TY - JOUR AU - Ghoulbzouri, A. El AU - Khamlichi, A. AU - Almansa, F. Lopez AU - Vera, M.A. Parron AU - Cintas, M.D. Rubio PY - 2011 TI - Evaluating some Reliability Analysis Methodologies in Seismic Design JF - American Journal of Engineering and Applied Sciences VL - 4 IS - 3 DO - 10.3844/ajeassp.2011.332.340 UR - https://thescipub.com/abstract/ajeassp.2011.332.340 AB - Problem statement: Accounting for uncertainties that were present in geometric and material data of reinforced concrete buildings was performed in this study within the context of performance based seismic engineering design. Approach: Reliability of the expected performance state is assessed by using various methodologies based on finite element nonlinear static pushover analysis and specialized reliability software package. Reliability approaches that were considered included full coupling with an external finite element code and surface response based methods in conjunction with either first order reliability method or importance sampling method. Various types of probability distribution functions that model parameters uncertainties were introduced. Results: The probability of failure according to the used reliability analysis method and to the selected distribution of probabilities was obtained. Convergence analysis of the importance sampling method was performed. The required duration of analysis as function of the used reliability method was evaluated. Conclusion/Recommendations: It was found that reliability results are sensitive to the used reliability analysis method and to the selected distribution of probabilities. Durations of analysis for coupling methods were found to be higher than those associated to surface response based methods; one should however include time needed to derive these lasts. For the reinforced concrete building considered in this study, it was found that significant variations exist between all the considered reliability methodologies. The full coupled importance sampling method is recommended, but the first order reliability method applied on a surface response model can be used with good accuracy. Finally, the distributions of probabilities should be carefully identified since giving the mean and the standard deviation were found to be insufficient.