Lecture 9: Stability of Elastic Structures

Structural Mechanics 2.080 Lecture 9 Semester Yr Lecture 9: Stability of Elastic Structures In lecture 8 we have formulated the condition of static equilibrium of bodies and structures by studying a small change (variation) of the total potential energy. The system was said to be in equilibrium if the first variation of the total potential energy vanishes. The analysis did not say anything about the stability of equilibrium. The present lecture will give an answer to that question by looking more carefully what is happening in the vicinity of the equilibrium state. To illustrate the concept, consider a rigid body (a ball) siting in an axisymmetric paraboloid. shown in Fig. (9.1). Π Π C>0 C=0 Π C<0 u u u δu Figure 9.1: Illustration of stable, neutral and unstable equilibrium. In the case of a rigid body the total potential energy is just the potential energy Π = mgh = Cu2 (9.1) where u is the horizontal displacement of the ball from the resting position. Let’s calculate the first and second variation of the function Π(u) ∂Π δu = 2Cuδu ∂u δ 2 Π = δ(δΠ) = 2Cδuδu δΠ = (9.2) (9.3) At the origin of the coordinate system u = 0, so the first variation of Π is zero no matter what the sign of the coefficient C is. In the expression for the second variation, the product δuδu = (δu)2 is always non-negative. Therefore, the sign of the second variation depends on the sign of the coefficient C. From Fig. 9.1 we infer that C > 0 corresponds to a stable configuration. The ball displaced by a small amount δu will return to the original position. By contrast, for C < 0, the ball, when displaced by a tiny amount δu, will roll down and disappear. We call this an unstable behavior. The case C = 0 corresponds to the neutral equilibrium. One can formalize the above consideration to the elastic body (structure), where the total potential energy is a function of a scalar parameter, such as a displacement amplitude u. The function Π(u) can be expanded in Taylor series around the reference point uo Π(u) = Π(uo ) + dΠ du (u − uo ) + u=uo 9-1 1 d2 Π 2 du2 (u − uo )2 + · · · u=uo (9.4) Structural Mechanics 2.080 Lecture 9 Semester Yr The incremental change of the potential energy ∆Π = Π(u) − Π(uo ) upon small variation of the argument δu = u − uo is ∆Π = dΠ 1 d2 u (δu)2 = δΠ + δ 2 Π + · · · δu + du 2 du2 (9.5) For the system in equilibrium the first variation must be zero. Therefore, to the second term expansion, the sign of the increment of Π depends on the sign of the second variation of the potential energy. We can now distinguish three cases    > 0, Positive (stable equilibrium) 2 δ Π = 0, Zero (neutral equilibrium)   < 0, Negative (unstable equilibrium) 9.1 Trefftz Condition for Stability In 1933 the German scientist Erich Trefftz proposed the energy criterion for the determination of the stability of elastic structures. We shall explain this criterion on a simple example of a one-degree-of-freedom structure. Consider a rigid column free at one end and hinged at the other. There is a torsional spring mounted at the hinge. Upon rotation by an angle θ, a bending moment develops at the hinge, resisting the motion M = Kθ (9.6) where K is the constant of the rotational spring. The column is initially in the vertical position and is loaded by a compressive load P , Fig. (9.2). In the deformed configuration, the force P exerts a work on the displacement u 2 θ u = l(1 − cos θ) ∼ =l 2 (9.7) The total potential energy of the system is 1 1 1 Π = M θ − P u = Kθ2 − P lθ2 2 2 2 (9.8) Upon load application the column is of course rigid and remains straight up to the critical point P = Pc . The path θ = 0 is called the primary equilibrium path. If the column were elastic rather than rigid, there would be only axial compression along that path. This stage is often referred to as a pre-buckling configuration. At the critical load Pc the structure has two choices. It can continue resisting the force P > Pc and remain straight. Or it can bifurcate to the neighboring configuration and continue to rotate at a constant force. The bifurcation point is the buckling point. The structure is said to buckle from the purely compressive stage to the stage of a combined compression and bending. The above analysis have shown that consideration of the equilibrium with nonlinear geometrical terms, Eq. (9.7) predicts two distinct equilibrium paths and a bifurcation 9-2 Structural Mechanics 2.080 Lecture 9 Semester Yr (buckling) point. Let’s now explore a bit further the notion of stability and calculate the second variation of the total potential energy δ 2 Π = (K − P l)δθδθ (9.9) The plot of the normalized second variation δ 2 Π/δθδθ is shown in Fig. (9.2). Figure 9.2: Stable and unstable range in column buckling. It is seen that in the range 0 < P < Pc , the second variation of the total potential energy is positive. In the range P > Pc , that function is negative. A transition from the stable to unstable behavior occurs at δ 2 Π = 0. Therefore, vanishing of the second variation of the total potential energy identifies the point of structural instability or buckling. Physically, the test for stability looks like this. We bring the compressive force to the value P ∗ , still below the critical load. We then apply a small rotation ±δθ in either direction of the buckling plane. The product δθδθ is always non-negative. Figure 9.3: A discrete Euler column in the undeformed and deformed configuration. For equilibrium the first variation of the total potential energy should banish, δΠ = 0, which gives (K − P l)θδθ = 0 (9.10) 9-3 Structural Mechanics 2.080 Lecture 9 Semester Yr There are two solutions of the above equation, which corresponds to two distinct equilibrium paths: • θ = 0 – primary equilibrium path • P = Pc = K – secondary equilibrium path l P θ 0 Figure 9.4: Two equilibrium paths intersects at the bifurcation point. And so is the second variation of the total potential energy (length AB in Fig. (9.2)). When the lateral load needed to displace the column by δθ is released, the spring system will return to the undeformed, straight position. We repeat the same test under the compressive force p∗∗ > Pc . The application of the infinitesimal rotation δθ will make the function δ 2 /δθδθ negative. This is a range of unstable behavior. Upon releasing of the transverse force, the column will not returned to the vertical position, but it will stay in the deformed configuration. It should be pointed up that the foregoing analysis pertains to the problem of stability of the primary equilibrium path. The secondary equilibrium path is stable, as will be shown below. To expression for total potential, it can be constructed with the exact equation for the displacement u rather than the first two-term expansion, Eq. (9.7) 1 Π = Kθ2 − lP (1 − cos θ) 2 (9.11) The secondary equilibrium path obtained from δΠ = 0 is P θ = Pc sin θ (9.12) The plot of the above function is shown is Fig. (9.5). For small values of the column rotation, the force P is almost constant, as predicted by the two-term expansion of the cosine function. For larger rotations, the column resistance increases with the angle θ. Such a behavior is inherently stable. The force is monotonically increasing and reach infinity at θ → π. 9-4 Structural Mechanics 2.080 Lecture 9 P Pc Semester Yr P Pc 1.0 0 θ ⇡ 2 0 ⇡ 2 π θ Figure 9.5: Plot of the secondary equilibrium path. 9.2 Stability of Elastic Column Using the Energy Method The Trefftz condition for stability can now be extended to the elastic column subjected to combined bending and compression. The elastic strain energy stored in the column is a sum of the bending and axial force contribution  Z l 1 1 U= M κ + N o dx (9.13) 2 2 0 It is assumed that the column is fixed at one end against axial motion and allow to move in the direction of the axial force. N̄ u w Figure 9.6: Initial and deformed elastic column. To maintain generality, no static or kinematic boundary conditions are introduced in the present derivation. The work of external forces is W = N̄ uo The first variation of the total potential energy is Z 1 l δΠ = δ(U − W ) = (δM κ + M δκ + δN  + N δ)dx − N δuo 2 0 9-5 (9.14) (9.15) Structural Mechanics 2.080 Lecture 9 Semester Yr For linear elastic material M = EIκ, δM = EIδκ (9.16a) N = EA, δN = EAδ (9.16b) By eliminating the bending and axial rigidity between the above equation, the following identities hold M δκ = δM κ (9.17a) N δ = δN  (9.17b) Therefore, the expression (9.15) is reduced to Z l δΠ = (M δκ + N δ) dx − N̄ δuo (9.18) 0 Now, recall the strain-displacement relations in the theory of moderately large deflection of beams κ = −w00 , δκ = −(δw)00 1  = u0 + (w0 )2 , δ = (δu)0 − w0 δw0 2 Substituting the increments δκ and δ into Eq. (9.18) yields Z l Z l Z l δΠ = − M δw00 dx + N w0 δw0 dx + N δu0 dx − N δuo 0 0 (9.19a) (9.19b) (9.20) 0 In the above expression the axial load is unknown but constant over the length of the column. Therefore the load N could be brought outside the integrals. Consider now the last two terms in Eq. (9.20) Z l x=l N δu0 dx − N δuo = N δu − N δu = N δuo − N δuo = 0 (9.21) x=0 0 With the above simplification we calculate now the second variation of the total potential energy Z l Z l 2 00 δ Π = δ(δΠ) = − δM δw dx + N δw0 δw0 dx (9.22) 0 0 According to the Trefftz stability criterion δ 2 Π = 0, Z l Z l 00 00 − EIδw δw dx + N δw0 δw0 dx = 0 0 (9.23) 0 The buckling load N = Nc is then Z l − δw00 δw00 dx N = EI Z 0l δw0 δw0 dx 0 9-6 (9.24) Structural Mechanics 2.080 Lecture 9 Semester Yr Let’s express the out-of-plane deflection of the column as a product of the amplitude A and the normalized shape function φ(x). The shape function should satisfy kinematic boundary condition of a problem w(x) = Aφ(x) (9.25) We can calculate now the first and second derivatives of the function w(x) and their variations w0 = Aφ0 , δw0 = δAφ0 (9.26a) w00 = Aφ00 , δw00 = δAφ00 (9.26b) Substituting the above expression into Eq. (9.24), we get Z l 00 00 0 0 Z l δAφ δAφ dx Nc = EI Z0 l = EI Z0 l δAφ δAφ dx 0 φ00 φ00 dx (9.27) 0 0 φ φ dx 0 where Nc = −N is he compressive buckling load. The above equation for the critical buckling load of a column is called the Raleigh-Ritz quotient. The Trefftz criterion does not provide the shape function but for a given shape calculates the approximate value of the buckling load. This is always an upper bound. Should the shape function coincide with the exact buckling shape, the Raleigh-Ritz quotient will give the absolute minimum value. As an illustration, consider the pin-pin supported column and assume the following buckling shape φ(x) = x(l − x) (9.28) which satisfies identically kinematic boundary conditions φ(x = 0) = φ(x = l) = 0. The first and second derivatives of the shape function are φ0 (x) = 2x − l (9.29a) φ00 (x) = 2 (9.29b) After straightforward integration, the calculated buckling load is Nc = 12EI l2 (9.30) Can the accuracy of the above solution be improved? Let’s try and assume as a shape function the solution for the pin-pin beam under the uniform line load φ(x) = x4 − 2lx3 + l3 x 9-7 (9.31) Structural Mechanics 2.080 Lecture 9 Semester Yr The above function satisfies the simple support boundary condition at both ends. The slope and the curvature of the deflected shape are φ0 (x) = 4x3 − 6lx2 + l3 00 2 φ (x) = 12x − 12lx (9.32a) (9.32b) Because the curvature at both ends vanish, so does the bending moment. Also the slope at mid-span is zero. This means that the static (zero bending moments) boundary conditions are also satisfied. The previously considered shape function, Eq. (9.28) led to the constraint curvature, meaning that the static boundary conditions were violated. After 1680 = 9.88. There was over straightforward calculation, the numerical coefficient become 170 20% improvements in the accuracy of the solution Nc = 9.88 EI l2 (9.33) Can the solution be further improved (lowered)? yes, but not by much. Assume a sinusoidal shape of the buckling shape πx φ = sin l π πx 0 φ = cos l l  π 2 πx 00 φ =− sin l l (9.34a) (9.34b) (9.34c) The expression for the critical buckling load becomes Z l πx π 4 sin2 dx EI( ) l l Z0 Nc = π l πx ( )2 cos2 dx l l (9.35) 0 Both integrals are identical and the solution becomes Nc = π 2 EI l2 (9.36) Because π 2 = 9.86, the sinusoidal solution is slightly lower than the previous polynomial solution. This is the lowest possible coefficient meaning that it must be an exact solution to the buckling problem. To prove it, it is sufficient to check if the local equilibrium equation is satisfied EIwIV + N w00 = 0 (9.37) Indeed, substituting Eqs. (9.34) and (9.36) into the equilibrium equation brings the left hand side of this equation identically to zero. 9-8 Structural Mechanics 9.3 2.080 Lecture 9 Semester Yr Effect of Structural Imperfections Consider the same discrete strutter as in Section 9.1. This time the rigid rod is not straight but is rotated by the angle θo before the vertical load is applied. Upon load application the column is subjected to additional rotation θ, measured from the theoretical vertical direction, Fig. (9.7). P θo Mint l θ K K 0 θo θ Figure 9.7: The initial inclination angle θo is a measure of structural imperfection. The problem will be solved by means of local equilibrium. The external bending moment at the base is Mext = P l sin θ, for θ > θo (9.38) where l sin θ is the arm of the force P . In the case of small angle approximation Mext = P lθ. The internal resisting bending moment is Mint = K(θ − θo ) (9.39) Equating the external and internal bending moments P lθ = K(θ − θo ) (9.40) For a geometrically perfect column θo = 0 and from Eq. (9.40) P = Pc = K l (9.41) Equation (9.40) can be re-written in terms of the normalized compressive force P/Pc P θ = θ − θo Pc Solving this equation for θ yields 1 θ = θo 1− 9-9 P Pc (9.42) (9.43) Structural Mechanics 2.080 Lecture 9 Semester Yr P ) is called the Pc magnification factor. It predicts how much the initial imperfections are magnified for a given magnitude of load. When structural imperfections are present, there are no primary and secondary equilibrium paths. There is only one smooth load-deflection curve called the equilibrium path. The plot of the above function is shown in Fig. (9.8). The term 1/(1 − P Pc 1 1.00 0.98 0.96 θo 0.94 0.92 0.90 0 20 θo 40 60 80 100 θ Figure 9.8: A family of equilibrium paths for different values of imperfections. It is interesting to note that with smaller and smaller initial imperfections, the equilibrium paths are approaching the bifurcation point but never reach it. This type of behavior is common to all imperfect structures. As another example of an imperfect structure consider a pin-pin elastic column. The following notation is introduced: • w̄(x) – shape of initial imperfection • w̄o – amplitude of initial imperfection • w(x) – actual buckled shape measured from the vertical (perfect) position • wo – central amplitude of the actual deflection The internal bending moment is Mint = EI∆κ = −EI(w00 − w̄00 ) (9.44) where ∆κ is the change of curvature from the initial curved (imperfect) column. For a simply supported column, the end (reaction) moments are zero so the external bending moment is Mext = P w (9.45) 9-10 Structural Mechanics 2.080 Lecture 9 Semester Yr P w̄(x) M w̄o wo w(x) w(x) Figure 9.9: A continuous imperfect column and the free body diagram. Equating the internal and external bending moments one gets EIw00 + P w = EI w̄(x) (9.46) This is a second order, linear inhomogeneous differential equation, where the right hand side is a known shape of initial imperfection. The solution to this equation exists in terms of quadratures, but the integrals are difficult to evaluate for complex shapes of imperfections. Let’s consider the simplest case of a sinusoidal shape of imperfections. It can be shown that the solution w(x) is also of the sinusoidal shape w(x) = wo sin λx (9.47a) w̄(x) = w̄o sin λx (9.47b) The kinematic boundary conditions are w(0) = w(l) = 0 which implies that sin λl = 0 → λl = nπ (9.48) Substituting Eq. (9.47) into the governing equation (9.46) −EIλ2 (wo − w̄o ) sin λx − P wo sin λx = 0 (9.49) P wo = EI(wo − w̄o )λ2 (9.50) which is satisfied if For a perfect column w̄o = 0, and Eq. (9.50) yields (Pc − EIλ2 )wo = 0 or Pc = EIλ2 = 9-11 n2 π 2 EI l2 (9.51) Structural Mechanics 2.080 Lecture 9 Semester Yr For an imperfect column P wo = Pc (wo − w̄o ) or solving for wo 1 wo = w̄o 1− P Pc (9.52) (9.53) The form of the magnification factor is identical to the one derived for the district column. The only difference is that a continuous column has infinity buckling mode where n = 1 corresponds to the lowest buckling load. The buckling load corresponding to the second buckling mode is four times larger and so on. 9.4 Stability in Tension For some materials instability in tension manifest itself by a development of a local neck, Fig. (9.10). Solid bar Thin-walled tube Figure 9.10: Necking in a solid section bar and thin-walled tube under tension. Consider a round bar of the initial cross-sectional area Ao subjected to a tensile force P . The bar becomes longer and because the Poisson effect its cross-section shrinks to a current value A. The present analysis is valid for materials that are incompressible, that is do not change volume but only shape. Certain polymers, rubber and metals (in the plastic range) are incompressible. The volume of an infinitesimal length l is V = lA 9-12 (9.54) Structural Mechanics 2.080 Lecture 9 Semester Yr The increment of volume for the incompressible material must be equal to zero δV = δ(lA) = δlA + lδA = 0 (9.55) Take the logarithmic definition of the axial strain  = ln l δl ; δ = lo l (9.56) δl δA =− l A (9.57) From the above two equations δ = Integrating both sides  = − ln A + C At A = Ao ,  = 0 so C = ln Ao . Therefore, the expression for the axial strain becomes  = ln l Ao = ln A lo (9.58) We conclude that axial strain can be determined by either measuring the change in length or the change in cross-sectional area. The true (Cauchy) stress is defined as the load divided by the current cross-section A P σ= (9.59) A Let’s construct the total potential energy and its first variation Z δΠ = σδ dv − P δu (9.60) V Before instability occurs, the deformation and stress (uniaxial tension) is uniform across the section of the bar of the length l u = l = l ln δu = −l Ao A (9.61a) δA A (9.61b) Thus, from Eqs. (9.60) and (9.61) Z δΠ = σδ dv + P l V δA A The second variation of the total potential energy is Z δAδA 2 δ Π= δσδ dv − P l A2 V (9.62) (9.63) Applying the Trefftz stability condition δ 2 Π = 0 one gets lAδσδ = P lδδ 9-13 (9.64) Structural Mechanics 2.080 Lecture 9 Semester Yr P δ = σδ A (9.65) or δσ = and finally δσ =σ (9.66) δ The incompressible bar is losing stability in tension when the local tangent to the stressstrain curve becomes equal to the value of stress at that point. A graphical interpretation is shown in Fig. (9.11). σ σ d d✏ ε Figure 9.11: The construction of Considere’ who was the first to derive Eq. (9.66). At what strain an instability develops for an elastic material? In uni-axial stress σ = E dσ =E d (9.67a) (9.67b) Equation (9.66) is satisfied if  = 1. For metals such strain is not attainable in the elastic σy ∼ −3 range because yield will be reached at the strain y = = 10 . However, for rubber and E similar polymeric materials the Young’s modulus is four orders of magnitude smaller, so necking is of common occurrence. The derivation of the instability condition (9.66) was done without any assumption on the stress-strain relation of the material. Therefore this condition is valid for an elastic as well as plastic material. This brings us to the next topic which is plastic buckling of columns. 9.5 Plastic Buckling of Columns Let’s consider the pin-pin column for which the critical buckling load is Pc = π 2 EI l2 (9.68) The corresponding critical buckling stress σc is σc = Pc π2E I = 2 A l A 9-14 (9.69) Structural Mechanics 2.080 Lecture 9 Semester Yr where A is the cross-sectional area. Note that the stress is calculated over the pre-buckling, primary equilibrium path, for which there is no bending. Denoting by ρ the radius of gyration, Aρ2 ≡ I, Eq. (9.69) can be re-written in terms of the slenderness ratio β = l/ρ σc = π 2 E 1 β2 (9.70) The buckling stress is small for long, slender column and is rapidly increasing for short columns. At some critical column length, the yield stress of the material σy will be reached, Fig. (9.12). σC D σy 0 B C βC β A Figure 9.12: A hyperbolic dependence of the buckling stress on the slenderness ratio. The critical slenderness ratio at which the buckling stress reaches the yield stress of the material is obtained from Eq. (9.70) by setting σc = σy s E lc (9.71) βc = = π ρ σy To give you the feel of the critical slenderness, consider a mild steel column with E = 210 GPa, σy = 250 MPa and square cross-section A = h2 , for which the radius of gyration is ρ2 = h2 /12 s   E 1 l =π · √ ≈ 30 (9.72) h c σy 2 3 Columns more slender than the critical will buckle elastically before yielding (path AB). Shorter column or stocky column will yield before buckling. What will happen with such columns? They will deform plastically in axial compression and eventually buckle in the plastic range. Gerrard (1948) extended the predictive capability of Eq. (9.70) into the plastic range dσ by replacing the elastic modulus by the tangent modulus Et = d (σc )plastic = π 2 Et 1 , for β < βc β2 9-15 (9.73) Structural Mechanics 2.080 Lecture 9 Semester Yr For example, for a plastic material obeying the power hardening rule, σ = B · n dσ = nBn−1 d (9.74a) (9.74b) Substituting Eq. (9.74) into Eq. (9.73), the critical buckling strain c is c = π2n β2 (9.75) Using the hardening rule, the buckling stress is  2 n π n σc = B β2 (9.76) In the above equation B is the amplitude of the hardening law and n is the hardening exponent. For most structural steels n ≈ 0.1-0.2. σ Squash load σult Plastic buckling, Eq. (9.75) σy Elastic buckling, Eq. (9.69) 0 β βC Figure 9.13: Range of elastic and plastic buckling. Very short columns are beyond the scope of the elementary theory of thin and slender beams. They will never buckle but flatten as a pancake. 9-16 Structural Mechanics 2.080 Lecture 9 Semester Yr ADVANCED TOPIC 9.6 Mode Transition Moment Equilibrium Equation For a pin-pin supported column, the shape of the imperfection w̄(x) and the deformation w(x) satisfy the moment equilibrium equation EIw00 + P w = EI w̄00 (9.77) The solutions should satisfy the boundary conditions w(0) = 0 (9.78a) 00 w (0) = 0 (9.78b) w(l) = 0 (9.78c) 00 w (l) = 0 (9.78d) Of course, the solutions should also satisfy the continuous conditions: w(x) and w0 (x) are continuous along the entire length of the column, namely, no step or kink occurs in the solutions. We can expand the imperfection w̄(x) in Fourier series as w̄(x) = ∞ X An sin n=1 nπx l (9.79) The coefficients An can be determined by Fourier transformation of w̄: Z nπx 2 l w̄(x) sin dx An = l 0 l (9.80) The deformation w(x) under a load P can be written as a summation of a complete set of Fourier series ∞ X nπx w(x) = Bn sin (9.81) l n=1 where Bn can be determined by Eq. (9.77). Eq. (9.77) now becomes − ∞ X n=1 Bn  nπ 2 l sin ∞ ∞  nπ 2 X nπx P X nπx nπx + Bn sin =− An sin l EI l l l n=1 (9.82) n=1 To make the equation hold, the coefficients should satisfy −Bn n2 π 2 P n2 π 2 + Bn = −An 2 2 l EI l 9-17 (9.83) Structural Mechanics 2.080 Lecture 9 Solve for Bn , we get B n = An 1 1 − P̄ /n2 π 2 EI . So, the deformation w(x) is l2 ∞ X nπx 1 sin w(x) = An 2 l 1 − P̄ /n n=1 Semester Yr (9.84) here, we defined P̄ = P/Pc and Pc = (9.85) The solution tells us what is the shape of the deformation, but it does not tell us any thing about the stability of the equilibrium shape. If we want to study the stability, we have to use potential energy method. Potential Energy Method Under a load P , the total potential energy of the column system is (due to Eq. (10.26) in Lecture 10): Z Z P l 02 EI l 00 00 2 (w − w̄ ) dx − (w − w̄02 ) dx (9.86) Π= 2 0 2 0 Substitute Eqs. (9.79) and (9.81) into it, we have Z Z EI l 00 P l 02 00 2 Π= (w − w̄ ) dx − (w − w̄02 ) dx 2 0 2 0 #2 Z " X ∞ ∞  nπ 2  nπ 2 nπx X nπx EI l − Bn sin + An sin dx = 2 0 l l l l n=1 n=1  #2 " ∞ #2  Z l "X ∞     X P nπ nπx nπ nπx  − Bn cos − An cos dx (9.87)  2 0  l l l l n=1 n=1 ∞ π2P X 2 2 + An n 4l n=1 n=1 n=1 (∞ )  π 2 Pc X  = (Bn − An )2 n4 − P̄ (b2n − A2n )n2 4l π 4 EI = 4l3 ∞ X π2P (Bn − An ) n − 4l 2 4 ∞ X Bn2 n2 n=1 The orthogonality of Fourier series is used to simplify the integration. In order to obtain the equilibrium solution, we need the first derivative of potential energy ∂Π 1 = 0 → B n = An (9.88) ∂Bn 1 − P̄ /n2 which is exactly the same as the solution given by solving the equilibrium equation. To see the stability of the solution, we need the second derivative of potential energy ∂2Π >0 ∂Bn2 → P̄ < n2 We can see the following points directly from Eqs. (9.88) and (9.89): 9-18 (9.89) Structural Mechanics 2.080 Lecture 9 • The critical buckling load for the nth mode is Pc = Semester Yr n2 π 2 EI . l2 • The modes that satisfy n2 > P̄ are in stable equilibrium. • For the modes that satisfy n2 < P̄ , we can still solve for a value of Bn , but those modes are unstable and will snap into either plus or minus infinity. Examples 1 The imperfection w̄(x) consists of only the first two modes, namely w̄(x) = A1 sin πx 2πx + sin l l (9.90) If A1 = 0, the zero point is at the center of the column. If A1 6= 0, the zero point is displaced by a distance u. u is given by u= sin−1 π A1 2 l, 0 ≤ A1 ≤ 2 (9.91) The deformation amplitudes vs. load curves are plotted in Fig. (9.14) for the case A1 = 0.5, Fig. (9.15) for A1 = 1, and Fig. (9.16) for A1 = 1.5. 8 6 4 2 1 2 3 4 -2 -4 -6 Figure 9.14: The deformation amplitudes B1 , B2 vs. load P̄ curves, for the case A1 = 0.5. 9-19 Structural Mechanics 2.080 Lecture 9 Semester Yr 10 5 1 2 3 4 -5 Figure 9.15: The deformation amplitudes B1 , B2 vs. load P̄ curves, for the case A1 = 1. 10 5 1 2 3 4 -5 Figure 9.16: The deformation amplitudes B1 , B2 vs. load P̄ curves, for the case A1 = 1.5. Examples 2 The imperfection is in such a shape that the zero point is displaced while both sections are self-symmetric. Such a shape can be described as  πx  0 < x < ηl  sin ηl (9.92) w̄(x) = 1−η π(x − ηl)   − sin ηl < x < l η (1 − η)l where 0.5 ≤ η ≤ 1. When η = 0.7, w̄(x) can be expanded in Fourier series as: w̄(x) = 0.634 sin πx 2πx 3πx 4πx + 0.563 sin − 0.174 sin + 0.071 sin + ··· l l l l (9.93) As expected, in this case, the first two modes dominate, but there are still higher modes in the expansion. 9-20 Structural Mechanics 2.080 Lecture 9 Semester Yr We plot the mode amplitudes vs. load in Fig. (9.17). If the load P̄ > 4, for example, P̄ = 7.5, the amplitude of mode III becomes largest. So, the solved deformation shape looks more like mode III, although the initial imperfection seems having nothing to do with mode III. Nevertheless, this shape is very unstable; since P̄ > 22 , both mode I and mode II are in unstable equilibrium. Under such a load, mode I and mode II will amplify exponentially. Figure 9.17: The deformation amplitudes of the first three modes vs. load P̄ curves, for the imperfection shape described by Eqn. 9.92. END OF ADVANCED TOPIC 9-21