2020 年 86 巻 882 号 p. 19-00242
This study describes a simulation-based estimation procedure of the structural failure probability using an “all quadrants sampling method” newly proposed, in which a sample called as a real sample is generated in an arbitrary quadrant of the basic variable space and then the coordinates of the real sample point is transformed into those of pseudo sample points over all of other quadrants by applying the previously proposed “inter-quadrant relational expression.” That is, once a real sample is generated in an arbitrary quadrant, and then pseudo samples are determined in all other quadrants concurrently. All of real and pseudo samples thus determined are used for the simulation to estimate structural failure probabilities. Numerical examples of various simulation methods combined with the proposed “all quadrants sampling method,” that is, “all quadrant sampling Monte Carlo simulation,” “all quadrant sampling directional simulation,” etc. are presented showing that the proposed method is effective for shortening simulation time.
