The results of a study conducted to assess the performance of a D-cracked pavement rehabilitated between 1986 and 1989 are presented. The original pavement, located at Forbes Field in Topeka, Kansas, consisted of a thick jointed plain concrete pavement (JPCP) placed directly on a clay subgrade and had D-cracking in the top 102 mm (4 in.). The D-cracking was removed completely by milling before the placement of a bonded JPCP overlay. The major concerns were (a) the bonding condition between the overlay and the original JPCP, (b) the extent of D-cracking progression in the original JPCP, if any, after rehabilitation, and (c) the structural condition of the overlay and the original JPCP. These concerns were investigated using data from heavy weight deflectometer tests performed at the center, corner, and edge of the JPCP slabs. Cores from the rehabilitated pavement were also inspected. The results of the study indicate that, after 6 to 9 years of service, the rehabilitated JPCP is in good structural condition. Significant friction and bonding still exist between the overlay and the original JPCP layer. These results may be useful to pavement engineers considering a bonded portland cement concrete overlay for rehabilitation of D-cracked JPCP.

Illinois Highway Research Study IHR-532 was conducted to develop project-level guidelines for evaluation, rehabilitation selection, and rehabilitation design for existing asphalt concrete–overlaid portland cement concrete (AC/PCC) pavements in Illinois. Use of these guidelines depends on data that are not normally available for network-level analysis. A study was conducted to assess the feasibility of implementing these guidelines in network-level decision making and to assess the effects that implementing these guidelines for selection of rehabilitation methods for existing asphalt-overlaid concrete pavements would have on the overall rehabilitation budget requirements for a district network of Interstate pavements, the overall network condition, and pavement network remaining life. These objectives were met by comparing the IHR-532 rehabilitation strategy (which identifies one or more rehabilitation options) with the Illinois Department of Transportation’s current rehabilitation policy, which is an AC overlay thickness of 83 mm (3.25 in.). This feasibility analysis indicated that the selection of the preferred rehabilitation strategy was strongly related to the type and condition of the existing pavement. It also indicated that the IHR-532 rehabilitation strategy and the single rehabilitation strategy yielded similar networkwide results when compared on the basis of average condition at the end of the analysis period. However, the IHR-532 rehabilitation strategy indicated greater benefit than the single rehabilitation strategy, at any budget level, when compared on the basis of network average remaining life at the end of the analysis period.

A study undertaken by the Texas Department of Transportation to nondestructively detect stripping in the asphalt surfacing on I-45 in the Bryan district is described. The highway was constructed in the 1960s and 1970s with an initial portland cement concrete thickness of 200 mm. Since then, several asphalt overlays have been applied. Maintenance of this highway is a recurring problem, and it is known that in several locations moderate to severe areas of subsurface stripping are present. To plan the future rehabilitation of this important highway, the Bryan district investigated the ability of ground penetrating radar (GPR) to provide subsurface condition information. A GPR survey was conducted at close to highway speeds, and the data were interpreted before taking validation cores. The GPR was used to provide information concerning the section breaks along the highway on the basis of asphalt layer thickness and condition, the average thickness of the asphalt layer within each section, and the extent and severity of any defect in the asphalt layer. More than 60 cores were taken to correlate the GPR interpretation. GPR results and ground truth cores are compared. In general, the comparisons were good. The GPR equipment and interpretation schemes used were found to provide information of sufficient quality and accuracy to permit the district to make programming decisions. GPR is now being used on several additional projects in the Bryan district. The best use appears to be for both defect detection and thickness estimation before deflection testing and coring. GPR will not eliminate coring or deflection testing, but by using all three in a coordinated approach pavement designers will have more confidence in their design decisions.

Current methodologies for developing, evaluating, and improving longrange rehabilitation programs may be classified as manual or analytical. The major limitation of manual methods is that although they are practical, they are not cost-effective because of the difficulties of comparing potential cost/benefit trade-offs among alternatives. Analytical methods, on the other hand, are more cost-effective but often not practical, because they oversimplify the real problem into a mathematical model and do not allow engineers and managers to be involved in the decision-making process. A new methodology and new concepts to overcome these major deficiencies were developed and implemented. The first of the new concepts is the use of graphical data interpretation and “visual thinking.” The second is that pavement performance evaluation should be comprehensive and quickly accessible. The third is that human knowledge should be involved in the decision-making process through a user-friendly interface. The fourth is that the relationships among the components of the methodology should be flexible rather than rigid. Finally, the process should be efficiently repeatable so that a satisfactory long-term rehabilitation program can be developed. A user-friendly, Windows-based pavement management system, called Windows ILLINET, was produced in this study on the basis of the new methodology and concepts.