Method of pushunder analysis (PUA) for approximate seismic analysis of structures using various patterns of stepwise increasing base acceleration and its comparison with pushover analysis (POA)

Abstract

In this paper, a new simple semi-dynamic method is introduced for the approximate evaluation of the seismic behavior of steel structures which is called Pushunder Analysis (PUA). In this method incrementally increasing horizontal unidirectional and reciprocal base acceleration in saw-tooth patterns, which increases from zero to a specified maximum value is applied to the base of a model structure for analysis of its response using the method of Time History Analysis (THA). This method is compared with conventional Pushover Analysis (POA), where incrementally increasing code specified lateral load patterns are applied to the floors of a model structure. The main objective of this paper is to obtain which one of these two methods (POA and PUA) can better represent the seismic behavior of the structure more accurately. For this purpose, the results of the semi-dynamical PUA of a ten story steel structure model are obtained using various patterns of base acceleration. These results are: total displacement, relative drifts, shear story in linear range and P-Δ type curves in nonlinear range for each selected base acceleration patterns which are shown and compared with POA results. To verify the degree of the accuracy and the peculiarities of this method, the linear and nonlinear responses of the model structure under: static UBC-97 standard seismic loading, dynamic El Centro 1979 acceleration records and FEMA-461 cyclic displacement (which was converted to analogous base accelerogram) are compared with the similar responses obtained by the POA and PUA methods. In general, the proposed PUA method needs further evaluation and case studies to ascertain its merits; however, it shows considerable potential and advantages due to its consideration of dynamical and torsional effects and hysteresis properties of structures. The sensitivities of this method to the length of dynamic time steps of calculation and the forms and durations of the saw-tooth oscillations are discussed and it is found that by the proper choice of these parameters, better results can be obtained.