- Author
- Simiu, E. | Heckert, N. A. | Whalen, T. M.
- Title
- Mean Recurrence Intervals of Ultimate Wind Loads.
- Coporate
- National Institute of Standards and Technology, Gaithersburg, MD Purdue Univ., West Lafayette, IN
- Report
- OMAE98-1218,
- Book or Conf
- Offshore Mechanics and Arctic Enginering (OMAE), 17th International Conference. Proceedings. American Society of Mechanical Engineers. OMAE98-1218. July 1998, 1-6 p., 1998
- Keywords
- wind velocity | extreme values | wind engineering | statistics | structural engineering | ultimate loads
- Abstract
- The attention of wind engineers and extreme value climatologists has been focused in the last few decades on estimates of basic wind speeds, that is, wind speeds with 50- or 100-yr mean recurrence intervals (MRIs). Recently, however, efforts have been reported on MRI estimation for speeds including the design strength or causing member failure. These efforts have benefited from progress in extreme value theory, notably the development of 'peaks over threshold' methods used in conjunction with the Generalized Pareto Distribution. According to results based on such methods extreme wind speeds are best fitted by extreme value distributions of the reverse Weibull type which, unlike other distributions used in the past, have finite upper tails. We note results according to which the ASCE 7-93 Standard and its 1980s predecessors specify wind load factors that place structures in hurricane-prone regions at substantially higher risk than structures in non-hurricane regions, and comment on changes required in Standard provisions in this respect. We also review results on the effects of wind directionality. On average, estimates of wind loads with relatively short MRIs, obtained by disregarding wind directionality effects, are conservative in relation to estimates that take directionality into account. However, the conservatism decreases asthe MRIs increase and may become marginal for speeds associated with wind-induced failures. We report and comment on the result that estimated tails of distributions that best fit samples of hurricane wind speeds are somewhat shorter than estimated tails for samples of squares of hurricane wind speeds. An Appendix provides instructions for accessing electronic files containing wind speed data and computer program listings.