An Alternative Approach to Damage Models for Mechanistic-Empirical Pavement Design

Abstract

Abstract The concept of transfer functions is well entrenched in mechanistic-empirical (ME) pavement design and essentially represents the performance component of ME design. Initially, transfer functions were relatively simple linear log-log functions relating some critical pavement response parameter to the number of load repetitions that can be sustained at the level of the critical response parameter before a predefined level of damage is reached. Later, given more advanced computing software, the formulation of transfer functions evolved to include continuous, nonlinear models. However, all these models have one common characteristic in the sense that they all include the number of load repetitions in the model formulation. This requires the use of a “time-hardening” approach to accumulate damage under mixed traffic and environmental conditions in a recursive damage simulation process. Unfortunately, the time-hardening process is associated with large numbers of calculations to evaluate the equivalent number of load repetitions at every instance in which the traffic load or pavement variables change. This article presents an alternative approach to the formulation of damage models that is based on the Markov property of memoryless systems. The advantage of using the memoryless model formulation in a recursive damage simulation process is that the calculation of the equivalent number of load repetitions is eliminated from the process. The recursive damage simulation process can also be initiated at any level of damage without knowing the prior history that caused the damage. The formulation and calibration of memoryless damage models are illustrated for the fatigue and plastic strain damage of asphalt concrete mixes.