Framework Development of Performance Prediction Models for Concrete Bridges

Sound performance prediction models are essential for maintenance decision making in bridge management systems (BMSs). Markov chain models have been extensively used for performance prediction in BMSs. However, these models have some limitations, for example, being previously based on constant inspection time intervals and/or on the Markov property assumption. Furthermore, these models do not make reference to the region-specific nature of the available condition data that may affect their effectiveness and applicability. This paper identifies some limitations of the Markov performance prediction models used in many state-of-the-art BMSs and proposes a framework that can address these limitations. The developed framework incorporates the semi-Markov approach in addition to the widely used Markov chain modeling. This combined approach is expected to improve the effectiveness and applicability of performance prediction models.