INTRODUCTION TO THE FINITE ELEMENT METHOD 9.0 INTRODUCTION The purpose of this chapter is to expose the reader to some of the fundamental con- cepts of the finite element method (FEM), and therefore the coverage is only intro- dductory in nature, For further detail, the reader is urged to consult the references ited at the end of this chapter. The finite element method is a numerical technique, ideally suited to digital ‘computers, in which a continuous elastic structure (continuum) is divided (is- cretized) into small but finite well-defined substructures (elements). Using matrices, the continuous elastic behavior of each element is categorized in terms of the ele- rucnt’s material and geometric properties, the distribution of loading (static, dy- ‘namic, and thermal) within the element, and the loads and displacements at the nodes of the element. The element's nodes are the fundamental governing entities of the element, as it is the node where the element conaects to other elements, where clastic properties ofthe element are established, where boundary conditions are as- signed, and where forces (contact or body) are ultimately applied, A node possesses degrees of freedom (dot's). Degrees of freedom are the translational and rotational motion that can exist at a node. At most, a node can possess three translational and three rotational degrees of freedom. Once each clement within a structure is defined. locally in matrix form, che elements are then globally assembled (attached) through. their common nodes (dof's) into an overall system matrix. Applied loads and ‘boundary conditions are then specified, and through matrix operations the values of all unknown displacement degrees of freedom are determined. Once this is done, it is a simple matter to use these displacements to determine strains and stresses through the constitutive equations of elasticity, Since matrix notation and algebra 673674 CHAPTER & © _IxRODUCTION To THE Ferre ELaMENT METHOD are extensively used in this chapter, it is recommended thatthe reader fully under- stand the treatment of maizices presented in Appendix 1 before continuing beyond this point. ‘Many geometric shapes of elements are used in finite element analysis for spe- cific applications. The various clements used in a general-purpose commercial FEM software code constitute what is referred to as the element library of the code. Elements can be placed in the following categories: line elements, surface elements, solid elements, and special-purpose elements. Table 9.0-I provides some, but not all, of the types of elements available for finite element analysis. Since the finite element method is a numerical technique which diseretizes the domain of a continuous structure, errors are inevitable. These errors are: 1. Computational errors, "These are due to round-off ervors from the computer floating-point calcufations and the formulations of the numerical integration schemes that are employed, Most commercial finite element codes concentrate on reducing these errors’ and consequently the analyst generally is concerned with dis- cretization factors. ‘Sample firite element library ‘Number Shape of Nodes Appllntions = os niheae ° sm IS ne CHILL f , ‘Frame 2 Axial, torsional, end bending. ach... * Siem, sont + Beeman nyemen ‘on Mae = oe & Famers. tin rn tee lateral ® vending pence one, sous reco oe om im Se ts Pou | {foretan , moat commercial programs vse avlomatic bandwidh apsiizakon, which can signtconlly reFable 9.0-1 (concluded) 9.0 INTRODUCTION 67s ‘Element ‘Number Type ‘None ‘Shape of Nodes Applications: ‘Plane stross or strain, axisymmesry, ‘S-noded ‘thin plate or shell in bending. Swf Teanga 6 Pre quad whee posible, Used foc son of pats ed exam Al ‘ Si, isk pate ing i nite nodes) brick) i] otc ; Pentagon IN 6 Sofi, nek plate ng mieide Solid! (wedge) nodes). Used for zansitions. tenoded 4 ‘Solid, thick plaie (using midside ‘tecahedron ‘nodes). Used for transitions. (et) op 2 Free dpa for ar papel cee Ga ‘oem re prescribed extension gap Speci asjse Vai. Rigid sonsins een ses * These elomens are oo ovale wth midside nodes, 2 Discresization errors. The geometry and the displacement distribution of a true structure continuously varies. Using a finite number of elements to model the structure introduces errors in matching geometry and the displacement distibotion due to the inherer limitations of the elements. For cxample, consider the thin plate structure shown in Fig, 9.0-1(a), Figure 90-1(6) shows a finite element model of