Seite - 1107 - in Book of Full Papers - Symposium Hydro Engineering

4.3. RESPONSE SPECTRA FROM STATISTICAL ANALYSIS In this method, originally proposed by Kimball [12], a suite of strong motion records is statistically treated. These records should originate from earthquakes with similar distances to the rupture source and the magnitude should be of the same order as the target magnitude. Some corrections for differences in site conditions may also be needed. The steps are as follows: • Selection of suitable strong motion records having magnitudes and distances to the source similar to the target parameters of the earthquake. • Adjustment of the records for differences in magnitude, distance, faulting mechanism, and other parameters between site-specific conditions and the conditions existing at the site of the record • Performing a statistical analysis of the adjusted response spectra of the collected records to obtain the target values of the site-specific spectrum. The median (50th percentile) and the median plus one standard deviation (84 percentile) are then selected as DBL and MDL levels respectively. For Gambiri diversion dam sites, site-specific ground motions for distance ranges (about 30 km) similar to the target magnitude were calculated and statistically analyzed. “Fig. 3 and 4” show the horizontal and vertical design response spectra recommended for OBE, DBE, MDE and MCE for this project. “Fig. 5” shows a comparison between Horizontal components of response spectra estimated by statistical and probabilistic methods with together. The amplitudes of recommended OBE, DBE and MDE response spectra estimated by statistical method are good coordinated with 150, 500 and 1000 year response spectra estimated by probabilistic method respectively. 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0 0.5 1 1.5 2 2.5 3 3.5 4 period (Sec) MCE MDE DBE OBE Fig. 3 Response spectra based on statistical processing for horizontal component 1107