Multiple Points-In-Time Estimation of Peak Wind Effects on Structures

Abstract

One of the problems encountered in the estimation of wind effects on high-rise structures is the development of combinations of wind-induced translational responses in possible conjunction with rotational responses and/or of forces and moments that contribute to the wind-induced demand at various cross sections of individual structural members. In current wind engineering practice such combinations are developed in large part intuitively because phase information on the effects being combined is not readily available from frequency domain analyses. In contrast, full time series analyses can produce estimates of combined wind effects because they preserve phase information; however, such analyses can be overly time-consuming. In current wind engineering practice it is common to use the empirical point-in-time (PIT) procedure for the estimation of peaks of combined stationary stochastic processes. The procedure is applied to pairs of such processes, and consists of adding an estimate of the peak value of one of the processes to the estimated value of the second process at the time of the occurrence of that peak. Even if the full time histories of the two stochastic processes are used, errors inherent in PIT can be in some cases as high as 20% on the unconservative side. The purpose of this paper is to present the empirical multiple points-in-time (MPIT) procedure, which improves significantly upon the PIT approach. The MPIT procedure is illustrated by an application to a 60-story reinforced concrete structure. Results show that the MPIT approach produces remarkably accurate estimates of the peak combined wind effects by using a limited number of peaks from the time histories of the individual wind effects being combined. Those estimates are obtained far more economically in terms of computational time than conventional time domain estimates that use full time histories.