Seite - 1227 - in Book of Full Papers - Symposium Hydro Engineering

(a) (b) Fig. 4. Cloud of data of Dc-PGA in logarithmic scale: (a) regression line for all of data, (b) separate regression lines for the pulse-like and the non-pulse-like records. An important parameter of pulse-like motions that affects the structural response is the ratio of the pulse period to the fundamental period of the structure Tp/T1. It is particularly important for the first-mode dominated structures such as gravity dams. The response data of the dam-reservoir coupled model under the pulse-like motions are plotted versus the ratio Tp/T1 in Fig. 5. Because the fundamental period of the model is as low as 0.38s, the ratio of Tp/T1 is more than 1 for all selected pulse-like motions; this ratio becomes even as high as 34. The local variation of the data is represented using a local average employing the Nadaraya-Watson kernel-weighted average, with a Gaussian weight function. Also shown are the mean and median responses under the non-pulse-like motions; the mean response is totally higher than the median that shows a bias into the higher values. The response of the pulse-like motions may be higher or lower than the average response of the non-pulse-like records. As the ratio of Tp/T1 increases, approximately, lower response values are observed specifically for the damage dissipated energy which is a global measure of the imposed cracking damage. This trend is somewhat observed for the base local damage index, but it is not so clear for the neck damage index. No all of pulse-like motions would result in Λn, i.e. neck cracking; this parameter is observed for motions with Tp/T1 < 22 for only 49 out of 75 records. For the larger period ratios, the pulse period is so much high that it cannot excite the lower main vibration modes of the dam which cause large movements and neck cracking. y = 1.022x + 3.1674 R 2 = 0.6884 -1 0 1 2 3 4 -3 -2 -1 0 1 ln[PGA] pulse-like non-pulse-like y = 0.893x + 2.8889 R 2 = 0.6359 y = 1.3353x + 3.8713 R 2 = 0.7982 -1 0 1 2 3 4 -3 -2 -1 0 1 ln[PGA] pulse-like non-pulse-like 1227