finite element methods for structural_engineers pdf

Trang 1

a Basic Approach) ‘The book presents the basic ideas of the ite element method so that it can be Used as a

textbook in the curriculum for undergraduate and gradvate engineering courses In the presentation of fundamentals and derivations care had been taken not to use an advanced ‘mathematical approach, rather the Use of matrix algebra and calculusis made Furtherno effort

isbeing made to include the intricacies ofthe computer programming aspect, rather the material {s presented in a mannerso thatthe readers can understand the basic principles using hand caleuiations However, a list of computer codes is given Several lusraive examples are Presented in a delaied stepwise manner to explain the various steps in the application of tho method A fairly comprehensive references lst atthe end of each chapter isaiven foraddiional

information and furtherstudy

Wall N ALRifaie is Professor of Civil Engineering at the University of Technology, Baghdad, frag He obiained his Ph.D ftom the University College, Cardi, U.K in 1975 Dr Wail established the Civil Engineering Department at the Engineering College in Baghdad and was ‘the Head for neatly seven years He received the Telford Premium Prize from the Insltuion of Civil Engineering (London) in 1976 His main areas of research are: Box girder bridges, folded plate siructures, frames and shear walls including dynamic analysis, He isthe author of three books on structural analysisin Arabic

‘Ashok K Govilis Professor in the Depariment of Applied Mechanics, Molilal Nehru Regional Engineering College, Allahabad, India and was also Head of the same department for over five years He obtained BE degree in Civil Engineering (1963) from BITS, Pilani, India, and MS (1968) and Ph.D, (1977) from the University of owa, lowa City, U.S.A Dr Govils main areas of esearch are: Optimal design of structures, fll-safe design of structures, and finite element method He has written several research papers and technical reports, and developed many computer programmes for optimal design of structures including dynamic analysis and ‘uinerabilty reduction,

9784122034102)

|

.PvaUsrNG roRokE Wonta

NEW AGE INTERNATIONAL (P) LIMITED, PUBLISHERS sy Delhi = Cent» Cochin «

Trang 2

Finite Element Method for Structural Engineers (A Basic Approach) Wail N Al-Rifaie Ashok K Govil fal

Punusiva ron one WORLD

NEW AGE INTERNATIONAL (P) LIMITED, PUBLISHERS lerabad

Trang 3

No part of this book may be reproduced in any form, by photosta, microfilm, xerography, or any other means, or incorporated into any information retrieval system, electronic or mechanical, without the writen permission of the copyright owner

Branches:

Near Starline Hotel, Guwahati, © (0361) 2543669

‘© No 105, Ist Floor, Madhiray Kaveri Tower, 3-2-19, Azam Jahi Road, Ninboliadda, Hyderabad, © (040) 24652456

‘© RDB Chambers (Formerly Lotus Cinema) 106A,1st Floor, $.N Banerjee Road, Kolkata, © (033) 22275247

© 18, Madan Mohan Malviya Marg, Lucknow ® (0522) 2209578 © 142C, Victor House, Ground Floor, N.M Joshi Marg, Lower Parel, ‘Mumbai © (022) 24927869 * 22, Golden House, Daryaganj, New Delhi, © (011) 23262370, 62368 ISBN : 978-81-224-2410-2 Rs 160.00 C-08-05-2573

Printed in India at Nisha Enterprises, Delhi PUBLISHING FOR ONE WORLD

NEW AGE INTERNATIONAL (P) LIMITED, PUBLISHERS 4835/24, Ansari Road, Daryaganj, New Delhi-1 10002

Trang 4

PREFACE

‘The finite clement method is one of the most popular numerieal tech- niques used for obtaining an approximate solution of complex problems in various fields of engineering In the beginning, the method was developed as an extension of matrix methods for the analysis of structural engineering problems However, later it has also been recognized as a most powerful ‘method for analysing problems in other fields of engineering, such as uid mechanics, soil mechanics, rock mechanics, heat low, ete The generality of its application coupled with the availability of high speed electronic digital computers, has put finite element method in wide use It has also been included in the curriculum of engineering colleges

The aim of this book is to present the basic ideas of the finite element method so that it can be used as a text book in the curriculum for undergraduate and graduate engineering courses Tn the presentation of fundamentals and derivations, care has been taken to usc matrix algebra and calculus only, rather than an advanced approach It may be mentioned further that the computer programming aspect has not been included but the material is presented in such a manner that the readers can understand the basic principles using hand calculations However, a list of computer codes is given in Appendix G

‘The book is divided into eight chapters The first Chapter introduces the basic concepts of the finite clement method and in Chapter 2, basic equations of elasticity are presented Chapter 3 discusses the structural idealization and also describes the commonly used elements in structural analysis In Chapter 4, methods for determining stiffess characteristics und transform= ation of matrices are given In Chapters 5 and 6, element stiffness properties are derived In Chapter 7, general formulation of the finite clement method is presented with the help of suitable examples Finally in Chapter 8, examples are presented to illustrate the various basic steps in the appl tion of the method,

The references listed at the end of cach chapter are those in which readers can find additional information or detailed developments

Trang 5

CONTENTS

Preface itt

Chapter 1_ INTRODUCTION TO FINITE ELEMENT METHOD 1 L1 Introduetien and General Description 1 12 TiusrativeExample 2 References 8 Chapter 2_ BASIC EQUATIONS FROM LINEAR ELASTICITY THEORY 8 2A_Introduction_9

22 Stress and Strain Components _9 2.3 Equations of Static Equilibrium 11 2.4 Strain—Displacement Equations 13 2.5 Compatibility Equations 13

2.6 Generalized Hooke’s Law (Constitutive Equ 2⁄1 Plane Strain and Plane Stress 15 References 18 Chapter 3_ STRUCTURAL IDEALIZATION » 34_Intraduction 19 lization or Discretization 19 3.3 Types of Structural Elements 22 ences 26

Chapter 4 METHODS OF DETERMINING STIFFNESS

PROPERTIES AND ITS TRANSFORMATION 27

4.1_Introduction 27

Trang 6

vì Contents

4.3 General Approach—Using Displacement or Shape Functions _29 44 Transformation of Reference Coordinate Systems 33

Reference 39

Chapter 5 DERIVATION OF STIFFNESS PROPERTIES FOR ELEMENTS

&1 Pin-Jointed Plane Bar Elements 40 5.2 Plane Ream Elements 49

53 Pin-Jointed Space Bar Elements 68 5.4 Space Beam Elements 71 References 75 Chapter 6_DERIVATION OF STIFFNESS PROPERTIES FOR ‘TWO-DIMENSIONAL ELEMENTS 76, ‘Triangular Plate Elements (In-Plane Forces) 76 6A

62 Rectangular Plate Elements (In-Plane Forces) 84 63 Triangular Plate Elements (In-Bending) 92

64 Rectangular Plate Elements (In-Bending) 99 References 108

Chapter 7_FORMULATION OF FINITE ELEMENT METHOD 109 ‘L1_Disoretization of Continuum 109

12 Generation of Basic Data 109

23 Determination of Transformation Matrix 113 74 Assembly of the Overall Stiffness Matrix 116 15 Elimination of Restrained Degrees of Freedom 122 7.6 Determination of Equivalent Applied Nodal Ferees _125 7.1 Calculation of Nodal Displacements, Forces and Stresses 130

Chapter §_APPLICATION OF FINITE ELEMENT METHOD 133 8/1 Analusis of Pin-Joimed Structure 133

Trang 7

Contents vii 8.3 Analysis of Deep Beam Problems 162 157

References 167

Appendix A_ MATRIX ALGEBRA 168

Appendix B_ REACTIVE FORCES FOR RESTRAINED BEAM ‘ELEMENTS 78

Appendix C_ PROPERTIES OF SECTIONS 182

Appendix D_ SOLUTION OF SIMULTANEOUS EQUATION — 186 Appendix ESI UNITS FOR STRUCTURAL ENGINEERS 192

Appendix F BIBLIOGRAPHY 04

Appendix G_COMPI

Trang 8

CHAPTER 1

INTRODUCTION TO FINITE ELEMENT METHOD 4.1 INTRODUCTION AND GENERAL DESCRIPTION

‘The finite element method represents an extension of matrix methods for the analysis of framed structures to the analysis of the continuum structures The basic philosophy of this method is to replace the structure or the continuum having an unlimited or infinite number of unknowns by a ‘mathematical model which has a limited or finite number of unknowns at certain chosen discrete points The method isextremely poworful as it helps to accurately analyse structures with complex geometrical properties and loading conditions In finite element method, a structure or a continuum, as shown in Fig 1.1() and (b), is discretized and idealized by using a mathematical model which is an assembly of subdivisions or discrete elements These discrete elements, known as finite clements, are assumed to be interconnected only at the joints called nodes Simple functions, such as polynomials, are chosen in terms of unknown displacements (andjor their derivatives) at the nodes to approximate the variation of the actual displacements over each finite element The external loading is also transformed into equiva~ lent forces applied at the nodes Next, the behaviour of each element independently and later as an assembly of these elements is obtained by relating their response to that of the nodes in such a way that the following basic conditions are satisfied at each node:

1 The equations of equil

2 The compat 3 The material constitutive relationship

Trang 9

2 Finite Element Method for Structural Engineers Toil boom (2) ldealization of helicopter tall boom [1-1] Orginal strcture sypical node ypleat element Idesllsed structure fenton ga Section of damm Discretizetion of the dam secton (®) Idealzetlon of đem [12] Fig 141 Structural idealization 1.2 ILLUSTRATIVE EXAMPLE

Trang 10

Introduction to Finite Element Method 3

2A simple linear spring aystem,

‘This system has two lincar springs connected in series with spring sifnesses ky and ky The left-hand end is rigidly fixed while the right-hand ends free to move, Further, itis assumed that under the action of applied load, these springs can have displacements in x-coordinate direction Thus, forces, displacements and spring stiffnesses are the only parameters in this system, Also fora linear spring, the applied force F is proportional tothe resulting ‘displacement A and may be expressed as

F=KA ap

where k is the constant of proportionality which defines the stiffness of the spring Thus, knowing the value of & and F in Eq (1-1), the displacement will be given by

(2)

which is sufficient to describe fully the deformed state of an elastic spring Now considering the system in Fig 1.2, the first step in the finite clement method is to subdivide the system into discrete elements Defining each spring to be an element, the idealized linear spring system consists of ‘wo elements and three nodes as shown in Fig 1.3 The forces (Fy, Fy Fs) and the associated displacements (A,, Ay, 43) at the nodes are also shown in the figure For convenience the direction of the forces and the displace ments shown in Fig 1-3 are taken as positive

TT ttement (6 Element (ep)

nose(?) Nede(2) Node (3)

Fig 1.3 Diseretization of the system,

Next, in order to obtain the response of each clement, free body diagrams of isolated spring elements are considered, as shown lá, It may be noted that for these elements, interpolation functions to describe the displacements over the clement are not needed and may be obtained directly using Eq (1 2) Now, assume that node (1) is displaced by A, due to force Fy, with node (2) fixed in position as shown in Fig 1.5(a) Using Eq, (1.1), the force-displacement relation for the node (1) may be expressed

Trang 11

4 Finite Element Method for Structural Engineers Fath ky Fate đuôi he WWW t—AVWw—+” ĩ w ‡ 7 oO Fig, 1.4 Free Body diagrams of isolated spring elements Fa = kab a3) Applying the equilibrium of forces (ZF, = 0) for the spring (s) one obt Fut Fax a4 Fay “` as

where Fy is the reactive force at node (2) due to Ay at node (1)

Similarly, now assume that node (1) is fixed and node (2) is displaced by Ay due to force Fyy applied at node (2), as shown in Fig 1.5 (b) Then, the foree-displacement equation at node (2) may be expressed as Ai of ge NHƯ: Fz tp ' š ?

(6) Superooerhon of cose(6) and case(b) Fig 5 Forcedieplacement relationship for lement (4)

Fa kay a6 Applying the equilibrium condition (SF, = 0), we get

Trang 12

Introduction to Finite Element Method 5

Fas

—hây (18)

where Fig is the reactive force at node (1) due to A, at node (2)

Now using the principle of superposition, one obtains the case shown in Fig 1.5 (c), i-e., adding up algebraically the two cases shown in Fig

1.5(a) and (b) Thus, we obtain

Fo= Fut Fai (1.9)

and

F=Fut Fa (1.10)

where F, and F, are the forces at node (1) and (2), respectively Substitut-

ing the Values of Fy, Fas Fay and Fea in Eqs (1.9) and (I-10), we obtain the

forces in terms of displacements as Tị BÀI — hide ay Frank bit kbs (1.12) I matrix notation, these equations may be rewritten as Fi) kom Ai { }- mộ [mm] lá 5 { Ì a3) and in condensed form {FY = [g*9(a)e0 (19)

Equation (1.14) defines the element force-displacement or equilibrium equation for element (e,) In this equation, square matrix [A}*? is known as clement stiffness matrix; the column vectors (A}° and (F)( are defined as nodal displacement and nodal force vectors, respectively -The superscript defines the element number

Following a similar approach as for element (c,), the force-displace- rent relationship for element (e,) may be expressed as

Ff he — he] (as

fob) eS Lm md las as

condensed form

(Ee = [agen gaye (116)

Trang 13

6 Finite Element Method for Structural Engineers

Although, the matrices J? and (Ff? are of the same order, they may not be added directly as they relate to different sets of displacements For this simple example, the equilibrium equations may easily be expanded by inserting rows and columns of zeros in such a way that both sets of equations are related to all the possible displacements (ic As, Ay and 1) of the system Thus, Eqs (1.13) and (1.15) after expansion, may be expressed as: For element (e) Fi) ky —k 07 (Ay Ft=|—k kị 0| JA,kand aan LF, o 0 of by For element (e,) Fy) [0 0 01 (Ai Fate] —k| 4A; (18) mi lo —k, kel Uy

Using the principle of superposition and applying the rule of matrix addi- tion (see Appendix A) we obtain

Ay ok 07 (iy

BÍ =|—k k+th — ie in (1.19)

FH) Loo —k hed la

It may be pointed out that the procedure of expanding element equilibrium ‘equations is lengthy and hence direct superposition is used as explained in Chapter 7

In condensed form, Eq (1.19) may be expressed as

()=[K] tà] (120)

Trang 14

Introduction to Finite Element Method 7 Now to obtain the solution of the problem, we use Ea, (1.19) or its ‘condensed form Eq, (1.20), It may be noted that so far no limitations has been placed on any of the displacements A, 4, and As Hence, application ‘of any external loading will result in moving the system as @ rigid body ‘Thus, before solving for unknown displacements, Eq (1.19) needs to be ‘modified to incorporate boundary conditions so as to prevent the rigid body motion of the structure For the linear spring system shown in Fig 1.2, node (I) is fixed, i.e A, 0; hence rewriting Eq (1 19) in partitioned form

Ry ‘A, =0:

Fy Ay q21)

Fs ‘As

Equation (1.21) contains two unknown displacements (A, and A.) and one unknown reactive force (F,) Forces F, and F, are known as applied forces and are equal to 0 and P, respectively Thus from Eq (1.21) and noting that A; = 0, we obtain Jak a1 ñ 2 tF)=Ink A (122

0) [kathy Ke] (Ái

PES PSL med Ro lay

Solving Eq (1.23) for unknown displacements, we obtain

a= Pky (129

Aạ=P (Œ + kạ)[kska) q25)

Knowing displacements 4, and Ay, the reactive forces F, may be obtained using Eq, (1.22) Thus,

Ra=-P (1.26)

Finally the internal forees in the elements may be determined using

forces-displacement relations Eqs (1.13) and (1.15) If P., and P,: are the

{internal forces in springs (e,) and (e), then

Pam hy (My — 4) (1.27) Ta =k, (lý — Aj) (1.28)

Trang 15

8 Fioite Element Method for Structural Engineers

‘This completes the solution using finite element method It may be mentioned that the finite element method involves extensive computations, ‘mostly repetitive in nature Hence the method is suited for computer pro- ‘gramming and solutions of the problems can be obtained easily using pro- ‘gramming on electronic digital computers However, in this book, without going into the intricacies of computer programming, the basic concepts and the development of the method are presented in a simple manner Further for easy understanding of the various steps in the method, illustrative examples with hand calculations are given in Chapter 8 These examples are taken from the field of structural engineering, nevertheless the method is general and can also be applied equally well to other fields of engineering

REFERENCES

1.1 Govi, A.K., 1.8, Arora and E.J., Haug, “Optimal Design of Frames with Substructuring”, Computers and Structures, An International Journal Vol 12, No 1, 1980

1.2 Clough, R.W., “The Stress Distribution of Norfork Dam,” Structures and Materials Research, Department of Civil Engineering, University of California, Berkeley, Series 100, Issue 19, 1962

Trang 16

CHAPTER 2

BASIC EQUATIONS FROM LINEAR ELASTICITY THEORY

2.1 INTRODUCTION

To provide a ready reference for the development of the general theory Of the finite clement method applied to the problems of structural analysis, some of the concepts and basic equations of linear elasticity theory are summarized in this chapter The equations are given without derivation or proof and apply to homogeneous isotropic materials 22 STRESS AND STRAIN COMPONENTS

2.2.1 Stress Components

A state of stress exists in a body acted upon by extemal force These external forces are, in general, of two kinds which may act on a body If

Trang 17

they act over the surface of the body, they are called surface forcesand are expressed in terms of force per unit area; if they are distributed throughout the volume of the body, they are called body forces and are expressed in terms of force per unit volume For example, a force such as the hydro- static pressure which is distributed over the surface of a body, is called a surface force; while gravitational and centrifugal forces, which are distributed over the volume of the body are called body forces

‘The state of stress, which exists in a body acted upon by external forces, is completely defined in terms of six components of stress as shown in Fig 2.1 Ín vector form, it is expressed as «| 1 + @1 | J

where en Gy & = components of normal stresses Tay Tyee Tex = Components of shear stresses

It may be pointed out here that we have designated only three components of shear stress because only these are independent In the next section by

considering the equilibrium of an elemental volum “hy Tyee oe tay and Tye = Tae

Notation Used ‘Components of normal stresses (¢z, 624) carry asingle subscript which indicates that the stress acts in the direction of subscript and on a plane whose outernormal is in the direction of subscript Components of shear stresses (tp) ye tz) carry a double subscript The first subscript denotes the plane on which itacts and the second subscript denotes its direction

‘Sign Convention

Trang 18

Basic Equations from Linear Elasticity Theory 11 22.2 Strain Components

Corresponding to the six stress components, the state of strain ata point ccan be divided into six strain components In a vector form, the state of strain is expressed as ce | | Je | @= Ễ (2.2) le J

where tan Ye nd Yor = components of shear strains ex, ¢y@nd ¢s = components of normal strains

‘The notation and sign convention used for the strain components are the same as those for the stress components

23 EQUATIONS OF STATIC EQUILIBRIUM

The equilibrium of an elastic body in a state of stress is governed by three partial differential equations for the nine stress components These equations are derived by considering the equilibrium of forces and moments ‘acting on an elemental volume of a body Consider a small rectangular arallelopiped of a body shown in Fig 2.2 It is subjected to a general system of positive three-dimensional stresses as well as to Positive body force-components X, ¥, Z in x, 2 directions, respectively ‘Summing all forces in the x-direction, and using the condition 3 F, = 0, we get

(oct Mai dye og dys + (5+ Baye

ate ded + (ton + GE de) dxdy — nye dxdy +X dedyde = 0

Collecting terms, we obtain

doe 4 Bye tet oy te

‘Thus, the condition F, = 0 gives

ey, Boe On wt + tÝm0 (23a)

Trang 19

as (dedyde) is not necessarily equal to zero Similarly, summing forces in and z directions, we obtain =o, Sr 4 Sey By 3y 0, ng Bey Baye G3) (2.30) Fig 2.2 Stress and body force components on an elemental volume of a body

On Face OA4'0': op tyes te

On Face BCC: oy + 2% dy, tye + OE dy, oye + HE đy 7 Dy ” ấy CHỦ ấy

On Eiee OÁCB ï cụ sa tợ

*A'C!B!: bo, Ôn Brey

On Face OA CB's 044+ 8 dey see + 208 de, uy + ĐH ác

Likewise, a balance of moments about the three coordinate directions

shows that in the absence of body moments

Sey Tom Tự = tor 804 Toe = Tae G30 Equations (2.38, b,c) must be satisfied at all points of the body The stessts vary throughout the body, and at the surface oF boundary they ‘must be in equilibrium with the forces applied on the surface Let the

Trang 20

Basic Equations from Linear Elasticity Theory 13 where /, m, and n represent the direction cosines of the outward normal to the surface at the point of interest

24 STRAIN—DISPLACEMENT EQUATIONS

“The deformed shape of an elastic structure under a given system of loads can be described completely by three displacements u,v and w in x, ‘y and z directions respectively The positive directions of the displacements ‘correspond to the postive directions of the coordinate axes The relations between the components of strain and the displacement components are

ae a er ey]

5+ [G) +6) +G)]

im )+1|(}+(5}+}] Buêu „ 8n ôy „ aw aw œ3

Ấp tây + sp Tập apt ay

âm ôn , bu du , ar ay, awow

= Ot et aya t yee Baz

au , ow, buau , dor | wow

5z ấy Í ấy ấy Lân ax toe ox

For small deformations, the strain-diplacement ri

the expressions for the strain components, given by Eq (2.5), can be sim-

plified by retaining only the first order or linear terms and neglecting the

second order terms, that is ions are linear and aa gad, ga “ấy “ấy oF a 1T e8 Y= Bt Re em Bt Tay Tây

IF the body experiences large (finite) deformations or strains, higher order terms must be retuined as in Eq (2.5) These terms represent significant hangs inthe geometry ofthe Body and thus are called geometic non: linearites 2.5 COMPATIBILITY EQUATIONS

Trang 21

known as compatibility conditions, are obtained by eliminating dieplace- ment components in Eq (2.6) and can be expressed as follows BeBe Bo) en ye Be, Fee đê oe tae

Equations (2.7) are the six equations of strain compatibility which must be satisfied in the solution of three-dimensional problems in elasticity

2.6 GENERALIZED HOOKE’S LAW (CONSTITUTIVE EQUATIONS)

Trang 22

Basie Equations from Linear Elasticity Theory 15

‘The matrix [C]is termed as the material stiffness matrix, while its in- verse [D] is the material flexibility matrix Equation (2.9) or (2.10) represents the constitutive law for a linear, elastic, anisotropic, and homogeneous material For homogeneous isotropic elastic materials, only two physical constants are required to express all the elastic constants in Hooke's Law Hence, in terms of Young's modulus (E) and Poisson’s ratio (x), the matrices [C] and [D] can be expressed as fly» | v dời ‘<n | t1“trzy-z |9 0 @.12) oo lu and “yy 00 1 o 0 o | ¬ 0 0 | 2.13) 0 0202) 00 | 000 20+) 0 | i 0000 203w,

Sometimes matrices (C] and [0] are express in terms of modulus of rigidity (G) and modulus of volume expansion or Bul modules (K) Those constants, in terms of E and », are dfinsd as af + Cw srey Kear 0.1) 27 PLANE STRAIN AND PLANE STRESS

Trang 23

presented in a simplified form In both the cases, it is assumed that the body force Z is zero, and X and Y are functions of x and y only

2.7.1 Plane strain

It is state of strain in which cy Z tye = Yer =O and only strains ex, ey and yạy exist, The body suffers displacements in one plane only, ie., Z component of body force is taken as zero

Stress Components:

alee % SP (2.150)

sự =v(øx + ø;) (2.15b)

strain Components: =le 6 vaƑ 2.16)

Static Equilibrium Equations: G1) (2.18) Compatibility Equations: Bey _ Prey ae day 69 Constitutive Equations: (9) =f4]49 2.20)

Trang 24

Basic Equations from Linear Elasticity Theory 17 2.7.2 Plane stress

In plane stress problem, the components of stress normal to the x-» plane are zero Ìe., ơy = ter = ys =0- ‘Stress Components: )=f s soi 0.2 Strain Components: sla 6 oF G2) enero 628) Static Equilibrium Equations: te Fay t=O G39 Bray 4 Boy TH Â +Y=0 Strain-Displacement Relations: a =- a ond 629 au, a atx Compatibility Equations: 2.26) Constitutive Equations: =H 62 where G29

It should be pointed out that in a state of plane stress, ¢ is not equal to zero, but is given by

Wo +0)

Trang 25

18 Finite Element Method for Structural Engineers REFERENCES

2⁄1 Timoshenko, S„ and JN Goodier, Theory of Elasticity 3rd ed ‘McGraw-Hill Book Company, New York, 1951

2.2 Wang, C.T., Applied Elasticity, McGraw-Hill Book Company, New ‘York, 1953 23 Love, AEH, A Treatise on Mathematical Theory of Elasticity, Dover ions, New York, 1944 Mathematical Theory of Elasticity 2nd ed., MeGraw- Hill Book Company, New York, 1956

2.5 Sechler, EE, Elasticity in Engineering, John Wiley 1952 & Sons, New York 2.6 Southwell, R.V., An Introduction 10 the Theory of Elasticity, Oxford University Press, Oxford, 1936

Trang 26

CHAPTER 3

STRUCTURAL IDEALIZATION 34 INTRODUCTION

The first and most important step in the finite element method of structural analysis is to generate, using finite number of discrete elemeuts, ‘a mathematical model which should be as near as possible equivalent to the actual continuum Such a formulation of a model is referred to as structural idealization of discretization In this chapter, the underlying principles of structural idealization are discussed followed by the most commonly used structural clements, which are employed in the subsequent

chapters, are described

3.2 STRUCTURAL IDEALIZATION OR DISCRETIZATION

The continuum is a physical body,a structure or a solid which needs to be analysed The subdivision or discretization process of the continuum is essentially an exercise of engineering judgement These subdivisions are called elements, and are connected to the adjacent elements only at limi ted number of points called nodes (Fig 3.1) ‘Thus, in the idealization of the continuum, we have to decide the number, shape, size and configuration of the clements in such a way that the original body is represented by it as closely as possible Hence, the general objective of such an idealization is to discretize the body into finite number of elements sufficiently small so that the simple displacement models can adequately approximate the true solution, At the same time, it may be pointed out that too many small subdivisions will lead to extra computation effort No effort here is being made to discuss as to how many elementsshould be employed in any particular problem, rather it is suggested that wo or three cases with Aifferent number of elements or fineness of meshes should be considered The resulls thus obtained can be used in establishing the rate of con vergence and enhance confdenes in the idealization employed The structures, in general, may be divided into two categories

1 Skeletal structures—Trusses, beams, Frames, ete

Trang 27

pose no problem and can be done accurately as the assumed mathe ‘matical model is similar to the actual structure, The elements in the model formulation of these structures may easily be defined by the lengths between the two nedes as shown in Fig 3.1 In the second type of struc-

Typical nodes Typical elements

fa) Plone truss

JWeical neges Typical elements (01 Beam iypieat elements fle ‘nodes ⁄ (e) Frome Fig 3.1 Skeletal structures,

tures, mathematical modelling presents some difficulties as the elements used are multidimensional and continuously attached to adjacent elements as shown in Fig 3.2 This aspect is discussed in Chapter 6, However, in discretization of both the types of'structures, it is assumed that the elements are attached to the adjacent clements only at the nodal points and have constant material properties Also it is assumed that the elements of ske- let! structures are straight and prismatic; similarly the elements of conti- uous structures are assumed to have constant thickness Thus, the location of nodes for both the types of structures are located at places

Trang 28

Structural Tdealizaton 2L

{6} 0etoil of folded piate model

Supports

(6) Details of finite elomont ideatization (8.1) Fig, 3.2 Continuum structure

Trang 29

22 Finite Element Method for Structural Enginesrs

Node

(c) Abrupt change in section properties Steel copper

Nooe

(6) Abrupt chong in moterial properties Fig, 3,3 Natural subdivision of structure or continuum

It is evident from the above discusssion that the structural idealization is simply a process whereby a complex continuum is modelled asan assembly of discrete structural elements satisfying the conditions stated in Chapter 1, Now in the next sub-section, various types of structural elements are described

33 TYPES OF STRUCTURAL ELEMENTS

The structural elements are of various types The shape or configuration of these discrete elements depends upon the geometry of the continuum and upon the number of independent space coordinates (e.g x, oF 2) necessary to describe the problem Thus, based on the space coordinates, a finite element can be classified as a simple one-, twor, or three-dimensional element Here in this section, only the most commonly used one-, and two- dimensional elements with straight boundaries are described and in the, next chapter, stiffness characteristics for these elements are derived 3.3.1 Oned ensional Elements

A one-dimensional clement may be represented by a straight line whose ends, such as 1 and 2, are nodal points (Fig 3.4 a) These elements are referenced in a coordinate system Fnown as local or element coordinate syetem In this sytem, x-axis is defined by element axis which is a line Joining the two nodes of the elements One-dimensional elements are used ‘when the geometry, material properties and dependent variables such as displacements can all be expressed in terms of one independent space coordinate which is measured along the element axis Examples of structures using this type of element are skeletal structures such as trusses, frames, etc (Fig 31) It may be pointed out that for skeletal structures, element and global coordinate systems generally do not coincide

Trang 30

Structural Técalzation 23 3.3.11 Pin-Jointed Bar Element

‘The pin-jointed bar or truss clement shown in Fig 3.4 (b), is the simplest structural element and is assumed to be pin connected at both the ends The bar element is also assumed to have constant cross-sectional area (4) and modulus of elasticity (E) over its length (L), external loads are applied at the nodes and the effect of self-weight is neglected Thus for a pplane structure, this element has four degrees of frecdom, two at each node; ‘whereas for a space truss itis six, three at each node, This element carries & ‘one-dimensional stress distribution as it is assumed to resist only axial force,

1.2 Beam Element

‘The beam clement shown in Fig, 3.4 (c) is also known as frame element For a plane structure, this element has six degrees of freedom, three at ‘each node i.c., axial and in-plane transverse displacements, and in-plane rotation, whereas for a space structure, it has twelve degrees of freedom, six at each node, i.e three displacements and three rotations in the three coordinate directior 1 — VẢ TRE NH Noge (a) Typical reoresentation of one-cimensionat ‘element ores {b) Sin jointed plane bor or truss element

Trang 31

24 Fite Slement Method for Structural Engineers 33.2 Two-Dimensional Elements

‘The elements shown in Figs 3.5 and 3.6 are two-diinensional elements ‘These elements are of constant thickness and with straight boundaries ‘Many problems in solid mechani:s such as plane stain, plane stress, plato bending, eto., could be idealized using these elements It may be noted that for small deflections, the in-plane and transverse deformations can be un- coupled [3.5] Thus, these elements for in-plane forces and in bending are ‘considered separately Y 00Fz6 a x {a} tangutar plate element (in-plane forces)

Fig 3.5 Two dimensional elements with in-plane forces, 33.2.1 Triangular Plate Element (In-Plane Forces)

Trang 32

Structural Idealization 25

33.2.2 Rectangular Plate Element (In-Plane Forces) ‘Rectangular plate element can be obtained by combining two triangular plate elements This element has four nodes with two degrees of freedom at each node Thus each rectangular element with in-plane forces has eight degrees of freedom as shown in Fig 3.5 (b)

3.32.3 Triangular Plate Element (In-Bending) In this case, elements are subjected to only bending, i , out-of-plane forees The element has nine degrees of freedom, three at each node ie., transverse displacement and rotations about x and y axes (Fig 3.6 a

Trang 33

26 Finlte Element Method for Structural Engineers

3.3.24 Rectangular Plate Element (In-Bending)

Like in the previous case, only bending is considered, The element has twelve degrees of freedom, three at each node,as shown in Fig 3.6 (b) Tn the next Chapter, methods for deriving element stiffuess properties are presented and in Chapters 5 and 6 these properties are derived for the elements discussed in this chapter

REFERENCES

3.1 Rockey, K.C., and H.R Evans, “An Experimental and Finite Ele- ‘ment Study of the Behaviour of Folded Plate Roofs Containing Large Openings”, International Asso for Bridge and Structural Engi- neering, Vol 36-I of the “Publications”, Zurich 1976

3.2 Desai, S Chandrakant and John F., Abel, Introduction to the Finite Element Method—A Numerical Method for Engineering Analysis Van Nostrand Reinhold Company, New York, 1972

3.3, Gaylord, Jr Edwin H., and Charles N., Gaylord (Edited by), Struc tural Engineering Hand -Book, Second Edition, McGraw-Hill Book ‘Company New York, 1979

3.4 Zienkiewicz, O.C., BIM., Irons, J Ergatoudis.S Ahmad and F.O., Scott, “Isoparametric and Associated Element Families for Two- and three-Dimensional Analysis” FEM Tapir The Technical University of Norway, Trondheim 1969 43.5 Timoshenko, S.P and S., Woinowsky-Krieger, Theory of Plates and Shells, McGraw-Hill Book Company, New York, 1959

Trang 34

CHAPTER 4

METHODS OF DETERMINING STIFFNESS PROPERTIES AND ITS TRANSFORMATION 4.1 INTRODUCTION

It has been mentioned earlier (Chapter 1) that the basis of finite element method is the representation of a continuum or a structure by an assemblage of discrete elements called finite elements These elements are assumed fo be interconnected only at node points In order to determine isplacement relationship or stiffness characteristics of the whole is required in this analysis, we must obtain first the stiff-

In’y'': Local or Element coordinate system

Trang 35

28 Finite Hlement Method for Structural Engineers

‘ness properties of individual elements These stiffness properties are deter- ‘mined, herein, using the following two approaches 1, ‘Elementary Approach—using basic definition

2 “General Approach—using displaceinent or shape functions

In this chapter, the two approaches for obtaining clement stiffness properties are discussed; and later in subsequent Chapters 5 and 6, these methods are applied to pin-jolated bar elements, beam elements, trian gular and rectangular plate clements The general principles discussed may also be used for deriving the required stiffness properties of other types of

lements

It should be mentioned that it is convenient to develop stiffness charact- ics particularly for one-dimensional elements in reference to local or clement coordinate system (Fig 4.1) Analysis using finite element method requires that these characteristics ‘ot orient in the direction of local coordinate system However, itis always be referenced in global or overall coordinate system which in general does for the entire structure or contiauum should

possible to transform element stiffness characteristics from one coordinate system to another coordinate system The general procedure for achieving this transformation is also presented herein

4.2 ELEMENTARY APPROACH—USING BASIC DEFINITION Consider a prismatic element shown in Fig 4.2 with known geometric and material properties The clement is assumed to be fully restrained at both the nodes which are denoted by / and k It is referred with respect to local coordinate system and has twelve degrees of freedom, six at each node The numbers written beside the arrows indicate the possible displace- ‘ments The single-headed arrows denote translations, whereas the double- headed arrow denote rotations Thus, at node j the translations are num- dered 1.2, and 3 and the rotations are numbered 4, 5, and 6 Similarly, at the other node k, numbers 7, 8, and 9 denote translations and 10, 11, and 12 denote rotations

Now, in order to obtain the element stiffness characteristics by this approach, unit nodal displacements are imposed one ata time while all other nodal displacements are retained at zero The resulting forces exerted ‘on the element are obtained by determining the values of the restraint for- es required to hold the distorted member in equilibrium These restraint forces due to unit displacements define the element stifnesses Finally using the principle of superposition, the desired element stiffness characteristics are more commonly known as element stiffness matrix ig obtained, ‘The elementary approach discussed above uses the basic definition of stiffness It enables the reader to have a clear picture of the force-displace- ‘ment concept and to obtain the element stiffness matrix without much ‘mathematical manipulations This method is applied to obtain the stiffness

Trang 36

Methods of Determining Stiffess Properties and its Transformation 23

Fig 4.2 Restrained element,

matrices of pin-jointed truss and beam elements For other types such a5 triangular or rectangular plate elements which are more complex, this elementary approach becomes cumbersome Hence a general approach is considered

43 GENERAL APPROACH —USING DISPLACFMENT OR ‘SHAPE FUNCTIONS

‘The general derivation approach was first used by Turner at cl [4.1] 0 derive the stiffness matrix for @ triangular plate clement for plane stress problems (sce Chapter 2) This approach is general and can be used for any other type of element as well There are four essential steps in this derivation, which can be expressed as follows

STEP I: For each element, choose a set of functions that defines displacements uniquely within the element These functions are called shape functions, displacement functions, or displacement fields Express these displacement functions in terms of the nodal displacements

STEP 2 : Introduce the strain-displacement equations and thereby deter- ‘ine the state of element strain corresponding to the assumed displacement field

STEP 3: Write the constitutive equations relating stress to strain These ‘equations introduce the influence of the material properties of

the element

STEP 4: Write force-displacement relationship and identify the element stiffness matrix

Trang 37

30 Finita Element Method for Structural Engineers

Now using the steps presented above, stifness matrix for a general finite clement is derived 43.1 Derivation of Element Stiffness Characteristics Using General ‘Approach ‘Consider a typical element (¢) connected to other elements at the nodes 5y m ete STEP 1: Choose a displacement function (f(x, », 2} a8 { sắc 2) Ì G3) = „ for 2D elements (40) x2) and us, 3, 2) {fla ¥,2)) =} vs, 9» 2) f, for 3D elements (42) (0 2)

where 1, v and w are displacements within the element at (x, y,2) Hereafter the arguments (x, ¥, z) shall not be used for ease in writing ‘Now to obtain displacement function in terms of nodal displacements, it can be expressed in matrix form as

{f) = (Ia) (4.3)

where [N] = matrix of shape functions

{A} ~ column vector of nodal displacements

‘These quantities depend upon the type and dimensions of element In ‘expanded form, Eq (4.3) can be expressed as [#2 | t= [iva ov ad ‡ 69 ao | tAm L9)

where [Mj,[N,] = submatrices of matrix [N] if m = nodes ofa typical clement (@)

{04} {8;} = subvectors of nodal displacement vector (A) STEP 2: With displacements within the element known, the strains at any point can be determined In matrix form, strain-displacement

Trang 38

Methods of Determining Stiffess Properties and Ís Trandormaien 3L

fe) = [8] {8} (4.5)

In the above equation, matrix [8] is given by

l= IE) 46)

where [L] matrix of a suitable linear operator

It may be pointed out here that in case [Z] is not linear operator, strain vector cannot be expressed by Eq (4.5)

‘Asan example of a linear operator, consider a plane stress element For this element, the strain-displacement equations are (see Chapter 2): pe % F2 67 a (fe & AE ° hy an vr a) |e wl J ola ax In the above equation, the linear operator matrix [£] is defined as ra " ø] a tị =| 9 5 69 a | ay a

STEP 3: Assuming general linear elastic behaviour of the material of clement (¢), the relationship between stresses and strains will be linear and of the form

(6) = i14 49)

‘where [¥] = elasticity matrix containing the appropriate material properties

For homogeneous and isotropic element experiencing plane stress, the stress-strain equations may be written as

Trang 39

In the above equation, the matrix [9] is defined as

w= 410)

STEP 4: For writing the force-displacement relationship, we consider the virtual work approach It states that for an elastic body to be in equilibrium, the total virtual work is equal to zero Mathemati

cally, it can be expressed as

3W, + |, 3% dv =0 (4.12)

where 817, = Virtual work due to extemal forces

317, = Virtual work of internal forces per unit volume Now in order to obtain the element stiffness characteristics, 17, and BW, are expressed in terms of force and displacement components, and then substituted in Eq, (4.12) Finally, the clement stiffness matrix is id fied The derivation procedure is straight forward and is given below with~ ‘out much detailed explanation

Let (F} represent a vector of nodal forces corresponding to nodal

displacement vector (A); the virtual work 817, may be given by

3H, = (BNF (F) 413)

where (84) = vector of virtual nodal displacements

‘The virtual work of internal forces per unit volume is given by B= BaF (0) (41) Using Eq (4.5), vector of virtual ed = rain may be expressed as, [8] {BA} 4.19)

‘The stress vector defined by Eq (4.9) can also be expressed in terms of nodal displacements by using Eq (4.5), as

{0} = 14) [8] {4} (4.16) Substituting in Eq (4.14) the values of (8) and (ø} from Eqs (4.15) and (4.16), respectively, we obtain

3/,= ~ (BA)? [BI lạ] (BI (8) (1) Fivally, substituting in Eq (4.12) the values of #, and ðW from Eqs (4.13) and (4.17), respectively, we obtain

Trang 40

‘Methods of Determining Stiftaess Properties and lis Transformation 33,

ear J, ay erate (a) av —0 19)

that the nodal displacement vector is independent of x, y and z inates, we obtain after rearranging the above equation,

cor {o-[f, arma] ap-0 a implies Since virtual displacements are not necessarily equal to zero, that (=1) (420) where the stiffness matrix denoted by [K] is identified as wxi= Í, têP tl18147 đ20

‘Thus, the force-displacement relationship and the stiffness matrix for an element is given by Eqs (4.20) and (4.21), respectively ‘The dimension cf clement stiffness matrix [X] will depend upon the degrees of freedom associated with the clement

44 TRANSFORMATION OF REFERENCE (COORDINATE SYSTEMS

In the preceding chapter, various types of elements were described and it was seen that for one-dimensional elements, local and global reference coordinates are different, whereas for two-dimensional elements, these y Os | |e cea acct en ? onvz:

Định dạng
Số trang	205
Dung lượng	9,77 MB