Comparative Study of Random Matrices Capability in Uncertainty Detection of Pier’s Dynamics

Because of random nature of many dependent variables in coastal engineering, treatment of effective parameters is generally associated with uncertainty. Numerical models are often used for dynamic analysis of complex structures, including mechanical systems. Furthermore, deterministic models are not sufficient for exact anticipation of structure’s dynamic response, but probabilistic models are far more efficient in modelling uncertainties. To determine the uncertainty effect on dynamic characteristics, the laboratory samples of a pier structure were created. Stochastic finite element method was used for numerical study of pier’s dynamic behavior. For this purpose, the concept of variable distributions of Gaussian and Wishart random matrices for modeling dynamic characteristics of the system was used. At last, by comparing the output frequency response function obtained from empirical experiments and numerical simulation of stochastic finite element method, it is shown that probability density function of Wishart matrix evaluates and tracks uncertainty parameters in system’s dynamic response far better than Gaussian matrix.