RELIABILITY-BASED MODEL FOR PREDICTING PAVEMENT THERMAL-CRACKING

Thermal cracking of asphalt pavements is a serious problem in Canada a nd the northern parts of the United States. On many occasions, either at the design stage or during service, highway agencies demand a forec ast of pavement performance rating, which is highly sensitive to the i ntensity of cracking. There are two failure modes of thermal cracking: low-temperature cracking and thermal-fatigue cracking, Low-temperatur e cracking is caused by accumulated thermal stresses in the pavement l ayer during cold winters or spring thaws. Thermal-fatigue cracking is caused by daily cyclic thermal loading. Classical probabilistic approa ches have been applied to pavement design systems, including thermal c racking predictions, during the past three decades. Advances in reliab ility analysis, however, have proven that classical reliability method s are inconsistent, and mandate that current design procedures should be revised accordingly. This paper presents an improved reliability mo del for predicting thermal cracking. The proposed model accounts for t he variability in the component design variables and the correlation b etween the two failure modes. The model results were verified using Mo nte Carlo simulation, and the sensitivity of the predicted intensity o f cracking to various design variables was examined.