Probabilistic characterisation of damage characteristic curve of asphalt concrete mixtures

ABSTRACT

Due to its efficiency, viscoelastic continuum damage mechanics has been used for modelling asphalt concrete behaviour widely. In this approach, viscoelastic and damage response are used to obtain a damage characteristic curve. Under ideal conditions, these damage characteristic curves obtained under different loading conditions should collapse to a unique curve. Due to inherent variations during specimen fabrication and testing, significant scatter are found in damage characteristic curves even under well-controlled laboratory conditions. Owing to deterministic nature, present day viscoelastic continuum damage models fail to account this scatter in damage characteristic curves. This paper presents a probabilistic approach to describe the scatter in damage characteristic curves. Several specimens were tested for their viscoelastic properties and damage response. These test results were used to construct damage characteristic curves. The damage parameter values obtained at a particular normalised pseudostiffness values were fitted with normal, lognormal and Weibull distribution. It was observed that damage parameter values (at a particular normalised pseudostiffness) could be best described using Weibull distribution when compared to lognormal and normal distribution. Due to its flexibility, three-parameter Weibull distribution was found to fit better than two-parameter Weibull distribution. Further, significant differences were found between probabilistic damage characteristic curves developed in this research and conventional approach. The proposed methodology combines advantages of continuum damage mechanics as well as probabilistic approaches. These probabilistic fatigue curves can be conveniently used for reliability-based pavement design.

KEYWORDS:

• Damage characteristic curve
• asphalt concrete
• probabilistic approach
• viscoelastic continuum damage
• beam fatigue
• pseudostiffness
• dynamic modulus
• phase angle
• mastercurve

Acknowledgements

The authors are grateful to Prof. Jo S. Daniel for allowing use of mixture testing data acquired at University of New Hampshire in this research. Also, first and second authors are grateful to Ministry of Human Resources and Development, India and National Science Foundation, U.S.A. for their financial support, respectively.