Jorge A Prozzi

USING DURATION MODELS TO ANALYZE EXPERIMENTAL PAVEMENT FAILURE DATA

Predicting pavement performance under the combined action of traffic and the environment provides valuable information to a highway agency. The estimation of the time at which the pavement conditions will fall below an acceptable level (failure) is essential to program maintenance and rehabilitation works and for budgetary purposes. However, the failure time of a pavement is a variable event; terminal conditions will be reached at different times at various locations along a homogeneous pavement section. A common problem in modeling event duration is caused by unobserved failure events in a typical data set. Data collection surveys are usually of limited length. Thus, some pavement sections will have already failed by the day the survey starts; others will reach terminal conditions during the survey period, whereas others will only fail after the survey has been concluded. If only the failure events observed during the survey are included in the statistical analysis (disregarding the information on the events after and before the survey), the model developed will suffer from truncation bias. If the censoring of the failure events is not accounted for properly, the model may suffer from censoring bias. An analysis of the data collected during the Road Test sponsored by the American Association of State Highway Officials (AASHO) is presented. The analysis is based on the use of probabilistic duration modeling techniques. Duration models enable the stochastic nature of pavement failure time to be evaluated as well as censored data to be incorporated in the statistical estimation of the model parameters. The results, based on sound statistical principles, show that the failure times predicted with the model match the observed pavement failure data better than those from the original AASHO equation.

EFFECT OF PERFORMANCE MODEL ACCURACY ON OPTIMAL PAVEMENT DESIGN

In the first part of this paper, an analysis of the data collected during the American Association of State Highway Officials (AASHO) Road Test, based on probabilistic duration modeling techniques, is presented. Duration techniques enable the stochastic nature of pavement failure time to be evaluated as well as censored data to be incorporated in the statistical estimation of the model parameters. The second part of this paper presents the use of economic optimization principles for determining the optimal design of flexible pavements. We study the effect of deterioration model accuracy on optimal design and lifecycle costs, by comparing three models. The first is a simple regression model developed by the AASHO, which forms the basis of design standards in use today. The second is a regression model that was developed with the same AASHO data set, but that includes a correction for data censoring. The third model is the probabilistic model developed in the first part of this paper. The results show that the AASHO model, when used as an input to lifecycle cost minimization, produces a pavement structural number that is lower than that produced by using the other two deterioration models. This results in shorter pavement lives and higher costs due to more frequent resurfacing. The savings in lifecycle cost accrued by using optimal structural number are shown to be quite significant, offering a sound basis for revising current design practices.

Calibration of "Mechanistic–Empirical Pavement Design Guide" Permanent Deformation Models: Texas Experience with Long-Term Pavement Performance

The performance models in the Mechanistic–Empirical Pavement Design Guide (MEPDG), developed under NCHRP 1-37A and 1-40D, are calibrated with sections throughout the United States. Hence, it is necessary to calibrate these models for specific states and regional conditions because of the differences in materials, environmental conditions, and construction practices. In general, a pavement design based on the nationally calibrated MEPDG will result in either an overestimate or underestimate of the pavement layer thicknesses because of systematic errors arising from local differences. This deficiency calls for local calibration of the performance models in the MEPDG so that they can be used to design pavements at a regional level. The calibration procedure described in this paper concentrates on finding two bias correction factors for the asphalt concrete (AC) permanent deformation performance model after values derived from expert knowledge have been assumed for the subgrade permanent deformation calibration factors. Pavement data from the Texas Specific Pavement Study (SPS)-1 and SPS-3 experiments of the Long-Term Pavement Performance database were used to run the MEPDG and calibrate the guide to Texas conditions. The regional calibration factors were obtained by minimizing the sum of squared errors between the observed and predicted surface permanent deformation. In this case, a simultaneous joint optimization routine was applied because it was theoretically sound. Finally, an average of the regional calibration coefficients for AC and subgrade permanent deformation was computed to obtain the set of state-default calibration coefficients for Texas.

Quantification of the Joint Effect of Wheel Load and Tire Inflation Pressure on Pavement Response

Most pavement design and analysis procedures predict performance on the basis of expected pavement damage under traffic loads expected during design life. Some failure criteria are primarily dependent on wheel loads and almost independent of contact stresses. Others are primarily dependent on normal and shear stresses, not on load magnitude. Wheel load is used as a proxy for tire pressure to account for the effect of contact stresses indirectly. In most pavement design methods, tire-pavement contact stress is assumed to be equal to tire inflation pressure and to be uniformly distributed over a circular area. A methodology that explicitly accounts for the effect of tire inflation pressures and the corresponding contact stresses on pavement response is not available. In this research, pavement responses of typical pavement structures under the combined actions of variable wheel loads and tire pressures were evaluated. A multilayer, linear-elastic computer program was used to estimate three critical pavement responses: longitudinal and transverse tensile strains in asphalt and compressive strains in the subgrade. The differences of the strains estimated by the two models were statistically analyzed to quantify the effect of the assumption of uniform stress over a circular shape. The traditional model proved to be reliable to estimate compressive strains in the subgrade layer. The tensile strains in the asphalt layer under actual contact stress, however, were quite different from those under uniform constant stress. Contrary to initial expectation, for the general case, the assumption of uniform stresses is a conservative approach.

Analytical Study of Effects of Truck Tire Pressure on Pavements with Measured Tire-Pavement Contact Stress Data

Truck tire inflation pressure plays an important role in the tire-pavement interaction process. As a conventional approximation method in many pavement studies, tire-pavement contact stress is frequently assumed to be uniformly distributed over a circular contact area and to be simply equal to the tire pressure. However, recent studies have demonstrated that the tire-pavement contact stress is far from uniformly distributed. Measured tire-pavement contact stress data were input into an elastic multilayer pavement analysis program to compute pavement immediate responses. Two asphalt concrete pavement structures, a thick pavement and a thin pavement, were investigated. Major pavement responses at locations in the pavement structures were computed with the measured tire-pavement contact stress data and were compared with the conventional method. The computation results showed that the conventional method tends to underestimate pavement responses at low tire pressures and to overestimate pavement responses at...

Optimum statistical characterization of axle load spectra based on load-associated pavement damage

Traffic is indeed one of the most critical inputs for pavement design; traditionally the one that is associated with the highest uncertainty. In the most comprehensive mechanistic–empirical (M–E) design approaches, traffic is accounted for by axle load distribution instead of equivalent single axle load (ESAL) as in the traditional empirical approach. Research has already been conducted concerning the statistical characteristics of axle load distribution, however, with focus on the goodness of fit of the data. Little of the past research directly accounted for the traffic load-associated pavement damage. To address this particular issue, this study develops a comprehensive statistical methodology that includes not only improved fitted axle distribution functions, but also sound statistics representing load-associated pavement damage. Mixed lognormal distributions are employed to fit the observed axle load spectra. Two fundamental advantages of the fitted functions are: (1) both the physical and statistical meanings of the load spectra are properly accounted for and (2) the load spectra data are well captured and the fitted distribution can be statistically evaluated. In particular, the load-associated pavement damage based on axle load distributions is investigated through the concept of moment statistics. The moment order (or power) is generalized to both integer and non-integer conditions, an important advantage of the lognormal distribution. Different power values are examined concerning varying load-associated pavement distresses or responses. In the case study presented, the results indicate that R 2 is not an adequate statistic to evaluate fitted functions from the perspective of using load spectra in pavement design. It is, therefore, recommended that assessment of fitted distribution be based on moment statistics. Of particular note, it is demonstrated that, due to relatively larger fit errors, higher moment orders should be adopted to evaluate load spectra fit functions in the context of pavement design. To address this issue, optimized parameters are estimated by jointly considering axle load distribution characteristics and load-associated pavement damage. Consequently, both efficient and precise traffic load spectra inputs for pavement design are established.

A NON-LINEAR MODEL FOR PREDICTING PAVEMENT SERVICEABILITY

A recursive non-linear model was developed for the prediction of pavement performance as a function of traffic characteristics, pavement structural properties and environmental conditions. The model highlights some of the advantages of relaxing the linear restriction that is usually placed on the specification form of pavement performance models. First, a functional form that better represents the physical deterioration process can be used. Second, the estimated parameters are unbiased, owing to a proper specification and the use of sound statistical techniques. Finally, the standard error of the prediction is reduced by half that of the equivalent existing linear model. This improved accuracy has important economic implications in the context of pavement management. The model developed as part of this research enables the determination of an unbiased exponent of the so-called power law and of the equivalent loads for different axle configurations. The estimated exponent confirms the value of 4.2 traditionally used. However, it should be noted that this exponent is only to be used for determining damage in terms of serviceability. On the other hand, equivalent loads estimated for different axle configurations tend to differ from traditionally used values, especially in the case of single axles with single wheels.

Effect of Traffic Load Measurement Bias on Pavement Life Prediction: A Mechanistic-Empirical Perspective

The new Mechanistic-Empirical Pavement Design Guide (MEPDG) requires comprehensive traffic inputs to predict pavement performance. Axle load spectra play a critical role in the impact of traffic on pavement performance. Weigh-in-motion (WIM) systems are becoming widely used as an efficient means of collecting traffic load data for mechanistic pavement design. The quality of the WIM-based data, however, remains a concern among pavement engineers. Previous research showed that WIM equipment calibration bias may lead to significant bias in the estimation of equivalent single-axle loads. This study investigated the effect of traffic load measurement bias due to WIM measurement errors on pavement life prediction on the basis of the mechanistic-empirical approach proposed by MEPDG. The results of this study not only support but also advance the existing research in this critical area. The findings of this study can be used to estimate pavement life prediction bias when inaccurate WIM data are used. They can also serve as guidelines for state highway agencies for the selection of WIM equipment and the establishment of criteria for equipment calibration.

Effect of Measured Three-Dimensional Tire—Pavement Contact Stress on Pavement Response at Asphalt Surface

Top-down cracking results from horizontal strains at the pavement surface that are caused by high wheel loads. When horizontal strains are calculated, the tire–pavement contact stress is assumed to be equal to the tire inflation pressure and to be distributed uniformly over a circular area. In reality, tire–pavement contact stress is distributed nonuniformly over a noncircular area. Actual contact stresses vary along the longitudinal and transverse directions. For precise assessment of the effect of actual three-dimensional (3-D) contact stress on pavement response, this study evaluates the horizontal strains at the surface of the asphalt layer produced by measured 3-D nonuniform stresses. The analysis evaluates five wheel loads, five tire inflation pressures, and 12 pavement structures. A multilayer linear elastic computer program, CIRCLY, was used to estimate horizontal strains in the longitudinal and transverse directions. CIR-CLYs ability to handle normal and shear stresses at the pavement surface makes it a preferred tool compared with finite element models, which are more data and computational demanding. Strain distributions due to uniform stress were also calculated for comparative purposes. The results show that horizontal strains at the pavement surface are compressive within the contact area and tensile at the edge of or outside the contact area. The vertical contact stress and transverse contact stress have significant effects on the longitudinal strains. The vertical stress dominates the developed transverse strains. Two models were developed to estimate the surface critical horizontal strains in the longitudinal and transverse directions. These equations could serve as a tool for determination of critical combinations of asphalt thickness, wheel load, and tire pressure that the uniform stress model would grossly misestimate.

Evaluating the Effect of Natural Gas Developments on Highways

The natural gas reserve in the Barnett Shale geological formation is one of the largest onshore natural gas reserves in the United States. The development of a natural gas well is a traffic-intensive operation that involves high volumes of truck traffic; saltwater traffic generated during the production phase is a major contributor to truck volume. The effect of traffic related to the natural gas industry on Texas highways is quantified. The oversized and overweight database maintained by the Texas Department of Transportation (DOT) was used to quantify several key characteristics of the rig traffic. In addition, personnel from the Texas DOTs Fort Worth District provided critical information about construction and saltwater traffic, including truck types used, gross vehicular weights, number of trips, and haul distances. Pavement sections were located along the busiest trucking corridors to evaluate damage caused by the truck traffic associated with natural gas development. Results indicated the approximate damage due to rig traffic was 1.6%, to construction traffic was 13%, and to saltwater traffic was 6%, relative to the damage caused by design traffic in terms of rutting. Additional damage caused by the natural gas truck traffic translated into reduced service life for pavements in the region. Results indicated a reduced service life of approximately 5.6%, 29%, and 16% associated with rig, construction, and saltwater traffic, respectively, in terms of rutting.