Abstract:Preventive maintenance (PM) is recognized as a potentially useful tool to reduce pavement network operation costs. To identify the lowest-cost maintenance and rehabilitation strategies to keep pavement conditions at their desired levels, agencies need to predict conditions on the network accurately. This need presents a challenge because pavement deterioration is a complex mechanism that involves not only structural damage but also the effects of interactions between traffic, material, environment, and time. Earlier studies have based cost analyses on deterministic pavement deterioration prediction models, even though these models lack the capability to capture the random part of true pavement field behavior. An alternative approach is to use probabilistic models, which are built on the basis of observations of in-service facilities and allow a more realistic representation of pavement performance. In this research study, pavement crack initiation models were developed on the basis of stochastic hazard rate modeling techniques. Significantly, model development included data from the California Department of Transportation Pavement Condition Survey so that the effects of a wide range of facilities and environmental conditions could be represented. This study also compared several pavement maintenance scenarios with the use of rehabilitation only as the control case. The study's performance results were then coupled with available cost information to develop life-cycle cost comparisons. The research results confirmed that, compared with rehabilitation alone and maintenance at advance levels of cracking pavement, PM practices significantly reduced long-term costs of roadway maintenance.