Numerical simulation of wind effects

Abstract

Modern structural systems are becoming increasingly complex and numerical simulation of the potential loads with which they will be affected is critical for analysis, design, and optimization of safe and reliable structures. Monte Carlo analysis approaches are often used, which involve the input of loads into a structural model and the output of responses. Besides being necessary for numerical analysis, digitally simulated data is also necessary to drive computer controlled test facilities. Both approaches necessitate an ensemble of input signals that accurately represents what the structure may expect to experience during its lifetime. Therefore, simulation of time histories of wind velocity, pressure, and force fluctuations are necessary, in addition to simulation of structural response, which allows assessment of attendant functionality and safety under service and design loads, respectively. Random processes simulated for analysis purposes are often assumed to be Gaussian and stationary for simplicity. Many wind events, however, are characterized by non-stationarity and non-Gaussianity. Therefore, simulation methodologies are necessary for univariate and multivariate processes, unidimensional and multi-dimensional fields, Gaussian and non-Gaussian data, stationary and non-stationary processes, and conditional and unconditional cases. In order to accomplish this task, methods based on the time, frequency, and timefrequency domains are employed. This paper summarizes a historical perspective, recent developments, and future challenges for simulation. Also included in the discussion are computational tools employed for data and response analysis. Examples are presented to illustrate some of the topics discussed.