Seite - 1226 - in Book of Full Papers - Symposium Hydro Engineering

The model is first loaded statically under the self-weight of the dam and the hydrostatic pressure of the full reservoir. Then, it is dynamically analyzed under the selected earthquake ground motions. The Rayleigh damping is employed for structural elements with the coefficients such that produce 5% critical damping in the first and third vibration modes of the intact model. Based on the mentioned properties, the first five natural periods of the dam-reservoir system are: T1=0.38s, T2=0.33s, T3=0.30s, T4=0.26s, T5=0.22s. The considered EDPs are the crest maximum relative displacement Dc, and the peak hydrodynamic pressure P imposed to the dam heel. Because the dam body has un-symmetric geometry, the crest relative displacements can be separately monitored into the upstream DcU, or the downstream DcD directions. Two local damage indices are defined which are directly related to the cracking response of the dam body. They are computed in the cracking-susceptible sub- regions of the dam body shown in Fig. 2(a) as base and neck sub-regions. These local damage indices, Λ, for both sub-regions are defined as """" | | | norbi A Ad ie e ie eet i      (2) where dt|e is the tensile damage of element e with area of Ae. The computation is done on the entire sub-region i. Subscripts b and n show the base and neck sub- regions, respectively. These damage indices are a measure of the amount of damage that the dam may locally experience. Another EDP can be defined as the energy dissipated through cracking process, EC, which can be considered as a global measure of damage imposed to the dam body. 5. RESULTS The dam-reservoir coupled system is analyzed under the selected pulse-like and non-pulse-like records, and the EDP values are computed. The difference between the pulse-like and non-pulse-like records can be investigated by assessing the cloud of data of, say, ln[Dc]-ln[PGA] shown in Fig. 4. Considering all of data as response under the near-field earthquake records (Fig. 4(a)) would result in regression line with slope of 1.022 and goodness of fitting R2=0.69. But, separating the data as response under the pulse-like and the non-pulse-like records (Fig. 4(b)) causes an increase in slope and R2 value for the non-pulse- records that shows better correlation for this kind of near-field records. An opposite trend is observed for the pulse-like records. The same trends are observed for other IMs. It is concluded that the pulse nature would lead to more scattered response results. Therefore, it is important to separately investigate the response parameters considering the pulse nature of the earthquake records. 1226