Artificial Wind Generation and Structural Response

Abstract

With modern computational means, the use of deterministic methods to predict the aerodynamic behavior of structures is becoming an attractive alternative to frequency domain methods. Wind speed samples can be generated by various techniques and fed into the system's equations of motion. The use of deterministic approaches, however, requires the artificial wind to be realistic and the simulation technique efficient. A study is undertaken to investigate the merits and drawbacks of commonly used simulation techniques. The latter can be divided into two classes, i.e., wave superposition methods and methods based on linear filtering. Techniques for simulating single and multiple time series are reviewed. Artificial wind speed is generated, and the respective responses calculated. The effects on the response of the simulated samples' different statistical properties are investigated. Results show that dynamic response is affected by the intrinsic characteristics of simulated fluctuations and their degree of correlation. Linear filters are computationally more advantageous and, in general, lead to larger response parameters compared with wave superposition methods. The use of autoregressive‐type simulation methods is recommended.