Abstract An innovative prediction method for the combination of failure modes of an arch bridge is proposed for its probabilistic risk assessment and is compared with the conventional system reliability analysis method. The suggested method reveals various combinations of failure modes in significantly reduced time and efforts in comparison to the previous permutation method or the conventional system reliability analysis method. Additionally, the suggested method can be used for the verification of the system reliability with more specific predictions of the failure modes.

Component reliabilities of girders have been evaluated by using response surfaces of the design variables at the selected critical sections, i.e. the maximum shear and negative moment locations. Response Surface Method (RSM) is successfully applied for the reliability analyses of complex structure with relatively small probability of failure. Previous methods such as Monte Carlo Simulations or First Order Second Moment method cannot easily calculate the derivative terms of implicit limit state functions. Hence, the target bridge system is modeled as a series-connection system for the analysis of its system reliability. The upper and lower bound of the probabilities of failure for the structural system have been evaluated and are compared with the suggested prediction method for all possible combination of failure modes.69804

§c  2007 Elsevier Ltd. All rights reserved.

Keywords: Combination of failure modes; Arch bridge; Risk assessment; System reliability; Response surface method

1. Introduction

The demands for systematic and efficient risk-safety assessment/management in each step (planning, design, construction and maintenance) of large-scale construction projects involving bridges and subways are increasing to prevent possible disasters subsequent to the construction project. Recently, effective risk assessment techniques for special facilities such as nuclear power plants, gas facilities, and pipe lines, etc. has drawn wide attention and interest. Several methods related to the investigation of potential risky accidents and effective risk assessment such as Hazard Operability Method [1] have been applied. A Fault Tree Analysis (FTA) technique is also used to identify all possible events that may fail in construction [2]. Event tree analysis (ETA) [3] is a method of illustrating the sequence of outcomes which may

arise after the occurrence of a selected initial failure event. The aim of event tree analysis is to identity the undesirable incidents following a triggering event. ETA provides a precise way of recording accident sequences and defining the relationships between the initiating event and the subsequent events in combination that result in an accident. Ranking the accidents considering the order of events needs a permutation calculation, which produces a large number of cases for predicting the failure scenarios.

Thus, the authors suggest an innovative quantitative risk assessment method by considering the combination of failure modes, composed of critical component’s failures. The combination of component failure is used for the system analysis of failure scenarios in this study instead of the factorial or permutation method as used in ETA. As an illustration of this point of using the combination method, the same  probabilities

can be obtained only in 26 − 1 (=63) cases, which results in over 90% time–cost saving compared to 6! (=720) cases when 6 failure modes of the components are considered.   The

supporting and dissenting arguments for the suggestion are discussed with a numerical example.   The suggested   method

0143-974X/$ - see front matter §c 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.jcsr.2007.05.004

1562 A.S. Nowak, T. Cho / Journal of Constructional Steel Research 63 (2007) 1561–1569