APPLICATION OF MULTIVARIABLE REGRESSION MODELS FOR PREDICTION OF COMPOSITE NANOSILICA/POLYMER ASPHALT MIXTURE OBC

In this research, the effects of nanosilica particles and polymer on conventional properties of hot mix asphalt have been investigated. The study also investigates the application of various regression models for the prediction of optimum binder content (OBC). The proposed models use values for stability and flow obtained from Marshall test results. The asphalt binder was modified using polyethylene and polypropylene polymers with varying percentages of nanosilica. The fundamental mechanical and physical properties of composite nanosilica/polymer modified binder and aggregate-binder mixtures were estimated through penetration, softening point, rolling thin film oven tests (RTFOT) aging and Marshall test. The results show that application of nanosilica improves the stability, reduces optimum binder content (OBC), increases stiffness as well as strength characteristic of the asphalt mixtures. The regression models analyzed was found to yields good predicted values with a high coefficient of determination R2 and very low percentage errors of less than 5%.