| Περίληψη: | The Boundary Element Method (BEM) is ideal for solving engineering problems dealing with infinite regions. Such a problem is the estimation of low-frequency noise and soil waves generated by a wind turbine (WT) because of its structural dynamic behavior. However, due to the size of that problem, the BEM becomes insufficient requiring very time consuming computations and almost forbidden memory capacities. The goal of the present dissertation is the development of an advanced and fast BEM able to solve accurately and efficiently the aforementioned problem. To this end, the development of a black box fast and memory efficient boundary element solver for general multi-domain static as well as harmonic problems based on the hierarchical matrix methodology combined with the algebraic low rank approximation technique, known as Adaptive Cross Approximation (ACA), is proposed. For the efficient algebraic compression of the utilized hierarchical matrices, different versions of the ACA algorithm have been implemented and tested, while for the solution of the assembled final linear systems of equations a preconditioned GMRES iterative solver has been employed. Next, the above-mentioned ACA/BEM is employed for the solution of a large-scale acoustic, elastic and acoustic-elastic interaction problem dealing with the noise emission and micro-seismicity related to the operation of a Wind Turbines (WT). Since these problems include a large number of interfaces the resulting number of degrees of freedom is prohibitive for the conventional BEM. The application of the proposed here ACA/BEM and the solution of the aforementioned problem is the second objective of the present dissertation. Results that combine the airborne infrasound and the soil surface waves generated by a WT are reported for the first time in the literature.
|
|---|
