Non-canonical spectral representation of seismic activity is employed to assess the reliability of nonlinearly modeled buildings. Seismic impact is modeled using a random process, represented by simple functions with random parameters. We consider random processes with correlation functions expressed as a sum of cosine-exponential terms. Reliability, defined as the probability of failure-free operation, is determined using statistical testing methods. The reliability calculation algorithm is implemented in MATLAB. As an illustrative example, we calculate the reliability of a section of a one-story industrial building frame modeled by a nonlinear system. Failure is defined as exceeding experimentally determined permissible displacement limits. Our calculations involve up to 2000 realizations of the random process. We analyze histograms, empirical distribution functions, and reliability values of maximum fragment movements. We find that using 100 realizations of the random process yields satisfactory accuracy in determining reliability. This reliability calculation method is recommended for rapid reliability estimates across various structure types, including those employing seismic isolation systems. We also observe a correlation between displacement magnitudes calculated under accelerograms and a random process represented in a non-canonical form. Thus, we recommend this method for reliability assessments in multi-story buildings. Doi: 10.28991/CEJ-2024-010-08-019 Full Text: PDF
Read full abstract