The Finite Element Mlethot using MATLAB Young UW. won Hyochoong Bang CRC Press Boca Raton 14 igton, D.C.CRC MECHANICAL ENGINEERING SERIES Edited by Frank A. Kulacki, University of Minnesota Published Entropy Generation Minimization Adrian Sajaan. Duke University The Finite Element Method Using MATLAB Young W. Kwon. Naval Pastaraduate School Hyochoang Bang, Korea Aerospace Rasearch Insthiute To be Published Fundamentals of Environmenta! Discharge Modeling Lorin A, Davis, Crogan Stato University Mechanics ot Composite Materials Autar K. Kaw, University of South Florida, Norlinear Analysis of Structures M. Sathyamoarthy, Clarkson University Mechanics of Solids and Shots Gerald Wempter, Georgia Institute of Technology Viscoelastic Sottas Roderic Lakes, University of lowaLIMITED WARRANTY CRC Press warants the ptysical disketto(s) enclosed herein to be fee of defects in uateriak ad workmanship for a period of thirty days from the date of purchase. if within the wartanty period CRC Press receives writen nolification of éefects in materials oc workmanship, and sein notilicelion is destined by ORC Pres he come, CRE Fress Wl replace me detective atsectet ‘The ealis and caddusive lability and remedy for becac of dis Linsiled Waray shall Tinted wocpiacement of defective disketie(s} and shall wot nul ur extend to any claien for or right cover any ober damages, achdieg but nottimited 10, lost of prof, data, af use ofthe cofewave, ot special, incidanal, of csnsquential damages or other Amilar claims, even if CRC Press fas been specifically advised of the possibilty af ach damages Fr wo even wil the liabiy of CRC Press for any damages to you or any ofter person ever exceed the lower sugested ist pie 9 cual pfite paid for the software, regardless of any form of the claim. ‘CRC Press specifically disclaine all other weerantics, expeess ce implied, icuding but oi inited 0, any iplisd ‘warranty of metchantabiity 07 fimness for a particular purpose. Specifically, CRC Press makes no presestation o¢ ‘warranty thal the softwave is it for any particular purpose and any ixaplied warranty of mercuabillys bie Te CLIMY-day eoreTion ef the Liraied Warranty covering the physlcel askees) ony (aN Noche SCO) sd otherwise eapressty and specifically disclaimed, ‘Since noe Mules de nol allow the exclusion of invidentaf or consequent! dang on implied warranty last tonsa af the akove may not apply v0 you srthe lization on howe lag. DISCLAIMER OF WARRANTY AND LIMITS OF LIABILITY: Thc ator ofthis book hae ued her best efforts in preparing this material These efforts include the development, rescarch, and testing ofthe theories and programs to determine cheir effectiveness. Neitacr the author(s) nor the putliser make warantes of any kind, express or implicd, with regard to trese programs or the documentation conned i tis book, inculing wont limitation warrants of merchan‘ability oF tiness fora particular purpose. Ne lig Is arcpedin any event for any damages, tncluding incidental ur oonseyuentil Uareages. 109 profls, so of Ws Uae 9 EO MEET, OF otherwise in connection with of arising ot of the furnishing, peformance, o sof the programs in this book, Libcary of Congress Cataloging-In-Publleatien Data Kwon, Young W. The finite clerient method using MATLAB / by Yeang W. Koon and Hyocioong Sane, B. Ga — (The mechanical engincering series) Includes bibliographical references and index, IEDN 0-0493-7652-0 (alls. pepo) 1, Finite clement method—Tiata proesssing, 2, MATLAB — Bang, Hyockoone, TE Tithe IE Series: Advanced topics in mechanical enginesting series. TAMTFSKSS 1995 624.1 70285832090 96-1901 cp ‘This book contains information obtained from anthentc and highly reganled smrees. Reprint malta acted ‘with permission, and sources are indicated. & wie varcty ofreferences are ied, Reacomoblefiptshave been made to publish reticbte deta and information, but the author(s) and the publisher carnot asm respeulty fr Lae saldity of afl materials or for the consequences of their eae ‘Neither this book nor my part muy be reproduced or tansmilted in any for tr by any means, clcroe ot mechanical, inchuding photocopying, miczofilming, and recording, or by any infornatonsiomye ore system, without prior permission in writing from the publisher. ‘The consent of CRC Press LLC does aol extend to copying for general dveibution, for promotion, for erating new works, or for resale. Spesific pecmission must be obtained in writing fem CRC Press LLC for such copying, Disect al inquiries «9 CRC Press LLC, 2000 Comorate Bivd. N.W., Boca Reto, Horia 3431 © 1997 by CRC Pros LLC No elim tn original 18 Couersent work: Inreenational Standaid Book Number 0-8493-9653-0 Library of Congress Card Number 96-9091 Printed in the United Sites of America 34567890 Printed on scid-free paperTABLE OF CONTENTS CHAPTER 1. INTRODUCTION TO MATLAB La 2 13 4 15 La LT 18 19 1.10 LAL 1.12 113 au Lib Finite Elernent Method Overview of the Book About MATLAB Vector and Matrix Manipulations Matrix Functions Data Analysis Funccions Tools for Palynomials Making Complex Numbers Nonlinear Algebraic Equations Solving Differential Equations Loop and Logical Statement Writing Function Subroutines File Manipulation Boosie Input Output Functions Plotting Tools CHAPTER 2. APPROXIMATION TECHNIQUES a4 22 2.3 24 25 26 Ze Methods of Weighted Residual ‘Weak Formulation Piccawise Continuous Trial Function Gelerkin’s Finite Blenent Formulation Variational Method Rayleigh-Rita Method Rayleigh-Kite Finite Element Method Problems CHAPTER 3. FINITE ELEMENT PROGRAMMING BL 32 33 B34 35 CHAPTER 4. DIRECT APPROACH WITH SPRING SYSTEM 41 42 43 Overall Program Structure Input Data Assembly of Element Matrices and Vectors Application of Constraints Example Programs Prablems Linear Spring Axial Member Torsional Member u i 5 16 1 20 2 28 mu ay 34 35 a7 a2 4 a AG Bt 52 54 68 a a Wi 31 bL 144 Other Systems Problems CHAPTER 5. LAPLACE’S AND POISSON’S EQUATIONS 5. 5.2 5.3 54 55 5.6 5.7 58 59 5.1C DLL 5.12 Governing Equation Linear Triangular Element Bilinear Rectangular Elemert Boundary Integral ‘Transient Analysis Time Tategration Technique Axisymmetric Analysis ‘Phree-Dimensionel Analycio MATLAB Application to 2D Steady State Analysis MATLAB Application tu Axisymmetric Analysis MATLAB Application to Transient Analysis MATLAB Application to 3-D Steady State Analysis Problems CHAPTER 6, ISOPARAMETRIC ELEMENTS 61 6.2 63 6.4 6.5 6.6 One-Dimensional Elements Quadrilateral Elements ‘Triangular Elements Gauss Quadrature MATLAB Application to Gauss Quadrature MATLAB Application to Laplace Equation Problems CHAPTER. 7. TRUSS STRUCTURES a 7.2 73 TA 15 7.8 One-Dimensional Truss Plane Truss Space Teuse MATLAB Application to Static Analysis MATLAB Application to Eigenvalue Analysis MATLAB Application to Transient Analysis Problems CHAPTER 8. BEAM AND FRAME STRUCTURES at 8.2 8.4 84 8.5 Euler Bernoulli Boam Timoshenks Beam Ream Elemente With Only Displacement Degrees of Freedom Mixed Beam Element Hybrid Beam Element 7 80 83 86 90 92 96 38 jot 108 107 ng 124 M8 153 wT 160 168 170 1% 183 193 197 199 08 204 15 223 281 295 wz 45 zag 2a 83 187 1987 2358.6 87 8.9 ata BLL 8.12 B18 B14 Composite Beams Two-Dimensional Frame Element ‘Three- Dimensional ¥ramne Element MATLAB Application to Static Analyeis MATLAR Application to Rigenvalna Analysis MATLAB Application to Transient Analysis MATLAB Application to Modal Analysis of Undarpped System MATLAB Application to Modal Analysis of Damped System MATLAB Application to Frequency Response Analysis Problens CHAPTER 9, ELASTICITY PROBLEM g1 2 9.9 94 9.5 2.6 97 a8 3.8 Plane Stress and Plane Strain Force Vector Bnergy Method Three-Dimensional Solid Axisymmetric Solid Dynamic Analysis MATLAB Application to 2-1) Stress Analysis MATLAB Application to Axisymmetric Analysis MATLAB Application to 3-D Stress Analysia Problems CHAPTER 10. PLATE STRUCTURES 04 10.2 10.3 10.4 10.5 10.6 10.7 Classical Plate Theory Classical Plate Bending Element Shear Deformable Plate Element Plate Element With Displacernent Degrees of Freedom Mixed Plate Element Hybrid Plate Element MATLAB Application Problems CHAPTER 11, CONTROL OF FLEXIBLE STRUCTURES dl 11.2 3 14 W5 116 17 11.8 11.9 11.10 Introduction Stability Theory Stability of Multiple Degrees of Freedom Systems “Analysis of 2 Second Order System State Space Form Description ‘Transfer Function Analysis Contro) Law Design for State Space Systems Linear Quadratic Regulator Mudal Contre for Secund Order Systems Dynamic Observer 28 9 264 265 0, 284 288 298 29 304 r BL 313 31a 318 320 82L 334 356 36L 364 368 373 378 308 303 304 398 401 408 Atz 427 438 451 307 361 393.PREFACE ‘The finite element method has become one of the must important ead useful engineering (uoly for engineers atid scieutisis. This ook prevents introductory and some advanced topics of the Finite Element Method (FEM). Finite element theories, formulations, and various example programs written in MATLAB! ate presented. The book is written a9 s textbook for upper level undergraduete and loner level greduate courses, as well as a reference book for engineera and scientists who want to write auick finite element analysis programs. Understanding basic program structures of the Finite Element Analysis (FEA) is an important past for better comprehension of the finite element methad, MATLAB ja especially convenient to writo and underatand finite clement analyes programe because a MATLAB program manipulates matrices and vectors with ease, Thess algebraic operations constitute major parts of the FEA program. In adiition, MATLAB has buili-in graphics features lo Lely ceaiers visuslize Ue wumucrical results in Wwo- and/or three-dimensional plots, Graphical presentation of aumericel data is important to interpret the finite element results. Because of thes: benefiie, many examples of fimte element analysis programs are provided in MATLAB, ‘The baok contains extensive illustrative examples of finite element analyses using MATLAB program for most problems discussed in the book. Subroutines (MATLAB functions) are provided in the appendix and a computer diskette which contains all the subroutines and example problems is also provided. Chapter 1 has a brief summary of usefal MATLAB commands which can be used in programming FEA. Readers may refer to MATLAR mammals for additional information, However, this chapter may be a good etait for readers who have no experince with MATLAB Subsequent chapters are presented in a logical order, Chapter 2 diseuseoe the weighted residual method which is used for the formplation of PEA in the remaining Clayvery, Tritially, cumtinuuue iia) fynctions ace used by oblaiu appruxiatale sululious using the weighted residual method. Next, piecewise continuout functions are selected to achieve approximate sohitions. ‘Then, FEM is introduced from the concept of piecewise continuous functions. Finally, classical variational formulations are compared with the weighted residual formulations Chapter 3 shows the basic program structure of FEA using ordinary differential equations for an one-dimensional aystern. MATLAB programs are provided to explain the programming. Both program input and output as well aa internal program structure are fully discussed. A direct, FEM approarh using simple mechasies i presented in Chapter 4. This chapter.presents the basic concept of FEM using an intuitive and physical approach ‘MATLAB is a registered tradomark of The MathWorks, Ine. For additional informationa, contact: Phe MathWorks, Tue 24 Prime Park Way Natick, MA 01760 Phone: (08) 647-100U, fax: (US) 647-7001Finite element formulations for partiel differential oquations are presented in Chapter 5. ‘This chapter explains not only domain integration for computation of the finite element matrices but also boundary integration to compute column vectors Applications of Laplace's equation to two- and three-dimensional domains as well as an axially symmetric domain are presented for both stesdy-slale and transient problems. Chapter 6 shows concepts and programing of isoparametric finite elements Because a complex shape of domain with enrved bourdary can be eceily haedled using isoparametric finite clements, these clements ate very useful and common in FEA Both one-dimensional and two-dimensional isoparametric cements are presented. A numerical technique und ils programusiag concept are aluo divcusved. Ay 8 program example, Laplace’s equation is solved using isoparametric elements, Chapters 7 and 8 discuss truss and frame structures. Static, dynamic, and eigenvalue problema are solved. In addition, one-, two and three-dimensional structures are considered. As a result, coordinate transformation from local to global axes is explained. In particular, various formmlations fot the bea structure are compared; the relative advantages and disadvantages af each are cited. Modeling of laminated beams with embedded cracks is also discussed, Further, Chapter 8 presents the madal analysia and Fast. Fourier Transform Elasticity is studied in chapter 9. Plane stress/strain, axisymmetric and three dimensional problems are included. Both static and dynamic analyses ate presented. The finite clement formulatione are presented in terme of the weighted residual method. However, an energy method is also discussed for comparison. Plate bending is given in Chapter 10. Similar to beam formulations, diferent plate beading formulations are presented for comparison. Finaily, stractural control using FEM is presented in Chapter 11. This chapter ig intended to provide a broad understanding of the basic concepts of control law in conjunction with FEM. Due to limited space, only a few major control theories are presented. It is assumed that readers are already familiar with fundamentals of linear dynamic systems analysis. ‘This book contains more material than can be covered in a one-semester. Thus, materials may be selected depending on course objectives. For an introductory FEM course, Chapters 2 thrangh 9 ara recommended. Depending on the desired course contents, some sections may be deleted. We would like to thank individuals who have contributed to this book. ‘The authors would Ike to express out appreciation to Profeasor Alekandra Vinogtador for reviewing the manuscrips and providing us with many useful suggestions. We are also indebled to the staffs of CRC Press for their professional guidance ia the production of this book. Finally but not lastly the authors sincerely appreciate the lifelong support. and encouragement by their parents. Y.W. Kwon E. C, BangCHAPTER. ONE INTRODUCTION TO MATLAB 1 Finite Element Method In order to analyze an engineering system, « mathematical model is developed to describe the system, While developing the mathematical 1odel, some assumptions are made far cimplifestion Finally, tha gaveming rathernalirsl aepraccinn is dovalepard to describe the behavior of the system. The mathematical expression usually consiste of differential equations and given conditions. These differential eyuativas axe usually very difflaus tu obtain solutions which explain the behavior of the given engineering system. With the advent of high perfor mance computers, it has become possible to solve such differential equations. Various numerical solution techniques have been developed and applied to solve numerous en- gineering problems in order to find their approximate solutions. Especially, the finite element methed has been one of the major numerical solution techniques. One of the major advanteges of the finite element method is that a general purpose computer program can be developed easily to analyze various kinds of problems. In particular, any compley ahape of peahlem damain with prescribed eonditiona can be handled with ease using the finite element method. ‘The finite element method requires division of the problem domain into many subdomains and cach subdomain is called a finite element, Therefore, the problem domain consists of many finite element patches. 1.2 Overview of the Book ‘This book is written as a textbuok for engineering students as well as a reference book for practicing engineers and researchers. The book consists of two parts: theory and program. ‘Therefore, each chapter has initial sections explaining fundamental theories and formulations of the finite element method, and subsequent sections showing examples of finite element programs written in the MATLAB program. The collection of MATLAB function files (i.e. m-files) used in the example programs is summarized in Appendix A and provided in a separate computer disc. 12 Tntraduction to MATLAB Chapter L A brief summary of some of MATLAB commands is provided in the following sections for readers who are nol familiar with ther, Those are the commands which may be used in finite element programs. Especially, the MATLAB commands for matrix operation and solution are most frequently ueed in the programs, For visxal- ination of the finite element solution, some plotting commands are alsa explained. 1,2 Abuut MATLAB MATLAB is an interactive software which has been used recently in various areas of ongincaring and ecientifie applications. Il is not a computer language in the normal, sense but it does most of the work of a computer language. Writing a computer code ja not a straightforward job; typically boring and time consuming for beginuers, One aliractive aspech of MATLAB iy that it ix celatively easy to Jean. Tk is wcitien ow an intuitive basis and it does not require in-deptin knowledge on operational principle of computer programming like compiling and linking in most of other programmsing languages. ‘This could be regarded as a disadvantage smec at prevents users from understanding the basic prineiple in computer prograraming. ‘The interective 1uade of MATLAB may redace computational speed in some applications. ‘The power of MATLAB is represented by the length and simplicity of the code, For example, one page of MATLAB code inay be equivalent to many pages of other compnter langnage source codes. Numerical calculation ix MATLAR uses collections of well written scicntific/mathematical subroutines such as LINPACK and EISPACK. MATLAB provides Graphical User Interface (GUI) as well as tatee-dimensional graphical animation. Ta general, MATLAB is a useful tool for vector and matrix manipuletions. Since the majority of the engincering systems are represented by marrix ard vector equations, we can relieve our workload to a significant extent by using MATLAB, ‘The finite element method is a well defined candidate for whick MATLAB can be very useful as a solution tool. Matrix and vector manipulations are eseential parta in the method. MATLAB provides help menu so that we can type felp command when we need help to figure out a command. The Aelp utility is quite convenient for both beginners and experts. 1.4 Vector and Matrix Manipulations Once we get into the MATLAB, we meet a prompt >> called MATLAB prompt. This prompt receives a user command and processes it providing the ouput on the next Hine. Let us try the following command to define a matrix. >> A= [L,3,552.7,8;0,5,9] ‘Then the ontput appears in the next line as shown below. Az ane: cnnSection 1.4 Vector and Matrix Manipulations 3 ‘Thus, a matrix is entered row by row, and ezch row is separated by the semi-colon(;). Within each row, elements are separated by # space or the comma(,). Commands and variables used in MATLAB are case-sensitive. That is, lower case letters are distinguished from upper ease letters. The size of the matrix is cheeked with >> size(A) ans=3 3 ‘Transpoxe of a matrix In order to find the transpose of matrix A, we type >> a ‘The result is 120 ans= 3 T 3 629 Column or row components MATLAB provides columnwise or rowwise operation of a matrix. The following expression >> Al3) yields which is the third column of matrix A. In addition, er AL) represents the first row of A as ang=i 30 ‘We can also try >> A(1,:)+.4(3,:) as addition of the first and third rows of A with the result ane=1 6 15 Now let us introduce another matrix B as >> B= [3,4,5:6,7,2:8,1,01; ‘Then there seems to be no output on the screen. MATLAB does not prompt output on the sereen wlien an operation ends with the semi-colon(;) at the end. Tf we want by check the Boaalsin again, we simply type >> B ‘The screen output will be4 Introduetion to MATLAB Chapter £ & " mae nae Matrix addition Adding two matrices is straightforward Uke >> C=A4B 47 c=8 14 10 e449 ‘Thus we defined a new matrix C' as cum of the previous two matrices Matix subtraction In order to subtract matrix # from matrix A, we tye >> Ca A-B -2 -1 1 406 -8 2 9 Nate that Cis now a new matrix not the summation of A and 4 any more. Matrix multiplication Similarly, atzix multiplication can be done a8 >> Ca Awe 69 31 11 = 2 6 24 90 30 6 1.5 Matrix Functions Manipulation of matrices is a key feature of the MATLAB functions, MATLAB is a useful tool for matrix and vector manipulations. Collections of representative MATLAB matrix functions are listed in Table 1.5.1. Examples an¢ detailed explana- tions are provided for each function below. Matrix inverse The inverse of a matrix is as simple as >> inv(A) 408 1.8571 0.4286 —0.8071 0.8571 6.4288 = 0.1905, 0.287 0.1429 0.0476 In order to verify the answer, we can try >> Asinv(A);Section 1.5 Matrix Functions 5 Table 1.5.1. Basic Matrix Rimesions Symbol. Explanation: inv inverse of a matrix det determinant of a matrix rank rank of a matrix cond condition number of a matrix eye(n) the n by m identity matrix trace summation of diagonal elements of a matrix zeros(nym) the n by m matrix consisting of all zeros which should be a 3 by 3 identity matrix. Determinant of a matrix >> d =det(A) produces the determinant of the matrix A. ‘That is, asm Rank af a matrix The rank of a matriv A, which is the number of independent rows or columns, is obtained from prank); Identity matrix >> eye(3) yidlde Loo ans=0 10 ool eyo(n) produces au identity matrix of size n by m. This command is useful when we Initialize a satrix. Matrix of random numbers A matrix consisting of random numbers can be generated using the following MATLAB function >>rand(3,3) 0.2190 0.6793 0.5194 ans= 0.0470 0.9347 0.8310 0.6789 0.3835 0.03466 Introduetion to MATLAB Chapter 1 That is, rand(3,3) produces a 3 by 3 matrix whose elements consist of random nuuthers, The general usage is rand(n, mm). trace Summation of diagonal elements of a matrix can be obtained using the trace operator. For example, >> Call 39, 67% 8 -1 —4 Then, trace(G) produces 6, which is the sum of disgonal elements of C. axe umatrin >> aeros(S, 4) produces a 5 by 4 matrix consisting of all sero elements. In general, zetos(n, m) is used for an n by m zero matrix. condition number The command cond(A) is used to calculate the condision number of a matrix A. ‘The condition umber represents the degree of singularity of a matrix, An identity matrix ha a condition number of unity, and the condition number of a singular matrix is infinity. >>cond(ege(6)) ans = ! An example matrix which is near singular is 14 a= [i 1+ 10] The condition number is >>eond(A) 4000004006 Further matrix functions are presented in Table 1.5.2. They do not inelide all matrix fonctions of the MATLAB, but represent only a part of the whole MATLAB functions. Readers can use the MATLAB Reference’s Guide or help command to cheele when they need more MATLAB functions. Matrix exponential The expma(A) produces the exponential cf a mabrix A. Ia other words, >> A srand(3, 3) 0.2190 0.8793 6.5194 A= 0.0470 0.9347 0.8310 0.6789 0.3835 0.0346Section 1.5 Matrix Funetions 7 ‘Table 1.5.2 Basic Matrix Functions (Gonnued) Symbol Explanations expm exponential of w matrix —_ eig eigenvalues/eigenvectors of a matrix la LU decomposition of a matrix ovd singular volue decomposition af a matrix QR decomposition of a masrix used Wo solve a set of linear algebraic eq >>expm(A) 1.2448 0.0305 0.6196 ans = 1.0376 L5L16 13380 1.0187 0.1184 2.0652 Bigenvalues The cigeuvalue proviern uf « umria is defined os Aga dd where ) is the eigenvalue of matrix A, and ¢ is the associated sigenvector. vy e selg() gives the eigenvalues of A, and set) produccs V matrix, whose columns are eigenvectors, and the diagenal matrix D whose valuos are eigenvalues of the matrix A. For example, p> Aa[6 3% 146; 07 >> [V,D] seig(A) 04197 a.5999 9.0889 V = 0.5657 —0.7773 —0.6388 0.7217 O.1918 0.7680 >> [KD] 12.5361 0 0 D o (17486 0 4 Q —a.9R47 LU Decomposition ‘The LY decomposition command is used to decompose a mesrix into a combination of upper and lower triangular matrices, respectively. p>Aa[L 35 248 47 3h >> [L,U] =Iul A)8 Introduction to MATLAB Chapter 1 9.2500 1.0000 0 = 0.6600 6.4000 1.0000 1.0000 0 0 4.0000 7.0000 3.0000 = 0 1.2500 4.2500 a 0 4.8000 In order to check the result, we try >> LeU 136 ans=|2 4 8 473 ‘Phe lower triangular matrix Fie not perfectly triangular, There is another command available >> [FH P| =Int A) 1.6006 0 0 0.2600 1.0000 8 0.5000 9.4000 1.0000 4.9000 7.0000 3.9000 L = 0 12800 4.2500 0 0 4.8000 oon Pol 00 o10 Here, the matrix P is the permutation matrix anch thet ?+A= LeU. Singular value decomposition The evd command is used for singular value decompcsition of a matrix. For a given matrix, A=uEy’ where Dis a diagonal mattix consisting of non-negative values. For example, we define a matrix D like >> Dall 37 29 %& 28 5] ‘The singular value decomposition of the matrix is >> [V, Sigma, V] =evd(D) which results in 0.4295 0.8998 | - 0.0775 U = | 0.6629 0.6495, 0.6133 0.7564Section 1.6 Matrix Functions 9 1s.6a92 oD 0 Sigma = [ 0 4.1333 0 | 9 0 0.1891 U.7771 -0.5982 -0.1956 0.1905 -0.0725 9.9790 v= 0.5999 0.7980 —0.0576 QR decompouit! A matrix can be olso decompocad inte combination of an orthonormal matrix and an upper triangular matrix. 1 other words, AZ=QE where Q is the matrix with orthonormal columns, and 2 is the upper triangulac matrix, The Q2 algorithm has wide applications im the analysis of matrices and associated linear systems. For example, 0.0470 0.9347 9.8310 0.2190 9.6793 0.5194 A= [soma 0.3835 9.0346 Application of the ar operator follows as >> [@, 4] =arfA) yields 0.3063 -0.4667 0.8297 Q= | -0.0858 -0.8591 0.5076 0.0407 0.9101 0.9394 R= 6 1.0205 —0.9490 0.7149 -0.6338 —0.2456 8 0 —0.0011 Solution of linear equations ‘The solution of a linear system of equations is frequently needed ia the finite element method. The typical form of a linear gystem of algebraic equations is written as Azzy and the solution is obtained by >> ny (cA) or we can use \ sign as >>as AW For example p> Aa=|l3 4 5 7 8 23 ot10 Introduction to MATLAB Chapter 1 Table 1.0.1 Dasa Analysis Functions Explanations min( max) minimurn(maximurn) of a vertor sum aum of elements of « vector std standard deviation of a data collertion sort sort the claments of vector mean mean value of a vector used for componentwise operation of a vecior and >> y= [10; 9 a]; Let us compare two different approaches. >> [inv(A)ey Aud —4,2500 —4.2500 ans= 1.750C 1.7500 2.2500 2.2500 1.8 Data Analysis Functions Tip to now, wa diseased matrix related functions and operstara of MATLAB, MA'TLAG hse also data analysia functions for a vector or a columa of a matrix. Ia Table 1.6.1, come operators for data manipulation are listed Minimum (maximum) The min (mex) finds a minimum (maximern) valve of a given vector. Hor example, >> vail 28 78 26 49 92 28 23); >>min(u) yields ans = i >>max(v) ans = 99 sum The sum command produces the summation of elements of a vector. For example,Section 1.6 Data Analysis Functions ul Standard deviation The std command calculates the standard deviation of a vector, For example, >> std(l 4 0 -5 6 9 ~20)) ans = 10.4517 Sort a vector The sort command js used to sort a vector in the ascending order. >> sorti[l 4 10 -5 6 9 —20) ans = -0-51469 19 Moai valuc ofa vecton The uteau valvulates Ure sean yelus uf « veciun >> mean([I 410 -5 6 9 —20)) ans = 0.7143 Vector componentwise operation Let us define two vectors >> v= [1,5,6,75u2 = 0,235); Sometimes we want to multiply components of v; with the corresponding components of ug. ‘The operation is D> Ua = Pde ans = om 18 9% In other words, ( ).*{ ) represents the componeatwise multiplication of two vectors Another useful operator ie D> te = te/r with a= 0 04 05 0.7143 Note that the deta analysie tools explained in the above are applicable to matrices too. Pach matrix column is regarded 5 a vector for data analysis.12 Introduction to MATLAB Chapter 1 Table 4.7.1 Polynomial Functions Explanations converte collection of roots into = polynomial equation finds the roots of a polynorrial equation evaluates a polynomial for a given value multiply two polynomials decompose a polynomial into a dividend and a residual curve Biting of 2 given polyaomial 1.7 Tools for Polynomials Polynomials are frequently used in the analysis of linear systems. MATLAB provide some tools for handliag polynomials. The sumriary of polynomial funetions is provided in Table 1.7.1. Roots of a polynomial equation A polynomial equation is given by 12" fae" +o pant ang =0 The roots of the polynomial equation is solved using roats command roots([a1 az --- an aa4i]} For example, zt 442° —527+62—9 >>roots([} 4 ~—5 6 —9]) yields 5.23564 1.2008 0.0178 1.1963% 0.0178 - 1.19034 Generation of a polynamial equation nsing enats The poly commend takes the roots, and converts them inio a polynomial equation. Por instance, if we know Tris ran ota) in (erie ra) fer} = 2h beget + age pe tena ton thenSection 1.7 Tools for Polynomials a >ppoly([ri, %) -°>, tal) provides us the coefficients{ [a1,a2,--+, dn]) of the polynomial equation. Fer exaznple, >ppoly([-1 —24+24¢ —2—248 —547«¢ —5—T+i]) produces ans = i o15 196 498 968 §92 In order to check the result, we use roots command again. >>roots([! 15 146 498 968 592)) ‘The result should be [-1 —24+2*¢ —2~2ei -5+i#i -5-7ed. Polysumial value When we want ty calculate the value uf a polynomial ab a certain point, we can use polyval. >> yspolyval((1 34 -5 19) ans = 23 which evaluates the polynomial 6? + 367 4 4s -— 5 at Multiplication of two polynomials ‘he conv command is used to muliply two polynomials. For example, a(s)= se? 44s—1, (a) 2s? + 60-7 ‘The multiplication of a(s) and 6(s) follows as >>excomv(l 3-1, -26 -7) 130-27 7 In other words, we cbtain the coefficient vector c of the product of a/s) and 6(s) ion of a polynomial The deconv is used to decampose a polynomial! ae a multiplicend and s residue. Let a{s) = He)m(«) + r(s) That ia, the polynomial a(3} is reprosonted in terms of a nmltiplicand fa} and « residue r(s) via. B(1). The MATLAB command ‘s >> bm,7] =deconv(a,}) where the parameters are ccefficient vectors for given polynomials. An example is given by >» [m,7] =deconv(t -2 6 -7,[1 3-1) m=14 Introduction to MATLAB Chapter 1 1 6 r= 0 0 22 -12 If we change the order of polynomials, >> [m.r]=deconv({1 3 —1},[1 -2 6 -7) m= Polynomial fit The poly fit command ip to generate » polynomial carve which fits 8 given sel of data, The polynomial is obtained by minimizing the etror between the polynomial and the given data set. The synopsis is p =polyfit(z, y,2) where 2 and y are vectors of the given data cet in (2, y) form, and m is the order of the desired polynomial to fit the data set. The ouiput result is p, the coefficient vecter of ‘the fitting polynomial, An example is provided below, praes[l 2345 6); >>ys[-1 352 -3 1) >> p =polyfit(z,y,1) p= 21714 -1.268T A linear curve fitting is performed for data set (z,x). 1.8 Making Complex Numbers In order to make a complex number 243 ¢ i, we use ep D4 Bel or >> 2+ aed MATLAB tokes 4 and j as a pure complex number. In ctse# or j is defined already, >> i ssqrt(—1) #=0+ 1.0000Section 1.0 Nonlinear Algebraic Equations 15 ‘Kable 1.9.1 Functions for Nonunear Algeprate Equarions Symbol Explanations fmin finds minimum of » function of one vsrisble frero_| solves anonlinear algebraic equation of one variable abs, angle For a given coraplex number, we use abs and angle commands to find out the magnitude (abs) and phase angle (angle) of the given complex number. For example, if Does lth then >> abs(e} ang = 14142 >> angle(e) ang = 2.3562 Real, imaginary parts of a complex number The real and imag are used to toke the real and imaginary parts of a eomplen aumber, For oxample, W494i >> freak) ,imag(c)] ans =—10 9 ppe= Conjugate The conj command is used to generate a complex conjugate number. For example >> conj(-1+5*#) ans = 1-544 1.9 Nonlinear Algebraic Equations Nonlinear algebraic equations are frequently adopted in many different. areas, The nonlinear equations are different from linear equations, and there is no unique analysis too! to the nonlinear equations. MATLAB is equipped with some functions which can handle nonlinear aquations The list is precented in Tabla 1.01 Minimum of a function The MATLAD command fain minimises 9 funetion by finding out 4 value which minimizes the given function. The synopsis is Hmin(' func’, 22)18 Introduction lo MATLAB Chapter 1 ‘Table 1.10.1 Numerical Techniques for Ditferential Equations Explanations ede23 | solution using the 2nd/Srd order Runge-Kutta algorithm odet5 solution using the 4th/5th order Runge-Kutta algorithm: where “fune’ is the name of function to be minimised and #,(z2) represents & lower(upper) limit of the interval of the function argument. For example, >pfmin('z + 208(2)', -2,2) produces ana = —0.8603 Solution of a nonlinear algebraic equation When a nonlinear algebraic equation is written as 0 the MATLAB function frero can be used to find solution of the nonlinear algebraic equation. ‘The synopsis is 9434 sin{z) — >> sol =faero(’ function’, 20) where ‘function’ ic a MATLAB function subroutine and #0 is an initial condition veclur of the vatiables. For the given example, we write a function aubroutine fotn.w as fanetion [f] =fetn(2); foad4 2 + cos(z) — 42; Then, ne use fzero command 48 >> sol =frcro(! fetn’,—5) sol = 984 In order to hack the solution >>fotm(—2.1981) ans = 1.9198e-004 The error is due to the numerical dormat error, The number is trunested for screen display purpose, even if it is calculated using dowble precision format inside MATLAB, 1.10 Solving Differential Equations Linear and nonlinear differential equations can be also solved using MATLAB. List of numerical techniques solving differential equations is in ‘lable 1.10.1.Section 1.11 Loop and Logical Statement wv Runge-Kutta second and third order algorithm MATLAB uses the Runge- Kutta algorithm to sclve a differential equation or a set of differential equations. The general synopsis is [é,2] = ode23(' func ,t0, tf, x0) where *fune’ is a function containing the derivative information, 10 (tf) is the initial (final) time, and 20 is an initial condition vector. The outputs are i, which eontaine the returned time points, and x which is the integrated output For oxample, we want. te oolve H+ sin(z) =0 which can be rewritten as fay fq = —sin() where x, = @ and #z =#. The ’fune’ function should be provided as an indpendent function subroutine ax faxc.m in a directory, which MATLAB can locate. Now we execute the odezs command >> [i,2] sode23(' func’, 0, 10, 20); o-{3) is an initial condition and furc.m is provided as Sumctbucs [f] =faue(x); eros(2,1); where Rumge-Kutta Bourth and Pifth arder algorithm There ic snather Runge Katia algorithm ode45 which is more accurate than ode23. [t, 2] = edeas(’ func’, 10, ¢/,20) ‘The came calling synopsis as ode23 can be applicd to make uso of the odedS function. 1.11 Loop and Logical Statement. ‘Phere ove sume logical slatcments available in MATLAB which help us writing combinations of MATLAB commands. Furthermore, loop commands can be used as in other programming languages. In fact, we can duplicate the majority of existing programs using MATLAB commands, which significantly teduces thesive of the source18 Introduction to MATLAB Chapter L al Statements Table da. Loup aud © Symbol Explanatior for Joop command similar to other languages while used for a Joop combined with conditional statement if produces a conditional statement elsoif, clea used in conjunction with if command break breaks a loop when a condition is satisfied codes. A collection of loop and logical statements in MATLAB is presented in Table LIL, for loop The for is a loop command which eads with end command, In the above example, iis a loop index which starts from 1 and ends at 100. There may be oloo multiple Ioopa. >> for i 100 for j=1:50 for k= 1:60 ) = AGA) weld) ta end while The while command is useful for an infinite loop in conjunction with a conditional statement. The general synopsis for while command is a8 follows while condition statements end For example, 1 while (5 < 100) gait; end Another example of the while command is = 1000; var =];Section 1.11 Loop and Logical Statement 19 while (n > 0) n=nft—1; var = (ear, es end ‘The result ic vat = Cohumns 1 throngh 6 4,9900e+002 2.4850c+002 1.2325e4002 §.0625e+4001 2.9313e4001 1.3856e+001 Columns 7 through 9 5.8281e+000 1.914Je+000 -4.2969e-002 where we used [ ] iu order vo dedare au exupty matrix, if, elseif, else The if, elseif, and else commands are conditional statemenis which are used in combination. if condition #1 staternent #1 elseif condition #2 stuteruent #2 else atoternent #3 end For example, n= 10; if (rem(n,3) == 0) 2=0; elseif (rem(n,3) == 1) eal where rem(x,y) is used to calculate the temainder of z diviced by y. break The break command is used to exit from a loop such as if and while, For cxample, if(¢== 10) break; end end20 Introduction to MATLAB Chapter 1 Table 1.11.2 Loop oud Logical Statearenes Symbol Explanations twa conditions are equal = two conditions are not equal <=( >=) | oneis less (greater) than or equal to the other fone ia Teas (greater) than the other & and operator - two conditions are met not operalor I or operaice - either one condition is met Logical and relational operators The logical and relational operators of MATLAB are as listed in ‘Table 1.11.2. The above command sets are used in combination, 1.12 Writing Function Subroutines MATLAB provides a convenient tool, by which we can write a pregram using collections of MATLAB commands, ‘This approach is similar to other common programming languages. It is quite useful especially when we write a series of MATLAB commande in a text file. Thio text file ic edited and omved for lator noc ‘The text file should have fileneme.m format normally called m-jile. That is, all MATLAB subroutines should end with .m extension, so thai MATLAB recognizes them as MATLAB compatible files, The general procedure is to make a text Mle using any text editor. If we generate a file called func.m, then the file funed.ma should start with the following file header fanetion[ou ,ovg,...] = func (ins,ivp, where iu,,it,... ate input variables while ovy,oo2,... are output variables. The input variables are specific variables and the output variables are dummy variables, for which we can use any varinbles. For example, let us solve a second order algebraic equation. az? +br+e=0 ‘The sulution is given in analytical form as abt VPN Teaee QeSection 1.12 Writing Function Subroutines a We want to write an m-file with the name secrvot.m, which produces the analytical solution. fimetion [r1,r2]=secroot(a,t,¢); % % Pind Determinant —— Any command in MATLAB which starts with % % sign is a comment statement Det = 02 —4waxc; if (Det <0), baja agrt(—Det\/Ifa b= jx egrt(—Det})/2/a; disp('The two roots are complex conjugates’); disp('There are two repeated roots‘); dlse(Det > 0) Bt agré(Det))/2/a; —b— sqrélDet))/2/a; disp(‘The two roots are real’); end Some commands appearing in the above example will be discussed later. Once the secroot.m is created, we call that function as >> [Pl r2]=seeroot(3, 4,5) or >> Ppl. p¥] Ssecroot(3, 4,5) One thing important about the function command is to set up the mefle pathname. The mafile should he in th ditectory which is set up by MATLAB configuration set up slage. In the recent version of MATLAB, the set up procedure is relatively cacy by aimply adding a directory which wo want to accoea in a MATLAB configuration file. Another function subroutine fct.m is provided belov. function [f] =fet(x) fa 272) ‘The above function represents f(z) = (1—)®. In the MATLAB command prompt, we call the function as =fct(9); The function subroutine utility of MATLAB allows users to write their own subrou- tines. It provides flexibility of developing programs using MATLAB. >>22 Introduction to MATLAB Chapter 1 ‘Tablo 1.19.1 File Manipolation Commands. Symbol Explanations save gave current variables in a file load Joad a saved file into MATLAB environment diary save screen display output in Lext format 1.13 Pile Manipulation Manipulating files is another attractive feature of MATLAB. We can save MATLAB workspace, that is, all variables used, in a binary file format and/or text file format. The saved file can be also reloaded in case we need it Inter on. The list of fle manipulation commands is presented in Table 1.13.1. sava The eave command is naed to save variahles when we are working in MATLAB. ‘The synopsis is aa follows gave filename wary vers os where jilename is the filename and we want to save the variables, var, vary, .... The filename generated by save command has extension of -tmai, called a mat-file. If we do not include the variables name, then all current variables are saved automatically. In case we Want to save the variables in a standard text format, we use save filename vary vurg ..fascii/doutte Joad The load command is the counter-part of save. In other words, it reloads Uhe variables in the file which was generated by save command. The synopsis is ax follows load filename var: varz where filename is matyfile saved by save command, Without the variables name specified, all variables are loaded. For example, ppas[i 3 4}; >> b= 3; >>save test >pelear all % clear sll variables >>who ‘% display current variablea being used >>load test >>who ary Using diary comand, we cm capture all MATLAB teats including command. and answer lines which are displayed on the screen. The vexts will be saved in a file, 80 that we can edit the file later. For example, >odiary onSection 1.14 Basic Input/Output Functions 23 Table 1.14.1 Inpul-Outpal. Kunetions Symbol Explanations input save current variables ina file disp load a saved file into MATLAB environment format check the file status in the directory poashbadies pola be] >> daanh presgth >>diary off Now Wwe can use any 1exXt editor to modify ihe #tary Me, ‘The diary command 's useful displaying the past. work procedures. Also, it can be used to save datn in text format, 1.14 Basic Input-Output. Functions Input/output functions in MATLAB provide MATLAB users a friendly pro- gramming environment. Some input/output functions ate listed in Teble 1.14.1. imput The input command is used to receive a user input from the keyboard. Both pumerical and string inputs ars available. For oxample, >> age= imput('Houw Old are you?’) >> name =input('What is your name’, s') ‘The ’s? sign denotes the input type is string. disp ‘The disp command dispiays a string of text or numerical values on the screen. It is useful when we write a function subroutine in a user-friendly manner. For example, pe disp(‘This is a MATLAB tutorial’) >> disp('‘The computed value of ¢ turns owt 10 bc) pre format ‘The format cormand is used to display narsbers in different formats, MATLAB calculates floating numbers in the double precision mode. We do uot want to, in some situations, display the numbers in the double precision format on the sereen, For a display purpose, MATLAB provides following different forrnats ppr=l/o24 Introduction to MATLAB Chapter 1 Table 125-1 Mowing Commands Symbol Explanations plot basic plot command xlabel{ylabel) attach label to x(y) axis axis meanuelly scale x and y axes tont place s toxt an the spacific pasition of graphic eeroen title place a graphic title on top of the graphic ginput produce a coordinate of a point on the graphie screen gtoxt reteives a text fram mouse input grid add a grid mark to the graphic window pause hold graphic acroen until keyhonrd ia hit subplot breaks a graphic window into multiple windows x= O11 >>format short ¢ z= Lille—001 >>format long w= DPIILHI1LI1LI1 >aformat long ¢ AULLLU11101 Le — 001 pofurmat hex SfbeTe7cPle7le 1.15 Plotting Tools MATLAB supports some plotting tools, by which we cen display the data in a desired format. The plotting in MATLAB is relatively cnay with various options available. The collection of plotting commands is listed in Table 1.15.1 A sample plotting command is shown below. 0.4: 105 y= sin(t); >> plot(y) >> title plotty))Section 1.16 Plotting Tools 25 ‘The resultant plot is presented at the top of Fig. 1.16.1. >> b= 0:01:10; >> y= sin(t); >> plot, y) >> title('plat(y)’) The resultant plot is presented at the bottom of Fig. 1.15.1. In the above example, += 0: OL : 10 represents a vector ¢ which starts from 0 and ends at 10 with an interval of 0.1. We can use just y or both y and ¢ together. In the tirst case, the horizontal axis represents number of data, from 0 to 103. In the second case, the horizontal axis is the actual time scale ¢ in the plot(t,y) command. Plotting inultiple data We plot multiple data sets as shown below. 0:4: 100; sin(t). #4 ros(t) ets >> plot(tiyd, 4492," ) where ‘-’ and "—’ represent line styles. The line styles, line marks, and colors are listed in Table 1.15.2. For example, if we want to plot a data in a dashed blue line, the command becomes >> plot(y,’4 xdabel, ylabel The xlabel('text’) end ylabel(‘text’) are used to label the and y axia The axis command sets up the limits of axes. ‘The eynopsis is aaxishitmins Zmae: Ymins trae) text The text command is used to write a text on the graphic window at a designated point. The synopsis is text(s,y,{text contents’) ‘where 2, y locates the 2, ¥ position of the ‘text conzenzs'. In case we want 1 aormalize the graphic window size fom 0 to 1, we use toxt(2,ytext contents / se’) input This command allows us to pick up any point on a graphic window. The synopsis is (t,9) = ginput28 Introduction to MATLAB ol as} 04} oat pe potty) oe on} os oa} aa a9} 4, Figure 1.19.1 A Sample Plot Chapter 1 We can pick as many points as we want on the graphic screen, ‘The welor [2,9] then contains all the peints. gtext The gtext command is used to place a text on the graphic window using the mouse input. The synopsis is gtext(‘teat')Section 1.15 Plotting Tools x Table 1.26.2 Line, Mark, aud Golor Styles Style Line marks Color solid point. rel or dashed = *~’ star * aren dotted "2 cirde 0 blue b dackdot 7." plus 4 white _ x-tnark invisible i Once the above command is entered or read in a function subroutine, the cursor on the graphic window is activated waiting for the mouse input, #0 that the “tar? is located at the point selected by the mause. grid ‘The grid cemmand adda grids to the graphir window It is useful shen we want to clarify axis scales. An example plot constructed using some of the commands described above is presented in Fig. 1.15.2. ‘The following commands are used for the plot oxtput. e> #20:0.1:205 >> plottt, san(t)) >> xlabel('Time(see}’)} >> ylabel(’ydata') >> title('This is a plot example’) >> erid >> gtext('sin(:)') >> axds((0 20 - 1.5 1.5) pause This command is useful when we display multiple grapbie windows sequen tially. It allows us to display one at a time with the keyboard interrupt. subplot The subplot is used to put multiple plots on the same MATLAB figure window, The command is >> subplot(pgr) ‘The plot cise is adjusted by a p by g matrix on the whole-sige of the geaphie window. ‘Then the third index r picks one frame out of the p by q plot frames, Av exanple subplot is presented in Fig. 1.18.3 with the following commands entered. pe eS 0: 01 De piyy = sin(e)jx = c0a(s); >> subplot (222) ve plot(s,y) 3> title('x and y?} >> subplot(223) >> plot(z,2)28 Introduction to MATLAB Chapter 1 ‘Theis a pleteoargla Figure 1.15.2 A Plot Example With Some Commands xewy as| asl x * 7% and os] es| q a oa Figure 1.15.8 A Subplot Example >> title(x and 2’) >> subplot(224) >> plotiz,y/Section 1.15 Plotting ‘Lools >> tith('x and [y 7} where pi is an internally defined variable equivalent to 7.CHAPTER TWO APPROXIMATION TECHNIQUES 2.1 Methods of Weighted Residual Methods of weighted residual are useful to obtain approximate solutions to « differential governing equation, In order to explain the methods, we consider the —, Ojw (419) 0 3 0 2 7 -2| fas] Ju 6 a G Qo -2 te a74 Direct Approach With Spring System Chapter 4 Z 2 le rigid maselear plate w % 2k «eZ [v T ae BE 7 po Tigid massless plate rigid —* massless plate 3 y w | ——rigidt massione plate 5 Figure 4.1.3 A Mass-Spring System The matrix eqnation determines the displucements of the springs. 4.2 Axial Member ‘The Linear spring can represent vasious systems in ongineczing applications. Onc direct application is the axial member. Consider an axial member with length 2, nniform crosi- section A apd elastic modulus £. The elongation § of the axid meraber subjected to an axial force P is computed from PL aE (4.2.1) Rewriting Raq. (4.2.1) gives P= os (42.3) As a reault, che cquivalen: spring coustapt for the axial bar ie hea — Ag (42.3)Section 4.2 Auual Member cy Ark, ky = fe yt aa ——P 4 uy, LL, us ‘ Figure 4.2.1 Axiel Members Represented by Pauivelont. Springs Example 4.2.1 Axial members can be represented by serial and/or parallel Linear springs. For example, a bar of telescoped shape can be replaced by a sories of springs as shown in Fig. 4.21. There are three Sinear spring clemanta and each clement haa malvie expression ax shorn ix Pq (444) with proper spring constant &;. Assembling them gives hy why 0 0 ts fi kL with, kp 0 w{_jo 0 kh krtks —ks| Jus (=) o (424) oO oO fg hs us P where ky == APE! and f; io the unknown reaction force at the Ieft en! support, Inotead, u; = Uis given as the boundary condition. Solution of Hq. (42.4) with this boundary condition will result in fy = ~P which can be also obtained from the static equilibrinm. However, the finite element firmlation zlready includes equations of static equilibrium 90 that we may not ure additional equilibrium equations. These equations are redundant. & Bxample 4.2.2 Consider a statically indeterminaut syotcan 1m shown in Pig. 4.2.2. The axial member cam be replaced by two fincat oprings a8 secn in the figure. The flue element matrix equation for this systema is sk 05k 0 wy fi D5 15k Kk] dug b= 40 (42.5) oo -& | lee b where k = 42 aad f; and fy are the anknown forces at the supperts. Because the system is statically indeterminate, we cannot find the forces fem the siatic76 Direct Approach With Spring System Chapter 4 Z AE Z PF Z, L 2L E AE ey TE op ee 4E 6 on ate b= as aL ZL Pigure 4.2.2 A Statically Indeterminate Axial Member “ i, Rigid Plate Figure 4.2.8 A Spring Representing Statically Indeterminate System equilibrium equation only. However, the finite element formulation includes net only equilibrium but also geometric compatibility (compatibility of deformation} ‘Therefore, Eq. (4.2.5) along with boundary conditions ty = tis = @ can salve the deformation as well as the reaction forces. Another statically indetertainate oyetemt anade of axial sods can he repreccated ‘by a opting constant os shown in Fig. 4.2.3. 1Section 4.4 Other Systems 7 Figure 4.3.1 Torsional Members 4.3 Torsional Member A cirenlar rod subjected to a twisting momen produses an eugle of twist TL a 3.) in which @ is the angle of twist, T'is the applied torque, Lis he length of the member, G is the ahear modulus, aad J is the polar moment of inertia of the circular cra section. Rewriting the equation yields 39 ‘The equivalent spring constant is key = $4. Torque T corresponds to the spring farco F and angle of twist 9 to the spring displacement u. As examples, both statically determinate and indeterminate torsional members are shown in Fig. 4.3.1 along with their equivalent spring systems. 4.4 Other Systems Other frequently used engineering systems, which can be easily substituted with a spring systern, are heat conduction, siraple fluid flow along pipes and electric cir- cuits. For one-dimensional heat conduction, leat Hux is proportional to temperature78 Direct Approach With Spring System Chapter 4 _ fi, — ah ky i fe Figure 4.4.1 Mechanical Forces From Newton’s ‘Third Law difference. Phat ia, hoat flux gia q=-kAT (44) where k, is the heat conduction coefficient and AT is the temperature differences between two end points of a one-dimensional bar. The minus sign it: the equation denotes heat flux is in the opposite direction as the temperature increases along the positive axis. ‘The equivalent spring system has the spring constant keg — ~hz, the spring force ? = q and the spring displacement w =P. Fluid flow rate Uhrough a pipe of constant cross-section is proportional to the pressure difference of the two ends. That is, Q=-kAp (44a) Here, @ is the Dow rate, Ap is Uhe pressure difference, and &y is the proportional constant. For a laminar flow, the proportional constant can be expressed as (44.3) where p is the fluid viscocity, d is the pipe diameter, and J is the pipe length. The equivalent spring system bas the spring constant kg = —Ky, the spring fore: F = Q and the spring displacement w = p. Flectric current flow # through a resistance Ris v ing (4.4.4) where V is voltage. The equivalent spring system has pring constant keg = $, force F = 4, and displacement w= V. One may think what about ¥ =i R such that POV, beg = Rand w— i. Which oue is concet? Te dhe oquivalent wpeing wastent Hor #? One is right and the other is wrong. One way to select the right form is to understand the nature of spring force and to find the parameter equivalent to the force. Then, the zest of parameters can be determined accordingly. When two springs are separated each other, internal forces occur hetween the two aprings. These forres are equal in magnitnde and opposite in direction When the two springs are put together, the forees cancel each ather and became sero if there ia no external force applied at the joint as seen in Fig. 4.4.1. This is known as Newton's third lew. When considering clectric current, ito aum ot the joint of rosiatante (i.0. the middle point in Fig. 4.4.2) is also zero like the force. However, voltage docs not vanish at such a joint. Ap a cesult, electric current is similar to the mechenical force. ‘Therefore, the equivalent spring constant for the electric sirauit is 4 but not X.Section 4.4 Other Systeme ” Figure 4.4.2 Flow Through a Pipe System, Example 4.4.1 Consider a pipe system for finid flow as seen in 4.4.2, The ntessnre at the inlet is 200 while the flow rate at the miet is 50. The proportional constants ky in Eq. (44.2) are given in the fignre, All unite are asgumedi consistent. We want to determine the flow rate between node 3 and node 4. Using the near spring equivalency, the matrix equation becomes -1 1 0 O 0 Oo PL a 1-3 4 -8 0 of ly 0 0 4 -2 2 t¢@ of|tmt_} o o 8 2 -1 1 0 J tm f=) 0 (40) 6 0 16 1-49 32 Po 1 0 6 9 2% ~291 Loe = 60 Applying the known pressure 71 = 200 to this equation yields the following pacomare at cach neds, pz = 160, ps = 142.2, pa= 147.6, ps = 199.6, ps = 188 Therefore, the flow rate between nodes 3 and 4 is Qa-4= 2 x (147.6 — 142.2) = -108 Hence, the flow is unward with rate of 10.8. t30 Direct Approach With Spring System Chapter 4 200 NVA A OANA us Fignra P41 Problem 4.1 Problems 4.1 Find the system of equations for the spring systern shown in Fig. P41. Solve the matrix equation to find the displacements of the nodal points. 4.2 A cizcular shaft is made of two different materials and fixed at both ends. It is subjected to torsions as shown in Fig. P4.2, ‘The diameter of the shaft is 0.1m. Find the angies of twist at the nodal points. 4.3. For the given mass-spring system (see Fig. P4.3), (a) develop the system mass and stiffness matrices to determine natural frequencies of the system. (b) Apply the given boundary conditions to the matrix equations. 4.4 An electric circuit is shown in Fig. P44. Find the current flow using the equivalent spring system. 4.5. Hor a laminar pipe tow, the How rate is proportional to pressare difference Construct the system of equations for the given pipe flow as shown in Tig. P4.5 in order to find the flow rate through each pipe. 4.6 Heat conduction through a circular pipe is expressed as Qrkb 1 Parad in which & is the heat condhiction coefficient, D is the length of the cylinder, and , and ry are the onter and inner radii of the cylinder, respectively. Therefore, two concentric cylinders with radii ry < rg ~< rx vau be represcuted by two springs im serial connection, Find the equivalent spring constants for the two Yinear springs in Fig. P4.$ and construct the system of equations.Problems 8 Fred Diameter = 01 m Element 1 & 2G = 80 GPo Element 3G = 40 GPa Figure P4.z Problem 4.2 rigid bor ig ks ‘y AWM\- oy Figure P43 Problem 4.3 7 10 volts Figure P4.4 Problem 4.482 Direct Approach With Spring System Chapter 4 EER Soren SNe Leer | Figure P40 Problew 46CHAPTER FIVE LAPLACE’S AND POISSON’S EQUATIONS 5.1 Governing Equation Laplace's and Poisson's equations are common field governing equations to de- scribe various physical asturse. For oxamplo, these differertisl oqustions can reprevent heat conduction, potential flow and torsion of noncixcular members. Therefore, we study the finite element formulation of these equations. Laplace's equation is Vuso (5.1. while Poisson's equation is Vusg (5.1.2) Because Poisson's equation is more general than Laplace's equation as seen above, we consider Poisson's equation In the following formatazien, Poisson’s equation in terms of the Cartesian coordinate sysiem becomes au Gat on oy? = ery) int (B13) for the two-dimensional demain Q. The bonndary conditions sre a 6.1.4) and ae =e wl, 61.9 where @ and g denote known variable and flux houndary conditions, aud n in Eg. (5.1.5) is the outward normal unit yeetor at the boundary. Furthenunote, C4 and Ca are boundaries for essential and natural boundary conditions, respectively. For the well-posed boundary value problem, T.Uln=P (618) 83#4 Laplace’s and Poisson’s Equations Chepter 5 and Pent, = 619 in which U and N denote sum ad intersection respectively, and [is the total boundary of the domain 2. Integration of weighted residual of the differential equation and boundary condition is wars r fw Fe F4 ge.) an [ oar (51.8) In order to develop the weak formulation of Ea, (5.1.8), integration by patis is applied to reduce the order of differentiatian within the integral. The subsequent example shows the integration by parts. & Example 5.1.2 Consider a two-dimensional domain as seen in Fig. 5. First of all, we want to evaluate the first term of Eq. (6.18) [8 wan (5.1.9) ‘The domain integral can be expressed as fe yes dy (5.1.10) where 1 and yz are the minimum and maximum values of the domain in the sania an the otrip along the a-nie moves in the ydircction nanecn in Fig. 6.1. Integration by parte with respect bo 2 yields EE Fe eee = [fe] ay GL) Ba cs oe, FL and tewtiting the expression nsing the domain and boundary integrations 35 shown in Fig, 5.1.1 results in dw du au jn Be Be" fac f ujpneat (5.1.19) in which #1, is the s-component of the unit normal vecior which is aswumed to be positive in the outward direction as shown in Fig 511. Fisally combining the two boundary integals gives du du [a + f ofienat 6.113) where the boundary integral is in the counter- clockwise ditectionSection 6.1 Governing Equation 86 Yar ty) SH ar x x x Xo Figure 5.1.1 Two-Dimensional Domain Similarly, the second term in Eq. (6.1.8) can be written as ow Ou f ou [By Ot RT Gad) Adding Bap (54.18) are (6.118) produces au du | dw Gs -f geet 4 Bete an +fo (F ne + am at 6.115) Since the boundary integral can be written as é Bu a an oat (51.18) we can rewrite Eq. (5.1.15) a5 [oe 3+ Fp) = ~ Geet eae) fem (B17) ‘The symbol § to denote the line integral around 1 cloned boundary is replaced by f for simplicity in the following text, Equation (5.1.17) is known as Green's trcorem.86 Laplace's and Poissen’s Equations Chapter 5 Use of Eq. (5.1.17) to Eq. (5.1.8) results in - f, (28 + MB) aa finns [ eltn — 1n9 Phe fink velurue inteyral becomes asuatria term while both Ue second volune integral and the line inbegral become a vector term. In the context of heal conduction, the second volume integral is related to heat source or sink within the domain, and the line integral denotes the heat ux through the nalaral boundary. 5.2 Linear Triangular Element Diseretization of the domain in Eq. (5.1.18) is performed using selected two- dimensional finite elements. One of the simplest two-dimensional elements is the three-noded triangular element. This is also known as linear triangular element. The element is shown in Fig. 5.2.1. It has three nodes at the vertices of the triangle and the variable interpelation within the element is linear ins and y like usa tenet agy (B21) or walle nfs} (52.3) where a; is the constant to be determined. The interpolation function, Bq. (6.2.1), should represent the nodal variables al the three nodal points. Therefor, cubstituting the 2 and y values at each nodal point gives a ta vl (a {eff} 5 i] { node no, and j-> x ory gcoord(1,1)=¢. 9 (58.1)Section 5.9 MATLAB Application to 2-D Steady State 109 25; aeoord(2,2 55 wooord (S,2)=6.8; gcoard(4.1}-=3.75: ecoord(4.2}=0.0: geoard(5,1 wcoord(5,2)=6.0; gcoord(6,1}—0.0; geoord(62}=1.5; seovrd(T1)—1.285 geoord(7.2)—9-55 geoord(8,L}=2.5; geoord(8,2)=2.5 geoard(9,1)=3.75; geoord(9,2) ‘g609rd(10,1)=0.9; Ecvord(10,2)=2.95 geoord(11,1}=6.5; geoord(11 2)=5.0; gcoard(12,1)=1.25; gevord(12,2)=5.0; geoard(14,2)=2.5; gcaord(13,2)=8.0. geoord(14,1}=3.75; gooord(14,2) gcoord(28,1}=8.0; gcoord(8,2)=8.0; gcoard(16,1)}=0.9; gcoord(16,2)=7.5; gcoord(I7,1)=1.25; gooord(17,2)=7. geoord(18,1)=4 5; geoned(12,2)~7.5, gcoord(19,1)}=3.75; gooord( 19,2)=7. geoord(20,1)}=5.9; gcoord(20,2)=7.5; seuurd(22,1}=0.9, gemunl(22,2)=20., gcoard(23,1)=2.5; gcoord(23,2)-=10,; geoord( 24,1}=-3.75; gcoord(24,2)=10. geoord(25,1)}=5.0; gcoord(25.2)=10.; % &—-——_- % impat data for nodal connectivity for each element % noranG) whore iat clement no. and 5% crmnected wocten §—_____________ nodes(1,1)=1; nodea(2,2)=25 nodea(1,3}=7; nodes{2,1}=2; noden(2,2)=9, noden(2,0)=9; nodes(3.3}=9; odes(4,3}=26; nesdes(9,5}=55 nadea(S,1)=1; nodes(s,7, nodan(6,1)=2 nodes(9.1): nodes(20,1 nodes(11,3)=14; 0; nodes(12,3)=15; nodea(17,1}=211; nodes(27.2)=22; nodes(17,3)=27; nodes(18,1}=12; nodes(18,2)=13; nodes(18,3)=18; nodes(18,2)=13; nodes(19,2}=14; nades(19,3)=19;110 Laplace’e and Poisson's Equations Chapter § nodes(22,1)=12; nodea(22.2); nodes(28,1}=18; nodes(23,2)= nodes{24,2}=14; nodes( 24,2) nodeo(26,}a116; nodea(I6,2)=17; nodea(26,2)—225 nodea(26,2}=17; nodes(26,2)=18; nodes(26,3)—23; ngdes(27,1}=18; nodes(27,2)=19; nodes(27,3)=24; nodes(28,1)=19; nodes(z6,2)=20; nodes 28,3)=205 nodes(29,1)=15; nodes(29,2 nodea(30,1)=17; nodes(30,2 nodes(31,1)=18; nodes(#1,2: nodes( 32,1)=19; nodes(32,2; % %— —______. % input data for houndary conditions %— bedof(1}=1; 3% fist node is constrained beval(1)=0; % whose described value is 0 Vedot2}=2, 7% weouad soe is coueaed beval(2)=0; % whowe described value is 6 bodof(3}=3; % third node is constrained beval(3}=0; % whose described value is 0 bedof{d)=4; % Ath node is constrained beval(4)=0; % whose described value is & bedof{S}=5; % Sth node is constrained beval(5)=0; %e whose described valve ia 0 odo (6) E fth nave in comatrainad beval (@)= % whowe described value is 6 bedaf{7)—13; % Mth node ia constrained beval(7}=05 96 whore described value is bedof{8)}=16; % Lith node is constrained beval(s)=0; % whose described value ia © bedo(y}=31; % 2iet node is constimaned % whose described value is 0 ‘% Bad node is constrained 5 whone described value is 38.2683 % Bed node is constrained % whose described value is 70.7107 % 24th node is constrained % whose described value is 92.9880 % Bich node is conetrained, ‘% whose described value 8 100 % initialization of matrices and vectora %——___—_—_____- HE=zeros(sdet,2); % initialization of system force vectorSection 5.8 MATLAB Application to 2D Steady Stete ut Kk=zeroe(adoljadol); % initialization of aynteru matrix index=zeros(nnel*ndof,); lization of index vector % %—— eS % computation of element matrices and vectors and their assembly % — - — for id=ILnel % loop for the total number of elements % nd(1)=nodes(tel!}; 6 tet connected node for (sl)-th element nd(2)=nodes(iel,2); % Ind connected rode for fiel)-th element nd(3)=nodec(iel,3); % 3ed connected node for (el)-th element xi=geaord{nd(3),1); ylexgooord(nd(4),2); _% cverd values of Jet rode x2=gcoord(nd(2),1); ¥2=gcoord(nd(2),2); _ % coord ralaes of 2ad node x3=gcoord ind(3).1): vdsegeoordind(3).2); —% coord valuet of 3rd node % index=feeldof{nd,anel,udof}; ‘% extract system dofs for the element % k-felp2dtS(x1,y1 x2.y2.x3,3), ¥% compute element matrix. % kek—fesambl1 (ck, eamlex); 1% ssromble clement matrices % end % % — % apply boundary conditions % —_—— (kk, f] =feaplyc2(kk, fi, bedof heval); % k——_—- % solve the matrix equation % an faolakk\tt; % % % analytical solution smnode x=gcoord(i,1); s=geoord(i.2}; esol (i) aT 0*sim h (0.91415997%y)Fain( 0.3141 5927*%) /simh(3.1415927}; end ‘% print both exact and fem solutions % store=[unm’ fsal esol’) % %—12 Laplace’s snd Puisson’s Equations Chapter 5 fonction [kk]=feastobll (Ek,k,index) %—_— % Purpose: % Assembly of element: matrices into the spatem matrix % % Synopsis % [kk]=teasmbli(ki,kindex) % % Variable Description: % le - ayatem matrix % k- clement matri % index - Lo. vector associated with an element length (index); iledof HHindan(i); Wi (Lat) = KKH) F449 end ead x Annction [index]=feeldof{ndnnel.ndot) % Purpose: % Compate syatom dofe astociated with exch element % % Synopsis: % [index]=leeldof(nd,nnel,ndof) n % Vasiable Deseription: % index - system dof vector associated with element iei % nd - element rode aumbers whose system dofs are to he determined % anel - number of nodes per element % ndo€ - number of dofa ner node Gow ee % edof — anel¥ndof siart = (nd(i)-1)* alos, for j=tindof kek+1; index(k)=atart+isSection 5.9 MATLAB Application to 2-D Steady State 13 end end % Fonction [k}felp2dea(x1,51.22,92.x3,y3) %—__— 9 Poapores ‘% element matrix for two-dimensional Laplace’s equation % using thyee-node linear triangular element % % Synopsis: % [k]=felp2dta(ict y1x2,y2,x3,y3) % 9% Variable Description; % k- element, stiffuexs matrix (size of 3x3) % x1, yl x and y coordinate values of the first ode of element 9% x2, y2~ x and y coordinate values of the second node of element % xt, 72 x and y coordinate value of the third aod of dement %——— % % element matrix % A=O.5*(x2"pS+x1*y24x3"y]-x2*y 1x1 "5-13" 72), % % area of the triangle (1 1}=( (3-12) *(33-02)4(92-79)¥(y2-y8))/(4FA); (A 2) = ( (2:92 )* fod 43} 4 (92-93 Ey 3-y LATA: Ke(L,3) = (x3-x2)* (2x1) 492-79) Ey 1-72) )/ (AFA: 12,2) —( (aed x8} (ad x3) 4672-7 Pv. ))/ UA: (2,3) = (x1 -x3)* (x21) +{y3-71)* (yd-¥2))/ (AFA) k(3,1)=k(1,3); 4(,2)=K(2,8), (8,3) (2-1 )*(2-x]) Hy) 72)" Ly 2))/ 4A); yO ‘The finite element and analytical solutions are compared below: store = dof # fem sol exact 1.090 9,000 (0.0000 9% at x=0.00 and y=0.0 2.unuo 6.0000 0.0000 % at nat 96 ond y=0.9 3.0000 0.0000 0.0000 ‘% ot X=2.50 and y=0.0 4.0000 0.0000 0.0000 Ye ab x=3,75 and y=0 sui von —_g.9000 9 ac x=5.00 and ¥=0.0 6.000 0.0000 u.0000 ‘% at K=0.00 and y=2.5 7.0000 4.6896 2.8785 % at K=1.35 and y=2.5 a.unno =. 56B9 5.3187 96 at x=2.50 and y=d.514 Figure 5.9.2 Mesh With Bilinear Elemente s.onnn 10.000 ono 12.000 13.000 14.000 15.000 16.000 7.000 18.000 19.090 20.000 21.0uy 22.000 23.000 4.000 25.000 Laplace’s and Poisson's Equations a.au4e 9.0361 o.oou0 10.621 17.312 21.626 23,124 4.0000 iad 7.568 48.108 49.199 4.0000 38.268 T0711 92.386 99.00 & Example 5.9.2 using bilinear rectangular dlements. The mesh is shown in Fig. 5.$.2 and the MATLAB programs we lied below. 9.492 T5218 a.noua 7.6257 14.090 12.410 19,927 9.0000 17.324 32.010 41.823 45.289 0.000 36.288 70.711 oa.ze8 100.00 Chapter 6 Few x=ATS and y52.5 at Kat. 9 ab 20.00 and yo atx % ot x=5.00 ond y=1.5 Mat 0 and y=10, %at % at x=250 and y=i % at nat.78 and y=10. % ab x=.00 ond y=10. We solve the same problem as given in Beatnple 5.9.1 SS % BX$.9.2.mSection 5.9 MATLAB Application to 2.D Steady State 118 % to solve the two-dimensional Laplace equation given 28 % wx + ayy =00 element no. and j-> connected andes yx — nodes(1,1)=1; nodes(1,2)=2; rades(13 noden(2,1)—2; noden{2,2)—3; nodoa( 2 nodes(3,1}-3; nodes(3,2)—4; nodes(3,3 nodes(4,1)=4; nodes{4,2)=5 nodes(43)=10; nodes(4,4)=9; sundes(5,2)=7; suder(5,3)=125 woden(5.A)=15 13; nodes( 6,4): 4; nodes(7,4}: nodes(8,1)=9; nodes(8,2)=10; nodea(8,3}=15; noceni,4)=14; nodes(9,1)=11; nodes(9,2)—12; nades(9,3)=17; nodes(94} nodes(10,1}=12; nodes(i0,2)=13; nodes(10,3}=18; nodes( 10, nodes(11,1)=13; nodes(11,2)=14; nodes(11,3)}=19, nodes( 11, nnades(12,1)=14; nodeo(12,2)=16; nodes(12.3)=20; nodes(t24)=19, nodea( 12, nndea(13,2}=17: nates! 2; wodes(14,4)}=21; nades(14,1)=17; nodes(14,3} 8; nodes(14,3)=23; way ( ( nadea(15,1)=18; nodeo(16,2)=19; nodeo(15,3)—24; nodes 15,4)23, noden(161}—19; noden(16,2)—20; noden(16,2)—25; noden(16}4)—24; % —$ $$ ‘% imput data for boundary conditions ye bedof(1 % firot node is constrained beval(1 % whore deseribed value in bedof(2} % second rode is constrained beval(2} % whowe described value is 0 bedof(s 96 third node is constrained beval 3 % whos: described value is 0 bodof(4 $6 40h sode in coneixsined Deval(a % whowe described value ie 0 dedof(5) $6 50h code is constrained bevat(s: 9 whom devcsibed value fs 0 bedot(s) % 6th vode is constrained beval(6 % whose deverihed valne ie 0 % 11th ode is constrainedSection 5.9 MATLAB Application to 2-D Steady State ur boval(r)=0; % whose described value is 0 % 16th node is constrained % whose described vaine is 0 % Dat node is constrained % whose described value is 0 ‘% 23nd node is constrained % whose described valne is 38.2683 bedof{11}=33; % 28rd node is constrained beval(11}=70.7107; % whoon devcstbed value ia 70.7107 bodai{t2 % BALA node is constrained beval(12}=92.3880; % whose described value is 92,3880 % 20th node fs constrained % whose described valve is 100 fi=reroa(sdof.1); % initialization of system force vector Ve seroe(edrfindnf): OF initislixation of ayetom menlsie, ina ¥%G initialization of index vector % computation of element matrices and vectors and their asverbly $—=—$ for iel=1: % loop for the total wumber of elements % for ici:mnel % loop for aumber of nodes ner element nd (i)=nodes(ieli}s % extract connected node Sor (iel)-th clement x(2)=gcoord(nd (i) 1); % extract x value of the node vi=gocord(ed(),2)s % onteact 7 value of the sede ead % Neng = x(2)-4(1}; % length of Une element in x-axis yleng = y(4)-7(2); % length of the element in y-axis index=feeldof(nd,nnel, ndof); ‘% extract system dofs for the element % K=felp2ar4(xleng, yleng); % compute element matrix % kk=feanmbl (kk k.index}; % awemuble dement. mabrices % end % 4—___- % apply boundary vouditions ——$——— [kk ff} =feaplye2(kk,ff,balof bevel): %14g Laplace’s and Poisson's Equations Chapter 5 Yee % solve the matrix equation % feol IRA; % *4——_ % analytical solution %——_——_ for i=Imnode x=gcoord{i,1); y=gcoord i,2); esol (i) =1004 sin h(0.31415627*v)*sin( 031415927 *x) sinh (3.141927); end % % % print both exact and fem solutions % Purpors: % clement matrix for two-dimensional Laplace’s equation % using fonr-node bilinear reclangalar element % % Synopsis: % (k}=telp2dr4(aleag,yleng) % % Variable Description: % le - elomont stiffness mat % xleug - clement size in Ub (size of da) % yleng - element size iu Uhe y-axis %— % % elemens matrix % K(1,1}=(aleng*xleng-+yleng*yleng) /(3*xleng*sleng) K(1,2}=Gdeng*sleng-2*ylenu*sleng)/(6*xlene*ylens): K(1,3}—0.5*(1,1); k(1.4)=(vleng* yleng-2*xleng*xleng) /(6*xieng*ylenz); (2,1) ke(1,2); k(2,2)—K(1, 1); k(2,9)—KkC1 4); k(2 A) (1,9); KG1=K(1,3}; ki9,2)=K(2,9}; k(3, (4,1) (2,4); (4,2) =4(2,4}; C4,Section 5.10 MATLAB Application to Axisymmetry 19 ‘The nite element solution ia shown below. ‘The same number of wodes were used for this case as that in the previous example. Hy comparing the two finite clement solotions using either linear triangular elements or bilinear rectangular elements, we see that the rectangmlar elements produced more accurate solntion in the present exam ple_ dof # femsol exact 1.0000 0.0000 0.0000 2.0nn0 enna n.nooo 3.0000 0.0000 0,0000 % at x=2.50 and y=0.0 4.0000 2.0000 0.0000 % at xed.8 and 5.0000 0.0000, 0.0000 Mat 00 and ennin 0.0000 0.0000 ab xf. and 7=2.5 70000 2en8e 2.8 TBS % ab me.I aad y=25 8.0000 4.9683 6.8187 % ab x=2.50 and ya25 9.0000 6.4914 6.9493 % at x=3.75 and 10.000 7.0ze3 7.5218 as 11.000 u.UUeu 0.0090 hae 12.000 7.2530 7,6987 Yat 13.000 13.402 14.090 Mat 14.000 17.510 18.410 Rat 15.000 18.953 19.927 % at 16.000 0.0000 0.0000 ae 17.000 16.878 17.324 $i ob RHLI5 ond y=75 iso 71.189 2.010 % ak LEO and y~75 39.090 40.742 41.823 Hat 20,000 44.058, 45.269 Mat zi.oee 9.0000 a.av00 wae 22.000 38.268 38.268 Meat 23.000 70.711 70.711 hab 24.000 92.388 92.3688 ‘Mat 25.000 100.00 100.00 ‘% at x=8.00 and y=! 5.10 MATLAB Application to Axisymmetric Analysis This section shows au exauple of gu asisynuuetric sendy stele probies using MATLAB programs. Linear triangular elements are used. @ Example 5.10.1 An axisymmetric Laplace equation is solved using Knoar trisagular clemonte. The governing equation is given ia By. (572) for a cylinder whose inaide and outside radii are 4 and 6, and whose height ia 1. The finite element mesh used for the present aualysis is shows in Fig. 5.10.1. The boundary conditions are u = 100 at the mside of the cylinder and $¢ = 2)120 Laplace’ and Poisson's Equations Chapter § ae / | 0/0 —— { a a ? a t * PAL Ody Oty Oy Od Ot Figure 5.10.1 An Infinite Cylinder Modeled With Symmetric Boundaries at the outside of the cylinder, Both top and bottom surfaces of tue cylinder ie. lamlated. "Ten triangnlar elements with 12 woden are ned and the MATLAB programs are provided below. As seen in the main program, the constant ux at the outside surface is converted into the nodal ux at the onteide surface. Each node takes a half of the total flux over the element which is Qafigql = 240w where r=6, ¢=20 and /=1 as explained in See. 5.7 %*—__—_________- % EXS.10.1.m % to solve the axisymmetric Laplace equation given as Kort Urt nes =0,4or< 606261 % u(4,2) = 100, u,2(6,2) = 20 % uyz{x,0) = 0, wn(r1) =O % using linear triangular elemenis Sh(see Fig. 5.10.1 for the finite element mesh) % 9% Variable descriptions 36 k= clement matrie element vector % kk = system matrix % E = system vector coordinate valnes of each node nodal connectivity of each element a vector containing syatem dof sssaciatrd with each element a vector containing dofs associated with boundary conditions ne heomachery cemnditinn values aanorinted with (he dofs in dedof input data for control parametersSection 5.10 MATLAB Application to Axisymmetry 121 nel % number of elements and ‘% mumber of nodes per element dof: % number of dofs per nede anode=12: % total sumber of nodes in system sdol=nnodetndoty % total system dofs % %———____..--____. % input data for nodal coordinate values % gcoord( iJ) waere 1:> node HO. and f-> x Ory 6 gecord(1,1)=4.0; gevard(1,2}=0.0; geoord(2,1}=4.0; gcoord(2,2}=1.0; geoord(3,1)=4.4; gcoord(3,2}=0.0; geoord(4,1)=4.4; gcoord(4,2}=1.0; seoord(5,1}=48; gcoord(§,2}=0.0; geoord(6,1)==£.8; gcoord(6,2}=1.0; gccord(7,1}=5.2; geoord(7,2}=0.0; gcoord(8,1}=8.2; zcoord(8,2}=1.0; gcoord(9,1}=5.6; gcoord(9,2}=0.0; gcoord(10,1)=5.6; gcoord(10,2)=1.0; geoord(11,1)=6.0; gooord(11,2]=0.0; geaord (12, ccoord(12,2}=1.0; % — % input data for nodal connectivity for each element 1% nadeo(ii) where F> element na, aud j connected noves ¢——$ $$ nodea( nodes{¥, nodes, aodes(5, nodes(6, noded(®, nodes(0, aodes(10,1)=9; nodes(10,2)=11; avdeo( 10,3} % $—$———_____— % input data for boundary conditions %¥—_ bedof(1)=15 % first node ie constrained boval( 1}=100; % whose described. value is 100 bedof(2 % second node is constrained 13; Ddeval(2}=106: % whose described value is 100 ct Jp % initialization of matsioes and vectors %— etoe(sdof,1); % initialization of system ferce vector Ru=zeren(doh aot); 36 inisialination of xystean ra index=teros{nnel*ndof.1); % initialization of index vector % pisd*atan(1); % define pi422 Laplace's and Poiason’s Equations Chapter 5 RL} =120*pi;, 9% nodal flux at the oxtsde boundary £(.2)=120* Dis % nodal flux at the ovtstic boundary % %— % computation of element matrices and vectors and their assombly rt oe for iel=1:nel % loop for te lotal mumber of demenis % ndf1}=nodestied,t}; 1% Ast connected node for (iain lement nd(2)=nodes(iel,2); % 2nd connected node for (iel)-th element nd(3)=nodestiel,3); % 3rd connected node for (iel}-sh element coord(nd(i),1); 21 =gcoord(nd(3),2); % coudinate of Ist aode coord(ad(2),1}; 22=gcoord(na(2),3); ‘6 coordinate of 2nd aode coord (nd(3).1): 23ecoord(nd(3).2): % coordinate of 370 aode index=fecklof(nd,nnel,ndof}; % exctract system cols for the element % ke felpaxt3(r,21,12,22,28,23); "Ye compute eloment matrix, % ‘ee fenssnbld (Be hehe); % asweuble clement matcices % end % %—_—___——. ‘% apply boundary conditions Gece (kk, ft] =feaplyc2(kk.ff, bodof, beval}; % %—______-— % solve the matrix equation *—______—. faol=kk\E; % _— % analytical solution Mer for node r=gooord(i,1}; z=gco0rd i,2); esol (1) =100-6*20*l0g(4)+6*20*lox{r)s end % a % ptint both exact and fom solutions pn mum=I:L:sdof; store=[num’ so) e01"]Section 5.10 MATLAB Application to Axisymmetry 128 fonction (kJ=felpaxt3(r1,z1,r2,22,x8,28) Tan % Purpose: % element matrix for axisymmetric Laplace equation % using three-node linear triangular element % % Bynopsis: % P}felpant9(ct 22,12.22413,53) % % Variable Descriptio: % k - clement stiffness matrix (size of 3x3) % 11,21 -© and z coordinate values of the firet node of dement ‘% 12, 23 -t and 2 cuurdinate values of the second node cf element % 18, 2 and 2 coordinate values of the Lhird node of clement a a—eeor % % wlement matriy % AxO.5*(c2*28+r1*o2f 13421 22el-11*23 1322}; % area of the triangle ApxB+19)/35 % x coordinate value of the clement centroid twopirc=é*atan(L)*r¢; (18-2) (03-12)-+ (22-23) *(22-23))/{44 A); 8-1)" fel-es}-+(22-23)"(2d-24 ))/ (4°); rr) (Der! )+ (2223) *{21-22))/(4* A}: La) Ce tetera) (s8-01)*(e0-21)/ C42 1-13)" (r2-r1)+(23-21)*(21-22)} /(4*A); (1.3): (2.3)5 U(3,3}=( (22-11) * (211) + (61-22)*(21-22)) f(A): A—twopireths % 9 ———— ‘The results are store = doi# femsol exact. 1,000 190.000 100.000 % at 14.0 and 250.0 2.009 [email protected] 100.000 hat r=40 and =1.0 g.on00 | 111.413 111.437 Goad rata and 120.0 4.0000 111.444 111.437 Seat r=44 and o=1.0 5.0000 121.836.121.870 % ob co AG and 2=0.0 5.0000 121,889 «1ZL.8TS Ye ave=4.8 and 2=1.0 7.0000 «131.427 (131.484 oot r=5.3 and 2-0.0 6.9000 432.501 131.a8a % av r=5.2 end 2=1.0 s.0000 140,290 140.377 % avr=s.6 and 250.0 10.000 140.417 140.377 Yoatras6 and aslo Viago 148.510 148.656 %e ab e=26.0 and 9=0.0124 Laplace’s and Poisson’s Equations Chapter 5 Figure 6.11.1 Mesh With Triangwar Element 12000 4a.749 148.030 1% me r200 and 224.0 5.11 MATLAB Application to Transiont Analysic Examples ate given for some transient analyses using MATLAB programm. Forward difference, backward difference and Crank-Nicoloa techniques are usd, & Example 5.11.1 The transient Laplace equation as described below is solved using the forward difference technique for time integration. The differential eqmation iz du _ Mu Pu Bt Gxt ~ By? over a rectangular domain defined by 0 < 2 < § andO< y <2 The whole domain haa the initial vale of = 0, and euddanly theleft and sight houndizica Gc. clges with x0 aud a—8) ate maintained at = 100, Ow the wile: lent, the bop end botiom boundaries (ie. edges with y=0 and y=2) are ineulated (Le. 2 = 0). We want to find the solution as a function of time. A finite dement mesh of the domain is shown in Fig. 5.11.1 using 16 linear Liangnlar ceents ‘The finite element. analysis program is shown below. (5.11.1) a % EX8.11.1.m 36 bu wdve the Ureasieat owodiuenaiunal Lapletc’s eyuation Hugs = xx tuyy 0node nv. and }>x or y %— — geoord(2,1} gcootd(3,1}=2.5; geoord (3,2) geouel(4,1}=3.75; geuved (4,2)=0.0; acoord(8,1}=5.0; geoord( 5.2) gcoond(6,1}=0.0; gcovrd(6,2} gcoord(7,1)=1.25; gooord|7,2}=1.05126 Laplace's and Poisson's Equations Chapter § 75; gcoord(9,2}=1.05 gcoord(10,1)=8.0; acoord(10,2)=1.; gcoord (11 1}=0.0; geocrd(11,2} gcoord(12,1}-1.25; geoord(12,2)=2.05 eoord(13,1}—9.8; pooord (13,2}—2.0, geoord(14,1)—3.75; gooord(14,2)—2.0; gcoord(15,1}=5.0; gcoord(15,2)=2.0; % 4 % input data dor nodal connectivity for each element % nodes(ij) where i-> element nu. and j-> connected nodes : nodes(. nodes(2,1)=2; nodes(2, nodes(3, nodea(4,2)~5; nodes(5, nodes(8,1}=4; nodea(3, nodes(9,1)=5; nodes(9, nodes(10,1)=7; odes{10,2)—8; nodes{10,3)=13; nodes(I1,1)=8; nodes(1,2}-9; nodes( 11, nodes(12,1)=9; nodes(12,2}=10; nodes(12,3 nodes(13,1)=6; nades(13,2}—12; nodes(13,3)—11; nodes(14,1)=7; nodes(14,2)==13; nodes(14,3)=12; nodes(15,1)=R; nodes(1 nodes(15,9 nodes(15,1)=9; nodes(16,2)=15; nodex(16,3 % Bowe % input data for boundary conditions %— —_—_ bedof 1} {i Jat node is constrained beval(1)=100; % whose described ralne is 100 bedof{2} % Bth node is constrained beval(2)=100; % whose described value is 100 bedof{3}= 4% 6th node is constiained bevall3}: % whose described value is 100 bedof4} J 10th node is constrained eval (6) % whose deseribed value is 100 bodof{5} % AIL nodewin sonstaained beval(} % whove described valve is 100 hedesl(6) % Ath monde is constrained beval{)=100; 9% whuve devedbell alae is 100 % % Yb initialization of matrices and vectorsSection 5.11 MATLAB Applicetion to Transient Analysis it feer09(sdoL1}5 % inisiatization of system vector fn=peros(stof 1): % initialization of effective aystem vector teol=zeros{sdof,1): % sohition vector sol=zor0s(2,ntimet I); % vector containing time history solution Lmecrefodfido)s % initialivetion of nyotcan eantein mmn=zeros{sdof sdof)}; % initialization of systern matrix index-zeros(anel*nduf,1); % initialization of index vector % %. =. % computation of element matrices and vectors and their assembly ——__ or for iel=L:nel % loop for the total number of elements % 1d(1)=nodes(el,1); % Ist. connected node for fiel)-th element 1d(2)=nodestiel,2); % 2nd connected node for {iel)-th element 24{3}—rodea($al,2); Ged commuted mode fox [ial)-Uh lennon xL=geoord(nd(1},1} x2=gcoord(nd(2),1)}; ¥: geoond(ad(1),2); 9% coord values of Ist node gcvond (ad(2},2), Weoord values of 2nd node, sSagcuord(al(3),1), yS=guvid(ad([3},2), 96 cord values uf Srd aude % index=feeldof(nd,nnel,adof}; % extract system defs for the clement % kefelp2dt (x1 y1,x2.92,%3,93); % compute eletent matrix ansielpl243(x1 y1e2,92,x3,99); % compute element matrix % kk=feasmbl1(kk,kindex}; ‘% assemble element matrices avnfeasmbll (mm,m,index)s % ascemble element matrices % ond * % % loop for time integration % initial condition % store time history solution for noite no. 8 % store time history solution for node no. 9 % etart loop for time integration (n=delU*f+(mmdeltt*kk)"fsol; ‘WG compute effective column vector % [mm,fa}=feaplye2(mm.fn,bedofbeval); _% apply boundary condition % teol=mum\in; solve he malzix equation128 Laplace’s aud Poisson's Equations Chapter 5 % sol(L.it+1}=ts0l(8); % store time history eolution for node no, 8 sol(2,it-+T)fenl (9); UZ ktove tite hiatory snlntion for node na, 9 % end % %— % plot the solution at nodes 8 and 9 Yoon —____— time=0:deltt:ntime*deltt plot(time,sol(1,:),™',time,tol(2,:}, xlabel( Time’) ylabel(’Solution at nodes’) % Fe function ftal=felpt213(x1,51,x2,y2.x3,¥3) % clement matrix for tronsiant teem of twro-dimencinnl % Laplace's equation using linear triangular element % 3% Syavpuis: % (mjafelpt203 (x1 y,x2,y2,23,y3) * % Variable Description: % m - element stiffness matrix (size of 3x3) % x1, yl - xand y coordinate values of the first node of element % x2, y2 - x and y coordinate values of the second node of dement % x3, y3- x and y coordinate values of the third node of element ya % Y%, clement matrix % ‘AxO.S*(x24yStalty2- 4x3" yl-x2"yL-xl¥yB-x3%y2); Foaxea of the tiangle w m = (A/12)" (2115 121; 112) %4—_—_—________. ‘The finite dement solutions are plotted in Fig. 6.11.2 aad Fig. 5.113. Time history of nodes 8 and 9 in Fig. 5.11.1 is plotted in both figures. While At = 0.1 woo used for Pig. 5.11.2, Af — 0.19 was need for Fig. 6.113, As noticed, the finite clement solution is unstable whea At = 0.12 is used bucausy the forward difference technique is conditionally stable. tSection 5.11 MATLAB Application to Transient Analysis 139 Figuvo 6.11.2 Finite Flement Solution With At—0.12 # Escurmple 511.2 The same example ap Beample £111 i oolved wang bilinaar rectangslar slemanta, The mosh is the game an thet shown in Fig. 5.11.1 except that 8 rectangular elements are used instead of 16 iieugular elements, % % EXS5.11.20 % to solve the transient tio-dimensional Laplace's equation Hut =urx toy, 0node no. ard j->x ory $$ gcoord(1,1}=0.0; groord(1,2}=0.0; gooord!2.1)=1.25: zcoord(2,2)=0.0: acoord(3,1)=2.5; gcoord(3,2)=0. gcoord(4,1)=3.75; gcoord(4,2)=0. necord(6,1)—6.0; geaord(5,2)—0.0; geoord(6,1)=0.0; gcoord(6,2}=1. geourd(7,1}=1.25; govord (7,2}=1.05 gevurd(8,1)=2.5; yeuurd(6,2)=1.0; gcoard(9,1)=3.75; gcoord(9,2)=1.0; geoord(10,1)=5.0; gcaord( 10,2 geourd(11,1}=0.0; gcoord(11,2}=2.0;Section 5.11 MATLAB Application to Transient Analysis gcvurd(12,1)=1.28; geoord(12,2}=2.0; acoord(13,1}=2.5; gcoord(13,2}=: acoord(14.1)=3.75: geoord(14.2)=2.0; geourd(15,1}=5.0; gcoord(15,2)=2.0; % % - ‘% input date for nodal connectivity for cach clement % nodes(ij) where i-> slement no. and j-> connected nodes %———_ - nodeaf 1,1)=1; nodes(1,2)=2; nodes(1,3)=7; aodes(1,4=6; nodes{2,1)=2; nodes(2,2)=3; nodes(2,3)=8; aodes(2,4=% 0; nodes(4,4} nodes(B.1)=7; nodes(6,2)=8; nodes(6.3)=i3; nodes(§.t)=: nodes(7,1)=8; nedes(?,2)}=9: nodes(7,3)=14; nodes(7 4) nodes(2,1)—0, noden(2,2)—10; nodea(#,2)—15; nodes(¥,d}=1; % %— % iuput date for boundary conditions %——_ bedof(1 bedof{2) % Sth node is constrained beval(2 %e whose described value is 100 bedof(3)=6; % tl node is constrained beval(3}=100; %h whose described value is 100 hedaf(A! % eth modes constrained beval(4} % whose described valve ix 100 bedut(S)=11; 4 1th node io constrained Veval(5}—100; 9% whuse desctibel value is 100 bedol(6) % 1th node is constrained deval(6} % whose described value is 100 % %— fl=zeros(sdof,1); % initialization of sy ster vector fn=neros(sdof.1): Yoinitialization of effective aystem vector frol=neros(sdof,1); % solution vector sol=zerus(2,nlime} 1); % vector vontaining time history solution Ki-azroa{edofedof); % jnitbaliention of ayelots stale mnm=zeros(sdof,sdof % inistabization of system matrix index=seros(nnel*ndof,); % initialization of index vector % % — % Ist node is constrained beval(1}=100; % whose described value is 100 % computation of element matrices and vectors and their assembly 131132 Laplace's and Poisson’s Equations Chapter 5 for iel=L-nel % loop for the total nnmber of elements % nd(1 )=nodestiel,1): % Ist connected nede for (jel) th element nd(2)=nodes(iel.2); % 2nd connected ncde for (il)-th element nd(3)—nodes(iel,a); % 3rd connected nude for ((el)-th element nd(4}—nodea(iel,3); % 4th connected node for (el)4h clement xL=gcvord(nd(1),1}; yl—gecord(nd(1),2);-% coord values of Ist node xd=gcoord(nd(2),1); y2=gcoord(nd(2),2}; _ % courd values of tnd node xd=gcoord (ad(3),1)3 YS=Kcoord(nd(3),2); _% coord values of Sra node x4=geuord(ad(4),1); y4=gcoord(ad(4),2), _% coord valuss of Sth node aleng=x2x5 % element size in x-axis yleng=y4-y1; % element size in y-axis % index=fecldofind.anel.ndat):% ‘k=felp2dr4(aleng,yleng): % Vimne-independent element rairix m=felpi2r4(xleng,yleng); % taansient. dement matrix % ‘kk=feasmbL (kk,k,index); ‘% assemble element matrices yam—feasmbl (mm,m,index); % assemble element matrices * end % &——_—_—$$____ % loop for time integration %— % for in=1:sdof feo end % 201(2,1) =f501(0)s % store time hintory solution for acde no, @ wol(2,1)=ts0l(3), % store time history solution for nodene. § % for t=t:nlime % start loop tor tame tegration % fasdeltt*f+(mm-deltt*kk)*fsol; 9% compute effective cola vector % {oum,fn]—feaplyc?(mea,fn, bedof,beval}: % apply boundary condition % ‘sol=mm\fn; % solve the matrix equation % eul(1 it-+-1)—feol(9); % elore time history solution for node ne. $ sol(2,it-+1] % store time bistwry solution for node no. $ % end % %*—__—_— Y% analytical solution at nade &Section 5.11 MATLAB Application to Transient Analysie 133 % pix4*atan(1); jeros(1.mtime+1); 0; for ti=O-deltt:ntimetdeltes esolfii) esol) 4A)" exxp(-its* pit pitti (lad) *singpitea/ ah end end es01=100-(100"4/pi)*es0l; % %. a % plot fem and exact solutions at node 8 ye timendidelttentimetdeltty ploi{ time, ol(1,:),™,time,esal,”}; xlabel(’Pime"} ylabel(’Solntion at nodes”) % % Ipt2r4(xleng,yleng) % Purpose: % element matrix of transieut term for two-dimensional Laplace's % eauation using four-node bilinear rectangular element % % Syropsis: % [mJ—felptdr4(alonsgyloms) % % Variable Desaiption: 1% m - eloment stiffness matix (skze of 6x4) % sheng - elememt size in the x-axis % yleny - element size in the y-axis %&——_____._______. h % dement matrix % m=(vlengtyleng/36)*[4 2 19; The finite element solution obtained using rectangular demerts is comparable to that obtained asing triangular clements. Figure 5.114 compares the ficite134 Laplace’s and Poisson's Equations Figure 5.11.4 Anatytical and Finite Element Heralts at Node 8 senate Figure $11.6 ‘Tie History of Nodes 8 and 9 Chapter 6 oloment solution to the svart wolufion at node ® (see Fig 8.i1.1). ‘the fiaite clement solution approaches the steady state solution dower than the exact solution. This is duc to the very crude mesh in the a-direction. i & Example 5.11.3 The present example solves tie same problem as that in Example 5.111 acing the backward difference technique for time iutgiation. Because this technique is unconditionally stable, we use a time stepsire Ai 04 which exceeds the critical time step size for the forward differerce teclniqne. Figure 5.11.5 shows the tme-history of nodes & and 9. %— % EX$.113.mSection 6.11 MATLAB Applicetion to Transient Analycia 135 % to salve the transient two-dimentioxal Laplace's equation Ut = UK tugy OC XS OC KZ % boundary conditions: % w(0,yst) = 100, a(6,y,4) = 100, % wy(x.0,4) = 0, ny(,2,t) = 0 % inildel condition: % a(x,y.0) = 0 over the domain % asing linear triangular elements and backward differeace method Ylsee Fig, 9.1.1 (OF the finite element mest) % % Variable descriptions % k = dement matrix for time-independent term (1,xx + u,yy) % m = element matzix for time-dependeat term (x,t) % {= cement vector % kk = system matrix of k % mm = system matrix of m % = cystom vector % in = affective system vector ‘% ‘sol = solution vector 76 vol = Gune hivtony culubion of selects wudbes % geoord = coordinate values of each node % aodes = nodal connectivity of each elament % index = a vector containing system dots agcociated with each element % bedol = a vector containing dofs associaled with boundary conditions 4% boval =a verter containing boundary condition values associated with % the dofs in bedof %— clear %--—___ % input data for control parameters * nel=16; % number of elements ancl=$; % umber of nodes per clement adof=t} ‘Ye tumber of dofs per node anode=15; % total wamber of nodes in system sdof=nnode*ndof, ‘% total system dofs % time step size for transicut analysis % initial time % termination time % number of time increment % input data for nodal coordinate values % goootd(i,j) whete i->node no. and }>x or y Soom geoord(1,1 coord 1,2)=0.0; coord (2,1) gcoord(2,2)=0.0; gcoord(3,1}=2.8; gcoord(3,2}=0.05136 Laplace's and Poisson’s Equations Chapter 5 gcoord(4,1)=8.75; gcoord(4,2)=0.0; gcoord(5,1)=5.0; gevordi5,2}=0.0; gcoord(6,1 coord (6,2' Bi geoord(7,1)=1.25; groord(7,2}=1.0; geoord (8,1 geoord(®, 2 geverd(0,1)—2.75; geoord(0,2)—1.05 gcoord(10,1)=5.0; geoond (10,2) geoord(12,1)=0.0, goourd(11,2)=! euerd(15,1}=8.0; geoord (15,2) % 6—_—_—__—________— % imput data for nodal connectivity for cach clement % nodes(ij) where i-> element no. and j-> comnected nodes i nodes(1,1}=1; wodes(1,2}=2; nodes( 1,3) nodes(2,1}=2; nodes(2,2}=3; nodes(2,3}—B; odea( 3,1) 5 auden( 3,9) nodes(4,1) ; nodes( 4,3} nodes(5,1) 5 Nodes(5,4)=85 aodea( 6,1) nodes( 6,3} aodes(7,1) odes(7,3}8 aoder(8,1) 0; nodes, ; neodes(9,S}e125 nodes(19,1)=7; nodes(10,2)=8; nodes(10,3)=13; nedes(11,1 nodea( 1,2): acdes( 12,1 nodes(12,8 nodea(13,1 2; nodes(13,3) aoden(14,1 nodes(15,1 nodes(16,! %———$$ ‘% inpus data for boundary conditions $—— bedati(1) beval(1} moden(a4,2 ; nades(15, ; nodes( 16, Ist rode is comstrained % whose described vine is 100 % Sth node is constrained beval(2}=10 % wheste described value ia 100 hedaf(a)—6; % 61h ande in concteained % wheoe described velue is 100 % 0th. node is constrained ‘whine described value is 100 % Ath node is constrained i wehose described value is 100 % 15th noc is constrainedSection 5.11 MATLAB Application ta Transient Analysis 17 eval(6)=100; % whow described value is 108 % % % initialization of matrices and vectors % Si-mncron(sdof,1)s $% imitiabeation of ayatem vector fn=zer0s(sdof 1); % initialization of efective wystem vector fuol=zeroa(sdof,}); % solution veetor sol=zeros(Zntime+ 1); 9% vector camtaining time hiniory solution kkk=zer0s(sdof,sdof); % initialization of system matrix amm=seros(sdof,sdof}; % initialization of system matrix index=zeroe(nnel*ndof,1); % initialization of index vector % loop for the total anmber of elements ud(1)=nodes(icl1); % st connected nude fur [iel}-th element nd(2)=nodes(el.2); 1% 2nd conuerted node for [iel)-th element nd(3)=nodes(iel.3); % ard conmected node for [el)-th element xl=gcoord (nd(1),1); yl=gcoord(nd(1),2); _% coord values of Ist node scoord(nd(2),1); y2=gcoond (nd(2),2); Yh coord valves of 2nd node ‘% coord values of Std node index=feeldof(ad,nnel,ndof}; ‘% extract system dofs for the clement % Keafel p 2442x191 ¥2.y2.3 92): % enmpute element mateie m=felpt2A(xt yl x2.y2,x3,¥3); % compute element matrix % Seh—fisamsubl (ke sondeca) FR asscuMle desncnt nataices anm=feasm)li (min,m,index); Fe aseemble emeut matuices % end % % % loop fox time integration % initial condition sol(1,1)=t0l(8); % sol contains time histay solution of wode 8 sol(2,1)=fs0l(4); % sol contains time-histary solution of node 9 % kh=mm+dett*kk: % for iti mtime138 Laplace's and Poisson's Equations Chapter 5 Us=50 convection cosfficiant = 100 G.02m Figure 8.11.6 Finite Element Mesh faadeltt*#¢mmsel % compute effective column vector [Rote eplyc iff eval; % apply boundary condition picta\ tn: % solve the matrix eauation Salc43 at; % sol contains time-history solution of node 8 wol(Qyit-+1}—frol(0)s % sal containe time-history solution of ade 9 % plot the solution at nodes B and 9 &——$_- time=0:deltt:ntime*deltt; plot({time.sol(.:)*"time.sol(24).- xlabel(‘Time’) ylabel(Sohition xt nodes") % % HExomple G114 A plate ofoise 0.02 m by 0.01 m, whose hat conduction coefficient is 0,3 W/m0, is initially at a temperature of 300 C. While its left and right aides are maintained at the same temperature of 900 C, the bation side is insulated and the tup side is subjected to heat convection with convection ient of he = 100 W/m? C and the ambient temperature of 50 C. The material has aiso density =1600 Ky/m? and specific twat c=0.6 1/KgC. The coefSection §.11 MATLAB Application to Transient Analysis 139 gh oy 8 | a a aE oa Wigure 5.11.7 ‘Lime History Plot MATLAB program usng the backward difference technique is Usted below and ‘the mesh is showa in Fig. 5.11.6. The time-history of the solution of node 8 is given in Fig. §.11.7. %—-——_—_ % BXS.11 4m 9% to solve the transient two-dimensional Laplace's equation Hatat = axetuyy Ue < 0.02,0<¢y <001 % boundary conditions: % n(0,y,t) = 300, u(0.02,3,¢) == 300, % uy(x,0,t) = 0, a,y(x0.01,¢) = 20(u-50) 9% initial condition: % a(x,y\0) = 0 over the domain % osing bilimear rectangular elements and forward differeno: method Fé(see Fig. 5.11.6 for the Anite element mesh) % % Variable descriptions % k = clement matrix for time-independent term (uxx + usy) % m = clement matrix for time-dependent term (ut) % {= clement vector 9 kkk = system matrix of k % mm = system matrix of m 1% #F = aystom vector % fm = effective system vector % fool — solution vector Aimehistory solution vector of sulccted nadeo coordinate valucs of cach node nodal connectivity of each element = a vector containing system dots associated with mach clement 8 vector containing dofs associated with boundary conditions 4 vector containing boundary condition values associated with the def ia bedof140 Laplace’s and Poisson's Equations Chapter 5 % k1 = clement matrix dae te Cauchy-type flux % {1 = clement vector duc to flux boundary condition % index = index for nodal dofs with flax §—— dear ——————— % input date for conizel parameters % number of elements ‘% mumber of nodes per clement ‘% wamber of defa per ttode % total mumber of nodes in system mode"ndal; % total system dots deltt=0.1; % time step size for iransient aalysi stime=0.0; % initial time ‘time: % termination time ntime—fixd (ftime-stime) fault); % socrahoy of fia increment % covfficient for the transient term % number of clomert boundaries with dus Th number of wunkes por side of each clewewt % iwpot data for nadal coordinate values % geoord(i,j) where i> node wo. and > xor y %— —— Reoord (1,1)=0.05 geoord (1,2 coord(2,1)=0.005; geoord(2,2}=0. acoord(3,1}=0.010: gcoord(3,2)=0.0; Reoord(4,1)=0.015; geoord(4, geoord (8,1)=0.020; geoord(5, 0; geoord(6,2)—0.008; 005; gcootd (7, 010; geoord(8, gcoort(¥,1)=U.0i5; geoord(¥,2 geoutd(10,1)=0.020; gcoord 10, gcoord(11,1)=0.0; gcoord(11; coord (12,1)=0.005; geoord( 12, gcoord(1 010; geoord(13,2}=0. gcoord(14,1)=0.015; geoord(14.21=0.01; gcoord(15,1)=0.020; goootd(15,2)=0.01; % a % input data for nodal connectivity for each element % nedes(i.j) where i-> element x0. and j-> connected nodes j wodes(1,2)=2; nodes(1,3)=7; nodea(ly ; nades(2,2)=3; nodes(2,3)=8; nodes(2, nodes(3,3)=3; nodes(2,2)—4; nodes(3,3)=9; nodea(8,4)=3;Seetion 5.11 MATLAB Application to Transient Analysis nodes(4,1}=4; nodes(4, nodex{5,1)=6; nedes(5, nodes( 6,1) nodes(6, s nodes(7,1)s8; nodes(7,2)=9; nodes(7,3)=1 nodes(8,1)=0; nodes(8,2) % % % input data for boundary conditions % input for flux boundary ecuditions % ufix(ij) where i-> element ao. and j-> two side nodes Yn nfix(1,1)=11; nilx(1,2}=12; % nodes on Ist clement side with tux % nodes om 2nd element. side with fine ‘% nodes on rd element side with flux % nodes on 4th clem le with flux % wiviahation of matrices and vectors %—_—_ fi=zeren(stof,1); % system vector fa=zeros(sdof,i); % effective aysiom vector fsol=zeros(sdof,1); % solution vector eros(Lntime+1); % timetistory of a selected node cros(sdof.sdof); ‘% of eyetem matrix mam=zer0s(sdof,sdaf); 4 system matric indor—zoroe(nnal4ndof,1); % index vector f=zeros(anclstndaf, 1); % element fox vector K1=seroa(nnels*ndof,nnels*ndof); % flux matrix index] scrom(ustelstadufl); % ox index vector % 9 ——_——_—- % computation of clement matrices and vectors and their assembly % let undies constrained, whose described value is 300 % Sth wove is constrained % whose described value is 300 % 6th node is constrained % whove described value is $00 % 10th node is constrained 3% whose described value je 400 ‘We Ath node ie constrained % whose described value is 300 96 15Uh wode is combraime whose described value is 300 Mt142 Laplace’s and Poisson's Equations Chapter 5 % loop for the total number of elements des(iel,1); % Ast connected node for (iel)-th element destiel.2)s % Amd connected node for (iel)-th element desi 3); % 8rd connected node for (ie!)-th element des(iel, 4); % 4th connected node for (ie)-th element xlgeaordind(1),1}; y1=gcoord(nd(1),2); % coord vaines of lat node xPrgesord(nd(2},1}; y2—geoord(nd(2),2); % coord valor af Ind node xit=gcoord(nd(3),1); y3=gcoord(nd(#),2); % coord values of 3rd node xd=ycoord{nd(4),1); y4=gcoord(nd(4),2); % courd values of 4th node % clement size tn x-axis yleng=yt-yl} % clement size in y-axis % index=fecldof(ndnieladof); —_-% extract system dofe for the clement % p2dr4(xleng,yleng); % compute clement matrix m=a*felpt2r4(xleng, yleng); % compute clement matrix % Kr —fesemblt (HL inden esembL1(mm.an,index); 1% secemble slamant matsiene % ascemble element maiziees &—_»-______~.. nda())=nfix(ifk,1); % node with flux BG for [ifx)-th element anf x(ife,2);, % mode with flux BC for [ifs)th eloment xi=geoord(ads(1),1); ¥!—gevsrd(nde(1),2); % nodal courdinate a2=gevord(nds(2),1); y2=geoord(nds(2},2); ‘% nodal coordinate dlemymsucl((xea1(R22) +271) yey); % element ade lengths % indexl=Ieeldof(nds,nnels,ndol); % find related system dofs % ki=bMefixl2(deng); % compute element matrix due to ux fl=b*c*fefil(O,eleng); % compute clement vector due to flux % [kk,f]=feasmbl2 (kh, £1,41 index); % assembly % ead m §—— % loop for time %— for in=lisdof