Abstract
Civil infrastructure systems provide physical supports to a community’s functionalities and are expected to achieve acceptable safety levels subjected to extreme load effects. However, these systems may deteriorate with time as a result of aggressive environmental or operating conditions in service, implying that the system reliability may decline beyond the baseline as assumed for design. Moreover, the increasing trend of the external loads may also contribute to the reduction of the system reliability. In this paper, a semi-analytical method is proposed for assessing the reliability of aging systems subjected to non-stationary loads. The series system is considered, where the system failure is defined as the failure of any single component (structure) among the system. The application of the proposed method is illustrated using a representative series bridge network with several individual bridges. The role of parameters such as the variations in the load intensity, resistance correlation and number of components under attack in the system reliability are investigated.
Acknowledgements
The support from the International Program Development Fund from the University of Sydney is acknowledged. The author would like to acknowledge the thoughtful suggestions of two anonymous reviewers, which substantially improved the present paper.
Notes
No potential conflict of interest was reported by the authors.
1 The hazard function h(t) is defined as the probability that the structure fails within the subsequent unit time given that the structure has survived up to time t.
2 Note that other integration techniques such as the Gauss–Legendre quadrature with 5 points (Golub & Welsch Citation1969) and the point estimate method (Rosenblueth Citation1975) may lead to an improved efficiency in calculating . However, this study uses the rectangle method as a practical method for time-dependent reliability analysis of typical civil structures.
3 For the rest of this paper, all the simulation-based system failure probabilities are obtained using 100,000 simulations.
4 It is noticed, however, that when more information on the correlation coefficient between different component resistances becomes available, e.g. is known be no more than 0.7, the lower bound of with Equation (Equation21(21) ) can be re-examined and a stricter estimate than LB1 can be obtained. .
5 The variance associated with the discrete can be reflected by adding an additional random item at the right hand side of Equation (Equation24(24) ).