Reliable Design Loads for Natural Phenomena: Illustration with Wind Speeds

Abstract

A primary purpose of structural codes is to provide designs that produce safe and economical structures for their entire intended lifetime. When considering forces due to natural phenomena, our historical records are often marginal in terms of reliable data. This is especially true for the case of wind loads, where topography effects prevent the use of historical data except in the local region. Therefore, one may need to select wind forces based on a 50- or 75-year design life when the total record length is of that magnitude. The field of extreme value statistics is a powerful methodology that has not had significant use by the engineering community. It is specifically intended to provide the most reliable estimates for return periods of the order of the total record length. In this study, automatically recorded wind data from several sites in the Pacific Northwest region of the United States are used to provide design wind speeds. These speeds are compared with the raw data to indicate the sensitivity of the procedures. Extreme sensitivity of the design wind speeds is found, dependent on which probabilistic method is used to compute them. Use of design data from observed extremes, fitted probabilistic models, and extreme value statistical theory give very different results between sites and even within sites. Great caution is needed for design values longer than the observed data record, and even for return periods on the order of the length of the data record. While the paper uses the recorded wind data as a pathfinder, the approach is discussed in terms of the determination of design loads for other natural phenomena.