سال انتشار: ۱۳۹۰
محل انتشار: ششمین کنفرانس بین المللی زلزله شناسی و مهندسی زلزله
تعداد صفحات: ۸
a Soroushian – Structural Engineering Research Center, International Institute of Earthquake Engineering and Seismology (IIEES), Iran
Time integration is a versatile tool for structures seismic analysis, suffering also from considerable computational cost. A source of computational cost arises when the steps by which the strong motions are digitized are smaller than needed for analyses accuracy and stability. In these cases, additional computational costs are generally being spent because of the smallness of the strong motions steps. Recently a method for replacing the original excitations with excitations digitized at larger steps, such that time integration by either of the excitations results in convergence with order two, is proposed. Reduction of computational cost and tolerable loss of accuracy were the observations in the numerical examples. Nevertheless, the question that what would happen if the excitations at stations between the essential integration stations were simply omitted is still open. Considering the special case of seismic excitations, if, for excitations digitized at steps smaller than the steps essential for accuracy, we can omit the excitations at stations between the integration stations with trivial loss of accuracy, we would not need the new method. This is the main issue under study in this paper, carried out, as a comparison between simply omitting the inter-integration excitation stations and the recently proposed method. The numerical analyses consider two different integration methods, two problems, two different excitations, and different number of omission of excitations. To state the conclusion, in brief, even for seismic excitations, simply omitting the additional excitations may significantly affect the responses; the new method can cause more accurate responses.