STABILITY oF Se PY) MTR OT 3}ALLIED PUBLISHERS LIMITED Regd. Off.: 15. J.N. Heredia Marg, Ballard Estate, Mumbai — 400038 Prarthna Flats (2nd Floor), Navrangpura, Ahmedabad — 380009 3-5-1114/1 Kachiguda Cross Road, Hyderabad — 500027 16A Ashok Marg, Patiala House, Lucknow — 226001 5th Main Road, Gandhinagar, Bangalore — 560009 17 Chittaranjan Avenue, Calcutta — 700072 81 Hill Road, Ramnagar, Nagpur — 440010 13/14, Asaf Ali Road, New Delhi — 110002 751 Anna Salai, Chennai — 600002 © 1998, ALLIED PUBLISHERS LTD. No part of the material protected by this copyright notice may be reproduced or utilized in any form or by any means, electronic or mechanical including photocopying, recording or by any information storage and retrieval system, without prior written permission from the copyright owners. ISBN 81-7023-804-8 ASSOCIATE EDITOR JYOTI MEHROTRA PRODUCTION MANAGER, COVER Published by Sunil Sachdev and printed by Ravi Sachdev at Allied Publishers Limited (Printing Division), A-104 Mayapuri, Phase il, New Delhi — 110 064,CONTENTS Preface vin List of Figures Ix L__ FUNDAMENTAL CONCEPTS L1_Introduction 1.2_Elastic Structural Stability 1,3. Structural Instability 1.4 Analytical Approaches To Stability Analysis 2.__DISCRETESYSTEMS 2.1 Introduction 2.2_Mathematical Implication of Law of Minimum Potential Energy (2.3 Illustrations 2.4 Large Deflection Analysis 2.5 Effect of Small Imperfections 3,__COLUMNS 37 3.1 Introduction 3.2_ Implication of Law of Minimum Potential Energy 3.3. Critical Load for Standard Boundary Conditions 3.4_Elastically Restrained Perfect Columns 3.5_Columns with Geometric Imperfections 3.6_Column with Load through a Fixed Point 3.7_Orthogonality of Buckling Modes 3.8 Large Deflection Theory of Columns 3.9 _Eccentrically Loaded Column 4, BEAM-COLUMNS AND FRAMES 4.1 Beam-Columns — 42 Some Standard Cases 4.3 Continuous Columns and Beam-Columns 4.4 Column on Elastic Foundation 4.5 Single-Storey Frames _4.6, Frame Buckling using First Principles 4.7. Frame Buckling Using Stiffness Method LATERAL BUCKING OF BEAMS 135 5.1 Introduction 5.2. Use of Energy Method - Case of Symmetric I-Beams 5.3 Monosymmetric I-Beams 5.4 Code Specifications 5.5 Other Loading Conditions 5.6 Beam - Columns ISOTROPIC RECTANGULAR PLATES 149 6.1 Introduction 6.2. Governing Differential Equations 6.3 Plates Simply Supported on All Edges 6.4 Plate with other Boundary Conditions 6.5 Use of Energy Method 6.6 Large Deflection Theory of Plates LAMINATED COMPOSITE PLATES 7.1 Introduction 72 Classical Thin Lamination Theory 73 Special Cases of Stiffnesses 74 Total Potential Energy 7.5. Differential Equations and Boundary Conditions 7.6 Applications DYNAMIC STARILITY OF STRUCTURES 198 8.1 Introduction 82 Need for Dynamic Investigation 83. Discrete Systems 8.4 Lagrange-Hamilton Formulation for Continuous Systems 8.5. Stability of Continuous Systems 8.6 A General Method of Analysis 8.7, Use of Galerkin’s Method8.8 Rectangular Plate Under Follower Force 8.9 Pulsating Load on a Column 9. THIN CIRCULAR CYLINDRICAL SHELLS 246 9.1 Introduction 9.2, Donnell’s Differential Equations 9.3 Applications 9.4 Yailure of Cylindrical Shells Appendix 259 Subject Index 2131 Fundamental Concepts 1.1 INTRODUCTION The stability of a structure essentially means the stability of its equilibrium configura- tion or state. In a practical sense, an equilibrium state of a structure or a system is said to be stable if accidental forces, shocks, vibrations, eccentricities, imperfections, inhomogeneities or other probable irregularities do not cause the system to depart excessively or disastrously from that state. In a mathematical sense, stability is usually interpreted to mean that infinitesimal disturbances will cause only infinitesimal depar- tures from the given equilibrium configuration. The mere fact that a system is stable in this refined sense does not necessarily signify that it is safe from an engineering viewpoint. For example, a convex shell-like structure can be in a state of stable equi- librium, yet a jolt may cause it to snap into a badly deformed shape. In designing structures, care should be taken that the stresses developed do not exceed certain limits, which may otherwise lead to ‘failure’. Two kinds of failure can be associated with a structure: (i) failure with respect to material behaviour, (ii) form failure. In the first case, stresses exceed the given safe limits resulting in the formation of cracks which cause failure. In the second case, the stresses need not cross the safe values but the structure may not keep up the forms in which it is designed. Here, the structure does not fail physically but it may deform to some other shape (due to external disturbances) which may not be desirable. For most thin-walled structural members loss of stability is in the elastic range, while for slightly thick bodies inelastic stabilityStability of Structures investigations may be needed. Also, loads on a structure could be tensile and/or com- pressive. The loss of stability due to tensile loads falls in the broad category of material instability, whereas the stability loss under compressive loads is usually termed struc- tural (or geometrical or form) instability. Here the term instability is being used in a simple sense of pertaining to a state in which the slightest change causes still further change. Specifically, in structures this means a condition in which the slightest incre- ment of deflection results in a further increase, which thus leads to collapse/failure. 1.2 ELASTIC STRUCTURAL STABILITY It seems appropriate as well as convenient to speak of structural stability as the ability of the structure to remain in position and support the given load, even if forced slightly out of its position by a disturbance. In investigating the stability of the equilibrium configuration of a system, one may ask the following questions: (i) If the system, which is initially in an equilibrium state, is given an arbitrarily small disturbance, does it remain near the equilibrium state? (ii) Does it remain near the equilibrium state and, in addition, tend to return to the equilibrium? (iii) What bounds must be placed on the magnitude of the initial disturbances so that, given a disturbance within these bounds, the system will eventually regain its original equilibrium state? Consider for example, an initially straight, centrally loaded, slender elastic column If the external load is P, the average internal compressive stress 6 in the column is obtained as mI Ga (1.1) where A is the area of the cross-section of the column. This part of the analysis is the study of stress distribution in which eqn. (1.1) represents the equilibrium between the external lead and the internal forces of the column in its original straight position. If the average compressive stress 6 is less than the yield stress 6, of the material, the column is then said to be in the clastic range. To study the state of equilibrium in terms of Euler's concept, a small external disturbance is given in the form of an added lateral force so that the column moves from its straight unperturbed position to an adjacent bent perturbed position. For values of load P < P , the column gets back to its initial loaded position after the removal of the external disturbance, whereas for values of load P > P,, the column tends to diverge from the adjacent bent position. In the first case, the original straight position of the column is in stable equilibriumFundamental Concepts Figure 1.1. Stable (A), unstable (B) and neutral (C) equilibrium states. while in the second case it is in unstable equilibrium. The load P, is the critical one in the sense that it marks the changeover from a stable to an unstable equilibrium configuration. The equilibrium state of the column under the load P = P;, is defined as the neutral equilibrium state. ‘These concepts of stable, unstable and neutral equilibrium states can best be demonstrated by the system shown in Fig. 1.1. This system consists of a ball of some weight placed in position at different points on a surface with zero curvature normal to the plane of the figure. The points of zero slope on the surface (points A, B and C) denote the positions of equilibrium state, but the nature of equilibrium state at these points is substantially different. At position A of the ball, the character of equilibrium is stable because the ball, if displaced slightly from its given position, returns to its position subsequent to the removal of the disturbing force. On the other hand, the nature of equilibrium is unstable at B, while at position C it is neutral equilibrium. If the displacements are allowed to be of large magnitude, then it is possible for a system to be unstable when disturbed slightly but stable in the large [Point B in Fig. 1.2 (a)] or stable for slight displacement and unstable in the large [Point A in Fig. 1.2 (b)}. 1.3 STRUCTURAL INSTABILITY The loss of structural stability is termed structural instability, which takes place in different ways depending on the structural properties and loading conditions. The loss of stability in terms of structural behaviour can be expressed by the relationship be- tween the load and the characteristic displacement. Based on the load-characteristic displacement relationship, structural instability may be classified as follows: — Transition to adjacent equilibrium states — Transition to non-adjacent equilibrium states ~ Transition from a state of rest to a state of motion — Transition to non-equilibrium statesStability of Structures § (a) Figure 1.2. Stable and unstable in the large - Transition from the primary equilibrium path to the secondary equilibrium path. Fig. 1.3(a) shows the load-characteristic displacement (P-w) for a centrally-loaded straight column. For P wx) (d) : w w(x) Figure 1.3 (Contd). (¢) Column under follower force. (d) Transition to non-equilibrium states 6Fundamental Concepts av/v F Figure 13 (Contd). (¢) Transition from primary 10 secondary equilibrium path. and A V/V represents the volumetric strain. Line OA represents the primary of fun- damental equilibrium path of the unbuckled configuration of the cylindrical panel. Line BC represents the secondary path of the buckled non-cylindrical configuration of the panel. In such strutures a finite disturbance during the application of the load can force the structure to pass from the primary equilibrium path to the secondary equilibrium path even before the classical critical load is reached. In each type of loss of stability, there is seen a change in the geometry or position of the system due to the ‘appearance’ of the characteristic displacement. The geometry change in the system, a consideration of which is one of the typical features of struc- tural stability analysis, is the cause for either introducing additional new forces or changing the nature of the forces that existed prior to the loaded position. In terms of these new forces which appear duritig the loss of structural stability, there can be a further classification of instability as follows: ~ Flexural buckling ~ Torsional buckling = Flexural-torsional buckling — Snap-through buckling Consider a centrally loaded I-section column, If the external disturbance in the form of characteristic displacement is in the plane of the web, then in addition to compres- sion, bending moment appears, Loss of stability in this case is by flexural buckling in the plane of the web. On the other hand, if the external disturbance given is in the plane of the flanges, loss of stability is again by flexural buckling but in the piane of the flanges. However, if the external disturbance in the form of characteristic displace-Stability of Structures ment is given such that flanges bend in their planes but in opposite directions, the column will then be subjected to torsion in addition to compression, Instability, in this case, is by torsional buckling. As an example of flexural-torsional buckling, consider an eccentrically loaded I- column with no lateral support to its compression flange. Initially, the column deflects by bending in the plane of the web. At a certain load, however, the column may fail due to loss of stability by a combined mode of twist and lateral bending of the cross- section. The compression flange becomes unstable due to the external disturbance in the form of lateral displacement and tends to buckle laterally, at the same time the tension flange is stable and tends to remain straight. This phenomenon of torsional bending buckling is also known as lateral buckling. In the case of snap-through buckling, the loss of stability is due to transition to a non-adjacent equilibrium configuration with a sudden change in the nature of internal forces of the structure. A typical example is the Mises truss as shown in Fig. 1.3(b). Such types of instability are quite common in convex shell-like structures, for example in shallow arches. 1.4 ANALYTICAL APPROACHES TO STABILITY ANALYSIS Stability analysis involves determination of the mode of loss of stability and the cor- responding critical load under which the structure gets into a critical state. Except in the case wherein loss of stability is due to transition from a state of rest to a state of motion (and which is often called instability by flutter), the structure remains at rest before and after buckling. This is called instability by divergence. Analytical ap- proaches to stability analysis based on the static concept are = Equilibrium approach — Imperfection approach ~ Energy approach Stability investigation based on the dynamic concept is referred to as the vibration method. In fact this method presents a very general approach and can be used for the stability analysis of any type of structure. Equilibrium Approach Stability analysis by this approach is concerned with the answer to the following ques- tion: What are the values of the load for which a perfect system admits two or more different but adjacent equilibrium states? By different equilibrium States it is meant that the response of the structure is such that equilibrium can be maintained with 8Fundamental Concepts different deformation patterns. The fact that the system occupies an equilibrium con- figuration ‘close’ to the initial one enables us to consider the slope of the deflection curve (representing the adjacent equilibrium configuration) as small compared to unity. The result is that the expression for the curvature of the deflection curve can be linearized. The method, then, requires the solution of the governing differential equa- tion subject to some prescribed boundary conditions. It leads to an eigenvalue problem. This approach is also known as the Euler's approach or the method of adjacent equi- librium configuration. Imperfection Approach This approach is essentially for stability analysis of imperfect elastic structures, im- perfect either in the geometric form of the structure or in the condition of loading. Like the equilibrium approach for perfect systems, mathematical formulation of the imperfection approach is based on the equilibrium condition in the adjacent deformed configuration. However, the transition from an undeformed configuration to an adjacent deformed configuration is not due to any unknown small external disturbance but to the presence of imperfection itself. For example, consider a simple supported column (Fig. 1.4) loaded axially at the left end and with an eccentricity at the right end of the column. Equilibrium requires the presence of reactions P and P,/L as indicated. Because of imperfection in the loading condition in the form of eccentricity e, the column deforms from the straight position to some adjacent bent position, as the load P increases from zero, If we plot P vs 6 (mid-span deflection), as shown in Fig. 1.4, it is found that the deflection is no more proportional to load P. All the curves have a common horizon- tal asymptote at P = x? EJ/L*, This means that no matter how small eccentricity e is, the deflections becomes infinite at the above value of load which is the critical load of a perfect hinged-hinged column. These topics are dealt in detail Chapter 3. Energy Approach The method is based on the law of minimum potential energy which may be expressed as follows: ‘A conservative holonomic system is in a configuration of stable equilibrium if, and only if, the value of the potential energy is a relative minimum (relative with respect to its immediate neighbourhood). A mechanical system is said to be ‘conservative’ if the virtual work W vanishes for a virtual displacement that carries the system completely around any closed path, In other words, the virtual work W corresponding to any virtual displacement of a con- servative system from one configuration Xo to another configuration X; in space depends only on the terminal configurations X) and Xj. A conservative system thatStability of Structures p fit Length L & a=? = a aA sg yr cutoff (because of small deformation assumption) increasing eccentricity 6 Figure 1.4. Load ys. mid-span deflection for an imperfect column, can pass from one configuration to another in space which is connected, with no con- straints to follow a certain path, is called a holonomic conservative system. Accord- ingly, the virtual work W in a virtual displacement from a given configuration Xo to a variable configuration X_ is independent of the path. It may be denoted by W=- V (Xo, X) (1.2) If Xo is a fixed prescribed configuration, the function V (Xo, X)_ is called the potential energy of the system in configuration X, In fact, the virtual work w=wr+w where W; is the part of virtual work performed by internal forces and W, is the part performed by external forces of the system during virtual displacement. 10Fundamental Concepts ‘Using the ‘law of kinetic energy’, which observes that the work of all the forces (internal and external) that act on a mechanical system equals the increase of kinetic energy of the system, We+W.=T a3) where Tis the increase of kinetic energy. Let us now state the first law of ther modynamics are use it to define the work done by the external forces. If electromag- netic effects are disregarded, the law is expressed as follows: The work that is perfomed on a mechanical system by external forces plus the heat that flows in the system from the outside equals the increase of kinetic energy plus the increase of internal energy’, ie. We + O=T+U aa) where Q is the heat that flows into the system, and U_ is the increase of intemal energy. Therefore, Ww=Q-U d.5) A process for which Q = 0. is said to be ‘adiabatic’. For such a process, eqn. (1.5) leads to an important special case, Wea-U (1.6) It is easy enough to now show that V(X%,X) =U - We ‘The configuration Xo, called the ‘zero configuration’, merely gives rise to an additive constant in the potential energy and is irrelevant since only the change in potential energy is significant. Consequently, V(X) =U — We (7) For an elastic continuum, the potential energy depends on one or more functions of one or more variables representing the displacement field. In Euler's column problem, for instance, Vis so-called functional (i.e. a function of a function) of the form V[w (x)], where w (x) is an arbitrary admissible function. The term admissible means that all the kinematic constraints of the problem are satisfied. In the case of a column, admissible displacement fields or configurations are represented by the continuouslity of Structures function w (x) having continuous first derivatives and satisfying the kinematics (or geometric or force) end conditions. This requitement of continuity is essential to ex- clude fracture of the column. Let us denote the trivial equilibrium configuration of system by w(x) = 0 and non-trivial equilibrium configuration by w(x) # 0, where w(x) is an arbitrary admissible function sufficiently close to the trivial function w(x) = 0. We define the increment in potential energy as, AV = V[w # 0] - V[w = 0] where V [w = 0] = potential energy in the trivial (initial) equilibrium configuration V fw # 0} = potential energy in the non-trivial equilibrium state If Vpw = 0] = 0 then, V[w # 0] = AV (1.8) If AV is positive, that is V(w) is positive definite, then Viw = 0) is minimum and hence, according to the aw of minimum potential energy, the trivial equilibrium con- figuration w = 0 is stable. From calculus of variation arguments it can be shown that the requirement AV > 0 leads to the variational problem 8V = 0, restricted to admis- sible variations 8w (w) of w(x). A function 8w(x) is called an admissible variation of w(x) if it results in a new admissible configuration w(x) + 8w(x) in the vicinity of w(x). It conforms to the kinematic constraints and thus satisfies the kinematic bound- ary conditions. Apart from the sign, 8V may be interpreted as the virtual work done by the internal and external forces in an admissible virtual displacement 6. Our problem is therefore equivalent to one of finding the smallest load for which a non-trivial equilibrium exists. It follows that for the type of systems considered here, the energy approach is equivalent to the equilibrium method. It may be remarked here that one of the ad- vantages of the energy approach is that in the formulation, definitions of a coordinate m and a sign convention are deemed unnecessary, Only expressions for strain energy and the work done by the external forces are needed. Vibration Method The vibration method examines the motion of the elastic system in the vicinity of the equilibrium configuration and thus is based on the dynamic concept. This method is more general in the sense that the other approaches, based on the static concept, are special cases of this approach when inertia forces are neglected. Since the vibration 12Fundamental Concepts method takes into account the inertia forces in its formulation, the mass of the elastic system becomes as important as the elastic stiffness of the s response of the system, therefore, becomes a function of both the space and the time coordinates. The method consists of following stipulations and concepts: An unper- turbed state whose stability is being investigated, is specified. A perturbation is then applied to the unperturbed state so that it is transformed into a perturbed one. Certain characteristic, called norms, are emphasized, which define the state at any desired time. The change in norms during the transition from the unperturbed state to the perturbed one, under the influence of perturbations, is determined. Based on this behaviour, a conclusion can be reached regarding the stability of the unperturbed state or its in- stability, This is also the basic underlying idea in Liapunov's method of stability in- vestigation which is discussed in Chapter 8. In practice, tall structures like towers, guyed masts and multistoreyed buildings are subject to wind and earthquake forces which are random in nature. In analytical in- vestigations, the load P is then no more a constant but is a function of time. The stability of these systems requires some knowledge of random vibrations and stochastic stability which are beyond the scope of this book. 132 Discrete Systems 2.1 INTRODUCTION In this chapter, the stability behaviour of some simple structural models is inves- tigated. The discussion is limited to conservative systems. The large deflection analysis leading to the concept of postbuckling (in the context of discrete systems) is also introduction and explained through a few examples. It is also intended to highlight the effect of geometric imperfections on the critical load of such systems. The energy approach which uses the law of minimum potential energy (sec. 1.4) is employed throughout. This is done intentionally as it helps in better understanding of stability of elastic systems. 2.2 MATHEMATICAL IMPLICATION OF LAW OF MINIMUM POTENTIAL ENERGY The implication of the law of minimum potential energy is first developed for a single degree of freedom system and then generalized to a system having n degree of freedom. 2.2.1 System with Single Degree of Freedom Let Q denote the degree of freedom in the generalized coordinates. At a given value of the load P, the potential energy Vis a single-valued function of Q, ie. 14Discrete Systems V = V(Q). The idea is to explore the V-Q curve around Qo, the given equilibrium state. One sets Q = Qy + q and takes Taylor’s series expansion as Vio + 4) = ViQo) + 7 wl4 (2.1) Now V(Q) = Vy) + AV = Vy + [8V + # Bv4 OV (2.2) av ry FV > where 8V = ‘ot 4 BV = Oa are the first, second, ..., variations of the potential energy. The first variation must vanish for the maximum or minimum to exist; this yields qv do 0 (2.3) which gives the equilibrium position of the system. Then, the condition for the mini- mum is that &°V > 0; this yields the following condition for the equilibrium position to be stable dv >0 (2.4) a Obviously, the condition 8°V < 0 implies unstable equilibrium while 8°V = 0 refers to neutral equilibrium. 2.2.2 System with n Degrees of Freedom At constant load P, the potential energy Vis a single-valued function of all the generalized coordinates Q,(i = 1...) ie. V = V(Qi, Qn...Qn) (2.5) To explore the potential energy hypersurface in the V—Q; space, set Q,= Qis + qi and take the Taylor’s series expansion asStability of Strnemres . ny V(Q) = VQu +4) =VOn +> ate DD ahaa oat fat Zi 2 90:22), - (2.6) Proceeding as in the case of a single degree of freedom system, the conditions of equilibrium are obtained by setting 8V = 0 i. wv 3g, 79 8= 12, . oil (2.7) Further, for stability, the second variation of V_ must be positive definite at the position of equilibrium, ie. s ~~ _aV sv= YD 300g (aur? (2.8) izt jai For this condition to hold, the determinant av gi av 90200 av _é 90,00; 90,00 ay well as all its principal minors must be positive (principal minors are subdeter- minants whose principal diagonals coincide with the principal diagonal of D). = | ove (2.8a) 9000; 2.3 ILLUSTRATIONS, The application of the energy approach will now be illustrated through examples. Example 1: A rigid bar of length L is hinged at one end free at the other. The hinged end is provided with a linear rotational spring (spring constant c). The free end is loaded with a force P in the direction of the bar. It is assumed that the direction ofDiscrete Systems Figure 2.1. Example 1: single degree of freedom. the force remains unchanged (Fig. 2.1). What is the value of the force P up to which the given straight configuration is in stable equilibrium? Let the rigid bar be displaced through a small angle @. Then the strain energy stored in the spring is The work done by the force P in going through the virtual displacement ¢ is W. = pee PL = PL. (1 = cos) = “6 (for small 6) Therefore, the potential energy of the system is given by ce - Pre (2.9) -3 V=U-W= Using the condition (2.3) for equilibrium (nothing that @=Q) we get from eqn. (2.9) 7Stability of Structures @(c - PL) =0 (2.10) The above equation yields that either © = 0 (which is the trivial case) or P = c/L whenever 84 O (the nontrivial case). The condition for stability, (eqn. 2.4) in this case gives (c- PL >0 which implies that so long as P < c/L, the given configuration of the rigid bar is stable. The value P = c/L will be termed as the critical value of the load. Example 2: Consider the system shown in Fig. 2.2. The linear rotational springs have the spring constant c. The lower bar is hinged at the bottom and is hinged to another bar which is free at the top where the force P is applied. The aim is to find the value(s) of P for which the given system is stable. Let @; and 6 be the small rotations about the hinges. It is straightforward exercise to shown that u= 308i + 560: - 0) P | fe L 2 ce! fen L 2 % c Figure 2.2, Example 2: two degrees of freedom, 18Discrete Systems cos 0 - 5 cose = & (6F + 62) (for small rotations) As such the potential energy expression is given as ct + Se (0: ~ on ~ (of + of) 6 +5 (#-4) - P(e + 03), P= PL/4e 4) Application of the condition (2.7) to V in eqn. (2.11) gives (note that Qi = @,.Q = @) (2-2P)0-@=0 - 0+ (1 -2P)%=0 from which the existence of a nontrivial solution is assured if 4P - OP +1=0 which implies that P= 0.19, 1.31 In order to investigate stability, the condition (2.8) has to be applied which takes the following from av ae” ° ev ev (vv) } 06; 367 (06, 0% In the present case, these inequalities are Pcl (2.13) GP -6P+1)<0 [P< 0.19, P> 1.31] Obviously, the practical range of P for stability isStability of Structures P= Pe coo (2.13a) 4c Example 3:° Consider a system consisting of three rigid bars as shown in Fig. 2.3a. All hinges are frictionless and contain springs such that the moment resisting a rotation @is proportional to 8. The spring constants are cz and cz, as indicated. The aim is to investigate stability of the system under an axial load P Consider the deflected configuration of the system shown in Fig. 2.3b. The angles 6, and @; are adopted as the generalized coordinates. The strain energy stored in the springs is U= } [ev0t + 01 68 + co @ ~ 0) + c2(6: + OF] (2.14) Also, sin@, =, sind: =", sing, = 22% @ a 15a Hence sin @; = 2 osin @ — sin) or, 8s = 2 (@: — @,), for small rotations (2.15) Eqns. (2.14) and (2.15) yield Us Ge + #4) (@ + @)-2 Scr 0) @ (2.14a) The work done by the force P is = P(3.Sa ~ L) (2.16) in which the length L is L = acos® + acos® + 1.Sacos 0; 2 2 2 ~ Bho (1-¥) ers ( 2) (2.17) *Langhaar (1962), 20Discrete Systems (b) stable zone cyleg symmetric antisymmetric Figure 2.3. (a) Three rigid bars system, (b) Deflected configuration. (c) Stable and unstable zones. (d) Instability modes Using eqns. (2.17) and (2.15), eqn. (2.16) becomes 5 2 2 . g Pa (0 + 63) ~ 5 Pa & (2.16a) 21Stability of Structures Thus, the total potential energy of the system from eqns. (2.15a) and (2.16a) is ve(3 at Be re] (ai + 68) + G Pa - 2 30] 8 (2.18) For equilibrium, application of eqn. (2.7) yields 1 29 5 2 20 (3 + 7g c2 ~ ree + (3 - Pa =0 2, 20 Mg yO po Ba Oem 2 (F- Pals a(pe +g e sPejano 2.19) For a nontrivial solution, the requirement is that 2 2 (2 + He- re) -(5r0- Pa) =0 eee pally or, [ge + 502 - which implies that c+ en Pays. a+ Ze 7 3° As in the previous example, to consider the nature of equilibrium for the system in its undeflected (0, = @ = 0) position, we need to investigate stability by using the condition (2.8). In the present case, this simply leads to the following inequalities 3 Pax 50+ Ba (2.20a) Pa 4 The critical load, therefore, can be expressed as ate Q.21a) Pa=(3. 7 ae at ze (2.21b) It is easily seen from eqn. (2.19) that for the critical load given by eqn. (2.214), @1 = 02; and for load corresponding to eqn. (2.21 b), 8; = ~ @:. Accordingly, the mode of instability is symmetric if 3c, < 7c; and antisymmetric if 3c) > Ter (Fig. 2.3 d). 2.4 LARGE DEFLECTION ANALYSIS The stability investigation using the small deflection approach presented in the previous section only gives information about the critical load; no information is available about the behaviour of the system after the critical load is reached. In other words, this precludes the study of the postbuckling behaviour of the system. To study the postcriti- cal behaviour, we must use the large desplacement analysis. This is illustrated through examples restricted to single degree of freedom systems. 2.4.1 Examples Consider the problem of a rigid bar with a rotational spring (sec. 2.3, example 1, Fig. 2.1). This critical load was found to be P = ¢/L. Now, if the large rotations are allowed, the potential energy of the system becomes Vv pee - PL (1 ~ cos6) For equilibrium we must have dV/d@ = 0; the equilibrium path is then 68 ~ Tsin6 (2.22) This is ploted in Fig. 2 the stability of the equilibrium paths, we consider the second deri’ is It intersects the fundamental path at P = c/L. To study tive of V whichStability of Structures ‘post critical path i«—— fundamental path 8 ° Figure 2.4. Critical and posteritical equilibrium paths. dv ae ¢ — PLcos® (2.23) For the equilibrium path given by eqn. (2.22), eqn. (2.23) becomes OY ody 8 ay tan® which is always positive indicating that the postcritical path is stable. At the critical load (P = c/L), for the fundamental path (6 = 0), d°V/d8" = 0 from eqn. (2.23) Hence, to investigate the stability of the critical equilibrium state, it is necessary to examine the higher derivatives of V. It is easily seen that in the Taylor series expansion given by eqn. (2.1), 1 3! ae which at P = c/L becomes a 08 1 ay PLO = and is always positive. This implies that the equilibrium state at the critical point is stable. 24Discrete Systems Figure 2.5. Rigid bar with translational spring. In other words, the proceding discussion suggests that the stable fundamental path (6 = 0) bifurcates into ar unstable equilibrium path (corresponding to the given straight configuration of the bar) and a stable postcritical equilibrium path (correspond- ing to the deflected configuration of the bar). Consider now another system shown in Fig, 2.5. Here, one end of the rigid bar is connected to a linear translational spring (spring constant &). Using the small displace- ment approach, it is straightforward to show that the critical load is P = kL. Considering large rotations, the potential energy of the system is VsU-W.= $k (Lsin 8)? — PL(I — cos @) For equilibrium dV/d@ P=kL cos® (2.24) This is plotted in Fig. 2.6. It intersects the fundamental path (@ = 0) at P = kL. We now consider the second derivative of V: ev ae 0; the equation of the equilibrium path is then = kL? (cos’@ — sin’@) — PLcos® (2.25) For the equilibrium path given by eqn. (2.24), eqn. (2.25) becomesStability of Structures unstable — stable postcritical e{path i-— fundamental path 8 Figure 2.6. Critical and posteritical equilibrium paths. which is always negative, indicating that the posteritical path is unstable. As in the previous example, d?V/d is zero at the critical load and hence it is necessary to examine the higher derivatives of V. It is easy to show that in the Taylor series ex- pansion given by eqn. (2.1) 1dv| 3 ip] @ =o ° i ay Of = 5 PL ~ 4404 0° which at P = kL becomes — kL? 6°/8, and is always negative. It means that the equi- librium state at the critical point is unstable. Hence it can be concluded that the stable fundamental path (@ = 0) bifurcates into an unstable equilibrium path (corresponding to the straight configuration of the bar) and an unstable equilibrium path corresponding to the deflected configuration of the bar. Let us now consider a system shown in Fig. 2.7. The linear spring (spring constant 4) is inclined at 45° when the rigid bar is in the given straight configuration. Due to rotation @ of the bar, the strain energy stored in the spring is 1 U = FHA Saving 26Discrete Systems Figure 2.7. Rigid bar with inclined translational spring. Lv, 4 where As = 2 cos é - 3) and the work done is W. = PL(1 = cos @) Hence, the potential energy is v=U-W, if the series expansion around 8 = 0 is taken. From the condition dV/d8 =0, we get equations of equilibrium @=0 (2.262) (2.26b) which are represented in Fig. 2.8, To consider the stability of the equilibrium paths, we resort to the sign of the second derivative of V, We have aStability of Structures P --- unstable ‘ : — stable ‘ ' ' postcritical path fundamental path 8 Figure 2.8. Critical and postcritical equilibrium paths. #V 1 ys (y_ 59% =gkt (-»- q|7 Pet - For the fundamental path, given by eqn. (2.26a), eqn. (2.27) gives et (F-?) (2.27) de 2 from which we deduce that the equilibrium is stable for P < kL/2 and unstable for P > kL/2. Along the posteritical path, given by eqn. (2.266), eqn. (2.27) gives ay _ Ke do 8 (- 30 + 6°) which suggests that, as the linear term is dominant over the quadratic term around 8 = 0, the equilibrium is stable for @ < O and unstable for @>0 (Fig. 2.8). It is worth mentioning here that structures with postcritical behaviour of the type shown in Figs. 2.4 and 2.6 are called structures with symmetrical postcritical behaviour. On the other hand, all those structures which present an equilibrium path of the type shown in Fig. 2.8, with non-horizontal tangent, are called structures with asymmetrical postcritical behaviour. Finally, we consider a system (of one degree of freedom) with nonlinear precritical behaviour. It consists of two rods which can take only axial deformation. They make an angle ® with the horizontal. The load P is applied as shown in Fig. 2.9. It is assumed that the hinge can move vertically only. At some value of P, let the angle 28Discrete Systems Figure 2.9. Mises truss. which the rods form with the horizontal be @. Then the potential energy of the system ty ase zh (sy ~ Pe where k is the axial stiffness of each rod, As is the’elastic shortening of the rod and ¢ is the vertical displacement of the point of application of the load. It is clear from Fig. 2.9 that As = Lsec @ — Lsec® e = Ltan®@ ~ Luan from which the potential energy expression takes the form V = KL? (sec — sec @)* — PL (tan @ — tan@) By imposing the condition dV/d® = 0, we obtain the following equation of equilibrium = 2 sin @ (sec @ — sec 8) (2.28) which describes the equilibrium path and is shown in Fig. 2.10. The path is antisym- metric, vanishes for 9 =+ y, 0 and shows a maximum for @= + ®,. Hence by imposing the condition dP/d® = 0, we get cos’ 8 = cos A so that from eqn. (2.28) the following results 2»Stability of Structures BePrkL Figure 2.10. Stable and unstable equilibrium paths. P. = P/KL = tan’ @. To determine the stability of the equilibrium points, it is necessary to investigate the nature of the second derivative of V. We have a = 2kL? [sec 9 (1 + 3tan7@) — sec @ sec Gy (1 + 2tan? @) + 2P tan @ sec*O] (2.29) which, by eliminating @ using eqn. (2.28), becomes dv 21 iB og 7 AL - 2.29: agi = AL Sag (tan'e - P) (2.29) From Ey. (2.29a) it is clear that, for P< tr *@ the equilibrium is stable; for P > tan’ @ the equilibrium is unstable; for P = tan’ @ the equilibrium is critical. On the plane (P,6) the curve P = tan’® separates the region of stable configuration from the unstable region (Fig. 2.10). It intersects the path (2.28) at the limit points P = tan’ @,. The mechanical behaviour of the model (which is the Mises truss as mentioned in Sec. 1.3) can be better appreciated by referring to Fig. 2.11" which explains the snap- through phenomenon. As the load P increases from zero, the equilibrium point moves along the curve OC, characterised by configuration of stable equilibrium. If the system is unloaded from any point, say B, on the path OC, the representative point moves backward along BO. However, if the representative point reaches the maximum C, the *Pignataro etal. (1991). 30Discrete Systems P=PIkL Figure 2.11. Illustration of snap-through phenomenon position of equilibrium becomes unstable and the systems passes to E where it finds, at the same load, a position of stable equilibrium, A similar explanation holds good for negative values of P. 2.4.2 General Treatment ‘The preceding four examples have illustrated that, at least for single degree of freedom systems, there could be four types of posteritical equilibrium paths. In a general treat- ment of the analysis that follows, it is shown how all the types of equilibrium paths can be recovered by considering the derivatives of the potential energy function. The potential energy of a single degree of freedom system can be expressed as V=U ~ W. = UQ) - Pe(Q) where ¢ is the displacement in the direction of the load P,Q being the generalized coordinate. Let Po, Qo be the position of equilibrium. Now, the system is given a small perturbation such that P = Po + p, Q = Qy + qi with the result the potential energy of the system becomes V = U(Qo + q) ~ (Po + p)-e(Qo + 4) The Taylor's series expansion around the given equilibrium position is 1 > 1 VE Ut Ug t > Ung +3; Ung + 31Stability of Structures ) (2.30) 1 »>,1 Pot rote gt a eng + zea. in which, for the sake of shorthand notation, the subscript indicates the order of dif- ferentation with respect to q, i.e. d & a (n= ay Ons ape Om = Fs and the derivatives are evaluated at the equilibrium position (Po, Qs) Eqn. (2.30) can be rearranged as V = (Uo ~ Po ee) + (Ui = Poe) g + Un = Poe Gi + Lg Plotter. gt a eng +...) 2,1 , onV= Wt Vig ta Ma d+ 3, Ving +. -plere thagtteng + ) (2.30a) P rgt a eng tareng +... 23 The equation of the equilibrium path is obtained from eqn. (2.30a) by setting dV/dq = 0. This results in the following nonlinear equation 1 1 Vag +5) Ving + 3) Ying p= 3 (2.31) eteng+ Qe + If, now, ((Po. Qo) is a critical state of equilibrium, then Vi; = 0. at that position. Then, the following possible cases emerge. (a) Snap-through behaviour An approximate parabolic relation from eqn. (2.31) is 1 PS oy Vine gleDiscrete Systems (b) Asymmetric postcritical behaviour If e, = 0, then an approximate relation from eqn. (2.31) is ly pas Vn.g feng which implies that either g ot p = 5 Vin-a/en ymmetrical posteritical behaviour 0 and Vin = 0, then an approximate relation from eqn. (2.31) is 1 pay Yan. g/en.g which implies that either g = 0, 1 orp = a Vin. en The three cases above represent the four types of posteritical equilibrium paths men- tioned earlier and are explained in Fig. 2.12. P 4P 4P ' t 1 Ms P Aro 4.30 ee PO x prot fi in Wt \ gn ViP0 Vyge0's “a Vic » > @ case (a) case (b) case (c) Figure 2.12. Types of posteritical equilibrium paths 33Stability of Structures — —v (a) (b) (c) Figure 2.13. Single degree of freedom systems with initial imperfection. 2.5 EFFECT OF SMALL IMPERFECTIONS The aim in this section is to examine how the results obtained in problems in sec, 2.3 (Figs. 2.1, 2.5 and 2.7) are modified if imperfections are present. This could mean a small initial inclination of the bar corresponding to the unloaded spring, or a small eccentricity in the force P, or a small horizontal force acting on a point of the rod. Let us consider the case in which the imperfection is an initial inclination @ (Fig. 2.13). The procedure is to write the potential energy expressions which, for the three cases, are now of the following form: (Vv oe — ©)’ — PL (1 — cos®) + PL (1 — cos @) (2.32a) " (b) V $ k(Lsin @ ~ L sin @)* — PL( - cos@) + PL(1 — cos ®) (2.32b)Discrete Systems ‘ we G - & -26 a| - ee (2.32e) The equations of the equilibrium path(s) are obtained by setting dV/d® = 0; these for small values of @) are [in place of eqns. (2.22), (2.24) and (2.26), respectively), = 20 = 60) (a) P= ‘Lsind (2.33a) = @ _ & (b) P = kL (! -37 4 (2.33b) =-kL{,_35_ & (c) P= 2 (: 4° 3) (2.33c) Egns. (2.33) have been plotted in Fig. 2.14. In each case four families of curves are observed, two corresponding to > 0 and two to @ <0. When @> — 0, the equilibrium curves of the imperfect structure tend towards those of the perfect structure represented in the figure. The curve S, in each case, has been obtained by plotting the function d'V/d@? = 0. It separates the total domain into stable and unstable regions. It is interesting to note that for cases (b) and (c), there are limit-point loads P, on the paths of the imperfect structures. The magnitude of the, limit-point load may be obtained by setting dP/d® = 0; accordingly these values are Pr = kb (: - 3 w) for case (b) p= (1-5 a) for case (c) This suggests that structures with asymmetrical postcritical behaviour have a greater sensitivity 10 initial imperfections than structures with symmetrical postcritical be- haviour. Koiter (see, e.g. Thompson ané Hunt, 1973) has shown that the equation for the limit-point load has the following general form: (1 = A)" = OAL pl (2.34) In which 4, = P/P.» 1 is a parameter that represents the imperfection, ot is a positive constant, m = 3/2 for case (b) m= 2 for case (c). At the bifurcation point P =P, and %, = 1. Hence for limit-point loads sufficiently close to the bifurcation 35Stability of Structures ---- unstable __. stable Figure 2.14. Stable and unstable equilibrium paths for imperfect systems. point load in magnitude, A;= | and A, may be replaced by unity on the right-hand side of eqn. (2.34). Thus (1 Ae)" = oF Il for case (b) (=a)? =o Ip for case (c) The existence of these limit-point loads was verified experimentally by Roorda (1965), 363 Columns 3.1 INTRODUCTION The classical critical load theory of perfect columns is based on the following assump- tions (a) The column is perfectly straight (b) The column is centrally loaded i.e. the line of action of the compressive force coincides with the centroids at the two ends of the column. (©) The material is homogeneous and isotropic and it obeys Hooke’s law. (@ The assumptions of the theory of bending hold good for the critical load theory also, i.e. plane sections remain plane before and after deformation and the slope at any section is small compared to unity. (e) The bending stiffness EV is constant throughout its length, Obviously, these requirements describe an ideal coluinn that cannot be realized in practice. However the study of such a column is requisite to an improved understanding of column behaviour. 3.2 IMPLICATION OF LAW OF MINIMUM POTENTIAL ENERGY In sec. 2.2, the mathematical implication of the law of minimum potential energy was developed for discrete systems. This will now be applied to an axially loaded column 7Stability of Structures p Pp Py. al lengthLl ,EI constant w —>x Figure 3.1. Hinged-hinged column (Euler column). of uniform prismatic section. For such a column (Fig. 3.1), the strain-energy in bending (neglecting any effect of axial shortening) is expressed as while the work done by the load P is Pt (dw) ef (tj Thus, the potential energy of the column is HET Gyett Ge 0 We set the deflection as w(x) +en (x), (4) being an admissible function (i.e. con- tinuous, at least twice differentiable, satisfying forced boundary conditions) and € an arbitrary infinitesimal parameter, to study the variation in the potential energy in the light of the law of minimum potential energy. Thus, V becomes rot le) ~3)]+ ve ffm de 4 pte a a 38Columns “telle(@)--)]« =VWtvr+reV G2) In eqn. (3.2), the first term on the right-hand side is the potential energy in the given configuration, the second term and the third term, respectively, denote the first and the second order changes in the potential energy. For the maximum/minimum to exist, the first variation must vanish, i.e. vise dw fy _ paw dn) gy _ avee j J [ere a ae ae He =0 63) Using integration by parts eqn. (3.3) can be expanded to yield dw gy a dw a dw d a Tw ~ af # @ 1 @ - [a aw + pa «| nid) t a a a otsmertpafpoe|lats-rd)va- G4) For eqn. (3.4) to hold, the integral and all the boundary terms must vanish. For arbitrary n(x), the vanishing of the integral is guaranteed if dw, dw Boat PS The vanishing of the remaining terms in eqn. (3.4) yields the possible boundary con- ditions (end conditions) for the problem. This can be explained through following three standard cases: (i) Hinged-hinged column: The forced boundary conditions are w(L) = 0, w(0) = 0; hence n(L) = (0) = 0 which makes the third and the fourth terms in eqn. (3.4) vanish. For the first two terms to vanish, the requirements are 0 G5) (3.6) which are the, natural boundary conditions 39Stability of Structures (ii) Clamped-clamped column: The forced boundary conditions are w(L) dw dw (0) = 0, F° (L) = 0. F* (0) = 0. Hence n@=n1@=2uH=-2@- All the four boundary terms in egn. (3.4) vanish. (iii) Cantilever column: The forced boundary conditions are (at the clamped end x = 0 say) w(0)=0, ae (0) =0. This implies that (0) =a (0)=0. Therefore, from eqn. (3.4), the natural boundary conditions are dw a ay = El +P =o (3.6a) Thus, we have seen that the condition 8V = 0 leads to the solution of a fourth order differential equation subject to a set_a four boundary conditions. The solution for w (x) would give all the possible equilibrium paths. To investigate the stability of the equilibrium paths, we need to consider the second variation of the potential energy. On the other hand, if the aim is to find the critical load for a given problem, one can look for the nontrivial solutions of the governing differential equation only. This is typically an eigenvalue problem. In the next section, this approach is used to determine critical loads of columns with some standard boundary conditions. 3.3 CRITICAL LOAD FOR STANDARD BOUNDARY CONDITIONS Let us consider columns of standard boundary conditions and obtain expressions for critical loads. 3.3.1 Hinged-hinged Column (Euler Column) Fig. 3.1 shows an ideal column which is hinged at both ends and subjécted to an axial compressive force P. The critical load can be obtained by considering the state of equilibrium of the column in its bent form, caused by the disturbance. Taking the 40Columns coordinate axes as indicated and equating the applied external moment to the internal resisting moment, we obtain the governing differential equation of equilibrium as EI SS = - M=- Pw G7) de Since the buckling (bending) will occur in the plane of minimum-bending rigidity, the plane of symmetry is taken here as the plane of buckling and the minimum value of EL is to be used in eqn. (3.7). Introducing the parameter k* = P/E! the governing differential equation of equilibrium is rewritten as (3.8) The general solution of this equation is w = A; sinkx + Az coskx G.9) where A; and A2 are the unknown constants of integration. These constants of integra- tion are determined by requiring the solution to satisfy the prescribed boundary con- ditions, namely, w(0) = Ww) = 0 (3.10) Due to the first requirement in egn. (3.10), we obtain Ar = 0. In order to satisfy the other boundary condition we require A; sinkL = 0 G.11) If A; is set equal to zero, then w = 0 everywhere along the span. This is a trivial solution, meaning that the straight form of the column is the only equilibrium state under the force p and no bent equilibrium state is available. Since we are interested in determining the critical load, the corresponding neutral state of equilibrium requires the column under the force P to be able to remain in equilibrium both in the straight and bent forms. Hence, A; # 0, sinkL = 0 The solution of this equation is kL = nnn = 1,2,3,.. 41Stability of Structures _ WEL Py = BEE 3.12) or Z ( The smallest of these corresponds ton = 1. Thus, 5 pl, = EEL (3.13) E When the force P takes any one of the values defined by eqn. (3.12), sinkL = 0; Ay # 0, thereby indicating the equilibrium of the column under this force in its bent form, represented by eqn. (3.9) as nm Ay sin L (3.14) It may be pointed out here, that because of the condition that the nontrivial bent equi- librium state must be proximate to the straight equilibrium state in order to satisfy the basic assumptions stated earlier, Aj must remain a constant, the order of magnitude being small. Jt is worth mentioning that when force P is different from the values defined by eqn. (3.12), then to satisfy the relation in eqn. (3.11) A; has to be zero. Therefore only the trivial straight form of the column is available. When force P takes on any one of the values defined by eqn. (3.12), eqn. (3.11) is. satisfied both with A; = 0 and A, # 0. It means that at these values both the straight and the nontrivial bent equilibrium states are possible. Hence, these values are sometimes known as ‘bifurcation’ loads. Incidentally, the homogeneous differential equation of equilibrium [eqn. (3.8)] along with the homogeneous boundary conditions [eqn (3.10)] from a class of problems known as ‘eigenvalue problem’. The value Pc, defined by eqn. (3.12) are called eigen- values and eqn. (3.14) defines the eigenfunctions. Fig. 3.2(a) shows the buckled-mode shape of the column for the lowest critical load Pl, For n = 2, 3,... higher values of the critical load are obtained; the corresponding buckled modes of the column defined by eqn. (3.14) are shown in Figs. 3.2(b) and ©. 3.3.2 Cantilever Column Fig. 3.3 shows a cantilever column acted upon by a compressive force P at the free end. The external bending moment at any cross-section in the bent configuration of the column is azbee 1 L/3- —+~—_. L/3. —+-—___L/9. —+1 Figure 3.2. (a) First buckling mode. (b) Second buckling mode, (c) Third buckling mode. " lr A: w(x) w Figure 3.3. Cantilever column. Columns —r a3Stability of Structures M =~ P (A -w) and the differential equation of equilibrium of the column becomes (3.15) As before, the equation takes the form aw RweRA (3.16) de the general solution of which is w= A; sinkx + Az coskx + & where Ay and Ay = unknown constants of integration A = unknown deflection at the free end The constants Ai and A> are determined by requiring the solution to satisfy the prescribed boundary condition at the fixed end, namely w(0) = 2 (0) = There are satisfied if A, =0,A2=-A The deflected bent form of equilibrium of the column is given by w= A(I — coskx) G17) The condition at the free end is that the deflection w becomes equal to A forx = Lie. A= A(I — coskh) The solution of this equation requires either A = 0 or coskL = 0. If A = 0, the only state of equilibrium under the force P available is the trivial straight form of the column. Therefore, to ensure the nontrivial solution, we require 4gColumns coskL = 0 whose solution is kL = (Qn = 1) } a BSpcx or > TEL wr = (Wn - IP 3.18 Peon - 1 2 (3.18) The smallest for these corresponds to n = 1; thus ‘The buckled form of the column is defined by eqn. (3.17) as me y= ali - cos HE w ( cos } For n = 2,3, in eqn. (3.18) we obtain the corresponding values of the compressive forces as On? El 25n° EL Pi, = > Py 4 aL The deflected shapes of the column are given by eqn. (3.17) as 3nx Sax waft on) w= a(s on) These deflected forms of the cantilever column are shown in Figs. 3.4(a) and (b). These higher critical loads and their corresponding mode shapes are really of no prac- tical significance since they cannot be realized although they exist mathematically. The simply supported column and the cantilever column are the cases in which the structures are statically determinate. These two structures provide the fundamental cases of buckling of perfect columns. 45Stability of Structures y tf les (a) (b) Figure 3.4, Higher buckling modes for cantilever column, 3.3.3 Fixed-hinged Column The column shown in Fig. 3.5 is fixed at the end A and hinged at the other end B. It is subjected to a compressive force P. The external bending moment at any section of the slightly bent configuration of the column is M = Qs (L - x) + Pw where Qg represents the reaction at the end B. The governing differential equation of equilibrium in this case is dw Er XS =~ M =~ Qs (L- 9) - Pw dx Differentiating the above equation twice with respect to x, we get, det det _ de de (3.19) 46Columns Soe w(x) Ww Length L Figure 3.5. Fixed-hinged column. which is the same as eqn. (3.5). In fact this is the governing differential equation o1 equilibrium of all perfect columns in their bent configurations in the presence of an axial force P. The general solution of this equation is w = A, sinkx + Az coskx + As x + Aa (3.20) The boundary conditions which have to be satisfied are w (0) = #o = w(L) = on () =0 These requirements lead to four linear homogeneous algebraic equations in the con- stants Aj-Ag Az + Ay= 0 Aik + As = 0 A, sinkL + Az coskL + AxyL + As= 0 Ay sinkL + A, coskL = 0 A nontrivial solution exists if all four constants are not equal to zero. This is possible only if the determinant of the coefficients Aj (i = 1 — 4) vanishes, ie. 0 1 0 k 0 1 sinkL coskL L sinkL coskL 0 eco a7Stability of Structures tank t0kKL) Figure 3.6. Graphical solution of eqn. (3.21). which finally leads to tankL = kL (3.21) and is also known as the characteristic equation. This is a transcedental equation which can be solved either numerically or graphically as shown in Fig. 3.6 The lowest critical force is determined from the smallest root of eqn. (3.21) which is found to be equal to 4.493: kL = 4.493 Pl, = 20.19 a The other roots of eqn. (3.12) represent the higher critical forces P2, Ply... The buckled mode shape corresponding to the lowest critical force is given by 4.493, ws Ay sin x - 40 + cos 4.493 7 — 7 3.3.4 Fixed-fixed Column Consider a column fixed at both ends and subjected to a compressive force P as shown in Fig. 3.7. The external bending moment at any section is 48aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.Stability of Structures Length 7x Figure 3.9. Elastically restrained column, dw __ dw dw, dw)_— Bl Ge = Bo Gi - [BIGS + PSE = Gow The first equation represents the moment equilibrium condition while the second equa- tion represents the equilibrium condition for the shear forces at the left end. Similarly, the boundary conditions at the right end x = L are dwg dw. dw dw = a ee raed i id Introducing new parameters -Hg-b 4 -&, Bh G0 = py Bo = py = Ey BL By the boundary conditions become fem ndtiqe TiO) ~ Bo G-(0) = 0 , £E 0) + EM) + cw) =0 20 +p. Hu) =0 G24) , so + eigy = a wL) =0 52Columns Substituting the general solution of the governing differential equation in eqn. (3.24) leads to the following four linear homogeneous algebraic equations in Ai i = 1,2,3,4). Bo kA) + @ Az + Bo Ax = 0 Ar + PAs + OAL = 0 (kBi coskL - K sinkL) Ay — (By ksinkL + k coskL) Az + By As = 0 (cm sinkL) Ay + (oy, coskL) Az — (kK — o% L) Ay + u% As = 0 For the nontrivial solution the determinant of the coefficients of the A; must vanish, ie. 0 oO RB a O% Bok e Bo Oo} 9 ay, sin kL. ot, coskL aL-P o| ~ (Bi coskL — F sinkL) -BrksinkL + KcostL) B. 0 Denoting kL by u the characteristic equation is obtained in the following form ‘ ‘ , [1+ + (Bifano) +05 cH 05 cu B+ = pao] sinu s , + ‘ + G1) (Bo + Bi) Ty ~ co aul (Bo + Bu ~ 200,88 ‘| cost + 2a ou BB Ft = (3.25) The roots of this equation yield crit first Euler critical load al loads and the smallest root u!. represents the Pi, = (uty? Et (3.26) L Alll the earlier solutions of stability problems of columns can easily be obtained from the general eqn. (3.25) by substituting values for the spring constants Go, Oz, By and By, corresponding to various end conditions. 33aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.Stability of Structures the characteristic equation is simplified to 4 a sinu = 0, 2 So ui, =m and r= oe 3.4.1 Partial Frames The column members of frames are typical examples of elastically restrained columns. Consider the partial frames shown in Figs. 3.10(a) and (b). The difference between the two partial frames is the support conditions at the base A. As the load P increases, it is possible to reach some values of P for which the frame will buckle as shown in Fig. 3.10. The column AB is an elastically restrained column, the elastic restraints being provided by the beam BC. The flexural or rotational stiffness of the beam BC is 4(EN,; /b and the axial or extensional stiffness, normal to the direction AB is (EA); /b. In most practical cases, the axial stiffness is taken to be infinitely large. For such cases, column models for the partial frames are those shown in Figs. 3.11(a) and (b). (a) Figure 3.10. Partial frames. The mathematical formulations for cases (a) and (b) are based on the same dif- ferential equation of equilibrium, but different boundary conditions. Case (i) The boundary conditions are 56Columns aS ELL ee tn ELL es r A= MEM lb Ae 4(E1y/b (a) (b) Figure 3.11. Column models for partial frames. wo = 220) =0, w@)=0 and £1 4) +B way =0 From the first two boundary conditions, Ar = Aa =0 Using the third and fourth boundary conditions, A, sinkL + As L = 0 Ai (Bik coskL — sinkL) + As By = 0 In order to get the nontrivial solution, the determinant of the coefficients of A, and A; must vanish. This results in the characteristic equation Elb 1 = a7 3. cot2u = Feng 2 + dy G.27) where, now, 2u = kL. The smallest root of eqn. (3.27) represents the first critical load for the buckling of the partial frame in Fig. 3.10(a). If (EJ, = E/ and L=b the char- acteristic equation becomes Qu 22, 3. cot2u 4 ea Qu (3.28) The smallest root (2te) = (Ker = 3.83 37aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.Stability of Structures Figure 3.14. Graphical solution of eqn. (3.31). Using the general solution of the differential equation and applying the boundary con- ditions [eqn. (3.30)], we obtain the characteristic equation 2u 2ub El Bl 7 7 Lem @.31) tan2u = - where, again, 2u = kL. From Fig. 3.14 we see that, depending on the value of b EW/L (ED), the critical value of 2u varies between 7/2 and m, as expected. The characteristic egn. (3.31) can also be obtained from eqn. (3.25) with SED = 2%, By = 2% a, = 0 and B. = SOF As a special case, if b EI = (EDj L, the characteristic eqn. (3.31) reduces to Qu b tan 2u = - (3.32) and (2i)er = (kL)or = 2.71 from which P,, = 7.344 Ef Bz Case (ii) The boundary conditions are 60aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.Stability of Structures he devised a plot from which P! = P,, can be determined with the help of experimental data. This is known as the Southwell Plot. Assume that a # 0 and all a, = 0 for n = 2,3,4,... Then, from eqn. (3.39) we get l= which leads to a linear relation between 5/P andd (3.40) a a P Since, as P -> P' (ct ~» 1), the first term in the series feqn. (3.39)] is the predominant one, it can be safely assumed that eqn. (3.40) holds for large values of P (but < P'). Thus, if we plot /P vs at higher values of 8, the relation is linear and the intercepts give a, and a) / P! (Fig. 3.17) 6/P x experimental a,/p! data ay Figure 3.17. A typical Southwell plot. 64aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.Stability of Structures Now, consider any two buckling modes, say w(x) and w(x) of a column whose bound- ary conditions are specified. Since these are solutions of the basic column eqn. (3.5) corresponding to the buckling loads P, and P, respectively, (3.42) (3.43) Multiplying eqn. (3.42) by wy and eqn. (3.43) by w; and integrating over the length of the column Loray r Jot S fer G8) », S| ae = 0 td ae de fw lS fer 22), 2 2] eo olde de)" de . Integration by parts yields 4 dy, dwn, Fn _ dw; y dui] a dni ua] «fy BL Oe TG | =O (3.420) Lot dw; dw; ,, dw, _ dui, dw, and {n ya uw) + J [4 Bd ~ GP Gal a= 0 (3.43a) d dw dw dw where V = -(é (<2) * oe M=- EIS Subtracting eqn. (3.43a) from eqn. (3.42a), n “ dw; dw; dw; dw, emi ft de [oy Ws vy = oy 9 a) The terms within brackets on the right-hand side vanish for any combination of the most commonly encountered boundary conditions, e.g. 68aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.Stability of Structures 2 2 wu=s[i+() é +3) ee] (3.50) Eqn. (3.50) shows that if kL < 7, there is no solution other than @ = 0. However, if kL > %, there are deflected equilibrium forms of the column. Consequently, a bifur- cation occurs at AL = 1, ie. at P = n° EI/L? = Pl, which is eqn. (3.13). Further, if a itself is taken to be very small, ie. a = 0/2, then from eqn. (3.50), _ WEL oe pat [i+§ which indicates that the post-buckled path (similar to postcritical path discussed in chapter 2) is an upward rising curve. However, the column sustains only a very small increment of load above the Euler buckling load. For example, the load corresponding to o = 30°, as obtained from eqn. (3.49) is only 1.03 P,. 3.9 ECCENTRICALLY LOADED COLUMN Consider a column subjected to a compressive load applied at an eccentricity ¢ at both ends (Fig. 3.21). By taking a coordinate system shown in the figure, it is obvious that the behaviour of the column is governed by eqn. (3.8) with boundary conditions w =e at x = 0, L.It is a simple to show that, in this case — cos kL wee Tin kL sinks + cosh] Therefore, the mid-span deflection 8 is B= when = ef Length L w(x) Figure 3.21. Eccentrically loaded column. nRaa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.Stability of Structures nae awe ds aR ig gece R+ Or ds~ K+ q(xyds=0 aM, ds dR M+ a ds ~M~ G+ Fa + 8) ome + 3{+ +H acs Hine 0 (42) 2 ds Since the terms (dR/ds) ds and (dH/ds) ds are negligibly small compared to R and H in the above equations, these three equilibrium equations reduce to du a9 dR a + q(x) =0 aM cos 8 + H sin = 0 43) ‘The curvature and the radius of curvature R are related to the deflection w (x) of the beam-column as Py fat = dw/de (44) o= A Ro (1 + (dw/dxyy? ‘The preceding relationship is obtained using the definition of slope @. tan@ = 5", 8 = tan dx _ Pwl dee ( 1+ (aw/as) dw -1 dy d. ers The arc length ds is approximately given by ds = [1 + (dw/dx)']'? dx When the deflection is small and the slope @ (in radians) is negligible compared to unity 76aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.Stability of Structures Mb Leh) _ 3El, [ew + i |-° Since M + 0 the characteristic equation becomes Lah y+ Bae A! we) +7 b (441) Making use of eqn. (4.31), eqn. (4.41) is rewritten as lbh uy cot Qu = 1+ 3 FF Quy (4.41a) The solution of eqn. (4.41) provides the critical load. Case (i) h=h hehe The characteristic eqn. (4.41a) yields 2u, cot 2u; = 1 + 5 (2a the smallest root of which is 2u) = 3.725, i.e. EL P., = 13.90 cL Case (ii) =m h=h=l In this case, the characteristic equation becomes uy cot Quy = 1 + 2 Qu; (4.41b) the smallest root of which is 2m, = 3.508 96aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.Stability of Structures B = 4n'EI/L*, the buckled column will have an inflection point at the middle. By increasing B, we obtain conditions in which the number of half-waves is n = 3, 4, As B continues to increase, the number of half-waves also increases. Then, when 1 is neglected in comparison with n, the condition given by eqn. (4.58) becomes (4.60) P,, is twice as large as for a hinged column of length L/n. This can also be seen from Fig. 4.20 where the load vs. modulus curves come closer and closer to the dashed curve which corresponds to the special condition & = r. This critical load, as given by eqn. (4.60), is consistent with eqn. (4.56) for the special case of k = r. 4.5 SINGLE-STOREY FRAMES The results for the beam-columns can easily be utilized to determine the buckling load for a frame member. As an example, consider the frame shown in Fig. 4.21(a). The loads P are assumed to remain vertical and retain the eccentrically e, regardless of deformations. Due to symmetry, the free-body diagram of the column can be repre- sented as in Fig. 4.21(b). Using results from sec. 4.2.6, it can easily be shown that for the column AB, # eee ab & =e (+ soc i}* > ( cot kL t} while for the beam Myb » 9« = oF "The expression holds as long as the axial forve in the horizontal member's neglected. 104aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.Stability of Structures Let us now assume that sidesway is allowed. Although the frame is loaded sym- metrically, some inevitable imperfections and disturbances cause it to move laterally and the frame buckles in an unsymmetrical mode. To locate this point of bifurcation, the possibility of an infinitesimally small lateral deflection of the joint A is first as- sumed. Having done this, one can find an equation [eqn. (4.64) to follow] relating all combinations of P and H which admit lateral movement. At incipient sidesway buck- ling, both the symmetrical and unsymmetrical configurations must be in equilibrium. Therefore, P., and its corresponding, Ha must be such as to satisfy eqns. (4.63) and (4.64) simultaneously. To obtain eqn. (4.64), one may proceed as follows:” (i) express @4 at the column ends in terms of the load, reactions, section properties and the in- finitesimally small sidesway; (ii) express the rotation at the beam ends (also @,) in similar terms; (iii) eliminate , to obtain one equation in other quantities. This equation with contain terms with and without sidesway. Since the aim is to find the conditions under which sidesway can start, the latter terms are of no interest and may be dropped. From the remaining terms, the conditions under which even a small sidesway is ad- missible, may be written as the desired equation. 2PL nt t X m_HL n x (n/m) BEE 4 ce Brgge [ce Borge Bie BE oe Bigeye Bi = 0 (4.64 b [! 3, ot | cot Pb pp cot Sn? 0 (4.64) P, can, now, be found through the simultaneous solution of the two transcendental equations (4.63) and (4.64). For the frame under consideration (J; = fh, b = L and d = b/3), such a procedure gives m = 2.36. This means that Bow 1 _ LIISEL “ (2.36L)° vc ‘The results are summarized in Fig. 4.22 (d). 4.5.2 Haarman Method Haarman suggested a semi-graphical approach for obtaining the buckling load of straight members. The method is based on the observation, if one draws the elastic buckled configuration of an axially-loaded, originally-straight bar having any end con- ditions and then establishes a set of rectangular coordinates with the origin at one flex point and one axis directed through the other, the elastic curve will be a simple sine wave in the coordinates. © *For details see McGuire (1968), 108aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.Stability of Structures Figure 4.28, Moment equilibrium at a joint my Ma Mi my, Ma, Mu 0 Ly * La * Ly if El of each member is the same. Continuity at a Joint (Fig. 4.29). The total clockwise rotation of the member ij is 6) - Wy. Hence, for continuity at the joint i, 0 — Wa = Six — Wa = Ox - Wa = Figure 4.29. Continuity at a joint Le (481) unbuckled shape 2-77 dbuckled I shape 1 imaginary member Figure 430. Compatibility of deflection.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.Stability of Structures Wiz = Was = W (say) Also, since pr = 0, far = 1/3 and gu = — 1/6. Therefore, the relations in (a) lead to the following three independent relations Or-gm = 0 — Ou-fm=0 On + m/6 = From (d) we get 82) — 823 — Wi2 = 0. Finally, considering the equilibrium of the frame in its buckled configuration (noting that y is small), we get the following relation Pry-m=0 The preceding five relations are sufficient to obtain the nontrivial solution. The re- quirement turns out to be that the following determinant should vanish, i.e. 100 0 -¢ o100 + 00 1 0 1/] 20 o1-1 -1 0 00 0 po =1 which leads to 1-1 pn - fos Eon and simplifies to cot Vp _ 1 Pir 6 whose lowest root is Vpi2 = 1.35. Hence, the buckling load is, Example 4.3 (Fig. 4.33) Let us consider a buckling mode which is symmetric about the two centre lines of the frame. 120aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.Stability of Structures 585 Ws W668 (b) Ma P length L t Figure 4.34. (a) Sign convention, (b) End-forces and displacements. (c) Member with sway permitted. (kL = kL + sin kL “1 22 cos kL = kL sin kL GEA) Moreover, the axial force P is related to the axial displacement e of the member as (assuming that the axial shortening due to the bending curvature is negligible) p= FA, (485) Eqns. (4.83) and (4.85) can also be expressed in the matrix form as e El AI 0 0 e Mil=) 0 si si] | Oe (4.86) Mp 0 sy se] [Oe 124aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.Stability of Structures Kee R-0 ee (ee + kp) keke ke) 02 04 06 «08 10 F 08 10 ke I (ke Kop) He (e+ kbp) Figure 4.38. Wood's charts for estimating critical load of the column of a substructure. 3500 kN 7000 kN ] iz ie am | qi b 3m. 3m Figure 4.39. Framed structure—example 1. it has the same E/ and L values as the column 1-2 but carries twice the load and hence is the critical element in the structure. Consider the substructure consisting of beams 2-4 and 4-5 and the column 3-4. We have A = 238 (beams), a = 884 (columns) kn = (4 X 238 + 4 x 238), ke = 4 x 884 132aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.Stability of Structures (a es) Gs}, Mave (Meve) [Pcs £ P Tp =05 Figure 5.7. Positive and negative bending moments. (b) Typical stability criteria—interaction curves, values denote sagging and hogging moments respectively [Fig. 5.7(a)]. It can be seen that for sagging moments and p < 0.5, there is a single value of critical moment corresponding to the given value of axial force. However, in the case of p > 0.5 and sagging moments, the nature of the interaction curve is quite different. As long as P < Po (Po is the flexular-torsional buckling load for the column), there is a definite combination of P and M for the beam-column to be stable. When the axial compressive load exceeds Po, the monosymmetric beam-column is unstable unless accompanied by adequate applied moment value within the stable region [i.c. between points C and D, for example, in Fig. 5.7(b)]. The applied moment serves to stabilize the member by decreasing the compressive normal stresses in the smaller flange, which is further away from the shear centre, thereby decreasing the destablizing torque caused by the com- pressive stress components as the section warps during buckling (the Wagner effect) For stability, the applied moment must be large enough for the resulting stresses in the smaller flange to become tensile to provide an effective restoring torque. On the other hand, when the applied moment is to large, the member will become unstable because the larger flange is not able to sustain the increased compressive stresses [above point C in Fig. 5.7(b)]. The same explanation is valid for beam-columns with P < 0.5 and subjected to uniform hogging moment. An interesting feature to note is that the shape of the stability criteria (interaction curve), for combined bending and axial force, shows that the member axial buckling capacity may be increased with the application of an applied moment. 148aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.Figure 6.3. Moments and shear forces on a plate element. Similarly, the moment equilibrium about the y-axis leads to OM: _ Ms 2 a ~ dy 27° Using eqn. (6.3) in eqn. (6.2) gives the single equation of equilibrium, (6.3b) (6.4) To solve eqn. (6.4), it is obviously necessary to have three more relations among the variable M., My, Mey and w (x,y). These are available in the form of moment-curvature relationships. These can be referred to in any standard text on the theory of plates, e.g. Szilard (1974), D (aw you az ay (6.5) 152aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.Stability of Structures equation in the present case is eqn. (6.7) which, with the assumed solution of w, reduces to an ordinary differential equation of the type 22 tat 2,2 me MEM oy (min Ne mn) 6 _ og 6.13 f oe +(e Bea \) (613) where the primes denote differentiation with respect to y. Using eqn. (6, 12), the general solution of eqn. (6.13) is £) = Asinh py + B cosh py + Csinny + D cos ny (6.14) in which p = ™ (4 1)* a = mE yp _ yy n=" K-10) and A, B, C, D are the constants to be determined from four boundary conditions on edges y = 0, b. Case (1) Edge y = 0 simply supported, y = b free The boundary conditons are The boundary conditons at y = 0 require f (y) to be an odd function in y; therefore B = D = 0 in eqn. (6.14). Using the boundary conditions on y = b, we get A(k + | — v) sinh wb - C (k -1 + v) sin nb = 0 A(k + 1)* (kK — 14) cosh wb — C (k - 1)* (kK +1 + v) cos Hb =O For a nontrivial solution, the determinant of the coefficients of A and C should vanish. Consequently, the following characteristic equation is obtained. tanh (mmvk + 1 b/a) _ Vk +1 kelley tanmn Vk -1b/a) Vk-T [k+1-v 158.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.Isotropic Rectangular Plates 2225, Two lorge buckles(antisymmeiric bucking) Figure 6.10, Symmetric and antisymmetric buckle patterns. pattern intersect at a/b = 2.05 and thus, from a/b = 1.0 to 2.05, the symmetric pattern gives the minimum critical load. Similarly, from a/b = 2.05 to 3.50 (where the next two curves of antisymmetric and symmetric patterns meet), the antisymmetric pattern governs the buckling, and so on. In Fig. 6.10 buckling configurations are shown for ab = 1.5 and 2.5 to illustrate the symmetric and antisymmetric types of buckle patterns, respectively. It is worth mentioning that the critical load of a plate is independent of the direction of the shear (positive and negative shear). However, this is true in general for isotropic plates only. The effect of the shear direction could be significant in anisotropic/com- posite plates (see. chapter 7). 6.5.3 Plates with Stiffeners A longitudinally compressed rectangular plate of large aspect ratio, supported on all four edges, will buckle in several half waves according to the degree of restraint along the unloaded sides of the plate. It is obvious that reinforcing the plate by transverse stiffeners will have little effect upon the buckling strength of the plate unless these are spaced very closely. The critical compressive stress of the plate will be increased 165aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.Isotropic Rectangular Plates w(ey) =A sin = sin 631) a a which exactly represents the deflection at the instant of buckling and should, therefore, be a fairly good approximation for deflection at least in the initial postbuckling range. Substituting eqn. (6.31) in the compatibility eqn. (6.29) yields a, ao Ext 2nx at rae cos 2% + cos 2B (632) axt * ay" The complementary solution 6. of eqn. (6.32) corresponds to the inplane stress dis- tribution just prior to buckling, i.e., N, = Ney = 0 and N, is uniform. Therefore, Me > mn 2 = Aly where G,, is the average stress acting on the plate. The particular solution 0, is of the form 2nx 2ny » = Az cos a A; cos re in which the constants Ax, Ay are determined by using eqn. (6.32). Thus, finally 2nx Qny)_ Gxx 2 == + cos - y (6.33 0 2 (6.33) (ex { ( The coefficient A in eqn. (6.33) can be determined using Galerkin’s equation, i.e J J @e.g0» aay =0 (6.34) where D.£. = left-hand side expression in eqn. (6.31) with w (x, y) defined by eqn. (6.31) and 6 (1, y) by egn. (6.33), my) = sin ™ sin 2 a Solving eqn. (6.34) imaa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.Laminated Composite Plates In eqn. (7.2), E1, E2 are Young’s moduli in the 1 and 2 directions, respectively (Fig. 7,1), Gn is the shear modilus in the 1-2 plane, vy is Poisson’s ratio for transverse strain in the j-direction when stressed in the i-direction. The stress-strain for a lamina of an orthotropic material (when the material axes are oriented at an angle @ with reference coordinate axes, as shown in Fig, 7.2) under plane stress are & On Or On] fer OF = ]Qr On Ox| & 73) 6] Qs Gs Oss} [Yo In eqn. (7.3), Qj, the transformed reduced stiffnesses, are expressed as Qn = Qu cos'@ + Qn sin'@ + 2 (ir + 20s) sin? cos*@ Qn = Qn sin'® + Oz cos'® + 2 (Qiz + 206) sin’ ® cos?O iz = (Qn + O22 — 4Qu6) sin? © cos?® + Qiz (cos + sin’ @) Qos = (Qu + O22 — 2Qi2 - 206s) sin? @ cos*@ + Qs (sin'@ + cos*®@) (7.4) O16 = (Qi: ~ Qi2— 2056) sin @ cos* @ — (Qz2 — Oy: — 2Q5) cos @ sin’ ® Gas = (Qi - Qx2 — 2046) cos O sin* 0 + (Qi2— Orr + 20e4) sin @ cos’ @ Eqn. (7.3) can be thought of as stress-strain relations for the kth layer of a multilayered laminate. Thus, it can be written as Figure 7.2, Onhotropic lamina in plane siress (material axes oriented at ante @). 177aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.Stability of Structures Symmetic Cross-ply Laminate (N odd) 1 30n(teq} +3 2s [- ] On 0 A] =h 1 1 1 (al 30 (I-q) +3 Sym (7.18a) By =0 (7.18b) : Qn 0 i 2 e w=4 o | (7.180) Oe Antisymmetric Cros: . Qu + On on 0 0 (7.19a) Ou 0 0 (7.19b) 0aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.9 Thin Circular Cylindrical Shells 9.1 INTRODUCTION The primary difference between a shell and a plate is that the former has a curvature in the unstressed condition whereas the latter is assumed to be initially flat. The presence of initial curvature affects the membrane behaviour of the surface significant- ly. Membrane action in a surface is caused by inplane forces which may be primary forces caused by applied edge deformations or secondary forces resulting from flexural deformations. In plate theory (at least in small-deflection theory), the membrane action due to secondary forces can be neglected, whereas it is quite significant in shell theory, regardless of the magnitude of deformation. In this chapter, the stability of only circular cylindrical shells is ueuted, as these are very common shell configurations. We shall make use of Donnell’s equations which are relatively uncomplicated and have been shown to give satisfactory results when used to deal with buckling problems. In deriving shell equations the following assump- tions are made. (a) The shell thickness is very small compared to its other dimensions. (b) Lateral deflections are small compared to the shell-thickness. (c) The shell material is homogeneous, isotropic and obeys Hooke’s law. (@) Lines normal to the middle surface before bending remain straight and normal during bending. 246aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.aa You have either reached a page that is unavailable for viewing or reached your viewing limit for this book.SUBJECT INDEX A Adjoint System 225 B Beams 135 lateral buckling 135 monosymmetric 140 Beam-columns 74,147 continuous 92 with distributed loads 86 with end couples 87 with lateral loads 78, 81, 85 with single couple 84 with symmetric lateral loads 85 Buckling 1 flexural 7 modes 42, 45, 48, 50, 67 of frames 112, 123 of plates 149, 175 of shells 7, 246 snap-through 7, 30, 32 torsional 7, 255 c Code specifications 143 Columns 37 cantilever 4, 42, 199, 215 continuous 92 eccentrically loaded 72 elastically restrained 51 Euler column 2, 40, 54 fixed-fixed 48, 54 fixed-hinged 46, 55 follower force 200, 233 geometrically imperfect 62 large deflection 70 on elastic foundation 99 rotationally restrained 90 pulsating load 242 Conservative systems 9, 222 D Discrete systems 14, 205 Donnell’s equations 247 E Effective length 50 Effective width 173 Energy approach 9 F Frames 74 buckling 112, 123 partial 56 portal 58 single storey 104 with primary bending 106 G Galerkin’s method 171, 223 a Haarman method 108 Hamilton’s equations for continuous systems 211 for discrete systems 205T Imperfection approach 9 L Lagrange’s equations for continuous systems 209 for discrete systems 205 Laminates 175 angle-ply 181, 192 cross-ply 181, 192 Large deflection analysis 23, 70, 169 Law of minimum potential energy 14, 37, 136, 160 Liapunov functional 213 Liapunov’s method 213 application of 215 M Mises truss 4, 30 N Non-conservative systems 221 P Plates 149, 175 anisotropic 190 dynamic stability 219 effective width 173 energy method 160, 183 follower force 238 isotropic 149 laminated 175 214 large deflection 169 other boundary conditions 157 orthotropic 188 simply supported 153 with shear loads 161 with stiffeners 169 Problems 259 R Ritz method 160 Ss Secant formula 73 Shells 7, 246 buckling 252 circular cylindrical 246 failure 257 Southwell plot 63 Stability 1 dynamic 198 of continuous systems 212 niLof. systems 15 single d.o.f. system 14, 31 stiffness method 123 structural 2 small imperfections 34 Structural instability 3 v Vibration method 12 w Wagner effect 140, 148THE BOOK LSE GCM eC RR CUr Neier myoes me mes author’s earlier book published in 1985. Chapters on Discrete Systems, Lateral Buckling of Beams and on Laminated Compos- ite Plates have been added. There is an additional appendix on Problems. The book is written primarily for students; others will find it useful as a quick reference for basic tools. The topics covered will form a useful postgraduate course for structural een eae CR CoE moles kee Com tC oa graduate students in aerospace, civil and mechanical engineer- ing departments of various engineering institutions. THE AUTHOR Dr, Ashwini Kumar is a Professor in the Department of Civil DRC eet eB cts em CC age cere oor he has been since 1975 after his postdoctoral experience at the University of Waterloo, Canada and the University of Cambridge, England. At IIT, Kanpur, he was Dean, Academic Affairs (1988- ERR ose Seca e rthos CRE ern CC coc Ph.D. and M.Tech. theses and has published extensively in the area of stability, large deformations, plastic analysis and the buckling /postbuckling behaviour of composite laminates. ISBN 81-7023-804-8 ALLIED PUBLISHERS LIMITED