Styliani TSANTAKI
A contribution to the assessment of the seismic collapse capacity of basic structures vulnerable to the destabilizing effect of gravity loads
This dissertation aims to provide a better understanding of seismic sidesway collapse of highly inelastic basic structure vulnerable to the destabilizing effect of gravity loads (i.e., the P-delta effect), and to enhance the prediction of their seismic collapse capacity with simple measures.
A major part of this dissertation is devoted to the quantification of the collapse capacity uncertainty of P-delta vulnerable single-degree-of-freedom (SDOF) systems. In particular, the reduction of record-to-record variability (RTR) of the collapse capacity is addressed through an appropriate choice of the intensity measure (IM) of the earthquake excitation. It is proposed to utilize an IM based on the averaged spectral pseudo-acceleration in a certain period interval. Furthermore, a relative IM is introduced that takes into account the structural parameters of the SDOF system affected by P-delta. That is, the 5% spectral pseudo-acceleration is read at the at the period of vibration of the system in the presence of gravity loads, and further normalized by base shear coefficient of the P-delta affected system. Through a parametric study it is shown that both IMs reduce significantly the RTR dispersion of the seismic collapse capacity compared to the widely used benchmark IM, i.e., the spectral pseudo-acceleration at the system period without considering P-delta.
The effect of uncertainty of the main characteristic parameter responsible for seismic induced global collapse, i.e. a post-yield negative stiffness ratio, on the median and the dispersion of the collapse capacity is quantified. The outcomes of the first-order-second-moment method are verified by results of the latin hypercube sampling (LHS) technique. Subsequently, the total uncertainty composed of the parameter uncertainty and the RTR variability is assessed with two approaches, using the square-root-sum-of-squares superposition rule, or alternatively, two-dimensional LHS. The latter approach allows the simultaneous consideration of both sources of uncertainty with the same computational demand compared to simulations considering the RTR variability only. It is shown that the parameter dispersion of the collapse capacity can be of the same order as the RTR variability. The inclusion of epistemic uncertainties flattens the fragility curves, and larger probabilities of collapse are predicted for small intensity values.
The impact of the characteristic structural parameters and various IMs on the collapse capacity and its RTR variability of P-delta sensitive multi-degree-of-freedom systems is studied. It is confirmed that the negative global post-yield stiffness ratio is the dominant parameter for collapse of P-delta vulnerable and non-deteriorating systems. The choice of an “average” IM that considers also higher mode effects results in the smallest dispersion for the considered generic moment resisting frames. Thus, it can be concluded that an equivalent SDOF system cannot reflect directly the collapse capacity dispersion of a multi-story building.
Design collapse capacity spectra and fragility curves with reduced RTR dispersion based on the proposed “average” IM, refined collapse capacity spectra and fragility curves based on a conventional IM, and analytical parameter dispersion collapse capacity spectra are provided through multiple regression analysis, aiming to enhance the simplified assessment of P-delta vulnerable systems within the general framework of the collapse capacity spectrum methodology.