چكيده به لاتين
Fatigue and rutting as major failures of asphalt pavements are complex processes that are subject to various factors. Among the influential factors, asphalt binder plays an important role in many functional aspects of pavement as a factor in the viscoelastic behavior of asphalt mixtures. Accordingly, predicting the properties of asphalt mixtures with the help of asphalt binder properties reduces the heavy laboratory costs and saves time. Despite the good performance of MSCR and LAS experiments in predicting the rutting resistance and fatigue life of asphalt mixtures, the heavy laboratory costs and the impact of different conditions on the behavior of asphalt mixtures continue to be one of the main challenges in the field of pavement.
The aim of this study was to investigate the rutting resistance and fatigue life of modified binders under the influence of effective factors such as basic binders characteristics, type and percentage of additives, functional temperatures of pavement, loading and aging levels. In this regard, three additives SBS, GTR and PPA have been used as binders modifiers and ANN, ANFIS, GMDH and GP algorithms and nonlinear regression to determine the binders behavior prediction model.
The results of the temperature sweep test showed that despite the relatively similar performance of the three additives at intermediate temperatures, they behaved quite differently with increasing temperature. As the temperature increased, binders containing SBS showed better resistance to permanent deformation than the other two additives due to its polymer bonding. The results of the MSCR test showed that by increasing the stress level, the positive yield of binders containing SBS was more tangible by increasing the stress level compared to the other two binders. Also, with increasing stress to 12.8 kPa, binders sensitivity to stress changes increased and nonlinear behavior in binders became apparent. This negative effect was more pronounced in neat and binders containing PPA than in binders containing GTR and SBS. LAS experimental results showed that despite the positive effect of additives on binders fatigue behavior at low strain levels, but with increasing strain level, the positive effect of additives on binders fatigue resistance decreased and at high strain levels, the fatigue life of modified binders was almost equal to neat binders. . Examining the effect of aging intensity and increasing stress level on the accuracy of predicting the rutting resistance of asphalt mixtures using Jnr values of MSCR test showed that the best relationship was observed at the stress level of 12.8 kPa and short-term aging of asphalt mixtures. Also, the evaluation of fatigue behavior showed that to increase the accuracy of predicting the fatigue life of asphalt mixtures using LAS test, the need for long-term aging on asphalt mixtures along with applying PAV aging to binders is quite felt.
The results of the modeling section for predicting binders rutting resistance showed that the extended generalized regression models, GMDH and ANFIS did not have sufficient strength and desirability, while ANN models with R2=0.993, VAF=99.3%, RMSE =27.5 and MAPE=13.7% values had the highest accuracy and lowest error in predicting the Jnr values and showed very good predictability in predicting the rutting resistance of modified binders. In addition, the performance of the developed models to predict the fatigue life of binders showed, ANN and ANFIS models with optimal structures had very high correlation coefficients and both models showed very good ability to predict the fatigue life. As the results analysis showed, the proposed ANN model had the lowest error and highest accuracy by presenting the RMSE=258956, MAPE=2.57, COV=71.37 and R2=0.994 values and showed a high ability to predict the fatigue life of binders under the influence of different aging conditions and strain levels.
