Uncertainty modeling of many correlated and skewed random variables
Abstract
This thesis presents a point estimate methodology directly suited to uncertainty modeling for systems with many correlated and skewed random variables. The methodology proposes two models. The models provide a simple but powerful procedure to approximate the joint probability density function of many correlated and skewed random variables. It is assumed that the only available information is the set of expected values, standard deviations, skewness coefficients, and correlation coefficients of the random variables. Examples and step-by-step procedures are provided to illustrate the models. Comparisons are made with the results of Rosenblueth (1975), Lind (1983), and Harr (1989) point estimate methodologies, the Monte Carlo simulation technique, and also with the exact solution for a special case. Application is also made to the flow code LLUVIA for estimating the groundwater travel time in an unsaturated medium.
Degree
Ph.D.
Advisors
Harr, Purdue University.
Subject Area
Civil engineering|Statistics
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.