| This dissertation documents a long-term performance study of continuously reinforced concrete pavement (CRCP). The final result consists of models to predict the probability of survival of a CRCP, from visual condition survey data. The models were obtained using a data base consisting of 747 CRCP test sections, and they were calibrated using Survival Analysis, a theoretically sound method to estimate the reliability function of a structure or a device. The use of Survival Analysis in pavement performance studies was still unexplored, and it gave good results, in spite of the fact that only 91 sections had reached failure.; During the model calibration process, the significance of several variables affecting pavement performance was tested. Among these variables were the elastic modulus of the P.C. concrete and the modulus of reaction on top of subbase, obtained by back-calculation from deflection data. In the case of rigid pavements, this back calculation can be done by inverse application of either plate or layered theory. Since no comparison between both approaches was found in the literature by the time this part of the study was being done, the parameters were back-calculated with both layered and plate theory. A significant discrepancy was found between the results, and its possible causes were analyzed.; It is felt that the most important contributions of this study reside in the redefinition of the problem of predicting pavement performance, and in the innovative and theoretically sound technique applied to develop the models. The analyses of the problems of restricted inference spaces for and of error propagation in pavement performance studies draw attention to a crucial limitation suffered by most pavement performance models, which is often overlooked and practically impossible to remove.; It is thought 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 in the future, and it is recommended that future attempts to calibrate pavement performance models that capture the reliability be done with the statistical approach used in this study. |
