Statistics-Based Design of Experimental Framework to Formulate Ternary RAP Binder Blends

Abstract

The conventional design approaches for reclaimed asphalt pavement (RAP) mix follow blending rules to select the grade of the virgin binder for hot recycled bituminous mixtures. The interactions of RAP binder with the virgin binder are highly complex in the presence of the rejuvenator. Identifying the optimal proportion of RAP binder, virgin binder, and rejuvenator is often a trial-and-error procedure, involving considerable experimental data collection. The current research study proposes the statistical-based design of experiments (DoE) framework to design ternary RAP binder blend, develop response surface prediction models, and optimize the proportions to meet targeted binder properties of viscosity grade (VG30) binder. A desirability-based multiple response optimization is carried out, and the optimal blend proportions with target binder properties are arrived at with different desirability levels. A total of 13 RAP binder blends are formulated using the I-Optimal algorithm in the current study. Experimental data such as softening point, high-temperature performance grade (PG), and linear viscoelastic parameters in the temperature range of 46°C to 70°C and frequency range of 0.1 to 50 Hz are collected. The binder blends are evaluated using master curve parameters and relaxation spectrum. The response surface models are developed for two measured parameters (high-temperature PG and softening point), three computed master curve parameters, and four relaxation spectrum parameters. These models capture the possible individual component effects and combined component effects on the final RAP binder blend through linear and nonlinear terms. A desirability-based simplex optimization is adopted to identify combinations of constituents at various levels of desirability that comprises only direct measurements, direct measurements and master curve parameters, and direct measurements, master curve parameters, and relaxation spectrum. Using this proposed approach, a wide range of solution zones are identified with various levels of desirability. Reclaimed asphalt pavement (RAP) consisting of aged bitumen coated on mineral aggregates is used for recycling in pavement construction. Many challenges exist related to the optimal use of such material. The aged binder may need to have the required properties suitable for recycling. Generally, rejuvenators are added to the RAP to enhance the properties of aged binder. However, how much rejuvenator should be added is unclear. Aged bitumen from different sources of RAP can have different degrees of aging. In addition, it may not be easy to estimate the extent of blending of the RAP with the virgin materials (aggregates and bitumen), as it is linked to the characteristics of aged binder and the choice of fresh bitumen and rejuvenator. Also, the aged bitumen, when mixed with a rejuvenator and virgin bitumen, may not necessarily result in a final mix with bitumen that meets the requirements. As can be seen, the choices are entirely diverse and varied. This research attempts to solve such issues using a statistics-based tool called design of experiments (DoE). Using such a tool, this study demonstrates how to optimally choose the proportions of virgin bitumen, RAP bitumen, and rejuvenator. This work presents a methodology that can be implemented efficiently using commercial software by any asphalt mix designer.