Chapter 11 · Section 6

Stability of Rigid Frames

Under vertical loads, frame buckling is usually of Type II (limit-point type). In engineering practice, the beam's vertical loads are equivalently replaced by column-top concentrated forces, turning the problem (approximately) into a Type-I one — and then the displacement method or finite-element method builds a structural stiffness equation with axial-force corrections, from which $F_{\mathrm{Pcr}}$ is obtained via $|\boldsymbol K + \boldsymbol K_{G}| = 0$.

How to use

Click any item in the left Contents to jump · Click the green "Next" · Each animation has a ▶ button.

11.6.1

Features of frame stability problems

Overview
Question · Type-I or Type-II buckling for a frame?

Under vertical loads, does a frame buckle as Type-I (bifurcation) or Type-II (limit point)? What is the difference?

Answer: typically Type-II (limit point) — a frame deflects laterally immediately upon loading; members are in bending, deflection grows nonlinearly, and the capacity peaks and drops (§11-1.5). In engineering practice, the problem is approximated as Type-I (axial deformation neglected, loads lumped at joints).

A frame under uniformly distributed vertical loads begins to sway immediately, and its members are in bending equilibrium. The axial-force-induced additional moment increases with sway, which grows nonlinearly — at the critical load a maximum point appears (see Animation 11.6.1-a), i.e. Type-II (limit-point) buckling.

Anim. 11.6.1-a
Engineering simplification: Type-II → Type-I

In practice, the vertical load on the beam (see Animation 11.6.1-a) is decomposed into concentrated forces at the joints of the beam ends (see Animation 11.6.1-a). Ignoring the bars' axial deformation, then only the columns carry axial compression — the frame stays straight until buckling, at which point a bifurcation occurs.

This converts the original "Type-II stability" problem to an approximately Type-I problem, now amenable to the displacement/finite-element method.

Stiffness change in axially loaded members

The larger the axial compression $F_{\mathrm P}$, the smaller the member's rotational stiffness and lateral stiffness — when $F_{\mathrm P}$ reaches the critical value, the stiffness drops to zero. Without axial force, a cantilever member in has rotational stiffness $S_{BA} = 3i = 3EI/l$ and lateral stiffness $k = 3i/l^{2} = 3EI/l^{3}$. Once either stiffness vanishes, the rest of the structure can no longer support the member.

11.6.2

Displacement method · Slope-deflection with axial force

Displacement Method

① Equilibrium of an axially loaded member

Consider the member in the figure: rotations $\theta_{A}, \theta_{B}$ at the ends, relative lateral displacement $\Delta$; end moments and shears as in the figure.

$$ EI\, y'' \;=\; -\,(M_{AB} \;+\; F_{\mathrm Q}\, x \;+\; F_{\mathrm P}\, y) \qquad \text{(a)} $$
(11.6-1)

The third term in parentheses captures the effect of axial force on bending moment. Let

$$ u \;=\; \alpha\, l \;=\; l\, \sqrt{\dfrac{F_{\mathrm P}}{EI}}$$
(11-34)

So equation becomes $y'' + (u/l)^{2}\, y = -(M_{AB} + F_{\mathrm Q} x)/EI$; general solution

$$ y \;=\; A\cos\dfrac{u x}{l} \;+\; B\sin\dfrac{u x}{l} \;-\; \dfrac{M_{AB} + F_{\mathrm Q} x}{F_{\mathrm P}} $$
(11.6-2)
Anim. 11.6.2-a

② Fixed-fixed member · Slope-deflection equation

Applying BCs $y=0, y' = \theta_{A}$ at $x = 0$; $y = \Delta, y' = \theta_{B}$ at $x = l$, we obtain

$$ \left\{\begin{aligned} M_{AB} &= 4i\,\theta_{A}\,\xi_{1}(u) \;+\; 2i\,\theta_{B}\,\xi_{2}(u) \;-\; 6i\,\dfrac{\Delta}{l}\,\eta_{1}(u) \\[2pt] M_{BA} &= 2i\,\theta_{A}\,\xi_{2}(u) \;+\; 4i\,\theta_{B}\,\xi_{1}(u) \;-\; 6i\,\dfrac{\Delta}{l}\,\eta_{1}(u) \\[2pt] F_{QAB} &= F_{QBA} \;=\; -\,\dfrac{6i}{l}(\theta_{A}+\theta_{B})\,\eta_{1}(u) \;+\; \dfrac{12i}{l^{2}}\,\Delta\,\eta_{2}(u) \end{aligned}\right.$$
(11-35)

Here $i = EI/l$ is the linear stiffness; $\xi_{1},\xi_{2},\eta_{1},\eta_{2}$ are the axial-force-modified stiffness coefficients (functions of $u$; see Table 11-2).

③ Fixed-pinned case

If end $A$ is pinned, use $M_{AB}=0$ to eliminate $\theta_{A}$:

$$ \left\{\begin{aligned} M_{BA} &= 3i\, \theta_{B}\,\xi_{3}(u) \;-\; 3i\,\dfrac{\Delta}{l}\,\eta_{3}(u) \\[2pt] F_{QBA} &= -\,\dfrac{3i}{l}\,\theta_{B}\,\xi_{3}(u) \;+\; \dfrac{3i}{l^{2}}\,\Delta\,\eta_{3}(u) \end{aligned}\right.$$
(11-36)

Table 11-2 also lists the end moments and shears induced by unit displacements.

④ Displacement-method stability equation · $D = 0$

For a frame with $n$ joint displacements $Z_{1}, \ldots, Z_{n}$, assembly gives the displacement-method canonical system:

$$ r_{11} Z_{1} \;+\; r_{12} Z_{2} \;+\; \cdots \;=\; 0,\qquad \ldots \qquad (\text{e}) $$
(11.6-3)

$Z_{i}=0$ is the trivial solution (frame in its original equilibrium); buckling requires a non-trivial solution, so the determinant must vanish:

$$ D \;=\; \bigl|\, r_{ij}\,\bigr| \;=\; 0$$
(11-37)
the stability equation of the frame in the displacement method; find the smallest positive root to get $F_{\mathrm{Pcr}}$

Example 11-7 · Critical load of Animation 11.6.2-b

The frame has two rotational DOF at joints $B$ and $D$ as basic unknowns $Z_{1}, Z_{2}$. Because both columns have the same size and axial force:

$$ u_{1} \;=\; u_{2} \;=\; u \;=\; l\sqrt{\dfrac{F_{\mathrm P}}{EI}} $$
(11.6-4)

Drawing unit moment diagrams $\overline{M}_{1}, \overline{M}_{2}$ (Animation 11.6.2-b,d):

$$ r_{11} \;=\; 11i \;+\; 4 i\, \xi_{1}(u), \qquad r_{22} \;=\; 8i \;+\; 4 i\, \xi_{1}(u), \qquad r_{12} \;=\; r_{21} \;=\; 4i $$
(11.6-5)
Anim. 11.6.2-b

Stability equation:

$$ D \;=\; \begin{vmatrix} r_{11} & r_{12} \\[2pt] r_{21} & r_{22} \end{vmatrix} \;=\; 0 $$
(11.6-6)

Substituting:

$$ \xi_{1}^{2}(u) \;+\; 4.75\, \xi_{1}(u) \;+\; 4.5 \;=\; 0 $$
(11.6-7)

Roots $\xi_{1}(u) = -1.307, -3.443$; from Table 11-2 the smaller-$u$ root gives $u = 5.46$. Hence

$$ F_{\mathrm{Pcr}} \;=\; u^{2}\,\dfrac{EI}{l^{2}} \;=\; (5.46)^{2}\,\dfrac{EI}{l^{2}} \;\approx\; 29.81\,\dfrac{EI}{l^{2}} $$
(11.6-8)
11.6.3

Finite-element method · Geometric stiffness matrix

FEM · K + K_G

In Ch. 8's matrix displacement method, the element force $\boldsymbol{F}^{e}$ and end displacement $\boldsymbol{\Delta}^{e}$ are related by the element stiffness equation:

$$ \boldsymbol{F}^{e} \;=\; \boldsymbol{k}^{e}\, \boldsymbol{\Delta}^{e} $$
(11.6-9)
$\boldsymbol{k}^{e}$ = element stiffness matrix

For stability analysis of a frame, incorporating the axial-force effect on element stiffness modifies this to:

$$ \boldsymbol{F}^{e} \;=\; \bigl(\boldsymbol{k}^{e} \;+\; \boldsymbol{k}^{e}_{G}\bigr)\, \boldsymbol{\Delta}^{e}$$
(11-38)
$\boldsymbol{k}^{e}_{G}$ = element geometric stiffness matrix (also called element initial-stress matrix) — depends only on the magnitude of the axial force
Anim. 11.6.3-a

Structural stiffness equation

Assembly using the same procedure as in the displacement method gives the global form:

$$ \bigl(\boldsymbol{K} \;+\; \boldsymbol{K}_{G}\bigr)\, \boldsymbol{\Delta} \;=\; \mathbf{0}$$
(11-39)
$\boldsymbol{K}$ = structural stiffness; $\boldsymbol{K}_{G}$ = structural geometric (initial-stress) stiffness

The RHS is zero, as the "free" terms in the displacement-method canonical equations all vanish.

Stability equation (FEM form)

At the critical state, Eq. (11-39) has a non-trivial $\boldsymbol{\Delta}$, which requires

$$ \bigl|\, \boldsymbol{K} \;+\; \boldsymbol{K}_{G}\,\bigr| \;=\; 0$$
(11-40)
The stability equation in the FEM

Because $\boldsymbol{K}_{G}$ scales linearly with the axial forces (which scale with $F_{\mathrm P}$), Eq. (11-40) is a generalized eigenvalue problem — the smallest eigenvalue is the critical-load factor. This approach is easily programmed for arbitrarily complex frames and is the standard modern stability-analysis procedure.

Note on second-order analysis

When internal forces in a frame are computed with very large axial forces (comparable to the critical load), the axial-force effect on stiffness cannot be ignored. Structural analysis that includes this effect (i.e. $P\Delta$) is called second-order analysis. For super-high-rise buildings or tall structures, second-order effects are often decisive.

Section summary
  1. Frames usually exhibit Type-II buckling; "concentrated joint loads" approximate the problem as Type-I;
  2. Displacement method: derive the slope-deflection equations with axial force (Eqs. 11-35/36), whose correction factors $\xi, \eta$ are functions of $u = l\sqrt{F_{\mathrm P}/EI}$;
  3. Displacement-method stability equation $D = |r_{ij}| = 0$ — find the smallest positive root for $F_{\mathrm{Pcr}}$;
  4. FEM: wrap the axial-force effect into the geometric stiffness matrix $\boldsymbol{K}_{G}$; the stability equation $|\boldsymbol{K}+\boldsymbol{K}_{G}|=0$ is a generalized eigenvalue problem;
  5. End of chapter — Chapter 12 turns to plastic analysis and ultimate load.