Abstract The adoption of mechanistic-empirical methods for asphalt pavement design has been taking place in various countries since the early 2000s. What these methods have in common is the use of robust databases capable of generating accurate transfer functions based on their calibration processes. This study evaluates the accuracy of the Brazilian mechanistic-empirical method (MeDiNa) for estimating cracked area evolution when applied to four different experimental segments of the Brazilian federal highway network, in the state of Bahia. Based on these analyses, the study proposes a methodology for obtaining a new transfer function, divided into three stages: (i) preliminary calibration, using pavement performance data from the federal highway pavement management system database; (ii) calibration check, using experimental sections from pavements that have been in service for less than 10 years and exhibit low cracking; and (iii) calibration consolidation, involving experimental segments built specifically for this purpose. The results of the initial simulations show that, while MeDiNa indicates accelerated cracking growth for the four sections evaluated, the field data revealed low levels of cracking over the same period. In particular, while the model predicted failure within 2 to 5 years, field data showed minimal cracking even after extended service. These findings demonstrate the necessity of refining the current crack transfer function to enhance predictive reliability, which can be effectively accomplished using the proposed calibration framework.
Almeida et al. (Wed,) studied this question.