Univariate and Multivariate Exploration of Resilient Modulus for Warm Mix Asphalt Mixtures

Abstract

Abstract This paper predicts the resilient modulus (Mr) for warm mix asphalt (WMA) mixtures prepared using aspha-min. Various predictor variables were analyzed, including asphalt cement types, asphalt contents, nominal maximum aggregate sizes (NMAS), filler content, test temperatures, and loading times. Univariate and multivariate analyses were conducted to examine the behavior of each predictor variable individually and collectively. Through univariate analysis, it was observed that Mr exhibited an inverse trend with asphalt cement grade, NMAS, test temperature, and load duration. Although Mr increased slightly with higher filler and asphalt content, the magnitude of this increase was minimal. Multivariate analysis revealed that the rate of change of Mr was highly dependent on NMAS and the thermo-rheological properties of the asphalt cement. Initially, a linear regression model was developed; however, it underestimated low Mr values and overestimated high Mr values. Moreover, the linear model resulted in negative Mr values, indicating an inadequate representation of the relationship between Mr and predictor variables. Consequently, a nonlinear transformed regression framework was employed to develop an equation that more accurately predicted the Mr values of WMA mixtures. The resulting predictive model exhibited a coefficient of determination (R2) of approximately 95 %. To validate the effectiveness of the proposed model, the remaining 30 % of the test data was utilized. The results demonstrated that the developed model effectively represented the observed data not used during the model-building process. This validation was supported by an R2 of 95.8 % between the predicted and measured Mr values of WMA mixtures.