An acceptable method to estimate the annual probability that specified wind speeds

DOE-STD-1023-95

(3) Determine control points and connect control points by straight line segments in log-

log space to form the spectrum plot.

Control

Frequency

Spectral Acceleration

Point

(Hz)

(g)

0.1

0.395 dmax/g

vmax/(2 dmax)

(vmax)2/(g dmax)

(g amax)/(2 vmax)

amax

8.0

amax

33.0

PGA

100.0

PGA

(4) The vertical design response spectrum is typically taken as 2/3 of the horizontal

spectrum provided that the site is not one of the "near-field" sites, i.e. closer than

approximately 15 km (9.3 mi) to a seismic source. For near-field sites, the guidance in

ASCE 4 (ASCE, 1998b) should be followed. Local site conditions should be

considered when developing appropriate vertical design response spectra.

A3.2.2-e.

An acceptable method to estimate the annual probability that specified wind speeds at the site

will be exceeded is included in Coats and Murray, 1985, and described by the following:

Step 1:

Select a data set of annual extreme wind speeds from a weather station near the

site of interest.

Step 2:

Correct the annual extreme wind speeds to an anemometer height of 33 ft (10

meters) above ground in flat, open terrain using appropriate methodologies. For

example, a power law (Simiu and Scanlan, 1986) could be used to make an

adjustment, if needed. No recorded wind speeds from anemometers located on

building roofs near the edges, sheltered by parapets or neighboring buildings, or

too close to the roof surface (less than 5 feet (1.5 meters)) shall be used.

Step 3:

Estimate the annual probability of exceedance of selected wind speeds with

associated uncertainty.

Data sets of historical extreme winds shall be obtained from weather stations close enough to

sites to represent the site conditions as described in DOE-STD-1022-94 If more than one

station is available, they may be combined, provided they represent the same conditions as

those at the site.

Several statistical models may be used to estimate frequency of winds. An estimate of the

models fitting the data shall be performed. If only one statistical model is to be used, the

Fisher-Tippet Type I extreme value distribution (also named Gumbel distribution) (Coats and

Murray, 1985) shall be used, unless justified otherwise. Additional guidance may be found in

Ramsdell, Elliott, et al. (1986).

A-8