Computational & Technology Resources
an online resource for computational,
engineering & technology publications |
|
Civil-Comp Proceedings
ISSN 1759-3433 CCP: 79
PROCEEDINGS OF THE SEVENTH INTERNATIONAL CONFERENCE ON COMPUTATIONAL STRUCTURES TECHNOLOGY Edited by: B.H.V. Topping and C.A. Mota Soares
Paper 293
Maximizing Gaps between Eigenfrequencies in Two-Material Structures N.L. Pedersen
Institute of Mechanical Engineering, Aalborg University, Denmark N.L. Pedersen, "Maximizing Gaps between Eigenfrequencies in Two-Material Structures", in B.H.V. Topping, C.A. Mota Soares, (Editors), "Proceedings of the Seventh International Conference on Computational Structures Technology", Civil-Comp Press, Stirlingshire, UK, Paper 293, 2004. doi:10.4203/ccp.79.293
Keywords: optimization, eigenfrequencies, two-materials, 2D scalar case.
Summary
One strategy for passive vibration control of mechanical structures is to design the structures so that eigenfrequencies lie as far away as possible from the excitation frequency. This paper exploits the possibility for using finite element analysis and the method of topology optimization to maximize the separation of two adjacent eigenfrequencies in structures with two material components. This study is restricted to 2D structures where the vibrations are governed by the scalar wave equation. Results related to the 1D scalar case can be found in [1].
When we optimize a 2D domain with respect to maximizing the gap between eigenfrequencies there are a number of extra difficulties we must deal with. The primary source of the difficulties is the possibility of multiple eigenfrequencies. The multiple eigenfrequencies can be calculated using e.g. the subspace iteration method [2]. The objective for the optimization is given by
where the gap between the eigenfrequencies of order and is maximized. If the eigenfrequencies of order and are both distinct, with squared eigenfrequencies and and corresponding eigenvectors and , no problems arise and we use the objective (81) directly since the sensitivities of a squared eigenfrequency with respect to a design parameter are given by
where is it assumed that the eigenvector have been normalized so that . In the case of multiple eigenvalues we cannot use (82) to find the sensitivities. Instead we can use an extended method presented in [3] and used more recently in [4] and [1]. It is possible to find the sensitivities of multiple eigenfrequencies. There is, however, still a problem because the sensitivities are given for specific eigenvectors that vary for each design parameter. It is therefore difficult to solve the optimization problem as formulated in (81). As an alternative formulation we propose to use a double bound formulation. We apply topology optimization with a material interpolation scheme where constant material properties is assigned to each finite element and associate these material properties with continuous design variables [5]. We choose one design variable per element and let it vary continuously between 0 and 1. We let the material properties in each element be a specified interpolation function of this design variable. Since the design variable vary continuously between 0 and 1 we may expect that in the optimal design we can end up with material properties that do not correspond to either of the two materials, but instead to some intermediate values. In order to ensure a well defined distribution of material 1 and 2 in the structure, we can manipulate the interpolation functions [6]. Especially when dealing with eigenvalue problems the choice of the interpolation function is important [7]. In the present study we can not use a simple linear interpolation of mass and stiffness, therefore a more involved interpolation is needed, and results obtained with a new interpolation function are shown. The results of the optimization of a square design domain with free-free boundary conditions is shown. The results show a clear separation of the two materials in the domain. References
purchase the full-text of this paper (price £20)
go to the previous paper |
|