Use of monitoring extreme data for the performance prediction of structures: Bayesian updating

Sensors of modern monitoring systems used in structural engineering provide data used for reliability assessment and maintenance planning. The storage and evaluation of sensor information are space and time-consuming activities. Therefore, it is necessary to process only the monitored data indicating a violation of defined performance thresholds. However, this process should not discard the knowledge gained from past monitored data and should allow the updating of prediction functions incorporating this knowledge. The objectives of this paper are to present: (a) a procedure for the effective incorporation of monitored data for the reliability assessment of structural components, (b) an approach for the updating of prediction functions and criteria for the interruption of monitoring, and (c) an effective use of the Bayesian approach for the incorporation of historical data in the structural reliability assessment. The proposed procedures and concepts are applied to the monitoring data obtained from the I-39 Northbound Bridge over the Wisconsin River in Wisconsin, USA. A monitoring program on that bridge was performed by the ATLSS Center at Lehigh University.