Development of pavement performance models for continuously reinforced concrete pavements in Texas

This report documents the final results of a long-term performance study of CRC pavement sections in Texas. An existing distress-index that describes CRCP deterioration as a function of punchouts and patches was used in conjunction with a CRCP data base spanning over 14 years and over 751 test sections statewide to develop three types of performance prediction models. The models were calibrated using Survival Analysis, a statistical technique still unexplored in the analysis of pavement performance, that is adequate for estimating the reliability of a design or a device. This approach is theoretically sound, and it gave accurate results. The models permit the remaining life of a test section to be estimated from visual condition survey data. During the calibration process, the significance of several variables affecting pavement performance was tested. Among these variables were the elastic modulus of the Portland cement concrete and the modulus of reaction on top of the subbase, obtained by back-calculation from deflection data. This back-calculation is generally done by inverse application of layered theory, performed by one of the many computer programs available in the literature for this type of calculation. There seems to be no consensus as to which program yields the best results. In addition, for rigid pavements, the back-calculation can be also done by inverse application of plate theory. The back-calculation phase of this study was used to compare results obtained with layered and plate theory. A significant discrepancy was found, and the possible causes were analyzed. It is felt that the most important contributions of this study are in the redefinition of the problem of pavement performance models and in the innovative and theoretically sound technique applied to develop the models. The complementary analyses of the problems of restricted inference spaces for and of error propagation in pavement performance studies are also very important, because they draw attention to a crucial limitation of most pavement performance models that is often overlooked. It is hoped that the models can be useful for CRC pavement management and that the findings of this study can contribute to a better understanding of CRCP deterioration. It is felt that the data collection procedures and the data base developed in this study can be useful for many other research studies. It is strongly recommended that future attempts at calibrating a pavement performance with this reliability be made with the statistical approach used in this study.