You are on page 1of 4

19 Computer procedures for finite element analysis

19.1 Introduction
A companion program to this book is available which can carry out analyses for most of the theory presented in previous chapters. In particular the computer program discussed here may be used to solve any one-, two-, or three-dimensional linear steady-state or transient problem. The program also has capabilities to perform non-linear analysis for the type of problems discussed in reference 1. Source listings and a user manual may be obtained at no charge from the authors internet web site (http://www.ce.berkeley.edu/rlt) or the publishers internet web site (http:// books.elsevier.com/companions). The program is written mostly in Fortran with some routines in C (see authors web site for more information on using C for user modules). Any errors reported by readers will be corrected so that up-to-date versions are available. The version available for download is called FEAPpv which is an acronym for Finite Element Analysis Program personal version. It is intended mainly for use in learning nite element programming methodologies and in solving small to moderate size problems on single processor computers. A simple management scheme is employed to permit efcient use of main memory with limited need to read and write information to disk. Finite element programs can be separated into three basic parts: 1. Data input module and pre-processor 2. Solution module 3. Results module and post-processor.

19.2 Pre-processing module: mesh creation


FEAPpv is mainly a solution module but provides simple data input and pre-processor capabilitites which permit generation of meshes using the multiblock schemes of Zienkiewicz and Phillips2 and Gordon and Hall.3 Alternatively the data may be input from neutral les written by other pre-processing systems (e.g., GiD4 ). Data input for the program consists of specication (or generation) of: (1) the coordinates for each node; (2) the element form and the nodal connection list for each element;

Pre-processing module: mesh creation

665

(3) boundary conditions and loads to be applied; and (4) material property data. The user manual describes the format for specifying the data to be used by FEAPpv.

19.2.1 Element library


As part of the input data it is necessary to describe the element formulation to be used in forming the stiffness matrix and load vector of each problem. This may be provided either by user written modules (see below) or using the element library provided with the program. Currently, the element library in FEAPpv includes: 1. Solid elements for two-dimensional linear elasticity. Forms are provided for the irreducible formulation described in Chapters 2 and 6; the three-eld mixed form described in Sec. 11.3 and the enhanced strain form described in Sec. 10.5.3. The elements permit consideration of elastic models which are isotropic or orthotropic as described in Chapter 6. (a) For the irreducible form the element shape may range from a 3-node triangle to a 9-node lagrangian quadrilateral. (b) For the three-eld mixed form the element shape may be a 4-node, 8-node or 9-node quadrilateral form. (c) For the enhanced strain model the element is restricted to a 4-node quadrilateral form. 2. Solid elements for three-dimensional linear elasticity. Only the irreducible form for a 4-node tetrahedron or an 8-node brick may be used. The 8-node brick may be degenerated into other forms by giving the same node number to nodes used to perform the degenerate shape (see Sec.5.8). The elastic material model may be isotropic or orthotropic as described in Chapter 6. 3. Frame (rod) elements for two- and three-dimensional elasticity. Conventional structural elements are provided to perform analysis of elastic two- and three-dimensional frame structures. While these forms have not been discussed in this text, except as suggested problems for solution, they are useful for use in general analysis. The theory is contained in standard references for structural analysis and also in reference 1. 4. Truss elements for two- and three-dimensional elasticity. Similar to frame elements, the FEAPpv system includes conventional truss elements which may be used to analyse plane and space truss structures. 5. Plate element for linear elasticity. A plate bending element for use in the analysis of plates which include the primary effects of transverse shear (so-called ReissnerMindlin theory1 ) is provided. The element form may be either a 3-node triangle or a 4-node quadrilateral. The theory is described in references 1, 5, 6. 6. Shell element for three dimensions with linear elasticity. A 4-node quadrilateral element form for use in modelling general shell forms is provided. The element includes membrane and bending effects only and, thus, may be used only for analysis of thin shells. The theory for the element is given in reference 7. The element form should be a 4-node quadrilateral.

666

Computer procedures for finite element analysis

7. Membrane element for linear elasticity. A general elastic membrane form is provided which is the same as the shell element but without the bending terms. The element form should be a 4-node quadrilateral. 8. Thermal elements for two- and three-dimensional Fourier heat conduction. The theory described in Chapter 7 for transient heat conduction is provided in elements which solve two- and three-dimensional problems. The Fourier model may be isotropic or orthotropic. 9. User developed elements. Users may develop and add element modules for any problem which can be formed by the nite element approach described in this book. Details for writing modules will be found in the Programers Manual available at the web sites.

19.3 Solution module


The main part of FEAPpv is a solution module which permits users to analyse a large range of problems formulated by the nite element method. Specic solution methods are prepared by the user using a unique command language, which is a sequence of statements which describe each algorithm. The current version of FEAPpv permits both batch and interactive problem solution. The commands provided permit specication of problems with either symmetric or unsymmetric stiffness matrices, selection of direct or iterative solution of the linear algebraic equation system, selection of different transient solution algorithms, and output of solution results in either a text or graphics format. Commands which permit solution of a symmetric generalized linear eigenproblem (see Chapter 16) using a subspace method8, 9 are also available as well as a feature to compute the eigenvalues and vectors for an element stiffness. While the main thrust of this book is the solution of linear problems, the system FEAPpv is capable of solving both linear and non-linear problems. The use of special loopcommands permits the construction of algorithms which require iteration or time stepping. In addition features to solve problems in which load following is needed are provided in the form of arc-length-type methods.1012 The solution of problems for which it is not possible to deduce an accurate stiffness matrix may be attempted using a quasi-Newton method based on the BFGS method.13, 14 The user manual available at the web site provides examples for several algorithms as well as a list of all available commands.

19.4 Post-processor module


As noted above the FEAPpv system contains capabilities to report results as text data written to an output le or in graphical form which may be displayed on the screen or written to les for processing by other systems. Files are written in PostScript format (in an encapsulated form which may be used by many programs e.g., TeX or LaTeX). The general features of graphical post-processing are limited to displaying two-dimensional objects. More complex forms require an interface to a separate pre-/post-processing system (e.g., GiD.4 ). The two-dimensional capabilities in FEAPpv include display of the mesh including node and element numbers, boundary conditions and loads. Contour plots for each degree of freedom of the solution system may be displayed as well as contours of

User modules

667

element values such as stress or strain components. The user manual provides a list of all commands for constructing graphical outputs. The available version for graphics is limited to X-window applications and compilers compatible with the current HP Fortran 95 compiler for Windows-based systems.15

19.5 User modules


A key ingredient of the FEAPpv system is the ability of a user to add their own modules to extend the capabilities of the program to other classes of problems, material models, or solution strategies. Some user developed modules are available at the authors web site given above and include element modules for other problem forms, an interface to other linear equation solvers, etc. Experienced programmers should be able to easily adapt these routines to include additional features. Programming additions to the system may be performed following descriptions in the Programmer Manual available at the web sites.

References
1. O.C. Zienkiewicz and R.L. Taylor. The Finite Element Method for Solid and Structural Mechanics. Butterworth-Heinemann, Oxford, 6th edition, 2005. 2. O.C. Zienkiewicz and D.V. Phillips. An automatic mesh generation scheme for plane and curved surfaces by isoparametric coordinates. Int. J. Numer. Meth. Eng., 3:519528, 1971. 3. W.J. Gordon and C.A. Hall. Transnite element methods blending-function interpolation over arbitrary curved element domains. Numer. Math., 21:109129, 1973. 4. GiD The Personal Pre/Postprocessor. www.gidhome.com, 2004. 5. F. Auricchio and R.L. Taylor. A shear deformable plate element with an exact thin limit. Comp. Meth. Appl. Mech. Eng., 118:393412, 1994. 6. F. Auricchio and R.L. Taylor. A triangular thick plate nite element with an exact thin limit. Finite Elements in Analysis and Design, 19:5768, 1995. 7. R.L. Taylor. Finite element analysis of linear shell problems. In J.R. Whiteman, editor, The Mathematics of Finite Elements and Applications VI, pages 191203. Academic Press, London, 1988. 8. J.H. Wilkinson and C. Reinsch. Linear Algebra. Handbook for Automatic Computation, volume II. Springer-Verlag, Berlin, 1971. 9. K.-J. Bathe. Finite Element Procedures. Prentice Hall, Englewood Cliffs, N.J., 1996. 10. E. Riks. An incremental approach to the solution of snapping and buckling problems. Int. J. Solids Struct., 15:529551, 1979. 11. K. Schweizerhof. Nitchlineare Berechnung von Tragwerken unter verformungsabhangiger belastung mit niten Elementen. Doctoral dissertation, U. Stuttgart, Stuttgart, Germany, 1982. 12. J.C. Simo, P. Wriggers, K.H. Schweizerhof, and R.L Taylor. Finite deformation post-buckling analysis involving inelasticity and contact constraints. Int. J. Numer. Meth. Eng., 23:779800, 1986. 13. J.E. Dennis and J. More. Quasi-Newton methods motivation and theory. SIAM Rev., 19:4689, 1977. 14. H. Matthies and G. Strang. The solution of nonlinear nite element equations. Int. J. Numer. Meth. Eng., 14:16131626, 1979. 15. HP Fortran home page. http://h18009.ww1.hp.com/fortran, 2004.

You might also like