Development and Validation of Performance Prediction Models and Specifications for Asphalt Binders and Paving Mixes Essay Example

Development and Validation of Performance Prediction Models and Specifications for Asphalt Binders and Paving Mixes Essay Development and Validation of Performance Prediction Models and Specifications for Asphalt Binders and Paving Mixes Dr. Robert L. Lytton Dr. Jacob Uzan Dr. Emmanuel G. Femando Texas Transportation Institute Texas AM University College Station, Texas Dr. Reynaldo Roque Dr. Dennis Hiltunen Dr. Shelley M. Stoffels Pennsylvania Transportation Institute Pennsylvania State University University Park, Pennsylvania .r_rrD Strategic Highway Research Program National Research Council Washington, DC 1993 SHRP-A-357 ISBN 0-309-05617-9 Contract A-005 Product No. 012 Program Manager: Edward Harrigan Project Manager: Harold Von Quintus Production Editor: Cara J. Tate Program Area Secretary: Juliet Narsiah October 1993 key words: calibration crack initiation crack propagation creep compliance elastic fatigue cracking fracture mechanics fracture toughness laboratory testing of asphalt concrete material properties mechanics microcracks pattern search method pavement performance prediction plastic resi lient dilatancy resilient properties rutting system identification method thermal cracking validation vermeer plastic properties viscoelastic viscoplasticStrategic Highway Research Program National Academy of Sciences 2101 Constitution Avenue N. W. Washington, DC 20418 (202) 334-3774 The publication of this report does not necessarily indicate approval or endorsement of the findings, opinions, conclusions, or recommendations either inferred or specifically expressed herein by the National Academy of Sciences, the United States Government, or the American Association of State Highway and Transportation Officials or its member states.  © 1993 National Academy of Sciences 1. 5M/NAP/1093 Acknowledgments The research described herein was supported by the Strategic Highway Research Program (SHRP).SHRP is a unit of the National Research Council that was authorized by section 128 of the Surface Transportation and Uniform Relocation Assistance Act of 1987. In a project as extensive and dem anding as the SHRP A-005 project, the utmost efforts were called for and received from the project staff. Of particular note are the contributions of W. W. Crockford, Ming-Lou Liu, T. Freeman, C. K. Estakhri, Fuming Wang, C. H. Michalak, J. N. Galey, N. Stubbs, O. J. Pendleton, M. Chau, C. E. Schlieker, S. Phillips, and S. Cain with the Texas Transportation Institute (TTI) team.TTI secretarial services were provided by B. Cullen, C. Bryan, G. Fattorini, and S. Pierce. Noteworthy contributors from the Pennsylvania Transportation Institute team were S. Arnold, M. G. Sharma, W. Butlar, T. Farwana, N. Kim, K. Knechtel, W. Lauritzen, S. Reddy, P. Romero, N. Tabatabaee, and V. Tandon. SHRP technical staff who provided valuable assistance were H. Von Quintus and R. Leahy. ..  ° 111 Abstract The objectives of SHRPs asphalt research were: to extend the life or reduce the life cycle costs of asphalt pavements; to reduce maintenance costs; and to minimize the number of premature failures.An important result of this research effort is the development of performance-based specifications for asphalt binders and mixtures to control three distress modes: rutting, fatigue cracking and thermal cracking. The SHRP A-005 project developed detailed pavement performance models to support these binder and mixture specifications and performance-based mixture designs. This report documents the findings of this extensive research effort and provides supporting data for the performance-based specifications and mixture design procedure called SUPERPAVE TM.SHRP Contracts A-002A (Binder Characterization and Evaluation) and A-003A (Performance-Related Testing and Measuring of Asphalt-Aggregate Interactions and Mixtures) accomplished three goals: 1) to identify relationships between asphalt binder properties and field performance; 2) to develop models for the estimation of pavement distress based the physicochemical properties of the asphalt binder; and 3) to produce test methods to measure these pavement performance factors. The final validation of these relationships, test methods and findings is the focus of long-term, controlled field experiments (Specific Pavement Studies[SPS]-9).This validation approach requires 10 or more years of investigation and observations, which is inconsistent with SHRPs objective to rapidly develop performance-based specifications. The validation process was accelerated by the SHRP A-005 contract, which developed and used a sophisticated, mechanistic-based pavement performance model to define the relationships between asphalt binder and mixture properties and pavement distress. The objectives of the SHRP A-005 Contract are as follows: 1. stablish, on the basis of documented field performance data, criteria that may be used in support of the asphalt binder specification and for the design of asphaltaggregate mixture systems develop performance prediction models for asphalt binder and for asphaltaggregate mixture systems 2. V A comprehens ive pavement of fatigue cracking, thermal time, using results from the environmental data and can (combination of binder and performance model was developed that predicts the amount cracking and rutting in asphalt concrete pavements with accelerated laboratory tests.The model uses detailed be used to determine the optimal mixture design aggregates) for specific conditions. The pavement performance models for each distress were also used to confirm the relevant binder and mixture properties established by other SHRP contractors using accelerated laboratory tests and laboratory torture tests; and to establish the degree of correlation between those asphalt binder and mixture properties. Results from these model studies were used to confirm or make recommendations for revisions to the asphalt binder specification.In general, these analytical studies confirmed the material properties and specific limits used in the performance based specification. The model is simple, runs on a microcom puter, and can be used to evaluate or design asphalt concrete mixtures. The model can minimize a specific distress or combinations of different distresses, or it can set specification limits for specific materials and environments. The model has three parts: 1) a mixture evaluation model, 2) a pavement response model, and 3) a pavement distress model. The mixture evaluation program calculates the relevant binder and mix properties from the accelerated laboratory tests.These properties are used with the pavement response program to evaluate the behavior of a mixture subjected to traffic and/or environmental loads. This mixture evaluation program calculates the non-linear elastic, viscoelastic, plastic and fracture properties of a mixture. The pavement response program calculates the stresses and strains in an asphalt-aggregate system from applied wheel loads and temperature changes. The pavement distress program uses the relevant mixture properties and the appropriate stresses and st rains to calculate the amount of cracking (from wheel loads and environmental loads) and rutting with time. i Executive Summary The objectives of SHRPs asphalt research were: to extend the life or reduce the life cycle costs of asphalt pavements; to reduce maintenance costs; and to minimize the number of premature failures. An important result of this research effort is the development of performance-based specifications for asphalt binders and mixtures to control three distress modes: rutting, fatigue cracking and thermal cracking. The SHRP A-005 project developed detailed pavement performance models to support these binder and mixture specifications and performance-based mixture designs.This report documents the findings of this extensive research effort and provides supporting data for the performance-based specifications and mixture design procedure called SUPERPAVE TM. The objectives of the SHRP A-005 Contract are as follows: 1. establish, on the basis of documented field perfo rmance data, criteria that may be used in support of the asphalt binder specification and for the design of asphaltaggregate mixture systems develop performance prediction aggregate mixture systems models for asphalt binder and for asphalt- 2.The successful development of performance-based specifications required the validation of those properties identified as important determinants of pavement performance. This validation effort is a three-stage process. The first two stages were completed under SHRP. The third or final stage will be directed by the FHWA through the SPS-9 projects of the ongoing Long-Term Pavement Performance (LTPP) program. The first stage of the validation effort was accomplished through the use of accelerated laboratory tests by the SHRP A-003A contract.The second stage was accomplished by the SHRP A-005 contract through the use of field performance data. This second stage established the degree of correlation between the field observations of distress and: 1. 2. those asphalt binder properties shown to significantly affect the performancerelated characteristics of asphalt-aggregate mixtures performance-related material properties of asphalt-aggregate mixtures Furthermore, the second stage validation effort provided the experimental results needed to set the specification limits for the relevant binder and mixture properties selected to vii ontrol pavement distress. These experimental results were also used to estimate the accuracy of the accelerated laboratory test methods employed to measure the relevant properties. This second stage validation effort relied heavily on the sampling and testing of asphalt-aggregate mixtures from the LTPP General Pavement Study (GPS) sections. A comprehensive pavement of fatigue cracking, thermal time, using results from the environmental data and can (combination of binder and performance model was developed that predicts the amount cracking and rutting in asphalt concrete pavements with accelerated labo ratory tests.The model uses detailed be used to determine the optimal mixture design aggregates) for specific conditions. The pavement performance models for each distress were also used to confirm the relevant binder and mixture properties established by other SHRP contractors using accelerated laboratory tests and laboratory torture tests; and to establish the degree of correlation between those asphalt binder and mixture properties. Results from these model studies were used to confirm or make recommendations for revisions to the asphalt binder specification.In general, these analytical studies confirmed the material properties and specific limitsused in the performance based specification. A brief list of the products from this research follows. Systems Identification Model The performance models are used to predict distresses or observations, such as cracking and rutting. The SID generates material properties for those models to minimize the error between the observed and predi cted pavement distress. Basically, the SID sets up a sensitivity matrix and solves for the required change when the change is larger than the prescribed acceptable value for convergence.This system was used in the calibration process for each of the pavement distresses, and to establish the specification limits for asphalt binders and mixtures. The SID can be extremely useful to mix designers and specification writers to theoretically establish limits for mixture properties under projectspecific conditions. It can also be used to update and revise the calibration coefficients of the pavement performance models. ? Micromechanics (MM) Model The micromechanics (MM) model is a system of material response equations for each material component used in asphalt concrete mixes.The model combines the measured response characteristics (in terms of creep-compliance) of the binder, mastic and aggregate to predict that same response of a combined mixture. This model will be useful to mix designer s to theoretically evaluate and initially optimize a mixture blend prior to any accelerated laboratory tests on that mix.  °Ã‚ °Ã‚ ° _III Pavement Distress Models A set of comprehensive models that predict the amount of fatigue cracking, rutting and thermal cracking with time are available.Each of these models use results from the accelerated laboratory tests to calculate critical material properties of the mixture for the stresses and strains imposed on the mixture from applied wheel and environmental loads. These stresses and strains are then used to calculate each type of distress. Fatigue cracking is calculated in terms of square feet per unit area and is based on fracture mechanics methodology. Thermal cracks are calculated in terms of linear feet per unit length and used linear viscoelasticity to determine thermal stresses and fracture mechanics principles to estimate the amount of cracking with time.Rutting is calculated in terms of an average rut depth, based on a 4-ft str aight edge. The model uses the concept of a material elasto-plastic formulation to calculate the initial plastic deformation due to both shearing and volumetric stresses and uses the concept of strainhardening to calculate the total depth rutting within each different season. Binder Performance-Based Specification Binder performance-based specifications limits for thermal cracking were found based upon patterns of binder stiffness and log slope of the compliance as they affect the extent of cracking.No such patterns could be found to relate binder properties to the rate of appearance of load-related distress, leading to the conclusion that binder properties alone do not control the fatigue cracking or rutting performance of asphalt concrete. Instead, specifications for both the binder and mixture properties are necessary for quality assurance against load-related distress. The calibrated performance prediction models provide a sound and reliable means for such mixture specification limits.Binder properties alone also do not control thermal cracking and mixture tests must be performed to evaluate the thermal cracking performance of a particular mixture. Finally, it is important to note that all of the quantities used for specification limits are material properties that are defined as they are used in mechanics. Its efficiency and simplicity in descriptions of essential and sensitive material characteristics make the discipline of mechanics a very useful and practical framework, not only for predicting distress but for the development of specifications that are directly related to performanceMaterial Property Relationships of pavement materials have been used in Several new methods for the characterization the load-related performance model. †¢ The resilient properties of asphalt were found to be identical to those used to characterize the materials in the supporting layers. Poissons ratio was found to be stress dependent. A tension in the asphalt binder w as found to be acting between ix I the aggregate temperature. †¢ particles. Both of these effects were found to depend strongly uponThe plasticity parameters which characterize asphaltic concrete include two friction angles, a cohesion, and a volumetric component, all of which are also commonly used in soil mechanics to characterize base course and subgrade materials. †¢ The accumulation of plastic deformations due to repeated traffic loading was shown to be related to the plastic properties and to the slope of the log creep compliance curve of the asphalt mix. The fracture properties of asphalt concrete were found to apply to both the crack initiation and crack propagation phases of crack growth in viscoelastic materials was found to be applicable to asphalt concrete. Pavement Performance Models Models of permanent deformation and cracking damage were developed and calibrated to field observations of these distresses on pavement sites distributed across the United States and Canada. Calibration adjustments were made to the predictions by a mathematical technique known as the Systems Identification (SID) method or by a nonlinear pattern search method. The adjustments were made in seven material properties that represent the two phases of fatigue cracking and in one coefficient of the rut depth.The calibration adjustments were not large nor did they vary much from one climatic zone to another, indicating that the prediction models developed for both fatigue cracking and rutting are sound. The calibrated models take into account the traffic, the temperature variations in all layers throughout the year, and the consequent variations in the material properties of each pavement layer in computing the accumulations of rutting and cracking with the seasons. Even though the pavement predictions were calibrated in sets grouped into climatic zones, the performance predicted for each pavement matched the actual performance very well.Temperature and moisture com putations were made with the FHWA Integrated Model of Environmental Effects, this project which showned to be capable of reproducing field measurements very well. The calibrated load-related model was shown to be well suited for the development of performance-related specifications and for the SHRP mix design model SUPERPAVE TM. X Field Experiments GPS Sites An experiment design was adopted to provide a wide variety of climatic exposure and rates of appearance of rutting and fatigue cracking.Very detailed information was obtained for each pavement section to permit a fairly complete characterization of the material properties of each layer. Cores were taken and tested in a variety of temperatures, stress and strain states. Material properties derived from these were found to be comparable to elastic and viscoelastic properties backcalculated from nondestructive deflection tests made with a falling weight deflectometer. Dynamic analysis was performed on about half of the pavement se ctions to determine the viscoelastic and damped elastic characteristics of each layer.It was shown that this backcalculation can produce reasonable values of m, the slope of the log creep compliance curve of the asphalt surface layer. This slope has been found to be the most important single variable in predicting rutting, fatigue cracking, and thermal cracking. Thus, the ability to measure it nondestructively in the field is a research result of this project that is of primary importance. The FHWA Integrated Model of Environmental Effects was shown to be very accurate in predicting seasonal variations of layer temperature, moisture, and moduli.This model, and the pavement performance prediction model form the computational elements of the SHRP SUPERPAVE TM mix design system. Laboratory Testing of Asphalt Concrete Mixtures A method of taking and testing cores in a variety of test modes and stress states was developed in this project. Among other testing modes, the equipment can run monotonic, repeated, and frequency sweep tests in tension or compression with triaxial testing equipment. Tensile tests have been avoided customarily because the samples fail in the grips.In this project the difficulty was found to be alignment, which was cured with a high production rate gluing jig when this device was used and 98 percent of the samples tested to failure parted in the center of the sample and away from the loading heads. Simulation of traffic loading rates in the tests, would have required much more sophisticated equipment, both to load and to record accurately the response of the sample. Instead, the objective of each test described herein was to measure the material response under a steady applied load, in a sample where the stress and strain patterns were uniform.This permited an accurate determination of the material properties that were then be used to determine the response of the material to traffic loading rates. Use of a cylindrical sample, taken from a fi eld core, permited the application of a variety of triaxial stress states that allowed a determination of how the material properties depended upon the stress state. The sample was placed in a temperature-controlled chamber to permit testing at the full range of temperatures to which an asphalt concrete layer is subject in the field.A specially designed metal chamber allowed the sample to be tested under a wide variety of confining pressures. A pair of LVDTs measured the †¢ xi i lateral strain of the sample, which made possible a complete and accurate characterization of the elastic, plastic, viscoelastic, viscoplastic resilient, dilatant, strength, and fracture properties of a material. Overall, the coring and testing apparatus developed in this project is reliable, accurate, and relatively simple both to construct and to operate.The A-003A project developed methods to analyze the test data for the Accelerated Laboratory Tests to produce the viscoelastic, resilient and plastic ity properties needed as input to the load-related performance model. In the course of all of the test material characterization and performance prediction work performed in this project, it became clear how important it is to measure the lateral strain as well as the vertical strain in any sample of pavement material tested, whether it is asphalt concrete, base course or subgrade. Pavement Performance PredictionThe performance prediction models developed in this project are developed for use on microcomputers and have been incorporated into the SHRP SUPERPAVE TM mix design procedure. The use of plasticity characterization to predict rutting makes it possible to consider both vertical and lateral permanent deformation in all of the pavement layers. The prediction of fatigue cracking is separated into two processes: crack initiation and crack propagation. During crack initiation, distributed microcracks form, grow, densify, and coalesce under repeated loads. Healing is assumed to occ ur primarily in the crack initiation phase.During crack propagation, cracks grow outward from the tensile zone where the microcracks form. These cracks grow from small but visible flaws into fulldepth cracks of the asphalt layer primarily in the shearing propagation mode. The predictions made with this approach were found to be unusually accurate even before they were calibrated to match the field observations. The calibration process was used extensively in this project as a form of nonlinear regression analysis. It was employed in a variety of ways to reduce laboratory data as well as to bring predicted distress into close accordance with the field observations.Two powerful techniques were used which should find many practical uses in the pavement field in the future: the Systems Identification SID method, and the pattern search method. Both methods satisfy a least sum of squared errors criterion. Calibration is done in this program by finding multiplying factors for the most sens itive material properties. These were found by sensitivity analysis to be the stiffness of the asphalt concrete, its tensile strength, the log slope of the creep compliance curve, and the two healing coefficients.The calibration method has another unique use which was developed and demonstrated in this project. Once a performance model is calibrated, the same calibration process can be converted to determine minimum values of those material properties which are xii needed to meet prescribed distress target levels. This use of the performance prediction programs to assist in setting specification levels for material properties is very promising and deserves further exploration. Validation of Relations Between Material Properties nd Pavement Performance While it is evident that the properties of the asphalt binder are an important factor in the service life of asphalt concrete, the lack of observed patterns of high and low rates of rutting and fatigue cracking show plainly that the se lection of the binder alone will not assure good load-related performance. It is clear that the disciplines of pavement analysis, design, and mix design, and the effects of construction and weather, also play important roles in the development of load-related distress.The pavement performance prediction model represents several important advances in pavement analysis and design and in mix design through its use in SUPERPAVE TM. The micromechanics model of the properties of a mix which was developed in this project needs to be studied in detail for its potential benefits to an understanding of the important interactions of the constituents in mix design. Without such an understanding, setting specification limits on the properties of the binder will have little or no effect on the retardation of load-related distress. ThermalCracking were drawn regarding the effectiveness of new SHRP binder which are based on the results of the field validation studies The following conclusions and m ixture specifications presented in this report. †¢ The indirect tensile creep and failure test at low temperatures (ITLT), that was developed at PTI as part of this contract is clearly suitable to support the new SHRP mixture specification for the control of thermal cracking. Excellent correspondence was obtained between observed thermal cracking in the field and thermal cracking predicted by using properties determined from the ITLT.Use of the time-temperature superposition principle to determine the viscoelastic properties of asphalt mixtures, and determination of fracture parameters from viscoelastic properties and mixture strength, appear to be valid and suitable methods to determine the low temperature mixture properties needed for the control of thermal cracking. The parameters S and m (stiffness and m-value at 60-second loading time) determined from the bending beam rheometer tests appear to be suitable for evaluating the relative thermal cracking performance of asphalt binders.S and m correlated reasonably well with observed thermal cracking in the field. †¢ †¢ xiii †¢ The limits on S and 29,000 psi and m gt; in this investigation, 45,000 psi and m gt; m defined in version 7G of the binder specification (S lt; 0. 35) appear to be too restrictive. Based on the data obtained the following limits appear to be more appropriate: $ lt; 0. 30. thermal cracking effect on the thermal ITLT test must be a particular mixture. †¢ The binder specification alone does not guarantee adequate performance. Mixture characteristics may have a significant cracking performance of a particular binder.Therefore, the performed to evaluate the thermal cracking performance of Aging The aging studies led to the following conclusions: †¢ For the range of mixtures evaluated in this investigation, it appears that the thermal cracking performance of asphalt mixtures can be reasonably evaluated on the basis of only the properties of the mixture after long- term-aging (i. e. without accounting for the rate of age-hardening). It appears that long-term mixture aging (as defined by the SHRP A003-A contractor) is needed to properly evaluate the thermal cracking resistance of asphalt mixtures.This assumes that the long-term mixture aging procedure results in aging levels comparable to those observed in the field. Based on limited data, it appears that aging levels comparable to those observed in the field, as reflected by changes in the low temperature properties of the binder, can be attained through TFOT followed by PAV aging as proposed in the SHRP binder specification. The effectiveness of the long-term mixture aging procedure proposed by the SHRP A-003A contractor could not be evaluated because the laboratory mixing and compaction methods used in this investigation did not represent field mixture and compaction.Low temperature properties of laboratory-produced mixtures were very different than properties of field cores of the same mixt ure. †¢ †¢ †¢ Moisture Effects studies led to the following conclusions: The moisture †¢ Moisture changes within the asphalt mixture have a significant effect on fundamental low temperature properties of the mixture and on thermal cracking performance. xiv †¢ Low temperature properties for the evaluation of thermal cracking resistance of a mixture using the PTI thermal cracking model must be determined on a dry mixtures (i. e. ither laboratory-produced mixtures or field cores that have been thoroughly dried prior to testing). Use of properties determined at other moisture levels may result in significant errors in thermal cracking predictions and thus in evaluating the thermal cracking resistance of asphalt mixtures. For the range of mixtures evaluated in this investigation, it appears that the thermal cracking performance of asphalt mixtures can be reasonably evaluated on the basis of only the dry properties of the mixture (i. e. , without accounting for ef fects of moisture changes).