Safety analyses of multi-component systems based on multidimensional seismic fragility investigations considering operational and age-related changes. Final report

The seismic safety of a nuclear power plant (NPP) is assessed based on reliable functioning of the entire plant. ln order to control the safety functions, the NPP is divided into several systems based on the functional or structural units. A commonly used method for the seismic safety analysis is the Seismic Probabilistic Safety Assessment (SPSA), see e.g. [RP Kennedy et al., Probabilistic seismic safety study of an existing nuclear power plant. Nuclear Engineering and Design, 59(2):315-338, 1980; YJ Park et al. Survey of seismic fragilities used in pra studies of nuclear power plants. Reliability Engineering & System Safety, 62(3):185-195, 1998]. In SPSA, classical event trees and fault trees are used to perform the system analysis or the accident sequence analysis for the plant. The structural and functional coupling among the components within a system or subsystem are presented in the fault tree using the logic gates 'AND' or 'OR', which are not capable of effectively representing the interaction among the components in a system. An alternate method for system analysis, capable of generating system or subsystem fragility curves, including the interaction among the components called the multidimensional fragility analysis, is proposed in this thesis. The method is based on the multidimensional performance limit state approach proposed by Cimellaro et al. [GP Cimellaro et al. Multidimensional performance limit state for hazard fragility functions. Journal of Engineering Mechanics, 137(1):47-60, 2010]. As opposed to the fault tree analysis, where the component fragilities are combined to derive the system fragility, in the multidimensional fragility analysis method, the system fragilities curves are generated by combining the probabilistic responses of the components and their limit states. The definition of the system performance limit state allows to include the interaction of the responses of the components, which is not represented in the classical fault tree analysis. Random sampling method such as Monte Carlo simulation (MCS) is the most commonly used method for probabilistic analysis of the components. MCS requires samples in the range of several thousands to give accurate responses, which requires high computational time. Therefore, to improve computational efficiency, the proposed multidimensional fragility method uses response surface method (RSM) which uses stratified sampling techniques, thus reducing the number of samples for the computation. The component and structural aging are generally neglected in the fragility calculation of the components. However, the structural aging might cause brittle failure of the structure or components, compared to that of the predicted failure of the structure or the system. Therefore, the proposed method is extended to account for degradation effects due to aging. The structural aging attributed to chloride attack from the extemal environment is considered for the analysis. For reinforced concrete structures, the concrete cover depth is one of the main deciding factor of the structural attribute for the reinforcement corrosion initiation due to chloride attack. The aging in components, with wear and tear are accounted by reducing the performance limit state of the components. The proposed method can also be incorporated in the fault tree analysis software programs. The algorithms developed for the multidimensional fragility evaluation, are using the program MATLAB@. This can be easily used by other fault tree analysis software programs such as CAFTA@, SAPHIRE@ etc. by using interfaces, or by subroutines. To apply and validate the proposed method, an example system is presented in this Thesis. For this, a reinforced concrete test structure, tested in the international benchmark study, SMART 2013, is selected. To this model, nonstructural components such as pipes and emergency generator are added, forming an example system. The development of the system fragility curves requires simulation models for the analysis of the system. Two structural models, first a linear model using multi degree of freedom (MDOF) and second a 3-dimensional model with simplified walls and nonlinear properties, are investigated for the accuracy and computational efficiency. For the validation of the multidimensional fragility analysis method, the system fragility curves generated are compared with that generated using fault tree with 'OR' gate. The results of the fault tree corresponds to the multidimensional fragility with the independent response, i.e. without considering the interaction of the components. A comparison of the fragility curves for the chosen example system using different interaction factors, shows increase in the median capacity with increase in the interaction factors. This implies that the interaction factors have a significant influence on the system fragility curves. In the given example, a significant reduction in the median capacity is observed for the multidimensional fragility curves as the aging year progresses, from the calculated corrosion initiation time. Therefore, the interaction and aging are represented well using the proposed method.