PRENTICE HALL INTERNATIONAL SERIES IN CIVIL ENGINEERING AND ENGINEERING MECHANICS William J. Hall, Editor ‘Au and Christiano, Fundamentals of Structural Analysis Au and Christiano, Structural Analysis Barson and Rolfe, Fracture and Fatigue Control in Structures, 2fe Bathe, Finite Element Procedures in Engineering Analysis Bere. Plamonte af Stenenival Dynamiive Chajes, Structural Analysis, 2ie Collins & Mitchell, Prestressed Concrete Structures Cooper and Chen, Designing Steel Structures Cording, et al., The Art and Science of Geotechnical Engineering Gallagher, Finite Element Analysis Hendrickson and Au, Project Management for Construction Holtz and Kovacs, Introduction to Geotechnical Engineering Humar, Dynamics of Structures Johnston, Lin, and Galambos, Basic Stee! Design, 3/e Kelkar and Sewell, Fundamentals of the Analysis and Design of Shell Structures MacGregor, Reinforced Concrete: Mechanics and Design, 2fe Mehta, Concrete: Structure, Properties and Materials, 2fe Melosh, Structural Engineering Analysis by Finite Elements Meredith, et al,, Design and Planning of Engineering Systems, 2ie Mindess and Young, Concreie Nawy, Prestressed Concrete Nawy, Reinforced Concrete: A Fundamental Approach, 2¢ Pfeffer, Solid Waste Management Popov, Engineering Mechanics of Solids Popov, Introduction to the Mechanics of Solids Popov, Mechanics of Materials, 2fe Schneider and Dickey, Reinforced Masonry Design, Sle Sennett, Matrix Analysis of Structures Wang and Salmon, Introductory Structural Analysis Weaver and Johnson, Structural Dynamics by Finite Elements Wray, Measuring Engineering Properties of Soils ‘Yang, Finite Element Structural AnalysisMATRIX ANALYSIS OF STRUCTURES ROBERT E. SENNETT Catifornia Polytechnic State University 105212 COLLEGE LIBRARY ATENEO DE DAVAO ENGINEERING PRENTICE HALL, Englewood Cliffs, New Jersey 07632Liprary of Congress Catalagingin-Publicaton Data ‘Maris anahysis of structures / Robert Senet cludes bibliographical references and fe. ISBN 0-13-36203-1 f suractra analysis (Engincesing)—Masx methods. 1. THe FAG SHE 199 G24. T1420 nda cP Acquisitions editor: Doug Hurmphrey Production editor: Jennifer Wenzel Copy editor: Sim Tully Cover designer: Maureen Eide Prepress buyer: Linds Behrens Manufacturing buyer: David Dickey Supplements editor: Alice Dworkin Beéitorial assistant: Susan Handy © 1994 by Prentice-Hall, Ine ‘A Simon & Schuster Company Englewood Cliifs, New Jersey 07632 Al sights reserved. No part of this book may be reproduced, in any fofm or by any means, without permission in writing from zhe publisher. Printed in the United States of America 987654321 se HES rsen o-23-shagd-¥'! #2 Anke i 08 Dasa get Prentice-Hall International (UK) ‘Limited, London Prentice-Hall of ‘Australia Pty. Limited, Sydney Prentice-Hall Canada Inc., Toronto Prentice-Hall Hispanoamericana, ‘S.A., Mexico Prentice-Hall of India Private ‘Limited, New Delhi Prentice-Hall of Japan, Jne., Tokyo ‘Simon & Schuster Asia Pre. Ltd., Singapore Editora Prentice-Hall do Brasil, Lida., Rio de JaneiroCONTENTS PREFACE xi ANALYSIS OF ONE-DIMENSIONAL BARS 1 1.1 Introduction 1 12 Nodes and Degrees of Freedom 4 13 Stiffness Definition 4 1.4 Individual Element Stiffness 8 15 Combination of Element Stiffness 9 1.6 Structures with Specified Non-Zero Displacements 15 1.7 Noti-Nodal Forces 17 1.8 Thermal Effects 21 19 Computer Formulation 22 1.10 Summary 26 Problems 26vi Contents 2 ANALYSIS OF TWO-DIMENSIONAL TRUSSES 3 21 Introduction 32 2.2 Coordinate Transformations 33 2.3 Global Stiffness Matrix 35 24° Support Movements 41 2.9. Temperature Changes and Fabrication Enors 46 2.6 Computer Formulation for the Truss 50 27 Summary 52 Problems 52 3 ANALYSIS OF TWO-DIMENSIONAL BEAMS AND FRAMES 57 3.1 Introduction 58 3.2 The Beam Elemental Stiffness Matrix $9 3.3. Stiffness Matrix for the Two-Dimensional Frame Element 61 34 The Transformation Matrix for the Frame Element 62 a Example Beam and Frame Problems 63 3.6 Non-Nodal Loads 68 ay Thermal Effects in Beams and Frames 74 3.8 Support Movement for Beams and Frames 76 a Computer Formulation for the Frame 77 3.10 Summary 79 Problems 80. 4 ANALYSIS OF GRIDS 86 4) Introduction 87 4.2 Development of the Grid Elemental Stiffness Matrix 88 4.3 Coordinate Transformation 90 44 Non-Nodal Loads 91 4.5 Example Grid Problems 91Contents vil 4.6 Computer Formulation 95 4.7 Summary 95 Problems 95 5 THREE-DiMENSIONAL TRUSSES 98 5.1 Introduction 99 5.2 Elemental Stiffness Matrix 100 Coordinate Transformation 100 Examples of Three-Dimensional Truss Problems 103 5.3 Computer Formulation for the Three-Dimensional Truss 106 5.6 Summary 106 Problems 106 6 ANALYSIS OF THREE-DIMENSIONAL FRAMES 109 6.1 Introduction 110 6.2 Development of the Elemental Stiffness Matrix 110 63 Transformation of Coordinates 111 64 Example of a Three-Dimensional Frame Problem 113 65 Computer Formulation 120 66 Summary 121 Problems 122 7 ADDITIONAL TOPICS IN THE STIFFNESS METHOD . 125 7.1 Discussion of Bandwidth 126 7.2 Combining Different Elements to Model a Structure 129 73 Blastic Supports 131 74 Inclined Supports 133 75 Hinges in Beam and Frame Elements 136 7.6 Static Condensation 141vill Contents 7.7 Axial Deformation in Frames 144 78 — Substructuring 147 7.9 Non-Uniform Members 150. 79a Linearly Tapered Bar, 150 7.0 Stepped Bar, 151 7.9¢ Linearly Tapered Bear, 152 7.10 Summary 154 Problems 154 VIRTUAL WORK AND THE PRINCIPLE OF MINIMUM POTENTIAL ENERGY 157 8.1 Introduction 158 8.2 The Principle of Virtual Work 159 82a Rigid Bodies, 159 8.2b Deformable Bodies, 162 8.2¢ Expressions for Internal Virweal Work, 164 8.3 Elemental Stiffness Using the Principle of Virtual Work 167 83a Shape Functions and Elemental Stiffness for an Axially Loaded Bar, 167 8.3b Shape Functions and Elemental Stiffness for a Beam Element, 170 8c Shape Functions and Elemental Stiffness for a Three-Node Axial Element, 173 8.4 Non-Nodal Forces Using the Concept of Equivalent Work = 174 8.5 Strain Energy and Force Potential 176 85a Strain Energy, 176 85 Force Potential, 178 8.6 The Principle of Minimum Potential Energy 180 8.7 Approximate Solutions Using Minimum Potential Energy 181 8.8 Determination of the Structural Stiffness Equation Using Minimum Potential Energy 184 8.9 Summary 184 Problems 186Contents 9 A BRIEF INTRODUCTION TO THE FINITE ELEMENT METHOD ot Introduction 189 9.2 Plane Stress and Plane Strain 190 9.3 Shape Functions for the Three-Node Triangular Element 191 9A Strain-Displacement Relationships and Strain Energy 193 9.5 Force Potential 194 9.6 Application of the Principle of Minimum Potential Energy 195 9.7 Example Plane Stress Problem 196 9.8 Summary 199 Problems 199 A REVIEW OF MATRIX ALGEBRA B MATRIX INVERSION ROUTINES C DERIVATION OF THE SLOPE-DEFLECTION EQUATIONS SELECTED REFERENCES ANSWERS TO SELECTED PROBLEMS INDEX ix 188 206 211 216 217 226PREFACE Mairix analysis of structures and the closely related finite element method have achieved wide acceptance and use in virtually all engineering disciplines. The use of these meth- ods, which require the solution of large numbers of simultaneous equations, has become a practicality due to the development of digital computers. Ever since the introduction of desktop microcomputers, these methods have been used in most engineering offices to solve a wide variety of problems. ‘The purpose of Matrix Analysis of Structures is to present to the student the dis- placement method of matrix analysis, Background required consists of basic strength of materials and introductory courses in structural analysis. The text should therefore be suitable for senior-level undergraduates, beginning graduate students, and practic- ing engineers in disciplines including civil, mechanical, architectural, and agricultural engineering. The text begins with the development of the matrix method for the one-dimensional ar element, Because the bar or rod element is very simple, it allows development of all the necessary procedures without involving topics such as coordinate transformations. These transformations are derived when required for the element being considered. Thus, two-dimensional coordinate transformations are introduced in Chapter 2 for the truss and in Chapter 3 for the frame. Three-dimensional transformations are discussed in Chapter 5 (trusses) and in Chapter 6 (frames). The distinction between matrix analysis of structures and the finite element method applied to structural problems is somewhat artificial since the same basic concepts and xixii Preface procedures are used in both cases. The primary difference between these methads lies in the procedures used for deriving the stiffness of the individual elements; that is, the determination of the force-displacement relationships for a specific type of element such as a bar, beam, plate, or shell element. In the matrix analysis method, these relationships are generally determined by using strength of materials and basic structural theory. Thus, the method is limited to simple shapes such as rod, beam, and frame elements. In contrast, the finite element method is thought of as using more complicated elemental shapes such as plate or shell elements, The procedures used for deriving the force-displacement relationships for these elements generally involve minimizing a functional such as total energy. OF course, these minimization procedures can also be applied to the simpler ele- ments. This text uses, in addition to the structural analysis approach, the principles of virtual work and minimum potentixl energy to derive the elemental stiffinesses and global stiffness equations. In addition, topics such as non-nodal forces and the assemblage of elements can be developed easily using these techniques. The author believes that computer programming is an essential part of learning the material presented. Naturally, the student is not expected to write commercial-quality code; however, the organization of program steps and understanding of algorithms is greatly enhanced by a hands-on programming problem. As a result, code fragments per- meate the text in areas where procedures and algorithms are discussed, and problems for most chapters include the writing and use of a computer program for analyzing structures using the Lype of element presented in that chapter. And, once one program has been ‘written, it can be easily modified and extended to deal with more complicated elements by simply addressing additional elemental degrees of freedom, coordinate transforma- tions, and the differences in the elemental stiffness matrices. In this way, the author hopes that the student will gain a firm grasp of the techniques used in the ever-increasing number of commercial structural analysis and design computer programs available, Although special topics and energy methods are presented in Chapters 7 and 8, respectively, some instrurtors may want to include sections of these chapters at earlier stages of development of the method. Thus, derivation of elemental stiffnesses and treatment of non-nodal loads using the principle of virtual work could be introduced in parallel with the development of material presented in Chapters 1 and 3, In addition, material presented in Chapter 7 such as elastic and inclined supports, and hinges in beam and frame members, could be considered when discussing the truss, beam, and frame chapters. Finally, Chapter 9 presents a brief introduction to the finite element method. This is accomplished by using work and energy methods to derive the stiffness matrix and structural stiffness equation for the three-node triangular element. ‘The author is indebted to a number of reviewers for their useful comments and valuable suggestions. Specifically, the author wishes to thank Dr. Jack H. Emanuel, University of Missouri, Rolla; Dr. William J. Hall, University of Illinois; Dr. Ronald S. Harichandran, Michigan State University; Dr. Eric Lui, Syracuse University; and Dr. Everett McEwen, University of Rhode Island. Robert SennettCHAPTER 1 | ANALYSIS OF ONE- DIMENSIONAL BARS2 Analysis of One-Dimensional Bars Chap. 1 1.1 INTRODUCTION Tn matrix analysis of structures. wo formulations are possible (1) the force method (flexibility method) (2) the displacement method (stiffness method) The difference between these formulations involves the selection of the variables used as unknown quantities. The force method, as you might expect, uses forces as the unknowns for which a solution is desired. Similarly, displacements are the unknowns in the displacement method of analysis, We an identify the classical methods of structural analysis with one or the other of these techniques. For example, when using the force method the equilibrium equations are first used. Additional equations are found by introducing compatibility conditions. Consider the beam shown in Figure 1-1 Ry Rey Figure f-1 ‘The beam shown is statically indeterminate to the first degree, That is, there are four possible reactions and only three equilibrium equations available. The possible reactions present al any support can always be determined by determining the displacements that are restrained. A reaction is possible in a given direction if the displacement in that direction is restrained. For the beam shown, the horizontal displacement is prevented at Point a and the vertical displacements are prevented at all three support points, Thus, four reactions are possible. Since we are considering a two-dimensional problem, only three independent equilibrium equations are available. If we write the equilibrium equations we find: SOR =0 Re SOF =O Ray + Ry + Rey 0. YM =0 — Rey(2l) + Riy(l) - PU/2) =0 Notice that the last two equations contain three unknowns. Clearly, we need an additional equation.Sec. 1.1 Introduction 3 ne classical technique for finding a solution to this problem is the method of superposition. As you will recall from your structural analysis coursework, the first step in applying this method is to select a redundant (or redundants if we have a multi-degree of indeterminacy problem) thal, when removed, results in a stable, determinate structure. ‘This is called our “primary” structure, Suppose we select the vertical reaction at point b as our redundant, We next draw a superposition diagram as shown in Figure 1-2. P . fyren * a ; 7 Ak, alk "FE a aR, t= Le =, 1 Figure 1-2 Superposition diagram. From this diagram we can see that in order for the superposition to be valid we must insist that the total vertical deflection at point 6 is zero in order to match the zero displacement at this point on the original structure. Thus, 6; + Ry x at? = 0. This is a compatibility of displacement equation that, when solved simultaneously with the two remaining equilibrium equations, yields the three unknown forces. Thus, the method of superposition is an example of the force method of analysis: ‘This problem can also be solved by other techniques such as slape-deflection or a technique derived from it, moment distribution, Again secall from your structural analysis courses that when using the slope-deflection method the unknowns are the joint rotations and translations. Equilibrium of the joints yields equations in these unknown displacements. Having found the displacements, the slope-deflection equations are then used to determine the forces (moments) at the joints, or nodes, of the members. Thus, the method of slope-deflection is an example of the displacement approach to solving structural problems. ‘The displacement method deals with kinematic indeterminacy rather than static indeterminacy, The degree of static indeterminacy of a structure is equal to the number ‘of equations, in addition to the equilibrium equations, necessary to determine completely the reaction and member forces. The degree of kinematic indeterminacy is equal to the total number of degrees of freedom of a structure that must be constrained to ensure zero displacements of the structure, excluding boundary constraints. ‘As an example, consider the two structures in the figure that follows. a @ ®4 Analysis of One-Dimensional Bars Chap. 1 I axial deformation is neglected, which is generally the case for beams, (a) is one- Agree statically indeterminate (4 reactions, 3 equilibrium equations), and three degrees kinematically indeterminate (3 rotations at the supports), Part (b) has four degrees of static indeterminacy (4 vertical reactions and 2 moments at the fixed ends), but only two degrees of kinematic indeterminacy (2 rotations atthe intermediate supports). The formulation of a structural problem is generally much simpler when the dis- placement (or stiffness) method is used. Note that the more highly statically indeterminate the structure, the fewer displacements need to be found. OF COtiLoe, ie Wuubet UE Uis- Placements that need to be found represents the number of simultaneous equations that Must be solved, Most commercial computer programs use the stiffness approach, and we will use only this method in this text. 1.2 NODES AND DEGREES OF FREEDOM As we have seen, the stiffness method uses displacements as the unknown quantities. Furthermore, these displacements are those of the nodes of the members, The nodes ate points at which equilibrium d and displacements found. They are generally located at the ends of the me most common structural shapes such As rods or beams. However, formulations are possible that place nodes at interior points of a member. We have noted in the previous section that the number of degrees of freedom at a node is equal to the number of possible displacements of that node, That is, the number of displacements that need to be specified in order to define uniquely the Position of a node equals the number of degrees of freedom of that node, ‘Therefore, for one-dimensional bar or rod element, where displacements are restricted to a translation the axial direction of the bar, each node has a single degree of freedom. For a two- dimensional truss element, each node has two degrees of freedom that correspond to translations in two directions (say x and Y). For a two-dimensional beam element, we have one translation and one rotation possible at each node, yielding two degrees of freedom per node. For « two-dimensional frame element, each node has the capability of translating in two directions and Totating about one axis, Thus we have three degrees of freedom for each node. Since displacements constitute the unknowns in the stiffness formulation of a problem, the number of degrees of freedom will be indicative of the Humber of simultaneous equations that we will eventually need to solve, Of course, each Structure has to have & number of restraints (such as supports) in order to remove any gid body motion that could be present. The actual number of equations that must be Solved is equal to the total number of degrees of freedom of the structure minus the number of restraints present. 1.3 STIFFNESS DEFINITION If'we have a simple linear spring fixed at one end with a force F applied at the other end, We know that the relationship between the applied force and the resulting displacement is F = ky where x is the displacement and k is called the spring constant or the springSec, 1.3 Stiffness Definition % srifiness. We have a Tinear spring since the force-displacement relationship is Tineat with © Notice that & is the force corresponding to @ unit displacement x = 1. Of course, we can expand the concept of force-displacement relationships to structures with multiple degrees of freedom, When we do this, we generally write these relationships in matrix form, (F] = (Ellul). For a two-degree-of-freedom system this matrix equation is expanded-as shown below: fF feu kn] fn) LR Ur i Le! where the fs are called the stiffness influence coefficients and form the stiffness matrix. Suppose that we let ny = Land ap =0. Then Fy = kin. That is, ky is the force at point 1 corresponding to a unit displacement at point 1 and only point 1. Similarly, yy is the force at point 2 corresponding to a unit displacement at point 1 only. In fact, in general, the stiffness element k;; is defined as the force at i corresponding to a unit displacement at j and j alone. ‘This force is that required to maintain equilibrium in the displaced configuration, We wit! now use this definition to determine the stiffness matrix for a simple structure, Consider the following problene: The structure shown in Figure 1-3 consists of two bar elements with individual stiffnesses (spring constants) ky and 2, and three nodes. The two elements are connected at node 2, We wish to determine the stiffness matrix for the entire structure using the definition of stiffness, Notice that the structure has three degrees of freedom; three translations, one at each node in the axial direction of the bars. Thus, we will generate 2.3 x3 stiffness matrix, 1 2 3 ___o_ 8 ky ky Figure 1-3. Two-element structure. ‘To accomplish our task it is necessary to introduce a unit displacement at each node, one at a time, and find the forces that must be present at the nodes in order to maintain equilibrium, Forces and displacements will be considered positive when acting toward the right. Introduce a unit displacement at node. 1, keeping all other nodes at their original positions, Figure 1-4 shows the free-body diagrams of the nodes and elements for this configuration. Bel RB Fy By a ky, KB, ibs kid Figure 1-4 Unit displacement at Node 1 Node 2 Node 3 node 1 ‘To maintain equilibrium of node 1, force Fi, which is the force at node 1 cor- responding to a unit displacement at node 1, or Ay, must be equal to kd). Thus,6 Analysis of One-Dimensional Bars Chap. 1 ky) = h18, = ky (Since 6; = 1), For equilibrium of nods 2, force F;, which equals ka, must be equal to £3). Thus ko, = yd) = hy Since there are no displacements of nodes 2 and 3, there is no force in element 2 and F5, which equals ks), is zero, Thus, kyy = 0. We next introduce a unit displacement at node 2, ‘The corresponding free-body diagrams are shown in Figure 1-5 be=1 F Fy By 0 -smmee- 2 em 0 kb, hots bd 6, 6, Figure 1-5 Unit displacement at Node t Nods 2 Noie3 node 2. Proceeding in @ manner similar to the previous case, realizing that 6) = 1, we find Fuk =-h, F Ay thy, and Fy = kay = hy, Finally, we introduce a unit displacement at node 3. Figure 1-6 shows the free body diagrams corresponding to this case. bs=1 Fy, Fy Fy Te mmmmes 6+ -pmemmes; 8 6, kb, Figure 1-6 Unit displacement at Node 1 Node? Node3 node 3. We find Fisky=0, h=hky =k, and Fy = hy = by ‘Thus the 3 x 3 stiffness matrix for this structure is ko kh 0 [K]=|-b bth -k 0 ohy ky ‘We can achieve the same result by superposition of the forces in Figures 1-4, 1-5, and 1-6 when arbitrary displacements 5, are imposed, To illustrate, from Figure 1-4 where a displacement 6, is introduced, we have Fi =k, y= —kb,, and Fy =0. For a displacement 5, Figure 1-5 yields Fi hid, Fr = (ki +h2)b2, and Fy = — kd),Sec. 1.9 Stiffness Definition ? Finally. for a displacement 63, Figure 1-6 gives Fy =0, Fy= keds, and = Since we are dealing with a linear system, superposition is valid. Thus, if we introduce all three displacements simultaneously, the nodal forces required to maintain equilibrium will be the sum of the forces obtained by introduction of each displacement separately. Adding the forces yields Fy = ky8) — kid2 + (08s Py Fy = (Obs ~ habe + ads 181 + (hy + ka): — hb Writing the above equations in matrix form yields F ky ky 07 8 n| -k kth | {a} FILO mk bt bbs (F}=(K18). or Suppose we attempt to solve this equation for the displacements 4; through 33. (a) = {KT F) Clearly, we need to calculate the inverse of the structural stiffness matrix [K), In order for n matrix to possess an inverse, its determinant must be non-zero (see Ap- pendix A), Calculating the determinant of [K] by using Cramer's rule and expanding using the first row we have kth kh -h -k why hy 0 hy ini=h| |-e09| = ky (baka +1 — G) + ka(krk) = 0 ‘Thus, the stiffness matrix is singular. This means that we have an infinite number of solutions to the equation. Referring to Figure 1-3, we note that there have not been any constraints imposed on the structure. Physically, with no support conditions specified, rigid body motion can occur, resulting in an infinite number of possible displaced con- figucations, As we shall see, once support conditions are introduced, « unique solution will be possible as long as the structure is a stable one. Note that the stiffness matrix is symmetrical and has positive terms on the main diagonal. This will always be the case in structural problems, Positive terms must be on the main diagonal since a positive displacement at a node requires a positive force at that node, Also note that the terms in each column add to zero. ‘This is a direct consequence of nodal equilibrium for unit displacements of each node individually.8 Analysis uf One-Dimensional Bars Chap. 1 1.4 INDIVIDUAL ELEMENT STIFFNESS In the previous section we derived the stiffness matrix for a two-member structure di- rectly. Since our ultimate goal is to apply the matrix analysis approach to systems with large numbers of degrees of freedom, we would like to generate the structural stiffness matrix for any structure by combining stiffnesses of individual members. Toward this goal we now derive the force-displacement relationships for a single bar element, That is, {F) = fi. Consider the single element shown in Figure 1-7. FR uy UR Figure 1-7 Single-bar element. Clearly, this element has two nodes and therefore two degrees of freedom. We need to determine the elements of the stiffness matrix (&] in the equation Fil [ku kn] fue FeS [ka ka] Lue Note that when up = 0 and wy, = 1, Fy =k; and Fe = hoy, which agrees with our definition of stiffness kj; as the force at i corresponding to a unit displacement at J. In addition, when wg = 1 and uz = 0, Fy = ky and Fy = kp. Therefore, to find the elemental stiffness terms, we need to introduce unit displacements at each end of the bar, one at a time, and find the corresponding forces. Recall from basic mechanics of materials that the change in length of a prismatic bar loaded by an axial force F at its end is given by the expression = FL/EA. Introducing a unit displacement at only the left end yields u, = F,L/EA. Thus, F, = EA/L = ky). Also, Fp = —F, = —EA/L = ky Next we introduce a unit displacement at only the right end giving ug = 1 = FRL/EA, Thus, Fy = EA/L = ky. FP, = —Fy = -EA/L = kx. The elemental stiffness matrix becomes _f BA/L -EA/L] _ l= w= | Eee pale] =e |_| Hl and the force-displacement relationship for this one-dimensional bar or rod element be- comes Fil _ 1-1) fu {ra}vene [a 7) {0} Note that in the force-displacement relationship it is important to keep the order of forces and displacements as left first, and then right, since this is the order for which the elemental stiffness matrix was derived.Sec. 1.5 Combination of Elemert Stifinesses 9 1.5 COMBINATION OF ELEMENT STIFFNESSES Figure 1-8 shows free-body diagrams of the members and joints of the two-member structure shown in Figure 1-3. Fy Fp Fy. Oe ne OE Fi Fy Fy Fa Figure 1-8 Free-body diagrams of Node 1 Node 2 Node 3 elements and nodes. Note that the F; forces are either applied nodal forces or reactions. From joint equilibrium, F) = Fy, Po = Fat Fan, and Fy = Fp. Also note that the forces Fy correspond to forces at the i" node of the if element, Since we cannot specify both 2 force and a displaceuont at the same node, at cach node we will know either the applied nodal force or the nodal displacement. To illustrate the combination of elemental stifinesses, consider the example shown in Figure 1-9. ‘1 Bbuly 2 EyAnl, 3 Figure 1-9 Two-element bar Element 1 Element 2 structure. Example 11 Referring to Figure 1-8 and using the elemental force-displacement relationships we have: Element 1: Fi / = (BA fal Aeeues Element 2: Fo) _ 1-1] fia {B}-teo[ ts] Letting £ Ai/Li =k: we have Fu) _,f 4 4) mye Li be } an) rlemLt Tt) 210 Analysis of One-Dimensional Bars Chap. 1 We want to form (] = (K]u) for the entire structure by combiving these individual clement stiffnesses. Since there ate three nodes and therefore three degrees of freedom for this structure, the force matrix {F| will be of order 3 x 1, the stiffness matrix [K] of order 3 3, and the displacement matrix (u) of order 3 x 1. The overall force-displacement relationship for the structure is constructed by placing the coefficients of 1, 2, and wx and corresponding Fi;’s into appropriate locations in the structural stiffness mattix [A]. For example, the first column of [&} in equation (1.1) multiplies wy. this ("}-[4 = 3] Next, the second column of equation (1.1) and the first column of equation (1.2) multiplies up giving Fy kok =] fu PutOn}=|-k hth -| fm to - -b -J}t- Finally, the second column of equation (1.2) multiplies wy giving Al Komk 0D fay Fat Fraps] hy ky tke ky |p ug Fa 0 = bd las Now, Fi = Fy, Fy = Fo + Fra, and Fy = F3 from the previous equilibrium equations. The final structural stiffness formulation becomes Ky koh OT fu Pop =| kt kith -k | pur KR 0 ak fod lag Also, remember that Fi, F), and Fs are either reactions or nodal applied loads. Naturally this structural stiffness matrix is identical to the one obtained eatlier for the same two-element structure by direct means, One procedure that can automate the construction of the structural stiffness matrix from the individual elemental stiffness matrices involves identifying the rows and columns of these individual elemental stffhesses with the global displacements associated with them, For example, using the stiffness matrices in equations (1.1) and (1.2) we have: Member 1: 1 2 ko -h 1 ak ky ]2 Member: 2 3Sec. 1.5 Combination of Element Stiffnesses W where the nimbers identifying the rows and columuas are those of the w displacements associated with each element. The structural stiffness matrix is then constructed in the following way: (1) Create a square matrix that is of order equal to the total number of degrees of freedom of the structure (the structure! stiffness matrix). In this case a 3 x 3 matrix is created. (2) Place the elements of each individual stiffness matrix in the rows and columns of the new’ matrix corresponding to the global displacement coordinates, (3) If there is more than one element to be placed in the same location in the overatl structural stiffness matrix, the elements are added at that location. For example, pp in the structural stiffness matrix will be ky +p since &y for member 1 and fy for member 2 both lie at global coordinates 2-2, ‘We can visually represent this assembly process by overkeying the individual elemental stiffness matrices on the structural stiffness matrix. Figure 1-10 illustrates this process. 12 3 Element | _.~ Figure 1-10 Combining elemental Element 2 stiffhesses For the problem being considered, ny = 0, Fy = 25t, and F; = —10*. The above equation becomes i ky -h «07/0 asthe) i hth be |} (3) -108 Ok kod Cus Note from this equation that wé have two unknown displacements, ua and u3. We next reduce this matrix equation to one that contains only the unknown displacements. We perform this task by noting that the first column of the matrix will multiply the known zero displacement and therefore can be eliminated. The first row can also be eliminated at this point since it will only be used to determine the reaction Fy after the two unknown displacements have been found, Another way to think about this process is to notice that if we would expand this matri: equation, the last two equations would contain the applied nodal forces and the two unknown displacements. After eliminating the first row and column we have 25) [ath —h {2} 7 a 14) -10 [ ky 2] 43 Ha Since the elements of the stiffness matrix in equation (1.4) are known, this matrix equation represents two scalar equations in the two unknown displacements 12 and 3. The left hand side of equation (1.4) contains the applied nodal toads.12 Analysis of One-Dimensional Bars Chap. | 29 x 103 ksi, Ay =2 in?, Ay = 1 in’, /10 in = 58 x 10° Win, and ky = 29 x 10? Equation 1.4 becomes | Adel 87x10? -29 x 10? ual ~10 | ~ 1-29 x10? 29 x 10? J us Solving equation (1.5) we find uy = .002586 in and ws = —.000862 in ‘We determine the reaction F, by expanding the first row of equation (1.4). This yields Fy = —kyug = —15k. The member forces are found oy using wp and 3 mn equations (1.1) and (12) as _ast [nse] 1] oxesse} = | is] Fy) _ al 1 “I arses | bal {me} <0 1 1] | -o008e2 } = { 108 Referring to Figure 1-11, the member free-body diageams are shown with the actual directions of the member forces. Remember that positive member forces act 10 the right along the axis of the member 10 in, Then te) k k 13 1st 10k 10" Figure 1-11 Member free-body + EE diagrams, To illustrate the process further, consider the following problem: Example 1.2 E+ 10 x 10%pi @Mor @wv @ w Member! Member? Member3. “Figure 1-12 Using the material and geometric properties given, we first write the elemental force- displacement relationships. Member 1: 1 2 joe Le a): 2 Member 2 3 (E]-eo [2 Ge Member 3: 3004 (R= ft aRSec. 1.5 Combination of Flement Siifinesses 13 Combining, using the labeled cows and columns, we have 1234 raio-1 0 07 fm)t = 342 | whe ag 0 -2 3 -1])wef3 ee) 0 o-t tstuasa Since both u) and i are 2ero because the ends of the structure are fixed, we eliminate rows | and 4 and columns | and 4 giving the following equation to be solved for the unknown displacements us and 43; Sp 3-27 fw {| [2 stn] = This equation is called the reduced structural stifness equation. Solving, we find ty = —.001 ia, and 13 = -.004 in, Erom the complete set of equations we next find the reactions Ry and Ry. Ry = 10x 10°(.e; —u2) = 1000 Ib = If, Ry = 10x 15 (03 +4) = 4000 Ib = 4, Note that these reactions creale overall equilibrium with the applied loads (see Figure 1-13). 5 10 Ss, a 4k Figure 1-13 Applied loads and | TO, reactions. We next determine the member forces. Member 1: Fal. ‘| 1 “yf 0 | _f 1000 {pai | 19x00 -1 01 oon} =[ -tcoe Member 2: Fo|_— i, sf 2 2) {00 =| o0o# pal =iox1e [a | “oo =| tne Member 3: Bley OY = [toms Figure 1-14 shows free-body diagrams of the nodes and members. 10008 1000# 6000# 6000# 400% 400i Sa a 5000# 10000# — — ee e+ + 1000# 1000# 10008 6000# ©6000 4000F 4000 4000# Figure 1-14 Free-body diagrams of members and nodes. We could also obtain our reduced structural stiffness equation, (1.7) from the com- bined stiffness equation, (1.6), by using a partitioning technique on equation (1.6).14 Analysis of One-Dimensional Bars Chap. 1 Consider rewriting equation (1.6) in the following way: Fy\ _ | &pp Kos] uy (18) F, Kop Kas | (its P,, = submatrix of applied loads Fs % u; = submatrix of known displacements, including support movements. where uomatrix of reactions ubmattix of unknown displacements Expanding equation (1.8) we have Fy = Kpptty + Kyat 9) By = Kaptly + Kesty (1.10) Note that the number of applied loads and unknown displacements are always equal. Thus, Kpp will always be a square matrix. From equation 1.9 ty = Keyl (Fy — Kpots) (ui We can solve equation (1.11) for the unknown displacements, If reactions are desired, equation (1.10) can then be used. For the previous example we rearrange equation (1.6) in the following way: 5t 3-2-1 07 (m 108 | 2 3 0-1) Jus R = 10x 10° “1001 olfu (1.12) Ry o-1 0 ts hu Referring to equation 1.8 we have (5)={} ={2} Kel=[2 3). wm=[% t] tk = [74 fl tkal= [5 ‘| two= {2}. w= {f}={9} From equation 1.11, since {u,) is null, we have (up) = [Kyp]"!{F,) or {e}-[2 ay"{.3} (1.13)15 Sec. 1.6 Structures with Specitiec Non-Zero Displacements Comparison of equation (1.13) with equetion (1.7) shows that the solution for uz and it will be identical. 1.6 STRUCTURES WITH SPECIFIED NON-ZERO DISPLACEMENTS Consider xample 1.1 presented in seviiuu 1.5. Instead of applying the 10k load at node 3, let us specify a displacement ny = ~.000862 in (the value calculated previously). _ Ky=58x10'Win ka= 29x 10" Win 1 2 3 Cie a ee use 000862" Blement 1 Element 2 Figure 1-15 Our matrix equation for the structure now becomes Ri ko kh 0 0 {= -k kth -+]{ uy | a4) Ry 0 =k — be} | 000862 For this simple problem we can expand the second row of this equation to obtain (ky + ka)ug — ke (—.000862). Solving for up yields w> = 002586 in as obtained previously. If we use the partitioning technique presented in section 1.5, we reorder equation ast 1.14 as shown below. 2st bth -k —k 1 { } . | = : ; { i } Ry Ry ky 0 ky J (~.000862 ‘Thus, Rp =tkithl, Ky =[-k ke) ky kh 0 woa(8). aff 8 = [254] Ht, = i, LO#16 Analysis of One-Dimensional Bars Chap. 1 {ams} From equation 1.11 we have: Of course, this is the solution obtained previously. There is also a numerical procedure that can be automated and is capable of treating complex problems. Since many matrix inversion routines make use of the symmetry of the global stiffness matrix, it would be beneficial for our procedure to maintain symmetry. The following method for handling non-zero displacements does this. (1) Eliminate the rows and columns corresponding to zero displacements as before. For the previous example we obtain 2st a [hee hy 1 Rsf~ | -k ky | | -.000862 (2) Ifa displacement uy is specified at coordinate 12, multiply ky, by a large number M and replace the force value in row n by az x MX Ein. We bave 25k [ith bh | fie us 000862 x tax MJ | ky Mx ho} Las ad (3) Solve for the dieplacements in the normal fashion. To illustrate with this example, we invert the modified stifiness matrix in equa- tion (1.15) to obtain Mbo/(Mkz(ky + a) — 8) kp/(same denom.) Koj(same denom.) ——_(k; + 2) (same denom.) tar = | Premultiply the force-displacement equation by [K]"! in order to solve for wy and us, We find: 25k — .000862K3 fale + - IM (25ko/M) — 000862(kyke + 13) hi +B - G/M In these equations, the last term in the denominator is very small due to the presence of the large number M. This term is neglected in comparison to the other terms. In addition, the first tetm in the numerator of the expression for #3 is also neglected in comparison to the other terms. Substituting the values for k; and ky of $8 x 10° ksi w= 13 =Sec. 1.7 Non-Nodal Forces 7 that > = 002586 in and 1; = —.000862 in (the and 29 x 10° ksi respectively, we fi specified displacement), 1.7 NON-NODAL FORCES Up to this point we have dealt with structures that have forces applied only at the nodes, + 8e need io be adic w address problems that have concentrated and distributed 16 shows such a structure, loads applied between nodal points. Figure Pox) —L x Figure 1-16 Non-nodal loads on an element. OF course, one approach that could be used to solve this problem would be to add several more nodal points—one at the concentrated load so that the load will be acting al a nodal point location, and several more in the region of the distributed load. We would then break up the distributed load into a series of concentrated forces acting on the additional nodal points in that region. The magnitude of the concentrated forces would generally be based on tributary length. For example, in Figure 1-17 the portion of the distributed load assigned to each node is indicated, The total load applied te a nods would be that load contained in a length half the distance to each adjacent node. — abo ° XX 0 The number of additional nodes required would be dependent on how rapidly the distributed load changes, that is, the gradient of the Joad, The faster the magnitude of the load changes, the more nodal points necessary to achieve a reasonable approximate solution to the problem. A major difficulty arises, however, when using this technique. Since we have added nodes we have added degrees of freedom and consequently increased the size of our force, stiffness, and displacement matrices. As a result, solution time and storage requirements will increase, We need to find a method that will enable us to determine nodal forces at the ends of the bar that are equivalent to the loads applied between nodes. The use of fixed end forces allows us to perform this task. Suppose we fix each node in position before applying the loads. After the loads have been applied, forces are required at the nodes in order to maintain zero displace- ments. These are the fixed end forces. If the nodal restraints are now removed, the18 Analysis of One-Dimensional Bars Chap. 1 be those caused by loads equivalent to the forces are removed w to displace, Th be due to lo: the fixed end farees but opp: efore, equivalent forces that must ed to a element in order to account for n fixed end forces. A a 4 Original: Fixed End Forces Equivalent Nodal Forces (a) (b) © Figure 1-18 Superposition diagrs 1 the nodal displacements we solve part (c} of the superposition di the displacements found in part using the equivalent nodal structure, However, we must add the forces in beth parts the nodes of the original structure. we can write ‘The solution of equation (1.16) yields the true nodal displacements, The true nodal forces acting on the members are found by using 0 — {Peauie} = OK (tls + {Fyined-end}; “where all subscripted terms refer to the element being considered, After determining the true nodal forces we can then construct the axial force dia- grams by considering free-body diagrams of the original structure with the actual loads applied. ‘The determination of fixed end forces and the use of equivalent nodal forces are illustrated in the following examples. Fe={k) . & Example 1.3 Determine the equivalent nodal forces for the uniformly distributed load shown. The cross- sectional area of the bar is constant. We shall determine the fixed end forces and then reverse their sense. P(x) = p g(constan’) Rp oe Rye t-30 Bar element with a uniformly distributed load.Sec. 1.7 Non-Nodal Forces 19 both nodes are fixed. Drawing a free-body diagram of a portion of the structure we have 7% foe ae | Rt — P(x) rns Figure El-3b Portion of bar from which P=) = Ry — pox. The strain at ps of the bar is zero, and is obtained by integrati 7 o= wae) (Ry = pox)dx = (1/AE) [aw ~ pol? /2] 0 from which Ry = pol/?. From overall equilibrium, Ry + Ro = pol. Thus, Ro = pol/2. Since both Ry and Rp are p. forces. Reversing their dir lds the equivalent nodal loads. Note that each equivalent force is equal to one-half the total load. for these Figure El-3e Equivalent nodal ?; V2. pecmmremmeresramren —— Py V2 igads, Exampie 1.4 Solve for the nodal displacements and member end forces for the structure shown in Fig- ure El-4a, Construct the axial force diagrams for each member. 1 kipfin ky 10in, — 25* 10 in 10k Element 1 Element 2 Figure Ei-4a Example 14. Since the uniformly distributed load totals 10 kips, 5 kips acting in the positive direction (to the right) are added to the applied nodal forces. This yields the nodal forces shown in Figure El-4b. Using values of #1 = 58 x 10° kin and kz = 29 x 10? k/in, and equation (1.16), the structural equation becomes . 0 2] ua R 58x 10? —58 x 107 0 30 P= | 58x 10? 87x 10-29 x 10 -# 0 -29x1P 29x10?20 Analysis of One-Dimensional Bars Chap. 1 Element 1 Element 2 Figure El-4h Total nod: ne the Rest eam: and onlimn Ive vor ing for member forces we have feud=s[ 4 “i]Looail + ‘Thus, Fy = -25 kips, Fay = +25 kips. frelee Ly c] [Sos] +15] Thus, Fox = 0, Fp = ~10 Kips. Note that the last terms in the above equations reflect the addition of the fixed end s free-body di odes and members, and Fi 10 = | we — Figure El-4e Free-body diagrams of members and nodes. “TI S110 Figure El-4d Axial force diagram, Before constructing the axial force diagrams, free body diagrams such as those of Figure El-4c should be drawn with the forces acting in the true directions, Remember that our matrix solution sign convention requires positive forces to act in the positive digection of the axis of the member independent of the end being considered. This is different from the standard strength of materials sign convention where tension in the member is considered positive. By constructing the free-body diagrams we can eas relate the actual force directions to the strength of materials sign convention for forces. Another very important reason for drawing the free body diagrams is that it gives us the ability to check equilibrium of the members and nodes. This is essential with any output from a computer program, since errors in input or the program itself can be found. ppereorvresrenseeihisllthagaheremenaseSec. 1.8 — Thermal Effects 21 1.8 THERMAL EFFECTS When a temperature change occurs in a bar element, an axial strain linearly proportion: to this temperature change takes place. This strain is generally expressed as 7 = o(AT), where @ is the coefficient of thermal expansion and has units of strain/?F or strain/*C, and AT is the temperature change. If the bar has a length L, the total change in length due to a temperature change over the entire bar is AL = pL = a(AT)L. We can treat thermal ettects in ros in the same way that we deait with non-nodal forces. That is, we will use the concept of fixed end forces and equivalent nodal loads. As in problems dealing with non-nodal loads, the equivalent nodal forces are the same magnitude as the fixed end forces but opposite in sense. After applying the equiv- alent nodal loads and solving for the unknown nodal displacements, we find the member forces by superposing the fixed end forces and the forces calculated using the nodal displacements. Since o = P/A = Ee = Eo(AT), then P = EAa(AT) and the fixed end forces for the rod are given by 1 FAa(AT) { 4 } Of course, since we are dealing exclusively with linear systems, the superposition prin- ciple is valid and the thermal effects can occur simultaneously with both applied nodal loads and non-nodal loads. Example 1.$ For the structure shown in Figure E1-Sa, determine the nodal displacements and member forces. In addition to the load applied at node 2, member 2 undergoes a temperature increase of 60°F. All bar areas are 2 in?, E = 29 x 10° psi, and a = 6.5 x 10-6 infiny*F 4 10,000# C r / “) 10 in isin member 1 memiber 2 Figure El-Sa Example 15. The fixed-end forces due to the temperature change in member 2 are found using P = EAA/L = (EA/L)a(AT)L = EAa(AT) = 22,620 lbs compression. The equivalent nodal forces (opposite of the fixed end forces) are shown in Figure E1-5b, 10000# ® ® — —— Figure E1-Sb Applied loads and For member 1: tal-se0 THEN KSUM-KSUM+1 [accumulate the number of restraints] NEXT I MKR-NN-XSUM [calculate the order of SKR] We next determine the numbers of the rows and columns that will be kept. 30 FOR I#1 70 NN IF KRES(I)=0 THEN JeJ+1:KEPT (J) =1 NEXT I In Example 1.1, KEPT(J)=(2 3}, which are the numbers of the rows and columns ‘we want to keep when constructing the reduced stiffness matrix SKR. Now we fill SKR and generate a new reduced force matrix that contains the known applied loads. These will be the forces of the original force matrix that are in the rows corresponding to the row numbers kept.Sec. 1.9 Computer Formulation 5 FOR I=1 TO NKk NeKEPT(T) FOR J=1 TO NER MeKEPT (J): S&R(I, J) =SK(W,¥) NEXT J: NEXT 1 T)*F(N) The solution of the resulting equations requires the inversion of SKR and multi- °, since (FR}=[SKR]{UR}, then (UR}= KR} {FR} where {UR} is the matrix of calculated unknown displacements. We must, however, place the calculated displacements in the correct row of the global displace- ment matrix {uv}, which contains all displacements, ineluding the ones that are zero or specified. This can be accomplished as illustrated below, where it is assumed that after inversion of SKR, SKR contains the inverse. In other words, the original elements of SKR have been replaced by its own inverse during the process of inversion. In the following code fragment, U represents the unknown displacements to be calculated and DU represents the complete global displacement matrix. After inverting SKR, c reduced force FOR I=] TO NKR: U(I)=0: NEXT I (zero all unknown displacements) FOR I=1 TO NKR FOR J=1 TO NER U(1)=U(I}+SKR(Z,J)*ER(J} — [eallculate the unknown displacements] WEXT J: NEXT I FOR I=1 70 NN: DU(I)=0.: NEXT I. [zero global displacements] FOR T=1 TO NKR: ND-KEPT(I): DU(ND)=U{I}: NEXT 1 Note that the code fragment above assumes that there are no nou-zero specitied displacements. In addition, we have included forces applied at only the nodes. Next, we calculate the member forces, Note that we can compute the values of the unknown reactions from the original SK matrix and DU displacements, or by using the member forces at the nodes with zero displacements. In calculating the member forces, we need only the force at one end of the member since the force is constant throughout the bar when loads are applied only at the nodes. We choose the right end of the member since a positive force at this end indicates tension in the member, which is consistent with the common strength of materials sign convention. : For clement j, with nodes i and i +1, Fj = kj(wi41 —))- The code becomes FOR I=1 10 NB [loop on number of elements] 2 (I)=(E (I) A(T) /L(Z))* (DU (E41) -DU(Z)) NEXT T P(D), of course, represents the Zorce in member I. We can now print out all displacements and forces. In general, it is always good procedure to print out all data that was entered into the program as well. If this is done, then it becomes much easier to check input data and, if necessary, to debug the program.26 Analysis of One-Dimensional Bars Chap. 1 1.10 SUMMARY cements was loads were derived by considering presented. Equivalent nodal forces due to he correct nodal ; however, the member forces required the addition of the fixed end forces he true end forces. We also o steps necessary to develop a computer program for solution of these one-dim problems, and we Presented algorithms to accomplish this task. In the chapter PROBLEMS ALL For the structure shown in Figure PI-1 R 5 E es fin —y— Sin —p— 10 in ‘Member 1" Member? | Member 3 ‘Areas for all members 2 5. n Figure PL-1 (1) Determine te structural sttiness matrix for the entire structure by using the basic define tion of stiffness (2) Determine the structural stitiness matrix by combining individual elemental stiffness, (3) Ifthe Jet end of the structure is fixed, solve forthe nodal displacements and member forces if Fa = 10 kips, Fs =—$ kips, and Fy = 8 kips. Use the p the method of removing rows and colurans from the original strctural stiffness matrix, E = 29 x 108 psi for all members. ~~Chap. 1 Problems ar 1.2 For the structure shown in Figure P1-2, solve for nodal displacements and member forces. Sin Bi Win 20in 20in B= Zy x 10% psi Member 1 2 3 Figure P1-2 1.3 Solve for nodal construct the axial for ements aad member end forces for the structure in Figure P1-3. Also 2 jin Z ae is Lint 2 ie ‘Loin 4 isin | 7 One ~q B= 30x 10° psi Member 1 2 Figure Pi-3 14 Referring p load at node 3 with a specified displacement of +0258 in. placements and member forces. Use the partioning method and the numerical method presented in section 1.6. Lyin eo 4 pn set tn — 2 Win 35% Win | oonse" Figure P1-4 1.5 Determine the equivalent nodal forces for the loads shown in Figures P1-Sa through PI-5f. abo Figure Pl-Sa x 7 “| Figure P1-Sh2B Analysis of One-Dimensional Bars Chap. 1 Figure PI-Se Figure Pl-Sd Figure PL-se a Figure PLS? Member 2 Figure P16 s cove fragments inthis chapter asthe basis for your work, write « computer program 'e one-dimensional bar problems. It should have the capability to use up to F ld calculate aodal displacements and member tine in Appendix B if desired. it is not the program to deal automatically with non-nodal loads or specified non-zeroChap.1 Problems 29 ‘ou obiained from problem 1.5, 1 body diagrams of each them. mber et directions and Mente 1 2 3 Figure PL-8 1.9 For the structure showa in Figure P1- ¢ the exact solution for the displacement as a function of position along the bar. Model the structure with 2, 4, and 8 elements, and use your computer program to solve for nodal displacements. Compare your computer results exact solution, A =2 in?, B = 30 x 10° psi. 2 Win Figure P19 1.10 Model the tapered bar shown in Figure P1-10 with 2, 4, and 6 elements, using the average with the exact solution, E = 29x 10° Figure PL-10 rease of 50°F. lacements and undergoes a temperatut Using E = 29 x 10° psi and @ = 6.5 x 10 infin/*F, find the nodal member forces, Draw free-body diagrams of each member. ber 2 member 1 mem a ——ee 10 ia 10in 2 sq, in 1sq. in Figure P1-1130 Analysis of One-Dimensional Bars Chap, 1 1.12 In addit member 1 distributed load show on the structure depicted in Figure P1-12, E = 10x 10° psi and a = 12.8 x slacements and member forces, All areas are 30008/in : 7] ¢ 4 I J 20in 20 in ae © Member 1 2 3 Figure P1-12 113 Determi (©) Beam (@) Frame Figure PI-13