Computational & Technology Resources
an online resource for computational,
engineering & technology publications
PROCEEDINGS OF THE NINTH INTERNATIONAL CONFERENCE ON COMPUTATIONAL STRUCTURES TECHNOLOGY
Edited by: B.H.V. Topping and M. Papadrakakis
Model Reduction in Finite Element Analysis for a Fluid Filled Pipe Using an Orthogonal Vector Set
R.J. Alkhoury1, M.H. Chikhalsouk1, R.B. Bhat1 and K.D.P. Nigam2
1Department of Mechanical and Industrial Engineering, Concordia University, Montreal, Quebec, Canada
R.J. Alkhoury, M.H. Chikhalsouk, R.B. Bhat, K.D.P. Nigam, "Model Reduction in Finite Element Analysis for a Fluid Filled Pipe Using an Orthogonal Vector Set", in B.H.V. Topping, M. Papadrakakis, (Editors), "Proceedings of the Ninth International Conference on Computational Structures Technology", Civil-Comp Press, Stirlingshire, UK, Paper 185, 2008. doi:10.4203/ccp.88.185
Keywords: finite element method, model reduction, orthogonal vectors set, Rayleigh-Ritz, pipes, Eigenvalue.
The finite element analysis of three dimensional structures is often computationally intensive in view of the large number of degrees of freedom. A model reduction approach is beneficial especially when only the first few modes of vibration are required. There have been many studies of the reduction of large eigenvalue problems, for example, the static condensation or Guyan reduction  which implies the selection of master and slave degrees of freedom. This method is very sensitive to the correct selection of the categories of the degrees of freedom. Another example is the dynamic condensation  which, beside the slave and master, requires inverting the stiffness dynamic matrix which is frequency dependant. Some more major studies on reduction methods are listed in a brief literature review in the introduction. In the present study a coiled heat exchanger filled with water is studied. The coordinates of the elements were obtained using MATLAB with a three dimensional curve representing the geometry of the coil. These coordinates were the input to ANSYS to build the geometry. The mass and stiffness matrices of the finite element model were extracted from ANSYS and transferred to MATLAB. A model reduction technique using a set of characteristic orthogonal vectors was used to reduce the size of the problem. In this method, the Rayleigh-Ritz analysis is employed to obtain the reduced mass and stiffness matrices. This method takes a set of vectors that are used as transformation matrix to a generalized coordinates system, which will reduce the model down to the same number of generalized coordinates. The method of generation of those vectors is presented. Results were obtained for two models, one with 10 generalized coordinates, resulting in 10x10 reduced matrices and the second with 20 generalized coordinates resulting in 20x20 reduced matrices. Tables containing the proprieties of the materials as well as the eigenvalues of the reduced and complete models are presented. It is concluded that this technique is able to predict up to 50% of the natural frequencies with negligible deviation from the finite element results without the need to supply either the coordinates or to choose any slave or master degrees of freedom.
purchase the full-text of this paper (price £20)