جلال اکبری

Prediction of collapse for buildings is a tedious task. The most common method for prediction of structural behavior is nonlinear dynamic analysis of structures. For accomplishing such analyses, selection and scaling of records are major issue. In order to preventing such difficulties, in this research an efficient method is used to predict collapse of frame buildings in earthquakes. The method involves steel and reinforced concrete moment-frame buildings and three types of ground motions (near-source ramp-pulse-like motions, long-period motions, and short-period motions). To predict whether a building will collapse for a given ground motion, the ground acceleration time history is first filtered using a Butterworth low-pass filter with an order and cutoff frequency; the order of filter depends on the type of ground motion, and the cutoff frequency depends on the natural frequency and ductility of system. Then, the peak value of the filtered acceleration record is compared with the maximum base shear obtained from the pushover analysis. If the peak of the filtered acceleration (PFA) exceeds the maximum pushover strength, then the building is expected to collapse. This method reduces the computational complexity and achieves good accuracy.