Scribd
Upload
10 views

The plastic methods of FEA

O livro fala sobre o método plastico do FEA.

Uploaded by

gabriel.emidiobap
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
10 views

The plastic methods of FEA

O livro fala sobre o método plastico do FEA.

Uploaded by

gabriel.emidiobap
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
You are on page 1/ 213
The Plastic Methods of Structural Analysis B.G. NEAL AHALSTED PRESS BOOK The Plastic Methods of Structural Analysis B.G. NEAL Professor of Engineering Structures and Head of Department of Civil Engineering, Imperial College of Science and Technology THIRD (S.1.) EDITION G LONDON CHAPMAN AND HALL ‘A Halsted Press Book John Wiley & Sons, New York First published 1956 Reprinted once Second edition 1963 Reprinted twice First issued as a Science Paperback 1965 ‘Third (S.L) edition 1977 Chapman and Hall Ltd 11 New Fetter Lane, London EC4P 46E ISBN 0 412 15400 5 (cased edition) ISBN 0 412 21450 4 (Science Paperback edition) Typeset at the Alden Press Oxford London and Northampton ‘and printed in Great Britain by Richard Clay (The Chaucer Press) Ltd., Bungay, Suffolk © 1977B.G. Neal This title is available in both hardbound! and paperback ‘editions. The paperback edition is sold subject to the condition that it shall not, by way of trade or otherwise, be lent, re-sold, hired out, or otherwise circulated without the publisher's prior consent in any form of binding or cover other ‘than that in which itis published and without a similar ‘condition including this condition being imposed on the subsequent purchaser Alll rights reserved. No part of this book may be reprinted, or reproduced or utilized in any form or by any electronic, ‘mechanical or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publisher Distributed in the U.S.A. by Halsted Press, a Division of ‘John Wiley & Sons, Inc., New York Library of Congress Cataloging in Publication Data Neal, Bernard George. The plastic methods of structural analysis “A Halsted Press book.” Includes bibliographies and indexes. 1. Plastic analysis (Theory of structures) 1. Tite. TAG52.N43 1977 624.71 77:24230 ISBN 0-470.99017-1 Contents Preface 1. Basic Hypotheses 1.1 Plastic hinge and plastic collapse concepts 12 13 14 Ls ‘Stressstrain relation for mild steel Elastic-plastic bending 1.3.1 Rectangular cross section 1.3.2 Section with a single axis of symmetry 1.3.3 Effect of upper yield stress Evaluation of plastic moment Plastic hinge assumption for other structural materials ‘Simple Cases of Plastic Collapse 24 2.2 23 24 25 2.6 27 Introduction Simply supported beam Fixedended beam 23.1 Direct calculation of collapse load 23.2 Behaviour on unloading Effect of partial end-fixity Rectangular portal frame 2.5.1 Principle of Virtual Work 2.5.2 Equilibrium equations by virtual displacements 2.5.3 Compatibility equations by virtual forces 2.54 Step-by-step analysis of proportional loading 25.5 Deflections by unit load method Invariance of collapse loads Plastic design Basic Theorems and Simple Examples 3. 3.2 Introduction Statement of theorems 3.2.1 Static theorem 3.2.2 Kinematic theorem 3.2.3 Uniqueness theorem Page xii 3.3 Illustrative example 3.3.1 Equilibrium equations by virtual displacements 3.3.2 Sway mechanism 3.3.3 Beam mechanism 3.3.4 Combined mechanism 3.4 Distributed loads 3.4.1 Maximum bending moment in a member 34.2 Mlustrative example 3.5 Partial and overcomplete collapse 3.5.1 Example of partial collapse 3.5.2 Example of overcomplete collapse 3.5.3 Continuous beams 3.5.4 Fixed-ended beam strengthened by flange plates Methods of Plastic Design 4,1 Introduction 42 Trial-and-error method 43 Method of combining mechanisms 43.1 Rectangular frame 43.2 Two-bay rectangular frame 43.3 Partial collapse 43.4 Disiributed loads 43.5 Lean-to frame 4.4 Other methods for determining the collapse load factor Estimates of Deflections S.1 Introduction 5.2. Load-deflection relations for simply supported beams. 5.2.1 Rectangular cross-section: ideal plastic material 5.2.2 Other cross sections and material properties 5.3 Effects of strain-hardening and shape factor 54 Estimates of deflections at point of collapse 54.1 Assumptions 5.4.2 Basic equations 5.4.3 Fixed-ended beam with off-centre load 5.4.4. Rectangular frame 5.4.5 Partial collapse Factors Affecting the Plastic Moment 6.1 Introduction 6.2 Variations in yield stress CONTENTS age SO 51 52 34 56 58 58 59 62 6B 4 66 68 3 B B 8B 8 82 87 89 95 7 102 102 103 103 107 108 ut i 2 113 7 120 126 126 126 CONTENTS 63 Effect of normal force age 128 6.3.1 Rectangular cross section 130 63.2 section bent about major axis 131 6.3.3 Effect of normal force in practical cases 132 64 Effect of shear force 133 64.1 Rectangular cross section 134 6.42 Isection bent about major axis 139 64.3 Combined effect of shear and axial forces 141 6.5 Contact stresses beneath loads 141 7 Minimum Weight Design 146 7.1 Introduction 146 7.2. Assumptions 146 7.3. Geometrical analogue and Foulkes’ theorem 147 73.1 Geometrical analogue: rectangular frame 147 7.3.2 Foulkes’ theorem 150 7.4 Methods of solution 152 8 Variable Repeated Loading 158 8.1 Introduction 158 8.2 Step-by-step calculations 159 8.2.1 Alternating plasticity 160 8.2.2 Incremental collapse 161 8.2.3 Behaviour when W = W, 166 8.2.4 Effect of eyelic loading on deflection 168 8.2.5 Experimental evidence 169 8.3 Shake-down theorems 169 83.1 Definitions 170 8.3.2 Shakedown or lower bound theorem im 83.3 Upper bound theorem 173 8.3.4 Observations on theorems 174 8.4 Methods of analysis, 174 84.1 Illustrative calculation 175 8.4.2 Partial incremental collapse mechanism 179 8.43 Other methods of calculation 182 844 Estimates of deflections 183 8.5 Relation to design 183 ‘Appendix A: Proofs of Plastic Collapse Theorems 190 Appendix B: Proofs of Shake-down Theorems 193, xii CONTENTS Answers to examples age 197 Author Index 201 Subject Index 203 Preface Since the first edition of this book was published in 1956, there has been a wide- spread acceptance of the concept of limit state design. It is also generally recog. nized that the appropriate ultimate limit state for many steel frames is plastic collapse, so that the design of such structures is based upon an assessment of the plastic collapse load, an appropriate load factor being provided. Whereas in 1956 the case for the use of the plastic methods had to be argued, this is no longer necessary, and the presentation has accordingly been shortened. The Principle of ‘Virtual Work has been used throughout to unify the treatment. The book is concemed with the plastic methods of analysis for beams and plane frames, which are based upon the simplifying plastic hinge assumption. It does not discuss the conditions under which members which have entered the plastic range fail by instability. Nor does it deal with other problems of import- ance in design, such as the behaviour of full-strength welded joints. Nevertheless, the plastic methods, as presented here, can fairly be claimed to be an essential ‘weapon in the armoury of any competent structural designer. Digital computers are now used extensively to solve structural problems, both of analyis and of design. Some programs which have been developed for frames, analyse their behaviour when the simplifying assumptions of the plastic methods are discarded, so that the actual properties of the members are taken into account. Others deal with the optimisation of designs subject to various forms of constraint. These developments have not been dealt with in this edition, although a few are referred to in passing. Only those techniques which are suit able for hand calculation are included; these need to be thoroughly understood as a prelude to the use of computer programs. Earlier editions of the book contained a comprehensive bibliography. This would now be inappropriate in view of the exclusion of a full discussion of computer-based developments, and so there are few references to the important work of this nature published recently. A selection of references to the classical work which established the basic theory has been retained. ‘The author is most grateful to Mr John Cima for his excellent work in preparing the illustrations, and to Mrs Fileen Wyatt whose capacity for the speedy production of an accurate typescript is unsurpassed. London June 1977 B.G. NEAL 1 Basic Hypotheses 1.1 Plastic hhinge and plastic collapse concepts The plastic methods of structural analysis are now widely used in the design of steel frames, which carry load by virtue of the resistance of their members to bending action. Multistorey, multibay rectangular frames and single or multibay pitched-roof portals are familiar examples of this type of structure, and the defi- nition also includes simply supported and continuous beams. For such structures Baker (1949) pointed out that the most economical and rational designs are achieved by the use of the plastic methods. The plastic methods also have the advantage of simplicity, The objective of the plastic methods is to predict the loads at which a framed structure will fail nt of excessive deflections. It is appropriate to begin by examining the behaviour of the simplest type of structure in this category, a simply supported beam carrying a central concentrated load. Fig. 1.1 shows the results of an early test carried out by Maier-Leibnitz (1929) on an F-beam spanning 1.6m. The beam remained elastic up to a load W of about 130 KN, when the yield stress was attained in the most highly stressed fibres beneath the load. At a load of about 1SOKN, the central deflection § began to increase very sharply for small increases in the load. The beam eventually failed cata- strophically by buckling at a load of 166KN, but before then collapse had already effectively occurred due to the development of unacceptably large deflections. ‘A slight idealizati i could grow indefinitely under a constant-load-of TSOEN, as shown by the broken line in the figure. This assumption disregards the small additional load- carrying capacity which the beam actually possesses above this load, and is therefore conservative. The assumed indefinite growth of deflection under con- stant load is termed plastic collapse, and the load 150 KN at which it occurs is the plastic collapse load, denoted by We This behaviour can be described on the hypothesis that a plastic hinge de- velops at the centre of the beam at the load W., when the central bending moment is 0.4 W, = 60kNm. The characteristic of this hing is that it can only undergo rotation when the bending moment is 60kNm, but while the bending ‘moment has this value the rotation can increase indefinitely, thus permitting an indefinite growth of deflection. The bending moment required to develop a 2 THE PLASTIC METHODS OF STRUCTURAL ANALYSIS -_ PLASTIC COLLAPSE Simm) Fig. 1.1 Test on simply supported beam (after Mater-Letbnitz) plastic hinge in this test, 60kNm, is termed the plastic moment of the beam, and is denoted by Mp. Itis related to the yield stress of the material, as will be shown in Section 1.3. The plastic methods of analysis, based on the plastic hinge assumption, enable the plastic collapse loads of quite complex frames to be found rapidly, as will be seen in Chapters 3 and 4, Their usefulness asa tool for designing steel frames depends on the fact that large deflections are unlikely to develop before the plastic collapse oad is attained. However, it may be necessary ‘to ensure that the deflections developed before collapse are acceptable, and methods for estimating these deflections are discussed in Chapter 5. The plastic methods should only be used for design if the avoidance of plastic collapse is the governing design criterion. There will be cases in which the pri ‘mary problem is to avoid other types of failure, for example by fatigue or brittle facture. These are outside the scope of the simple plastic theory. It is implicitly assumed throughout that no part of the structure will fail by buckling before the plastic collapse load is reached. The problems of buckling of ‘columns under the conditions actually arising in rigid frames when the members have partially yielded, and of lateral instability and other forms of buckling ‘under similar conditions, have been studied extensively. The pioneering work of J. F, Baker and his associates at Cambridge was presented in The Steel Skeleton, vol. 2(1956), and investigations carried out under the direction of Beedle at Lehigh were described in Plastic Desien in Steel (1971). The present position has been summarized by Home (1972) and Wood (1972). Rules are available which enable frames to be designed so that failure by certain types of buckling will not ‘occur before the plastic collapse load is attained, but their discussion is outside the scope of this book. BASIC HYPOTHESES 3 1.2 Stress-strain relation for mild ste! The plastic moment of a steel beam is directly related to the yield stress, as already stated. As a preliminary, it is necessary to review the stress-strain proper- ties of mild steel, the material which is commonly used in the construction of frames. The relation between direct stress @ and axial strain ¢ for a specimen of an- nealed mild steel in tension has the typical form shown in Fig. 1.2(a). The re~ lation is linear in the elastic range until the upper yield stress is reached at a. The stress then drops abruptly to the lower yield stress, and the strain then increases, at constant stress up to the point b, this behaviour being termed purely plastic flow. Beyond b further increases of stress are required to produce further strain increases, and the material is said to be in the strain-hardening range. Eventually a maximum stress is reached at c, beyond which the stress decreases due to the formation of a neck in the specimen until rupture occurs at d. The maximum stress is of the order of 400N/mm? and the strain at fracture is of the order of 05. (a) Fig. 1.2 Stressstrain relation for mild steel in tension (@) Behaviour up to rupture () Yield range The yield range Oab is of the most interest from the point of view of plastic theory. Since the strain at b is generally of the order of 0.01~0.02, the yield range can be examined more conveniently if the strain scale is enlarged, as in Fig. 1.2(0). In this figure the upper and lower yield stresses are defined as oy and 09, respectively, the slope of the initial elastic line Oa is Young’s modulus £, and the 4 THE PLASTIC METHODS OF STRUCTURAL ANALYSIS slope of the initial portion of the strain-hardening line beyond b is defined as E. ‘The strains at the yield point a and at the onset of strain hardening b are defined 8 0 and ey, respectively. If the stress is reduced aftr yield a relation such as ef is observed, the inital slope being Young's modulus. The deviation from lin- earity in such an unloading relation is associated with the Bauschinger (1886) effect. If the stress i increased again after a reduction of this sort, yield occurs at the lower yield stress along eb. This indicates the effect of cold.working in de- ‘stroying the upper yield stress, which only reappears after further heat “The values of the constants defined in Fig. 1.2(6) depend markedly on the composition of the steel and its heat treatment, except for the value of Young’s modulus, which shows very litle variation Data derived by Roderick and ‘Heyman (1951) from the results of bending tests on four annealed steels of dif- ferent carbon content are as shown in Table 1.1 Table 1.1 Effect of carbon content on properties of steel BC a ou & i N/mm?) Bo % EF 0.28 340 1.33 92 0.037 049 386 128, 37 0.058 074 498 119 19 0.070 0.89) 525 104 1s 0.098 1t will be seen thatthe effect of increasing the carbon conten sto increase the lower yield stress og while decreasing the ductility as measured by the ratio e] ¢o. For structural steel e i ofthe order 10 eo, and is ofthe order 0.04 &,s0 that the stress strain relation is very flat after yield. Its difficult to determine the actual tensile stress-strain relation of mild stel iin the elastic range near the yield point, because of unavoidable eccentricities of loading which cause significant bending stresses. However, Morrison (1939) showed that the inital departure from linearity usually observed below the yield point could be ascribed to yielding in the most highly stressed fibres caused by the eccentricity of loading. He therefore concluded that the yield point, pro- portional limit and elastic limit were all coincident. The tests also showed that the values of the upper yield stress showed no more variation from specimen to specimen of the same material than those of the lower yield stress. The unpre- dictable variations in the values of the upper yield stress reported by other observers were therefore concluded to be due to variations inthe ecoentrcity of loading It vas also shown that fora given stel the stessstrain relation in com- pression is practically identical with that for tension up to the point b where strain-hardening begins. BASIC HYPOTHESES 5 The yield phenomenon for mild steel is accompanied by the formation of Liiders’ lines making an angle of about 45° with the axis of the tensile specimen, showing that plastic low oc se planes where tl tes The material within the Liiders’ lines has undergone considerable amount of slip, corresponding to a jump in the strain from a to b in Fig. 1.2(b). The longi- tudinal strain in a yielded fibre therefore varies discontinuously along the fibre, and a stressstrain relation such as that shown in Fig. 1.2(b) only represents aver- age strains over a finite length. The stress-strain relation is often idealized by the neglect of strain-hardening and the Bauschinger effect on unloading, leading to the relation shown in Fig. 1.3(a). Although the upper yield effect is a very real one, it disappears on cold- working and is usually not exhibited by the material of rolled steel sections. Moreover, it will be seen in Section 1.3 that it has no effect on the value of the plastic moment. If it is disregarded, the stress-strain relation becomes that of Fig. 1.3(b), which is often termed the ideal plastic relation. (a) (b) Fig. 1.3 Stresrstrain relations neglecting strain-hardening (2) With upper yield stress (b) Without upper yield stress (ideal plastic) ‘The neglect of strain-hardening in these idealized relations may seem difficult to justify in view of the fact that the strains will certainly enter the strain- hardening range in many members in actual structures. However, by neglecting the increase of stress during strain-hardening, errors will be introduced which are ‘on the safe side, and it will be seen in Chapter 5 that these errors are usually very small, 6 ‘THE PLASTIC METHODS OF STRUCTURAL ANALYSIS 1.3 Elastic plastic bending For a homogeneous beam of given cross section, the relationship between bend- ing moment and curvature beyond the elastic limit can be derived from the stress-strain relation provided that the usual assumptions of the Bemnoulli-Euler theory of bending are made. These are: (a) The beam is bent by pure terminal couples, so that shear and axial forces are not present. (b) The deformations are small, so that stresses other than longitudinal nor- ‘mal stresses are nepligible. (©) The relation between longitudinal stress and strain isthe same in flexure as in simple tension or compression. (@) Originally plane cross sections remain plane, In addition it will be assumed that the stress-strain relation is of the ideal plastic type shown in Fig. 1.3(b), with no upper yield stress. It is further as- sumed that this relation is obeyed by each individual longitudinal fibre of the beam. In view of the discontinuous nature of the yielding process, this assump: tion requires experimental verification; several investigators, notably Roderick and Phillipps (1949) have provided evidence in its favour. Finally, it is assumed ‘that there are no residual stresses in the beam. The analysis is simplified con- siderably ifthe eross section is symmetrical with respect to an axis which lies in ‘the plane of bending, as happens in many practical cases. Suppose that the beam i initially straight, and is then bent into an arc of a circle of radius R by pure terminal couples M, say. It is shown in elementary texts on the Strength of Materials that the longitudinal strain ¢ at a distance y from a neutral axis is given by ean aay where x = 1/R is the curvature ofthe beam. This relation is derived from purely geometrical considerations, and is independent of the properties ofthe material. If the beam is initially curved, Equation (1.1) is still true provided that x denotes ‘the change of curvature produced by M. 1.3.1 Rectangular cross section Consider the rectangular cross section of breadth B and depth D which is shown in Fig. 1.4(@), with the bending moment Mf acting about an axis Ox parallel to the sides of breadth B. In this case the neutral axis will bisect the cross section, because of its double symmetry ‘The linear variation of strain across the section implied by Equation (1.1) is shown in Fig. 1.4(b). Here itis supposed thatthe strain inthe outermost fibres ‘exceeds the strain ey which corresponds to the yield stress 0g (Fig. 13(b)). The yield strain eg is attained at distances +z from the neutral axis. The correspond- BASIC HYPOTHESES. 7 6.A beam of solid rectangular cross section, breadth B and depth D, is bent about an axis parallel to the sides of breadth B. If the bending moment M is steadily increased to 0.88 Mp, find the depth of the elastic core, assuming no upper yield stress. If M is then reduced to zero, find the greatest residual stress, assuming elastic behaviour on unloading. Verify that M could then vary between 0.88 My and—0.453 M, without further yield taking place. 7. A uniform beam ABCD has a plastic moment M, and is simply supported at its ends A and D: AB = BC = J, CD = 2. It carries a concentrated load kW at B and a second concentrated load W at C. Find the value of W which would cause plastic collapse for the three cases k = 2,3. 8. A rectangular hollow section has all four sides of length B and thickness T. Find the plastic moment for bending about an axis XX passing through its centroid and parallel to two of the sides, assuming that 7 is very small as com- pared with B. Bending moments My and My are applied about XX and an axis YY perpen- dicular to XX, respectively, M, being greater than My. Find the relationship be- tween My and My, in the fully plastic condition. 8 THE PLASTIC METHODS OF STRUCTURAL ANALYSIS M, = bBD* oo aa) ‘My could also be found directly from the elastic theory of bending, for by definition M, = Zoo, as) where Z is the elastic section modulus, which for a rectangular cross section has the value BD? 6 ‘The curvature corresponding to this situation is denoted by xy, and from Equation (1.3) ky = 2eo/D. 6) Combining Equations (1.2)-(1.6), the bending moment-curvature relation ‘can be put in non-dimensional form as follows: M _ 1. _oe(s\" ™, =15 o,(2) > «a7 a result first obtained by de Saint-Venant (1871). Fig. 1.5 shows this bending moment-curvature relation, together with the elastic relation appropriate when M is less than M,. Its important feature is that M tends to @ limiting value 1.5 My as x becomes very large. In the limit, when ‘M= 15 My, « becomes infinite, and Equation (1.3) shows that z is then zero, so that the elastic core vanishes. The entire cross section is then plastic, and the corresponding bending moment is the plastic moment My. Using Equation (1.4), 5 My = 4BD* ap as) ost / | i ot 2 3 4 % Fig. 15 Bending moment-curvature relation for beam of rectangular cross section ‘The attainment of the plastic moment thus corresponds to the development of infinite curvature, which implies that a finite change of slope can occur over BASIC HYPOTHESES 9 ‘an infinitely short length of the beam. This is the explanation of the plastic hinge behaviour observed in steel beams. The stress distributions corresponding toMy and My are shown in Fig. 1.6. yh yh yh e o % oO 2 O 7 "a To a 2 2 Piano % o of Bending (a) (b) ©) Fig. 1.6 Stress distributions in beam of rectangular cross section (a) Cross section (b) At yield moment (©) At plastic moment In practice, the condition of full plasticity shown in Fig. 1.6(c) cannot be attained, Equation (1.1) shows that infinite curvature requires infinite strain, which is unattainable. Above a certain curvature, the strains in the outer fibres ‘would become sufficient!y large to cause strain-hardening. Suppose, for example, that strain-hardening commences when ¢ = 10 é9. From Fig. 1.4(b), this strain is, reached in the outermost fibres when z = 0.1 D/2, and it follows from Equations (1.3) and (1.6) that x = 10 ky. From Equation (1.7), M = 1495 M, = 0.997 My Thus the bending moment approaches Mp to within 0.3 per cent before strain-hardening begins. The plastic moment can therefore be regarded as an approximate indication of the bending moment at which something very much like a hinge action will occur in practice, with large curvatures developing at the cross section where this moment is attained. This simple theory assumes that the elastic-plastic boundaries are straight lines parallel to the sides of length B; this is not strictly true, as pointed out by 10 THE PLASTIC METHODS OF STRUCTURAL ANALYSIS Hill (1950). Moreover, at large curvatures additional radial stresses would be called into play, for a curved fibre subjected only to tensile or compressive forces at its ends would not be in radial equilibrium, Nevertheless, the plastic moment which corresponds to the fully plastic stress distribution is found in practice to be close enough for all practical purposes to the bending moment ‘which causes plastic hinge action. 1.3.2 Section with a single axis of symmetry Consider now a beam whose cross section only has one axis of symmetry, as shown in Fig. 1.7(a). O is the centroid of the cross section and Oy is the axis of symmetry, and it is assumed that the beam is bent in the plane containing the axis of the beam and Oy by terminal couples M. The axis Ox in the plane of the cross section is the neutral axis for elastic behaviour of the beam, Plane“ of benging fa) (b) () (a) fe) Fig. 1.7 Stress distributions in beam with single axis of symmetry (a) Cross section (©) Atyiold moment (6) Atattainment of yield stess on lower face {) Plastic zones spreading invatds from both faces (At plastic moment In this case yield first occurs at the upper surface of the beam, as shown in Fig. 1.7(b), where y has its greatest value y™**. The yield moment is given by ie = 209. ‘As M increases above My, a yield zone develops in the upper portion of the beam. Fig. 1.7(c) shows the corresponding stress distribution for the particular case in which the yield stress oy is just attained on the lower surface of the beam. The neutral exis no longer passes through the centroid O, but assumes a position dictated by the fact that the resultant normal force on the-crass section ‘ust be zero, BASIC HYPOTHESES W A further increase of bending moment causes yield to spread inwards from the lower surface of the beam, as well as spreading farther in from the upper sur- face, as shown in Fig, 1.7(4). Ultimately the two zones of yield meet, the distri- bution of stress then being as shown in Fig. 1.7(¢). This is the condition of full plasticity, and the corresponding bending moment is the plastic moment. 1.3.3 Effect of upper yield stress The theory was modified to include the effect of an upper yield stress by Robertson and Cook (1913), who assumed the stress-strain relation of Fig, 1.3(a). One consequence of this assumption is that the yield moment becomes Zou, rather than Zoo, and the (M, k) relation is also changed. However, as the curvature tends to infinity, the stress distributions still tend towards those shown in Fig. 1.6(c) and 1.7(@). The plastic moment, which is calculated from such fully plastic stress distributions, is therefore independent of the value of the upper yield stress. The simple theory of elastic-plastic bending which has been outlined in this section cannot be regarded as « completely accurate description of the behaviour of steel members. The discontinuous nature of the yielding process, as indicated by the development of Liders’ lines, invalidates some of the assumptions made. For a detailed study of the problem, reference should be made to the work of Leblois and Massonet (1972). 1.4 Evaluation of plastic moment ‘The value of the plastic moment can be calculated directly. Fig. 1.8 shows a cross section with a single axis of symmetry Oy which lies in the plane of bend- ing, Since the resultant axial force is zero the neutral axis in the fully plastic con- dition must divide the cross section lato two equal areas, so that the resultant axial Tensile and’ compressive forces are both equal to $Aag, where A is the total area of the cross section. If the two equal areas into which the cross section is divided have centroids G, and Gp at distances p, and 7 from the neutral axis respectively, as in Fig. 1.8, the resultant forces will act through G, and G, and the plastic moment will be given by My = 4A( + 52)00 as) Thus, if the plastic section modulus Zp is defined by the relation My = Zpao it follows that Zp = 4AGi +92). (10) For a rectangular section of breadth B and depth D, bent about an axis parallel to the sides of breadth B, the area A = BD and 9, =z =D/4, so that the plastic moment is 12 THE PLASTIC METHODS OF STRUCTURAL ANALYSIS Fig. 1.8 Fully plastic stress distribution for cross section with single axis of symmetry My = ABD? quay as found previously. This result was first obtained by de Saint-Venant (1864). As already pointed out (Equation (1.4)), the yield moment is M, = $BD? 09. For this cross section the ratio Mp/My is thus 1.5. In general, the ratio My/M, is termed the shape factor, and is denoted by v, so that =M_% Myr iit This ratio depends solely on the shape of the cross section. ‘A commercial Fsection can be idealized by regarding the flanges as rectangles of breadth B and thickness and the web as a rectangle of depth (D ~ 27) and thickness f, as shown in Fig. 1.9. For this idealized section it can be shown that for bending about the major axis XX, the elastic and plastic section moduli, Z and Zp, are given by (1.12) z= Suer +BT(D—T} +4(D—2T)] (1.13) Zp = BID~T)+iD—27F a4) ‘Taking as an illustration a 356 x 127 universal beam at 39kg/m, with D= 356mm and B= 127mn, itis found from section tables that the average flange and web thicknesses are T=10.7mm and Smm, respectively. From Equations (1.13) and (1.14) it is then found that Z = 569.3 cm? and Zy = 651.2 BASIC HYPOTHESES 13 cm®. The values given in section tables are $70.0cm? and 651.8.cm*, differing only slightly from those calculated because of the small departures of the actual section from the ideal assumed. The shape factor derived from either pair of values is 1.14, a value typical for a rolled Isection. Formulae for the values of Zz and v for some of the commoner structural sections are given in Table 1.2. Fig. 1.9 Idealized section If the axis of the applied bending moment is not parallel to, or perpendicular to, an axis of symmetry of the cross section, the plane of bending will not gener- ally be perpendicular to the neutral (equal area) axis. A general treatment of this problem has been given by Brown (1967), and the case of a rectangular cross section was dealt with by Harrison (1963). The plastic moment represents a definite limit on the value of the bending moment, regardless of the possible presence of residual stresses induced, for ‘example, by previous bending into the partially plastic range. This follows from the fact that the longitudinal stress cannot exceed go; on this basis the fully plastic stress distribution clearly corresponds to the greatest pi ‘moment which can be developed. Moreover, it is only when this distribution is attained that the curvature can become infinite, so that a plastic hinge can form, for with any other stress distribution there must be an elastic core with a corre- spondingly finite rate of change of strain with distance from the neutral axis. It follows that irrespective of any residual stress distribution across a section before loading, a plastic hinge can only form when the plastic moment is attained, as, pointed out by Baker and Horne (1951). ‘The foregoing analysis assumed that the only stresses acting were the longi- tudinal normal stresses due to bending. However, there will usually be shear and axial forces acting at a cross section in addition to the bending moment. These will modify the value of the plastic moment to an extent which is often negli- gible, but these effects may be calculated and allowed for where necessary. A 14 THE PLASTIC METHODS OF STRUCTURAL ANALYSIS ‘Table 1.2 Plastic section moduli and shape factors for structural sections Section zp P Solid = sao? 1s iar oF rectangular 5) *. 3 Rectangular BI(D—T)+4T(—27)|B=D hollow section —,— T=005D > 18 16 Solid D 4b? se circular in cr A Pee ail follow bce a _ |r 4ge7 ~ 1267" ‘making use of Equation (2.1). The behaviour of the beam can now be summar- ized on a diagram relating the load W to the central deflection 8, This load- deflection relation is shown as Ocb in Fig. 2.3. Oc is the behaviour in the elastic, SIMPLE CASES OF PLASTIC COLLAPSE 2 range, and cb represents plastic collapse under constant load, the increase of deflection from c to b being 10/2, asin the mechanism of Fig, 2.2(4). The hinge rotation and therefore the additional deflection developed during plastic collapse is indefinite. However, if very large deflections occurred, the change in geometry of the structure would affect the conditions of equilibrium, for example by enabling the load to be supported partly by direct tension in the two halves of the beam. The simple plastic theory does not concern itself with such effects; it predicts the loads at which large deflections are imminent, as at the point c in Fig. 2.3. as Fig. 2.3 Load-deflection relation for simply supported beam The broken curve commencing at a in Fig. 2.3 shows qualitatively the effect of taking into account the difference between the yield moment My and the plastic moment Mp. Elastic behaviour would cease at the yield load Wy when the central bending moment was M,, where » being the shape factor. Plastic collapse would still occur at the same value of W as before, but greater deflections would be developed before collapse. A more detailed study of this point is made in Chapter 5. For this simple example the ratio of the collapse load W, to the yield load Wy is equal to », the shape factor. The ratio of W, to Wy is always » for any stati- cally determinate structure, in which the greatest bending moment is pro- portional to the load and occurs at the same position regardless of the value of the load. Yield occurs when this greatest bending moment is equal to My, and collapse occurs when it is equal to Mp, for the introduction of a single hinge is always sufficient to reduce a statically determinate structure to a mechanism. It follows that the ratio of We to Wy is the same as the ratio of My to My, which by definition is the shape factor v. 22 THE PLASTIC METHODS OF STRUCTURAL ANALYSIS Equation (2.1) shows how the plastic collapse load was calculated by equating the maximum bending moment to the plastic moment. This is @ statical pro- cedure but the collapse load can also be found by a kinematical procedure, as first pointed out by Horne (1949). During collapse there is no change in the clastic strain energy stor since the bending moment distribution semains unaltered. The the loads during a small motion of the collapse mechanism is therefore equal to the work absorbed in the plastic hinge, since the motion is quasistatical. In the mechanism motion of Fig. 2.2(d) the oad We moves through a distance 10/2 and so does work Wel0/2. The rotation at the plastic hinge is 29, so that the work absorbed in the hinge is 2Mp0. It follows that }Wel0 = 2My6 Me W. 7 which agrees with Equation (2.1). fa) (b) () Fig, 2.4 Fixed-ended beam with uniformly distributed load (@) Loading (b) General form of bending moment diagram (©) General defected form SIMPLE CASES OF PLASTIC COLLAPSE 23 2.3 Fixedended beam The behaviour of a fixed-ended beam of uniform cross section and length 2, carrying a uniformly distributed load W, will now be considered. The end sup- ports are assumed to prevent rotation but permit small axial movements; if axial movement is also prevented large increases in the carrying capacity can occur, as shown by Haythomthwaite (1957). In what follows consistent sign convention will be used for bending mo- ments, curvatures and hinge rotations. Positive bending moments are those which cause tensile stresses in the fibres adjacent to the broken line in Fig. 2.4(a), and positive curvatures and hinge rotations correspond to tensile strains in the same fibres. The bending moment diagram has the parabolic form shown schematically in Fig. 2.4(b); a statical analysis gives the equilibrium equation MyM, = a 2) ‘The beam has one statical indeterminacy or redundancy; the separate values of ‘M, and M; cannot be found from equilibrium alone. The state of deformation depicted in Fig. 2.4(c) forms the basis of the sub- sequent calculations. Here the beam has developed a slope ~ @ at each end, and the entire span is presumed to be behaving elastically. An elastic analysis (by, for example, elementary beam theory) gives the following compatibility equation: my, = yn 28, 23) and it can also be shown that we 38487 —4lb,. (2.4) If W is increased steadily from zero the behaviour is at first wholly elastic, so that 6 =0. Equations (2.2), (2.3) and (2.4) then solve to give the elastic solution M, = — kW M, = kW we © 384ET Elastic behaviour ceases when M, =— Mp, so that plastic hinges form at each end of the beam. The yield load Wy is therefore given by Wy! = My 24 THE PLASTIC METHODS OF STRUCTURAL ANALYSIS 12My w, = 25) [At this value of the load, the state of the beam is as given in the first line of Table 2.1. Fig. 2.5(a) shows the deflected form of the beam at the load Wy, and ‘Table 2.1 Fixed-ended beam: proportional loading wi My Ms OBL BEL My My My Mgl Mp Fig. 2.5(b) shows the corresponding bending moment diagram. If W increases from Wy to Wy + AW, the plastic hinges at each end of the beam will undergo rotation while My remains constant at the value ~Mp. All changes occurring in this ‘step’ will be denoted by the prefix A. Fig. 2.5(c) shows the corresponding deflected form of the beam during this step, which is characterized by M,=—M,y, OM, =0, 6, <0. Equations (2.2), (2.3) and (2.4) become am, = *¥ C6) 8 0 = ~ aw 20, en AWE ete (2.8) as = AM 40, es) Since AM, is zero, there is only one unknown bending moment increment ‘AM, whose value is obtained immediately from the equilibrium equation (2.6) ‘The beam is therefore statically determinate in this step. However, there is now & new geometrical unknown A¢,. This is found from the compatibility equation 2.7) tobe Awe 6 = — Ter @9) Substituting in Equation (2.8) we sa (2.10) * 38487" SIMPLE CASES OF PLASTIC COLLAPSE 25 (a) () (©) “My @) 0 ; " muy Mp (e) Fig. 2.5 Behaviour of fixed-ended beam above yield load (2) Deflected form, W = Wy (b) Bending moment distribution, W = Wy () Deflected form, W= Wy + ai” (@) Bending moment distribution, W = We (©) Collapse mechanism Equations (2.6), (2.9) and (2.10) show that the incremental relations between AW, AM; , Ad; and Ab are those for a simply supported beam. This is because the conditions AM, =0, A¢, #0 correspond to simply supported end con- ditions. At the beginning of this step the value of M; is 0.5 Mp, as shown in Table 2.1, As AW increases, Mz increases in accordance with Equation (2.6) until it teaches the value Mp. The bending moment distribution is then as shown in Fig. 2,5(d). The corresponding value of AW is given by 26 THE PLASTIC METHODS OF STRUCTURAL ANALYSIS AW osm, + From Equations (2.9) and (2.10) the corresponding values of Ay and AB are a = —Mel EI a5 = Mol? ‘9687 ‘These incremental values are entered in the second line of Table 2.1, and the third line shows the resulting situation at the end of the step, with W=12M,/I+4M,/t= 16My/l. When W has this value, a plastic hinge forms at the mid-point, and the beam then collapses, the collapse mechanism being as shown in Fig. 2.5(¢). The col- lapse load We is therefore 16 Mp/t,a result first obtained by Kazinczy (1914). When W= We, but before the central plastic hinge has begun to rotate, the ‘beam is said to be at the point of collapse. The conditions at the point of col- lapse are those given in the lat line of Table 2.1, The hinge rotations — at each ‘end of the beam which are shown in the collapse mechanism of Fig. 2.5(¢) are additional to the rotations —M,i/6ET which have already developed at the point of collapse. ‘The load-deflection relation is shown in Fig. 2.6, in which Oy represents the clasticbehaviour up to Wy, yc represents the clastic plastic step and cb represents plastic collapse by the mechanism of Fig. 2.5(e). The broken curve commencing at a shows schematically the type of relation which would be obtained if the yield moment My was less than Mp. This load-deflection relation is typical for a beam or frame with one redun- dancy. When the first plastic hinge forms at the yield load (in this case a sym- metrical pair of hinges), the structure is rendered statically determinate for further increases of the load, and the plastic hinge rotations which then occur cause a reduction in the slope of the load-deflection relation. Collapse does not cour until a further plastic hinge forms, thus reducing the structure to a mech- anism. In general a finite increase in the load above the yield value will be required to bring the bending moment at the final plastic hinge position up to the plastic moment. ‘The behaviour of the fixed-ended beam is thus fundamentally different from the behaviour of the simply supported beam, for which the load-deflection relation was shown in Fig. 23. In that case the formation of a single plastic hhinge caused collapse, and the ratio of the collapse load We to the yield loed Wy ‘was the shape factor v. However, for the fixedended beam just considered the SIMPLE CASES UF FLAS IG CULE oe " yield load W, is 12Mp/vi, while the collapse load We is 16Mp/!, so that the ratio of We to Wy is 4v/3. The greater margin between the yield and collapse loads for the fixed-ended beam is a consequence of the single redundancy which exists in Fig, 2.6 Load-deflection relation for fixed-ended beam 2.3.1 Direct calculation of collapse load It can be seen by inspection that there is only one possible collapse mechanism for the fixed-ended beam, this being the symmetrical mechanism of Fig. 2.5(e). This enables the plastic collapse load to be calculated directly by either a statical or a kinematical procedure, The statical procedure consists simply of sketching the bending moment diagram at collapse, as in Fig. 2.5(4). It is seen that Wel Me a ww, = ‘ile ‘The kinematical procedure is based on the collapse mechansim of Fig, 2.5(e). Since the central deflection is 19/2, the average vertical displacement of the uniformly distributed load W is 10/4, so that the work done by the load during this mechanism motion is Wel0/4. At each plastic hinge the work absorbed must be positive, and is the product of Mp and the magnitude of the hinge rotation. Equating the work done to the work absorbed, WG = My(0) + My(20) + My(0) = 4My0 16tfy We 7 28 THE PLASTIC METHODS OF STRUCTURAL ANALYSIS 2.3.2 Behaviour on untoading If the load on the beam was removed after some rotation at the plastic hinges had occurred at the collapse load We, rotation at these hinges would cease and ‘the behaviour during unloading would, be wholly elastic, according to the as- sumed (M, x) relation of Fig. 2.1. Thus if W was increased to the point of col- lapse c in Fig. 2.6 and then removed, the unloading ine would be ed, parallel to the original elastic ine Oy. The residual deflection at d can be calculated directly from the information contained in Fig, 2.6 and Table 2.1, and is Ml? /24B1. The reason for the existence of this residual deflection is that the unloaded beam would contain residual bending moments caused by the plastic hinge ro- tations —MY/6EI at each end of the beam, which stay constant during the unloading process. These residual bending moments can be calculsted from Equations (2.2) and (2.3), with W=0 and ¢ J/GEI, and are found to be My = Mz = AMy, Itisreally verified that the same results are obtained by treating the unloading as a further step in which AW = —16Mp/l, 6, = 0. ‘The fact that residual moments can be induced in a structure by previous loading into the elastic plastic range shows that the Principle of Superposition cannot be applied in such cases. For instance, if the beam were reloaded from the point d in Fig. 2.6, the beam would of necessity behave elastically along de until the collapse load was reached at c, since the elastic behaviour during unloading is reversible. During such a reloading, the bending moments and deflections produced by a given load will be different from those arising during the first loading. 2.4 Effect of partial end-fixity Perfect end-fixity of the kind assumed in the foregoing example (Fig. 2.4) can- not be assured in practice. To examine the effect of partial end-fixity, consider a tuniform beam resting on four supports, as shown in Fig. 2.7(a). The central span is of fixed length f and carries a central concentrated load W, and the two outer spans are each of variable length AU. The non-dimensional parameter k specifies the degree of rotational constraint at the ends of the central span. With k=0 the central span becomes fixed-ended, while if kis infinite this span is effectively simply supported at its ends. The only possible collapse mechanism is as shown in Fig. 2.7(b). The collapse load We is found by the kinematical procedure to be given by We = My(0) + Mg(20) + M(0) = 4My0 SIMPLE CASES OF PLASHIL LULLArse ~ he w= The corresponding bending moment diagram at collapse is shown in Fig. 2.1(¢). The collapse load We can also be found by the statical procedure, and is given by W. Pa = Mp w, = She n=, agreeing with the kinematical calculation. fa) (b) ©) Fig. 2.7 Continuous beam on four supports ‘These analy that the collapse load Wis independent of the de end-fixity as specified by kits value will be 8M,/1, provided only that the con- ifions at each end of the central span are such that the plastic moment can be developed there. This illustrates one of the surprising features of the plastic theory, namely that plastic collapse loads do not depend or tual rigidity of ‘joints or supports. ‘The behaviour under a steadily increasing load can be analysed by the step- by-step process; details will not be given here. In the elastic range the greatest bending moment occurs beneath the load, and the yield load Wy and the central deflection 5y at yield are 30 THE PLASTIC METHODS OF STRUCTURAL ANALYSIS My peal T [3 +4k| Faber 24ET|3 + 4k) Above the yield load rotation occurs at the central plastic hinge. The bending moments at the two middle supports reach the value ~My when W= We, and the corresponding deflection 6¢ atthe point of collapse is, Mol? 5. = E+ aK), Load-deflection relations derived from these results are shown in Fig. 2.8. When k =0 (the fixed-ended condition), the yield load W,, coincides with the collapse load We, showing that in this special case all three plastic hinges form simultaneously. As X increases, both Wy and the slope of the load-deflection relation between W, and We are progressively reduced, so that the deflection 5. at the point of collapse becomes larger. However, plastic collapse always occurs at the same load 8Mp/I regardless of the value of k. For values of k in excess of, say, 3, unacceptably large deflections would develop before the plastic collapse load was reached. In such cases the theoretical 24618 Mg? Fig, 2.8 Load-deflection relations for beam of Fig. 2.7 collapse load would be of little interest to the designer, since the normal purpose of calculating the collapse load is to determine the load at which large deflections are imminent. This was pointed out by Kazinczy (1934), who discussed load- deflection relations similar to those in Fig. 2.8 for a beam with partial end-fixity SIMPLE CASES OF PLASTIC COLLAPSE 31 carrying a uniformly distributed load. When joint flexibilities are large, it is obviously necessary to supplement the calculation of the plastic collapse load by anvestimate of the deflections at the point of collapse. Methods for doing this are ven in Chapter e Tin the extreme case when K is infinite, the deflection 5¢ at the point of col- lapse also becomes infinite, and so the slope of the load-deflection relation be- tween yield and collapse becomes zero. The load-deflection relation for this case thus appears to correspond to a collapse load of only 4My/I, which would be the collapse load for a simply supported beam of span 1, whereas the calculated plastic collapse load is still 8Mp/I. However, this apparent paradox, which was pointed out by Stiissi and Kollbrunner (1935), is resolved when it is realized that the horizontal load-deflection relation which occurs in this case when W= W, is merely the limiting case in which the slope of the load-deflection relation between yield and collapse tends to zero as k tends to infinity Stissi and Kollbrunner carried out tests of this kind on small beams of section, 4.7 om x 3.6 cm. In these tests / was 60cm, and values of k of 0.5, 1, 2 and 3 were used. The average of their observations from two tests of this kind with & = 2are shown in Fig, 2.8, and it will be seen that the comparison between the observations and the theoretical relation is good. Further tests of a similar nature were carried out by Maier-Leibnitz (1936). A type of test which is similar in principle is obtained by applying a central vertical oad to a rectangular portal frame (as in Fig. 2.9(a) with H = 0), in which case the horizontal member functions as a partially fixed-ended beam. Tests of this kind have been described by Girkmann (1932), Baker and Roderick (1938), and also by Hendry (1950), who showed that increasing the height of the frame while leaving the span constant did not affect the collapse load but increased the deflections prior to collapse. Similar tests, but with symmetrical two-point loading, have been described by Rusek, Knudsen, Johnston and Beedle (1954), ‘The effect of partial end-fixity on the design of beams subjected to uniformly distributed loads, whose ends are encased in reinforced concrete or masonry, was the subject of a theoretical and experimental investigation by Kazinczy (1934). The behaviour of a full-scale portal frame whose feet were supported by short piled footings was investigated experimentally by Baker and Eickhoff (1955), who showed that for this frame the collapse load was not affected by the partial fixity of the feet to any appreciable extent. 2.5 Rectangular portal frame ‘The last structure to be considered is the rectangular portal frame whose dimen- sions and loading are shown in Fig. 2.9(a). All the members of this frame are uniform with flexural rigidity BY and plastic moment Mp. The joints at sections 2 and 4 are rigid, and the columns are rigidly built-in at their bases 1 and 5. The ‘sign convention for bending moment, curvature and hinge rotation is again that 32 THE PLASTIC METHODS OF STRUCTURAL ANALYSIS positive values correspond to tensile stresses or strains in the fibres adjacent to the broken line, (b) () Fig. 2.9 Rectangular portal frame (@) Dimensions and loading (6) Virtual beam mechanism (©) Virtual sway mechanism Within each of the four segments of the frame which are straight and free from external load, namely 12, 23, 34 and 45, the shear force must be constant The bending moment must therefore vary linearly along each of these segments. The values of the bending moments M,, Mz, My, M, and M, at the five num- bered cross sections therefore specify the bending moment distribution through- ‘out the frame, Moreover, since the bending moment cannot exceed Mp in ‘magnitude at any cross section, it follows that plastic hinges can only occur at the ends of these segments. Thus, the only possible locations of plastic hinges are the five numbered sections. (This excludes the special case in which the shear force is zero in a segment, so that the bending moment is constant along the segment.) The frame has three redundancies, for if a cut were made at any section, and the shear force, axial force and bending moment were all specified at this section, it would become statically determinate. It follows that there must be two ‘equations of equilibrium connecting the five bending moments. For the deter SIMPLE CASES OF PLASTIC COLLAPSE 33 mination of the bending moments which occur when the frame is wholly elastic there must therefore be three equations of compatibility. For situations in which some plastic hinges have developed, itis still possible to develop three appropri- ate equations of compatibility, and the two equations of equilibrium will still apply. The equilibrium and compatibility equations will now be derived using the Principle of Virtual Work. This procedure has certain advantages, and will be used again in later chapters. 2.5.1 Principle of Virtual Work ‘The Principle of Virtual Work for frames involves force systems (of loads and bending moments) which satisfy the requirements of equilibrium and displace- ‘ment systems (of deflections, curvatures and hinge rotations) which satisfy the requirements of compatibility. For the type of problem under consideration it takes the form Le Jaas + © me. (41) In this equation M is any distribution of bending moment satisfying the require- ments of equilibrium with concentrated external loads P. « denotes any dist bution of curvatures which is compatible with hinge rotations @ and deflections 6. The summation on the left-hand side covers all points of application of exter- nal loads. On the right-hand side the integral covers all members of the frame, distance s being measured along each member, and the summation covers all sections where there may be a hinge rotation. The equation is valid provided that the force system (P, M) satisfies the requirements of equilibrium and the displacement system (5, x, 6) satisfies the requirements of compatibility. In addition, the sign convention for 8 must be consistent with that for P, so that the direction of a positive force P must be the same as the direction of the positive corresponding displacement 5. Similarly, the sign conventions for both « and @ must be consistent with that for M. Equation (2.11) can be used in two ways. In the first of these the displace- ment system (6°, x*, $*) is virtual (throughout this Section, asterisks are used to denote virtual systems, either of displacements or of forces). This means that the displacements, curvatures and hinge rotations can be chosen arbitrarily, subject only to the requirements of compatibility, and need not be ascribable to any possible form of loading. This form of the principle is often referred to as the. Principle of Virtual Displacements, and its use generates equations of equilibrium. The other possibility is to use a virtual force system (P*, M*). Here the forces and moments are chosen arbitrarily, subject only to the requirements of equilib- rium. This is the Principle of Virtual Forces, and its application results in equations of compatibility. 34 THE PLASTIC METHODS OF STRUCTURAL ANALYSIS Both forms of the principle will now be used to develop the equations of equilibrium and of compatibility which are needed for the step-by-step analysis of the frame of Fig. 2.9(a. 2.5.2 Equilibrium equations by virtual displacements ‘The two equilibrium equations will be derived using the virtual displacement systems depicted in Fig. 2.9(b) and (c). These represent a beam mechanism and a sway mechanism, respectively. The hinges are not plastic hinges, but are intro- duced to permit the small displacements which are shown to take place while the members between the hinges remain straight. Because the curvature is every- where zero, Equation (2.11) reduces to Lp =D Mor. The equilibrium system consists of the loads H and V shown in Fig. 2.9(a), together with bending moments at the five numbered sections which are in equilibrium with these loads. Using this system in conjunction with the virtual displacement systems of Fig. 2.9(b) and (c) in turn gives Vay = Ma ¥) + Ms 20) + Ma) Hy = My(— V) + Ma (+ Y) + Ma(— 9) + Ms(+ Y). Cancelling } throughout gives the two equilibrium equations = -M, +2, —M 2.12) Hl = —M, +My — My +My 2113) ‘These are equations of equilibrium which must be obeyed regardless of whether the frame is behaving elastically or has become partly plastic. Equation (2.11) is independent of material properties, and these properties did not enter into the derivation of Equations (2.12) and (2.13). 2.5.3 Compatibility equations by virtual forces ‘The three compatibility equations are found by using virtual force systems in which all the external P* are zero. The bending moments are then referred to as residual moments and denoted by the symbol m” . Equation (2.11) becomes 0 = fmtads+ Emre. @.14) The actual curvature x at any section is related to the actual bending mo- ‘ment M at that section by the elastic relation x = MIET, so that Equation (2.14) becomes SIMPLE CASES OF PLASTIC COLLAPSE 35 mM 0 =| ast Lm (2.15) Within each of the linear segments 12, 23, 34 and 45, both m* and M vary linearly with distance s along the member. In these circumstances the integral in Equation (2.15) is readily evaluated for a typical uniform straight segment AB of length L, and is 4 mt diy ig _ [peg 88 = ge lMhOMa + Ma) + mp 2Ma + Mal. 2.16) Equation (2.15), taken in conjunction with Equation (2.16), can be used to generate three compatibility equations, provided that three linearly independent residual moment systems can be identified. To achieve this, it is noted that any set of residual moments must obey the two equations of equilibrium (2.12) and (2.13), with the loads V and H each zero, giving ms +2m3—ms 17) (2.18) Following Heyman (1961), the three systems used are as shown in the first three lines of Table 2.2. =m, +m, —mg + ms = ‘Table 2.2 Virtual force and actual displacement systems Section 1 7 3 4 5 Virtual force systems m* @ 1 1, os 0 ° w 0 0 0s 1 1 ii) 0 1 1 1 0 ue - ° 0 0 0 Actual displacement system Ek=M My M Ms Ms Ms ¢ os oe bo oa os It is readily verified that each of the three residual moment systems (i), (il) and i) satisfies Equations (2.17) and (2.18), and that they are linearly independent. ‘The actual displacement system is shown in the last two lines of Table 2.2. When used in conjunction with the virtual residual moment system (1), Equations (2.15) and (2.16) give IGM + 3M) + 2.Sty + 2M) + Uy + 0.5M] +61 + 05+ 050, 36 THE PLASTIC METHODS OF STRUCTURAL ANALYSIS and this reduces to 3M, +5.5M, +3My +0.5M + 6EI(, +02 + 0.56s)/l us) Similarly, the use of the residual moment systems (i) and (i) gives 0.5M; + 3M; + 5.5M, + 3M, + 6EHO.50; + 04 +¢s)/t (2.20) M, + 5M; + 6M; + SMe +My + 6EN62 +63 + Oa) (2.21) Equetions (2.19)-(2.21) are the three equations of compatibility. These equations, and the two equations of equilibrium (2.12) and (2.13), may be ‘written more compactly and in incremental form as follows: ° aM, AM AM ° AM, 1 avi ait =| 0 ° 0 2.22) where the prefix A is used to denote changes in bending moments, hinge ro- tations and loads. The two equilibrium equations are those whose coefficients appear above the broken line. “These five equations may be used to trace the behaviour of the frame when subjected to proportional loading, V=H=W, ‘up to collapse, starting from a condition in which the frame is unloaded and free from stress. Because the ideal (M, x) relation of Fig. 2.1 is assumed, either MI0 z Ag, = Ad, = Ads = Ads = 0. Equations (2.22) then have the solution AM, = —0.468AW1 AM, = 0.108AWI AM, = 0.342AW1 AM, = ~0.424AW1 Ags = 0.209AWI?/EI. During this step the number of redundancies has dropped from 3 to 2, be- cause one bending moment increment is now known, namely AMs = 0. There are therefore only four statical unknowns, AM;, AM; , AM; and AM. The two ‘equations of equilibrium therefore need to be supplemented by only two of the equations of compatibility to determine these four bending moment increments, However, there is now a geometrical unknown, Ads, and this is found from the third equation of compatibility. feo ro ero =O T 0 = 900 © 900 «ITO e9r0 0 0 er00~ 5400 wero ter cttO— 1 I 1 e900 isez coro = toro ete 0— ° 0 eo 880 066-0 uso 000 1 T= 90 $100~ uss ozo © oFv0 0 1900-600 S100 erro ur 1 660-0 OF0O— vere aN nn x W m oH Ta Ta Ta OW HW Wa ia iav Furpooy jouajesodoud round LynBubiooy ET AL. SIMPLE CASED UF FLAD TIL LULLAFSE 27 A plastic hinge of negative sign forms next at section 4, where the bending moment at the start of this step was —0.939Mp. The value of AW for the step is thus given by -0.939M, —0.424AWI = —My AW = 0.143Mp/l. With this value of AW the increments which occur in this step are shown in the second line of Table 2.3, and the resulting situation is given in the third line. During the next step rotations occur at the plastic hinges at both sections 4 and 5, so that My = —My, AMg = 0, $4 <0 Ms = My, AM, = 0, d¢;>0 Ag: = Ads = Ads = 0 ‘The corresponding solution of Equations (2.22) is AM, = -0.85AWI AM, = 0.1SAWI AM, = 0.S7SAWI Ay = —0.558AW0?/EI Ags = 0.258AWI"/EI. There are only three statical unknowns, AM , AMfs and AM; in this step, so that the number of redundancies has dropped to one. Only one equation of compatibility needs to be used in conjunction with the two equations of equilib- rium to determine these three unknowns, but the remaining two equations of compatibility are required to find the values of the two geometrical unknowns Ags and Abs ‘The next plastic hinge forms at section 3, where the bending moment at the start of this step was 0.776Mp. The value of AW for the step is therefore given by 0.776My +0.S75AWI = My AW = 0:390Mp/t. ‘The corresponding increments during this step are given in the fourth line of Table 2.3, and the fifth line shows the situation at the end of the step. ‘The ensuing step is characterized by Mz; = My, AM; = 0, 4g; >0 My = —My, AM = 0, Ags <0 40 THE PLASTIC METHODS OF STRUCTURAL ANALYSIS Ms = My, AM, = 0, A%5>0 46, = Abs = 0. The structure is now statically determinate; the two equilibrium equations suffice to determine AM, and AM,, and the three compatibility equations fur- nish values of Ads, Ag, and Ads. The solution is aM, = -2aW1 AM, = —AW Ay = 3.8331 /E7 B64 = —2.667AW17/ET Ags = 0.833AW07/EI. At the end of this step a plastic hinge forms at section 1, where the bending ‘moment at the start of the step was ~0.913Mp, The value of AW for this step is siven by —0.913M, ~2AWI = —My AW = 0.043Mp/i ‘The changes which oceur during this step and the final situation are given in the sixth and seventh lines of Table 2.3. The final Ioad is 3M,/l, and at this load there are four plastic hinges. This reduces the frame to the mechanism which is illustrated in Fig. 2.10. The col lapse load is thus We = 3Mp/I. Since Wy = 2.424Mp/I, the ratio of We to W, in this case is 3/2.424, or 1.24. If the effect of the shape factor » is allowed for, this ratio becomes 1.240, 2.5.5 Deflections by unit load method The analysis will be completed by the calculation of a particular deflection at the end of each step. This is achieved by using an appropriate virtual force system in conjunction with the actual system of displacements. If the deflection sought is 8, al that is necessary is to make the corresponding P* unity, with all ‘other external loads zero. This is the well-known unit load method; reference to Equation (2.11) shows that Mw 8 as+ EM", 223) ET where M* represents any distribution of bending moments satisfying the require- ‘ments of equtlibrium with the virtual unit load Forexample,leth be the horizontal deflection at section 4 (Fig 2.9(a)) come sponding to H. Examination of the equilibrium equations (2.12) and (2.13) SIMPLE CASES UF PLAS TIL CULLAPSE at shows that the simple system of bending moments Mf =—I, M3 = Mj = Mj = Mz =0, satisfies the requirements of equilibrium with loads H = 1 and V = 0. This system, which is entered in the fourth line of Table 2.2, can therefore be used in Equation (2.23) to calculate A. The integral in Equation (2.23) can be evaluated using Equation (2.16), with M* replacing m*, and the result which is obtained is . h = - 2am +t) W 024) Values of hare given in Table 2.3, and Fig. 2.11(a) shows the relationship between W and h for proportional loading. Each point at which a fresh plastic hinge forms is marked with the number of the relevant cross section, in accord- ance with Fig. 2.9(a). 26 Invariance of collapse loads If a structure is subjected to more than one load, the loads will only rarely in- crease in proportion to one another. Fortunately, wide variations can occur in the manner in which the various loads are brought up to their collapse values, without affecting the collapse load. Thus for the frame just considered, collapse occurs under proportional loading when H'= V=3M,/l. To take an extreme case of loading which is not proportional, suppose that a load V = 3Mp/Lis first applied, and then held constant while His increased steadily from zero, A step- by-step analysis shows that the load V = 3Mp/1 is borne by wholly elastic action, and that H must then be increased to the value 2.133Mp/I before the first plastic hinge forms. This hinge forms at section 4, although under proportional loading the first hinge formed at section 5. The load-deflection relation is shown in Fig. 2.11(b). When a fourth plastic hinge is formed at section 1 the value of His 3Mp/l; collapse then occurs by the same mechanism as before. Thus the collapse load condition for this case, H= V=3Mp/l, is the same as for proportional loading. The only difference between the two cases is in respect of the load- deflection relation prior to collapse. 42 THE PLASTIC METHODS OF STRUCTURAL ANALYSIS we My ° 0 Of 02 03 O4 ner © 01 02 03 04 05 me 0 01 02 03 04 05 06 MP Fig. 2.11 Load-deflection relations for frame of Fig. 2.9(a) (a) Proportional loading: H = W, ¥ = W (b) ¥ = 3Mp/l, followed by H = W (6) Feet spread apart, followed by proportional loading ‘The plastic collapse load is not affected by residual stresses, whether these be due to.welding, imperfect fit of members, plastic hinge rotations which have ‘occurred during previous loading or the movements of supports. To illustrate this point, suppose that in the frame of Fig. 2.9(a) the bases had spread apart, while the columns remained fixed in direction, the spread being just sufficient to cause the bending moments at the bases 1 and 5 to reach the value —Mp in the unloaded condition. An elastic analysis shows that the beam is then subjected to 4 uniform sagging bending moment M,/3. If the loads Hf and V are then in- creased proportionately, rotation of a plastic hinge at section 1 begins immedi- ately. However, the bending moment at section 5 undergoes a positive change, so that there is no immediate plastic hinge rotation at this section. The load- deflection relation is shown in Fig. 2.11(c). It is again found that collapse occurs when H= fl, although the load-teflection relation differs from the other two cases, and the sequence of formation of hinges is also quite different. ‘The fact that residual stresses have no effect on the value of the plastic col- lapse load for a given structure was pointed out by Karinczy (1938). Direct experimental confirmation of this point was provided by the work of Maier- Leibnitz (1928) and of Horne (1952) on continuous beams, in which the effect of initial lowering of supports on the collapse load was shown to be negligible Moreover, any welded frame will contain residual stresses due to the welding process unless a stress-elieving treatment is used, so that indirect confirmation hhas been supplied by the lack of any noticeable effect in the many tests which have been carried out on such structures. SIMPLE CASES UF PLASTIC COLLAPSE “4 The invariance of the plastic collapse load stems from the fact that collapse can occur only when a sufficient number of plastic hinges have formed to trans- form the structure into a mechanism. The rectangular frame just considered has three redundancies. If at any stage of the loading there are three plastic hinges, at each of which the bending moment is known, the frame becomes statically determinate, so that the remaining bending moments can be calculated in terms of the loads from the equations of equilibrium. If a fourth plastic hinge then forms, the frame is reduced to a mechanism, and the knowledge of the value of a fourth’ bending moment enables the loads to be calculated, provided of course ‘that the value of each load is specified as a multiple of a single load W. Thus ‘once the mechanism of collapse is known, the collapse load W. can be calculated. by considering only the equations of equilibrium, These equations are indepen- ent of residual stresses, the order of application of loads, the imperfect rigidity of joints or the sinking of supports, and so the collapse load is unaffected by such factors. To illustrate this point, consider the collapse mechanism of Fig. 2.10. The- ‘bending moments at the plastic hinge positions are M, = —My, Ms = My, Ms = —Mp, Ms = My. ‘Substituting these values in the two equations of equilibrium (2.12) and (2.13), with A= V= We, tis found that

You might also like