STATISTICAL ANALYSIS FOR BASIC WIND SPEEDS

Abstract

For structures, wind loading is a significant factor with respect to safety. To ensure the safety of a structure subject to a wind action, wind resistant design should be controlled by many provisions in a structural code and the first step is to give. a basic design wind speed at the site. Wind speed varies in a nearly random fashion'and can be approximately described as a stationary random process to the exclusion of the seasonal change. This makes it possible to predict a wind speed on statistical basis. The modeling ,of probabilistic behaviour of wind speeds is of interest on two counts. Firstly, the probabilistic models may be used to predict the maximum wind speed that'a structure may be subjected to during its life time, and secondly, the probabilistic models can be used easily for risk analysis. In the present study, an effort is being made_to___ examine the statistical approach to model the wind behaviour, its use to specify probabilistic wind speeds for design purpose, and the associated risks of failure, depending upon the projected life of structure. Statistical analysis of the wind data collected for 28 meteorological centres in India has been carried out. The analysis has been carried out with the help of computer program using extreme value type-I (Gumbel) and type-II (Freshet) distributions. Wind speeds are predicted for various return periods. Results have been presented in tabular form as well as graphically using Calcomp-1037 plotter. The risk analysis for the stations has also been carried out. Results have been presented in tabular form as well as graphically. The graphs consist of associated risk curves for different projected life of structure. These have been plotted on Calcomp-1037 plotter. Simulation of wind speeds has also been carried out. Extreme value analysis has been carried out for simulated wind speeds and results are compared to those obtained from extreme value analysis of observed wind speeds. Type-I1 (Frechet) distribution has been found more adequate to predict wind speeds for different return periods in Indian climate which is of tropical type. A length_ of record of 20 years or more may be considered adequate for Cir7J reliable predictions.