This paper outlines an estimation method of structural failure probabilitie
s based on a simulation within a limited sampling region. Samples are restr
icted outside the beta-sphere and within an effective distance range. The b
asic random variables are assumed to be independent normal variables and th
e reliability index beta to be known. Each sample for the limited sampling
region is produced by combining a random radius generated from a truncated
chi-square distribution with a random directional vector generated from sta
ndard normal variates. Numerical examples are presented to show the validit
y of the proposed method. (C) 1999 Elsevier Science B.V. All rights reserve
d.