Optimization of structural health monitoring and maintenance planning based on Bayesian joint modeling of time-dependent measurements and hazard functions

Structural Health Monitoring (SHM) provides useful information that reduces the uncertainty in the time-dependent degradation processes of civil structures, and hence facilitates improved decision making. However it comes at a price that is not always justified by its benefit. Therefore, the monitoring and maintenance strategy should be planned in a way that the life-cycle benefit gained by the SHM information is maximized, or at least justified by the cost of SHM. In this contribution, a framework for evaluating the value of SHM in order to optimize the monitoring and maintenance planning is proposed. First, a joint modeling of time-dependent measurements of the structural states and the hazard function is presented. The link between the time-dependent measurements and hazard model includes the current value of the structural state and the slope of its trajectory at the same time. A Bayesian approach using the Markov Chain Monte Carlo (MCMC) method is adopted and compared for the model parameter estimation. This joint model is then implemented to predict the future structural state and hazard rate in a dynamic manner. Further, a risk acceptance criterion and maintenance planning scheme based on the mentioned joint model are proposed. The candidate SHM strategies are, subsequently, defined (e.g. based on monitoring accuracy, duration, etc.) and the potential life-cycle benefit that can be gained by each candidate SHM strategy is evaluated based on Value of Information (VoI) theory and pre-posterior analysis. Finally, the problem of optimizing the strategy of monitoring and maintenance can be solved accordingly by identifying the SHM strategy that maximizes the life-cycle benefit of the structure