Transportation Research Record: Journal of the Transportation Research Board

Some of the design parameters in AASHTO’s Guide for Design of Pavement Structures require experienced engineering judgment to obtain adequate designs. The load transfer parameter for concrete pavements in the AASHTO Guide is reviewed. A set of equations was developed to assist in choosing a J-factor for various pavement conditions. With three-dimensional finite element analysis, factorial runs were conducted to find the relationships between the critical stresses and joint design parameters—that is, joint width, diameter, length, and spacing of dowel bars. Extended procedures that incorporate dowel parameters into the J-factor were proposed. Conclusions were made to clarify the load transfer concept in the current AASHTO Guide and the effects of joint parameters on pavement performance.

Transverse cracking is one of the more common distress manifestations in jointed concrete pavements. While the extent of transverse cracking is largely related to the specified joint spacing, there are several other primary design variables and distress mechanisms that can cause varying degrees of transverse cracking. These primary mechanisms and their associated variables are well-documented in the literature. However, all of these mechanisms often work on the pavement simultaneously over many years and, as a result, it has historically been difficult to calibrate prediction models with field data. The Strategic Highway Research Program’s Long-Term Pavement Performance (LTPP) program has collected a significant amount of condition survey data on more than 110 jointed plain concrete pavements (JPCP) and 65 jointed reinforced concrete pavements (JRCP) throughout North America over the last 7 years. The occurrence of transverse cracking in these sections is one of the principal distresses documented in the condition surveys and therefore provides an excellent data source for examining the relationships between the various primary distress mechanisms and the actual occurrence of distress in the field. Although it is premature to develop or calibrate purely “mechanistic” models based on the LTPP data, enough data have been collected to begin analyzing this distress and its association with the numerous prediction variables in the LTPP database. A complete analysis of the transverse cracking that has occurred in these LTPP test sections, along with their respective relationships with the primary prediction variables found in the primary distress mechanisms, is provided.

Proper consideration of traffic loading in pavement design requires knowledge of the full axle load distribution by the main axle types, including single, tandem, and tridem axles. Although the equivalent single axle load (ESAL) concept has been used since the 1960s for empirical pavement design, the new mechanistic-based pavement design procedures under development by various agencies most likely will require the use of the axle load distribution. Procedures and models for converting average daily traffic into ESALs and axle load distribution are presented, as are the relevant issues on the characterization of the full axle load distributions for single, tandem, and tridem axles for use in mechanistic-based pavement design. Weigh-in-motion data from the North Central Region of the Long-Term Pavement Performance study database were used to develop the models for predicting axle load distribution.

Methods of cost allocation for highway pavement rehabilitation and maintenance activities and pavement management estimations are based on empirical and semiempirical founded predictions that come up short, particularly when a roadway is subjected to heavy multiaxle vehicles. Additionally, materials used in constructing the pavement structure do not always behave in an elastic manner, and the ability to predict the pavement response in the presence of other than elastic material behavior is essential. Finally, prediction of pavement states of distress, based on empirical methods, and elastic material behavior is inadequate, particularly when traffic of heavy vehicles is involved. Battelle has been working on a mechanistic approach to address the issues and concerns at the core of current pavement design methods. The overall approach consists of combining three major software modules—a structural module that includes a general primary response model, a material characterization module, and a damage and distress module—are interconnected so that the influence of one on the other is continually updated. Four miles of highway pavement have been heavily instrumented with structural and environmental sensors so that pavement response can be monitored on both short- and long-term bases. Field test results from these pavements have been acquired and, along with laboratory data, have been used to partially validate and provide insight into pavement behavior under various loading conditions. The unique requirements for the design and implementation of the structural and environmental sensing elements are discussed. The mechanistic aspects in the software for the structural and material models are described, and predicted and field-measured results are compared.

Over the years, pavement engineers have attempted to develop rational mechanistic-empirical (M-E) methods for predicting pavement performance. In fact, the next version of AASHTO’s Guide for Design of Pavements is planned to be mechanistically based. Many M-E procedures have been developed on the basis of a combination of laboratory test data, theory, and limited field verification. Therefore, it is important to validate and calibrate these procedures using additional data from in-service pavements. The Long-Term Pavement Performance (LTPP) program data provide the means to evaluate and improve these models. A study was conducted to assess the performance of some of the existing concrete pavement M-E-based distress prediction procedures when used in conjunction with the data being collected as part of the LTPP program. Fatigue cracking damage was estimated using the NCHRP 1–26 approach and compared with observed fatigue damage at 52 GPS-3 test sections. It was shown that the LTPP data can be used successfully to develop better insight into pavement behavior and to improve pavement performance.

The interactive and computational features of EverFE, a new rigid pavement three-dimensional (3D) finite element (FE) analysis tool, are presented. To date, the use of 3D FE analysis has been hampered by (a) the difficulty of model generation and result interpretation, (b) the inability of many programs to adequately model joint shear transfer due to aggregate interlock and dowel action, and (c) the large computational requirements of conventional solution techniques employed by available programs. The development of EverFE is motivated by the desire to make 3D FE analysis feasible for routine analysis of rigid pavements. The intuitive graphical user interface employed by EverFE, which greatly simplifies model generation and result interpretation, is demonstrated using a sample problem. A novel technique for modeling aggregate interlock joint shear transfer that rationally incorporates nonlinearities is developed and presented, as is a new method for modeling dowel joint shear transfer. The solution strategy employed by EverFE that allows realistic 3D models to be simulated on desktop computers is briefly described, and its performance is presented.

A finite element model for nonlinear analysis of jointed concrete pavements is presented. The model allows for nonlinear representation of the properties of concrete, both in compression and in tension. It also accounts for the behavior under cyclic loading considering the nonlinear fatigue damage accumulation in concrete. An improved model accounting for the relative deformation between the dowel bars and the concrete slabs is presented to analyze pavement slabs connected with dowels. The sub-grade model is capable of representing pumping of the fine material with repetitive loading. Limited validation of the model is presented using data available in the literature.

A three-dimensional (3D) finite element (FE) model is developed to investigate whether the condition of plane sections remaining plane exists in concrete pavements subjected to nonlinear temperature gradients. This model is utilized to validate the analytical method proposed by Mohamed and Hansen. The 3D brick element is chosen so that the plane section condition is not imposed in the model, as compared with the model using the flat plate element. Furthermore, the possibility of loss of contact between the pavement slab and the subgrade is studied. The condition of full contact is investigated for a nonlinear temperature gradient that produces the maximum tensile stress in the slab according to the data used. Two slab lengths and two radii of relative stiffness are considered. It is found that plane sections remain plane for the entire slab except for a region very close to the free edges, which also establishes the boundary where solutions by Mohamed and Hansen are applicable. In both cases of the contact condition, the 3D FE model predicts no loss of contact between the slab and the subgrade.

A rational, three-dimensional (3D) finite element modeling technique was developed to predict the structural response of a jointed concrete airport pavement system. Model features include explicit 3D modeling of the slab continua, load transfer capability at the joint, explicit 3D modeling of the base course continua, load transfer capability across the cracks in the base course, and contact interaction between the slabs and base course. Environmental effects were not explicitly included in the model development. The model was applied to predict the response of an instrumented pavement at the Denver International Airport (DIA) to a test vehicle driven over the instrumented pavement under day and night conditions. The DIA pavement was modeled as a three-layer system with the presence of cracking in the base course as well as a variety of interface conditions between the slabs and base course considered. Complex response patterns caused by environmental factors were observed in the data from DIA, making separation of load-induced and environmentally induced response difficult. The general shape and form of the deflection- and stress-based load transfer efficiency predictions from the finite element models match those observed at DIA. Model predictions of stress-based load transfer efficiencies were generally more accurate than predictions of deflection-based load transfer efficiency. The model developed represents a significant advancement in the state of the art and features innovations that are compatible with the FAA’s advanced pavement design model requirements.

Continuously reinforced concrete pavement (CRCP) performance depends on, among other factors, the characteristics of early developing cracks caused by environmental loads. The primary objective is to evaluate effects of design, materials, and construction variables on the characteristics of cracks in CRCP when subjected to environmental loads. A mechanistic model is developed using finite element formulations. Concrete and longitudinal steel are discretized using the plane strain and the frame elements, respectively. Various bond stress and slip models between concrete and longitudinal steel and between concrete and the underlying layers are developed using the spring elements. The creep effect is also included using the effective modulus method. CRCP responses from the model vary depending on the concrete and steel bond-slip models. An accurate bond-slip model needs to be investigated further by experiments to increase the accuracy of the mechanistic model. Concrete creep has beneficial effects on CRCP responses. The thermal coefficient of concrete has significant effects on CRCP responses. Using concrete with a low thermal coefficient will improve CRCP performance. Longitudinal steel variables—the amount of steel, bar diameter, and steel location—are important design variables that influence CRCP behavior. For given environmental conditions, an optimum steel design can be developed using the model developed.

An experimental program was conducted in the laboratory to predict the behavior of model continuously reinforced concrete (CRC) slabs in tension. A tensile loading arrangement was devised using a Universal Testing Machine for this purpose. Model concrete slabs were cast for two types of reinforcement arrangements: a regular grid pattern, and an inclined grid pattern. Tests were conducted for different rebar spacings and orientations. In the numerical study, a two-dimensional model of the CRC slabs was developed using FLAC, a commercially available software package based on the finite difference method. Experimental and numerical results were compared and were found to be in close agreement. Finally, the average crack spacings for different rebar spacings and orientations from both experimental and numerical studies were compared with the results obtained from the Concrete Reinforcing Steel Institute design method.

In the structural design of continuously reinforced concrete pavement (CRCP), thermal stresses should be properly taken into account. Thermal strains and temperatures in concrete slabs were measured on test sections of CRCP. Measured strains were divided into axial, curling, and nonlinear components, and each component was examined. It was found that the curling component is predominant in terms of transverse stress, which is important in the structural design. However, the maximum thermal stress is reduced by 25 percent because of the nonlinear component. On the basis of the results, a procedure for estimating the thermal stress in CRCP was proposed.

A methodology for calibrating performance models for jointed plain concrete pavements (JPCP) is presented; it is based on statistical analysis of data from the Long-Term Pavement Performance (LTPP) database. The methodology provides calibration factors to pavements in four climatic regions (dry-freeze, dry-nonfreeze, wet-freeze, and wet-nonfreeze) for the JPCP performance models in HDM-4: joint faulting, transverse cracking, joint spalling, and roughness. The procedure allows calculation of global calibration factors, which does not affect significantly the quality of the prediction compared with the quality achieved through the use of regional factors.

Low-temperature cracking is a major distress mode in Alaskan pavements because of the extreme temperature conditions—which range, in some instances, from about −50°C in winter to more than 40°C in summer. The use of asphalt modifiers in Alaskan pavements occurred over the past 15 years. These modifiers include Styrene-Butadiene-Styrene polymers, Styrene-Butadiene-Rubber polymers, ULTRAPAVE, and CRM [both the dry process (PlusRide) and the wet process]. Field observations and laboratory studies in Alaska and elsewhere indicate that the use of these modifiers would improve the low-temperature cracking resistance of pavements. The degree to which these modifiers provide beneficial effects for Alaskan pavements needs to be evaluated. The objectives of this research were (1) To characterize asphalt and polymer modified asphalt from a number of selected sites using Superpave PG grading system and to conduct thermal stress restrained specimen tests (TSRST) and Superpave IDT laboratory tests on field specimens; (2) To compare low-temperature cracking performance using field surveys; (3) To verify the applicability of the Superpave thermal cracking model (TCMODEL) and other available models for predicting low temperature cracking; and (4) To recommend guidelines for predicting minimum pavement temperatures in Alaska. Results of this study indicate, in general, significant improvement in low-temperature cracking resistance when polymer modifiers are used. Comparisons between predicted and observed low-temperature cracking using available crack propagation models, including Superpave TCMODEL, were poor. An improved regression model was developed using minimum air temperature, TSRST fracture temperature and strength, and pavement age to fit the observed field data for both conventional and polymer modified sections.

Flexible pavements containing a thin asphalt-treated permeable base (ATPB) layer have been used in California for more than 15 years. The original philosophy upon which use of ATPB is based, the implementation of that philosophy, and observations regarding its effectiveness are examined. Laboratory tests for resilient modulus of ATPB are presented for the as-compacted state, after 3 and 10 days of soaking, and after repetitive loading in the as-compacted and saturated states. The results indicate a loss of stiffness after soaking, and failure caused by stripping and loss of cohesion between the aggregates when subjected to repetitive loading while saturated. The lab test results were used in simulations of fatigue life for pavements with and without ATPB. The results indicate that predicted pavement fatigue lives are improved when ATPB is included in the pavement structure as compared with when aggregate base alone is used, but the improvement diminishes if water damage occurs in the ATPB. Recommendations are made regarding future use of ATPB in flexible pavements.

A new rational mechanistic model for analysis and design of flexible pavement systems has been developed. Furthermore, a fundamental probabilistic approach was incorporated into this system to account for the uncertainty of material and environmental conditions. The system was integrated in a user-friendly Windows program with a variety of user-selected options that include widely used models and those recently developed in the Strategic Highway Research Program project. Three basic types of distress can be investigated separately or all together, including fatigue cracking, permanent deformation, and low-temperature cracking. The mechanistic approach makes use of the JULEA layered elastic analysis program to obtain pavement response. The system provides optional deterministic and probabilistic solutions, accounts for aging and temperature effects over the asphalt materials, variable interface friction, multiple wheel loads, and user-selected locations for analysis. Tabular and graphical results provide expected distress values for each month as well as their variability, probability of failure, and assessment of the overall reliability of the pavement relative to each type of distress for a user-selected failure criterion. Only the load-associated module of AYMA is presented; a separate work describes the low-temperature cracking analysis.

A concept for a mechanistic-based performance model for flexible pavement was developed that considers the interaction between vehicles and pavement. A dynamic vehicle model was used to estimate the dynamic wheel force, and a three-dimensional finite element nonlinear dynamic pavement model was used to determine the dynamic pavement response. The effect of pavement roughness on vehicle bouncing and the effect of vehicle bouncing on the progression of pavement roughness were investigated under different roughness levels, suspension types, and layer thicknesses. The increase in roughness after each load repetition can be calculated using basic material properties from which the pavement service life can be estimated. The number of equivalent 80-kN single axle load repetitions to failure was estimated under different conditions without the need for empirical observations. It was found that the number of load repetitions to go from one level of present serviceability index (PSI) to the next largely decreases as the PSI level decreases. The air bag suspension results in the longest pavement life, while the walking beam suspension results in the shortest pavement life. The total number of load repetitions to reach failure for thick pavement sections is 14 percent higher than that for medium-thick sections, and 63 percent greater than that for thin sections. The reverse of this analysis can be used to design the pavement section so that it would sustain a certain number of load repetitions before failure using a mechanistic procedure. The proposed concept for a mechanistic-based performance model developed in this study can be refined to increase the mechanistic portion of the model, reduce empirical involvement, and improve computational procedure.

Rutting is a major failure mode for flexible pavements. Pavement engineers have been trying to control and arrest the development of rutting for years. Many models are available to relate pavement rutting to design features, traffic loading, and climatic conditions. These models range from purely empirical to mechanistic models. Mechanistic-empirical models (the Asphalt Institute and Shell) were used to predict the development of rutting for 61 Long-Term Pavement Performance (LTPP) test sections. The rutting damage, calculated using these models, did not appear to be a good predictor of the observed rutting depth. A new rutting model was developed and calibrated using the data from the 61 LTPP sections. The model accounts for the plastic deformation in all pavement layers and allows the use of actual axle load and type, rather than the equivalent single axle load, in characterizing traffic.

In recognition of the potential of mechanistic-empirical (M-E) methods in analyzing pavements and predicting their performance, pavement engineers around the country have been advocating the movement toward M-E design methods. In fact, the next AASHTO Guide for Design of Pavement Structures is planned to be mechanistically based. Since many of the performance models used in the M-E methods are laboratory-derived, it is important to validate these models using data from in-service pavements. The Long-Term Pavement Performance (LTPP) program data provide the means to evaluate and improve these models. The fatigue and rutting performances of LTPP flexible pavements were predicted using some well-known M-E models, given the loading and environmental conditions of these pavements. The predicted performances were then compared with actual fatigue cracking and rutting observed in these pavements. Although more data are required to arrive at a more conclusive evaluation, fatigue cracking models appeared to be consistent with observations, whereas rutting models showed poor agreement with the observed rutting. Continuous functions that relate fatigue cracking to fatigue damage were developed.

The next AASHTO guide on pavement design will encourage a broader use of mechanistic-empirical (M-E) approaches. While M-E design is conceptually straightforward, the development and implementation of such a procedure are somewhat more complicated. The development of an M-E design procedure at the University of Minnesota, in conjunction with the Minnesota Department of Transportation, is described. Specifically, issues concerning mechanistic computer models, material characterization, load configuration, pavement life equations, accumulating damage, and seasonal variations in material properties are discussed. Each of these components fits into the proposed M-E design procedure for Minnesota but is entirely compartmentalized. For example, as better computer models are developed, they may simply be inserted into the design method to yield more accurate pavement response predictions. Material characterization, in terms of modulus, will rely on falling-weight deflectometer and laboratory data. Additionally, backcalculated values from the Minnesota Road Research Project will aid in determining the seasonal variation of moduli. The abundance of weigh-in-motion data will allow for more accurate load characterization in terms of load spectra rather than load equivalency. Pavement life equations to predict fatigue and rutting in conjunction with Miner’s hypothesis of accumulating damage are continually being refined to match observed performance in Minnesota. Ultimately, a computer program that incorporates the proposed M-E design method into a user-friendly Windows environment will be developed.

The design of flexible pavements is based on the empirical approach and in some cases on the empirical-rational approach. Funds are being invested in developing new design methods that will principally be based on the mechanistic-rational approach. A procedure for characterizing clayey subgrade materials to provide the properties needed by the new methods is presented. The procedure addresses the following elements: first, the material should be brought to the condition that prevails under pavements. Sample preparation and a soaking procedure to simulate field condition are described. Second, after about 10 days of sample conditioning, the material is tested under repetitive loading to provide both resilient and permanent properties. In clayey soils, two samples are tested under different deviatoric stresses and for 10,000 to 100,000 load repetitions. The stress level should correspond to the stress induced by overburden and traffic load. Tests are conducted on a swelling clay, and the results are analyzed. It is seen that (a) the proposed procedure for sample conditioning is appropriate for Israeli climatic conditions, but it can be changed to fit other climatic conditions; (b) the stress level in subgrades is relatively low, so the behavior of the clayey materials can be fairly well described by linear elastic theory; and (c) the total deformation increases either linearly or exponentially with number of load repetitions (in a log-log scale).

A study was initiated by the Wisconsin Department of Transportation to investigate the effects of spot diamond grinding on the performance and material properties of concrete pavement. A field survey was conducted to assess the conditions of selected spot-diamond-ground PCC pavement sites. Pavement distress data were collected on control and spot-ground sections on 22 highways and 34 locations in Wisconsin. In addition, microsurveys were completed for each of the spot-ground sections. Utilizing the pavement distress index, values and results of the microsurvey comparisons are made between sections that were spot-ground and those that were not. Conclusions are drawn concerning the effects of spot grinding on the performance of the concrete pavements; no practical differences were found between the spot-ground and control sections.

Pavement rutting is a problem that has unknown consequences from a safety-based point of view. It is assumed that the wheelpath fills with water in wet-weather situations, thus increasing the potential for a vehicle to hydroplane. A study was conducted to quantify how pavement rutting affects accident rates and to evaluate possible safety-based guidelines for the treatment of pavement rutting. Rut depth, traffic volumes, and accident databases maintained by the Wisconsin Department of Transportation for undivided rural highways were then analyzed to identify statistical trends and relationships. Accidents were categorized as rut-related if the prevailing conditions could be potentially associated with the occurrence of hydroplaning. Rut depth measurements were average values for both directions of 1.8-km (1.1-mi) segments and represent the average elevation difference between the tire paths and the high point between them. The results of the statistical analyses indicated that the defined rut-related accident rate begins to increase at a significantly greater rate as rut depths exceed 7.6 mm (0.3 in.). The cost-effectiveness of potential accident reductions associated with reductions in the relative amount of rutted pavement was also evaluated.

The Michigan Department of Transportation (MDOT) practice regarding the preservation, rehabilitation, and preventative maintenance actions for rigid, flexible, and composite pavements is presented and discussed. For each pavement type, the causes of distress and the corresponding MDOT fix alternatives are also presented. Examples of the MDOT practice regarding the selection of maintenance and rehabilitation alternatives for rigid, flexible, and composite pavements are also presented.

The results of a three-dimensional finite element analysis of the impacts of small utility cuts in urban street pavements are presented. The analysis was restricted to flexible pavements for cuts of the order of 915 mm (3 ft) square. Using ANSYS Solid 45 Version 5.2 on a one-eighth area, the excavation process was simulated using a stress relief approach, in which the material was assumed to be removed in successive layers. Starting with gravity loading on the uncut model, the element stress results from the previous removal were used to compute the new effective confining stresses of the respective layers. The in situ material properties were typical of flexible pavements in an urban street. The analysis presented a clear picture of the magnitude and extent of the distress induced on the pavement structure below and surrounding the cut. The analysis also showed how the restraining action of the asphalt layer is reflected in its arching up from heave and granular material thrusting, causing tension at the bottom and compression at the top. The results suggested that for unsupported depths of up to 1524 mm (5 ft), material distress may extend approximately 1068 mm (3.5 ft) into the pavement structure. Implications for methodology and economics of restoration are discussed, including determination of optimum cutback and recompaction of the affected area of pavement.

One of the most common types of pavement on the national highway system is composite asphalt concrete (AC) over portland cement concrete (PCC). With a large percentage of PCC pavements either approaching or at the end of their design lives, AC overlay of PCC pavements has become one of the most common methods of rehabilitation. This has resulted in several thousand kilometers of composite AC/PCC pavements. As the level of heavy truck traffic loading continues to increase on a majority of pavements, it is likely that the total length of composite pavements in the nation will continue to increase considerably in the coming years. A common type of distress that occurs on these composite pavements is reflective cracking. This occurs when the joints or cracks in the underlying PCC pavement reflect through the AC overlay. A performance model that can be used to predict accurately the amount of reflective cracks in composite AC/PCC pavements has enormous potential uses. The development of a mechanistic-based performance model for predicting the amount of reflective cracks in composite AC/PCC pavements is described. Data from the Long-Term Pavement Performance database were used to develop the model. Using the principles of fracture mechanics, it is illustrated that a mechanistic-based model can be developed that closely models the real-life behavior of composite pavements and predicts the amount of reflective cracks. Because of the mechanistic nature of the model, it is particularly effective for performance prediction for design checks and pavement management. Also, since the model can take into account the relative damaging effect of the actual axle loads in any traffic distribution, it has great potential for application in cost allocation.

A new approach to multiyear maintenance and rehabilitation (M&R) optimization programming for pavement network management is discussed; the approach can be used to help highway agencies make strategic decisions in choosing the optimal investment for their pavement networks. The M&R treatments are standardized in terms of costs, benefits, and performance impacts on the existing pavements. Each standardized pavement treatment strategy, ranging from minor and routine maintenance to major rehabilitation or reconstruction, is defined by its effect and improvement on the existing pavement serviceability. The optimization model is a cost-effectiveness-based integer M&R programming on a year-by-year basis. The objective of the optimization system is to select the most effective M&R projects for each programming year. The optimization system can also be used to calculate the minimum budget requirements for maintaining a prescribed level of the pavement network performance or serviceability. In such a case, sensitivity analysis can be performed to evaluate the annual budget effect on individual pavement performance. The prediction of individual pavement deterioration is modeled as a time-related (nonhomogeneous) Markov transition process. The investigation described was primarily concerned with integration of the performance prediction model, the standardized M&R treatments, and the network optimization process. The principle and methodology developed can be applied to different levels of pavement network management. Finally, a sample application of the integrated pavement optimization model is demonstrated.