Predictive SHM-supported deterioration modelling of reinforced concrete bridges

Abstract

Deterioration, increase in loading demand and change in utilization have induced an unknown level of risk in the use of transport infrastructure systems. Bridges being a vital element of such systems, due to their very nature as well as their exposure to harsh environmental conditions, should be effectively managed for the benefit of the overall transport network. Predicting the future condition and reliability of the bridges is vitally important in this process. Probabilistic models have been developed to estimate and predict the extent of deterioration in, for example, concrete bridges. However, the input parameters of these models are fraught with uncertainties, thus severely limiting their accuracy, particularly over longer time frames. On the other hand, continuous innovations in the sensing and measurement technology have lead to the development of monitoring instruments that can provide continuous (or almost continuous) data regarding the actual structural performance in the time frame. This information cannot be used directly for the prediction of future performance, first because it typically pertains to a small number of specific locations, and secondly because it needs to be combined with a whole host of other knowledge components. Furthermore, uncertainties in the instruments/measurements and in the future behaviour of the structure and its interaction with the environment (e.g. including the effects of deterioration) also hinder the predictive capability of current modelling tools. The potential benefits of improving performance prediction through the integration of health monitoring systems with probabilistic predictive models, and their implications on the management of deterioration prone structures are presented in this paper through the development of an integrated methodology. It is shown, through application case studies, that the confidence in predicted performance can be significantly increased through the use of SHM-supported modelling of deterioration and the major inspection and maintenance activities can be delayed on the account of increased confidence in the predicted performance. An example of such integration is illustrated in Figure 1 for various cases of sensor outputs including attainment of limiting value as well as (Graph Presented) confirmation of safety at various points in time during the service life. It is clear that the uncertainty is reduced with the availability of additional information and the level of this reduction depends on the quality and timing of information obtained through sensing equipment. A sensitivity study of various input parameters has concluded that the range of predicted performance is considerably reduced through the updating methodology presented in this paper. A typical result is shown in Figure 2, which quantifies the influence of the number of sensors on the coefficient of variation for the time to corrosion initiation at rebar revel for various hypothesized exposure conditions. The case with '0' sensor indicates the prior corrosion initiation times. It can be seen that the influence of various models that could be assumed for exposure conditions is minimized by the integration of data obtained through SHM into the predictive models. Finally a life-cycle cost analysis for various management strategies (with and without the use of SHM) highlighted the safety and cost benefits that can be obtained through the use of SHM-supported predictive models (Figure 3). It is clear from the figure that the LCC is minimized for the case where decisions are aided with predictive models updated through SHM. It is recognized that the above conclusions are obtained from a limited number of application case studies. Clearly more work is needed in this area including physical tests and field data collection to improve our understanding of the underlying phenomena and to reduce prior uncertainties, especially those related with modelling and measurement (epistemic components).