| Various analytical methods have been developed for the analysis of structures with stochastic material properties. However, they are either computationally-time-inefficient or applicable only to certain types of structures. As a versatile numerical approach, the finite element method (FEM) has been extended by many researchers to study structural problems with random material properties and is called the stochastic finite element method (SFEM).; A three story three bay frame with stochastic material properties was analyzed by Deodatis (1990) using SFEM. An assumption that the material property of individual members of the frame was uncorrelated, was made in his paper. In this thesis, the technique used by Deodatis is extended to the analysis of a stochastic beam problem by incorporating material property correlation between elements. To validate the modified method, the present results are compared with those of Deodatis and Shinozuka (1989), who used another approach to solve the problem. The beam problem study showed that the modified SFEM was accurate and computationally-efficient, SFEM is further extended to the analysis for conical/cylindrical shells having random properties.; The developed SFEM in this thesis is a combined application of standard FEM, Taylor series expansion and probabilistic theory to calculate the mean values of structural responses and to evaluate the response variations. Several numerical examples are considered. The results show the mechanism of structural responses to the stochastic material property under deterministic loading conditions. |
