An approach is presented for evaluating the vulnerability of electric cabinet in nuclear power plants. The method is based on the lognormal approach, including the maximum likelihood estimation and linear regression to establish the fragility curves. These procedures are applied for a cabinet considering various boundary conditions, which are expressed by restrained and anchored models at the base. The cabinet models have been built and verified by using the system identification technique. The results show that the fragility curves obtained for the anchored model are found to be closer to each other, compared to the fragility curves for the restrained model.

Deterministic seismic analysis of cabinet facility does not reflect the real behavior of the system with the uncertainties of material parameters. This paper investigates the effects of uncertain properties in the context of random field theory for the stochastic dynamic response of cabinet. The influences of different decomposition methods (i.e. Cholesky, Eigen, and modified Cholesky) with various correlation lengths of the spatial material distribution are explored. The responses are compared to the case where no variation is applied in the properties of the model.