Stochastic analysis of structural/acoustic systems

Finite element (FEM) and boundary element (BEM) methods are used extensively for simulating the behavior of structural/acoustic systems. In noise and vibration computations, the excitation and structural design parameters of a structural/acoustic system are predominantly considered to be deterministic. This assumption limits the accuracy and applicability of current structural/acoustic simulation techniques. Current simulation techniques cannot accurately account for spatially correlated random excitation or random variation in the structural design parameters. In this dissertation FEM and BEM are combined with the principles of stochastic analysis to account for these two types of randomness in structural/acoustic computations. A new method is presented for calculating the radiated noise of a structural/acoustic system subjected to a spatially correlated random excitation. The method is validated through comparison of numerical results with test data available in literature. Based on the new development, additional algorithms are derived for determining the radiated sound power and associated radiation efficiency of the randomly excited system. Sensitivity algorithms for the radiated noise and radiated sound power are also derived. Further, a new formulation is presented for calculating the probabilistic acoustic response in the presence of variability in the structural parameters of the system. The algorithm uses a readily available advanced mean value structural reliability method along with FEM and BEM to calculate the probabilistic acoustic response. The formulation is validated through comparison of numerical results and results from Monte Carlo simulation. The new developments presented in this dissertation are accurate and used to gain greater physical understanding of the structural/acoustic systems analyzed.