ABSTRACT
ABSTRACT: Directional simulation in the load space, due to considerable advantage of working in the space of lower dimension, is of remarkable importance. In this space, however, the variability of structural resistance parameters (which are normally taken to be time-independent random variables) leads the location of the involving limit states to be not constant. These locations therefore have to be assumed to be random variables. In the earlier works (e.g. Melchers 1992, Moarefzadeh, 1996), Gaussian distribution was proposed for this variability and then only two first moments were employed to describe the variability. In the paper presented by the writer earlier (i.e. Moarefzadeh, 2005), by assuming the structural resistance parameters to be independent, and by taking the other two moments(i.e. third and fourth moments) in addition to two first moments and employing “Hermite Moment Model” , it was shown that the assumption of Gaussianity may not be led to accurate results. In this paper, this investigation is extended for the cases in which the structural resistance parameters are not assumed independent any longer. In these cases, it is also shown that departure from Gaussian assumption may result in the considerable different outcomes.
