Optimizing Agro-based Natural Fiber Parameters to Address Binder Drainage in Open-Graded Asphalt Friction Course Mixes Employing Response Surface Methodology

ABSTRACT Open-graded asphalt friction courses (OGAFCs) offer benefits in terms of the absence of rainwater film, high skid resistance, low tire-pavement interaction noise, and better visibility under wet weather. Among the critical problems of binder draindown, raveling, and clogging encountered with OGAFCs, binder draindown is the one that can be largely resolved by using fiber with appropriate length, dosage, and type. With the growing interest in sustainable materials in road construction, the application of agro waste–derived natural fibers in pavements forms an innovative domain of research. In this study, four agro waste–derived biofibers, including banana fiber, sisal fiber, sunnhemp fiber, and pineapple fiber, were investigated as stabilizing agents to address the binder draindown challenge within OGAFC mixes. A response surface methodology (RSM) was used to optimize the fiber parameters of four agro waste–derived fibers for their application in OGAFC mixtures fabricated with unmodified and polymer-modified binders. RSM-based models were developed using three distinct input parameters: fiber lengths ranging from 3 to 12 mm, fiber dosages ranging from 0.15 to 0.45 %, and binder dosages ranging from 5.5 to 6.5 %, as well as one output parameter, binder draindown. The RSM-based developed regression models were statistically significant, with average R2 and adjusted R2 values higher than 0.9830 and 0.9695, respectively. Furthermore, the desirability function used to optimize fiber parameters yielded a desirability value greater than 0.84 for each developed model, indicating an exceptional level of predictive precision. Finally, during the model validation phase, an average percentage error deviation of 5.61 % demonstrated the efficacy of the RSM approach for fiber parameter optimization. The RSM-based optimization for each model with both binders, followed by parameter validation, yielded optimal values of 0.45 %, 9 mm, and 5.5 % for fiber dosage, fiber length, and binder dosages, respectively.