Autor/autori: Dinu BRATOSIN

Abstract: In any dynamic system, the coincidence between natural period and excitation period (the resonance) lead to extremely large dynamic amplification. For example a linear system with damping of order 25% the dynamic amplification factor has values between d / 2510 inamic static x x [4, 6]. The amplifications of this order can also occur in the building structures loaded in the resonant regime. And it is unlikely that a structure can support such a demand without major damages. The essence of the strategy of avoiding site – structure resonance consists in the correct evaluation of the both structural Ts and site natural periods Tg followed by the imposed conditionTs Tg . However, the site natural period Tg has no a unique value. The site materials have a mechanical behavior strongly dependent on strain, stress or loading level (manifested by dynamic stiffness degradation and increasing damping) [2, 4, 5, 6, 8, 11]. As a result, Tg becomes dependent on earthquakes amplitude, and this dependence can be observed in the seismic records [12, 13]. In the current practice for site natural period determination is usually used the "quarter length formula" Tg 4H / vs where H is the site depth and vs is the shear wave velocity [8, 13]. This method assumes the site as linear elastic space in contradiction with mechanical reality and gives a unique natural period value in contradiction with earthquake recordings. This paper proposes an evaluation method of the natural period nonlinear dependence assuming site materials as nonlinear viscoelastic materials modeled with a nonlinear Kelvin-Voigt model (which includes dynamic stiffness degradation and increasing damping) [3, 4, 6]. By using resonant column tests we can quantify the nonlinear dependence of the site natural period in the normalized formTn Tn •where Tn Tg /T0 and •PGA sau MGR . Then, from "in situ" information we can obtain the normalization value T0 and finally, the nonlinear site natural function result in the form: Tg •T0 Tn •. Validation of this method is provided by comparison between the function calc Tg evaluated by nonlinear calculus for a site with sufficient seismic records and the function rec Tg obtained directly from these records

Keywords: Soil dynamic degradation, Nonlinear site natural period, Nonlinear oscillating soilstructure system, Soil-structure resonance