Column buckling

Push down on a thin plastic ruler standing on end and, well before you get anywhere near crushing it, it suddenly bows sideways and gives way. That is buckling — a slender member under axial compression failing by sideways instability rather than by material crushing, at a load that can be far below its compressive strength. Column buckling theory, due to Euler, predicts exactly the load at which this sideways collapse begins, and shows how strongly it depends on the column's length, stiffness and end supports — not on how strong the material is. This is the closing topic of the ME10001 unit on stress transformation, combined loading and columns, and it is a genuinely different kind of failure from anything covered earlier in the unit: elastic instability, not stress at a point. Here we build the Euler critical load formula from the ground up, see how the four classical end conditions change it, and work through the numbers for a real steel column.

The core insight

A slender elastic column buckles sideways at the Euler critical load $P_{cr} = \pi^2 EI/L_e^2$ , which depends only on stiffness $EI$ and effective length $L_e$ — not on material strength — and the prediction is valid only while the resulting stress stays below the yield stress.

The core idea

Buckling is a stability failure, not a strength failure. A column loaded in pure axial compression can, in principle, be crushed by exceeding the material's compressive strength. But a long, slender column almost never gets that far: a tiny sideways disturbance grows under the axial load until the column bows out and collapses laterally, at a load set by geometry and stiffness rather than by how strong the material is.

The governing equation comes from bending, not compression. Model the column as initially straight, pin-ended, of length $L$ and bending stiffness $EI$ , carrying axial load $P$ . Any small sideways deflection $y$ creates a bending moment $M = -Py$ that the beam-bending equation $EI\,y'' = M$ must resist. Solving this differential equation shows that below a certain load the column simply straightens back out, but at a critical load a non-trivial bowed shape becomes possible — the column buckles. That critical load is the Euler buckling load:

P_{cr} = \frac{\pi^2 EI}{L_e^2}

where: $E$ is Young's modulus of the material, $I$ is the minimum area moment of inertia of the cross-section (about the axis the column bends about most easily), and $L_e$ is the effective length — the length of an equivalent pinned–pinned column that would buckle at the same load.

End conditions change $L_e$ , not the material. The same physical column buckles at very different loads depending on how its ends are held, because restraining rotation at an end effectively shortens the "free-to-bow" length:

End condition	Effective length $L_e$
Pinned–pinned	$L_e = L$
Fixed–free (one end built-in, one free)	$L_e = 2L$
Fixed–pinned	$L_e \approx 0.7L$
Fixed–fixed	$L_e = 0.5L$

Since $P_{cr} \propto 1/L_e^2$ , halving the effective length quadruples the critical load — end fixity is a far more powerful lever on buckling resistance than most students expect.

Figure

Four columns of equal length with different end supports and their buckled mode shapes, pinned pinned bowing in one arc, fixed free bowing as a quarter wave, fixed pinned and fixed fixed showing shorter effective lengths

Dividing by area gives a stress, and yield sets a validity limit. Dividing $P_{cr}$ by cross-sectional area $A$ , and writing $I = Ak^2$ with $k=\sqrt{I/A}$ the radius of gyration, gives the critical stress:

\sigma_{cr} = \frac{P_{cr}}{A} = \frac{\pi^2 E}{(L_e/k)^2}

where: $L_e/k$ is the slenderness ratio, the single number that governs whether Euler's formula even applies. The derivation assumes the material stays perfectly elastic right up to the moment of buckling, so the formula is trustworthy only when $\sigma_{cr} \le \sigma_{yield}$ . For a stubby, low-slenderness column, the formula would predict a $\sigma_{cr}$ above yield — meaning in reality the column crushes or yields before it ever gets the chance to buckle elastically, and Euler's formula simply does not apply there.

Derivation AElastic buckling load from beam bending

The goal of this derivation is to show precisely how the Euler critical load $P_{cr} = \pi^2EI/L_e^2$ falls out of ordinary beam bending theory once a column is allowed a tiny sideways deflection under axial load, and why the answer takes the particular form it does rather than some other combination of $E$ , $I$ and $L$ .

Step 1. Take a pin-ended column of length $L$ , bending stiffness $EI$ , carrying axial compressive load $P$ , and imagine it deflected sideways by a small amount $y(x)$ at position $x$ along its length. At that cross-section the axial load $P$ , now offset sideways by $y$ , produces a bending moment $M = -Py$ about the section — the compressive force is trying to bend the column further in the direction it has already started to bow.

Step 2. The ordinary beam-bending relation connects curvature to moment: $EI\,y'' = M = -Py$ , which rearranges to the differential equation $EI\,y'' + Py = 0$ , or $y'' + (P/EI)\,y = 0$ . This is the same mathematical form as simple harmonic motion, with $P/EI$ playing the role of the squared angular frequency, so its general solution is $y = A\sin(\sqrt{P/EI}\,x) + B\cos(\sqrt{P/EI}\,x)$ .

Step 3. Apply the pinned-end boundary conditions, zero deflection at both ends: $y(0)=0$ forces $B=0$ , leaving $y = A\sin(\sqrt{P/EI}\,x)$ ; then $y(L)=0$ requires $A\sin(\sqrt{P/EI}\,L)=0$ . For a genuinely bowed (non-trivial, $A\ne0$ ) shape to exist, the sine itself must vanish, so $\sqrt{P/EI}\,L = n\pi$ for some integer $n=1,2,3,\dots$ .

Step 4. The smallest, physically relevant load is the one at which the first ( $n=1$ ) buckled shape first becomes possible, since the column buckles at the lowest load for which any non-trivial bowed equilibrium exists. Setting $n=1$ and squaring, $P/EI = \pi^2/L^2$ , so $P_{cr} = \pi^2EI/L^2$ . Replacing the pinned–pinned length $L$ with the general effective length $L_e$ extends the same result to any end condition, giving the Euler formula $P_{cr} = \pi^2EI/L_e^2$ used throughout this article.

Worked example

A pin-ended steel column, length $L = 3\text{ m}$ , has a circular solid cross-section of diameter $d = 50\text{ mm}$ , with $E = 200\text{ GPa}$ and $\sigma_{yield}=250\text{ MPa}$ . Find the Euler critical load, check validity, and compare against the same column with both ends built in (fixed).

Step 1. Section properties: $I = \dfrac{\pi d^4}{64} = \dfrac{\pi(0.05)^4}{64} = 3.068\times10^{-7}\text{ m}^4$ , and $A = \dfrac{\pi d^2}{4} = 1.963\times10^{-3}\text{ m}^2$ . The radius of gyration is $k = \sqrt{I/A} = 0.0125\text{ m} = 12.5\text{ mm}$ , which for a circular section equals $d/4$ as a quick check.

Step 2. Pinned–pinned effective length is $L_e = L = 3\text{ m}$ , so

P_{cr} = \frac{\pi^2 EI}{L_e^2} = \frac{\pi^2(200\times10^9)(3.068\times10^{-7})}{9} \approx 67{,}300\text{ N} = 67.3\text{ kN}.

Step 3. Slenderness ratio and critical stress: $L_e/k = 3/0.0125 = 240$ , so $\sigma_{cr} = P_{cr}/A = 67{,}300/1.963\times10^{-3} = 34.3\text{ MPa}$ . Since $34.3\text{ MPa} \ll 250\text{ MPa}$ , the column buckles elastically at $67.3\text{ kN}$ , far below the load that would crush the steel.

Step 4. Now fix both ends: $L_e = 0.5L = 1.5\text{ m}$ , so $P_{cr} = \pi^2(200\times10^9)(3.068\times10^{-7})/1.5^2 \approx 269{,}200\text{ N} = 269.2\text{ kN}$ — exactly four times the pinned–pinned value, as expected from $P_{cr}\propto 1/L_e^2$ with $L_e$ halved.

The pin-ended column buckles at 67.3 kN; the same column with both ends fixed buckles at 269.2 kN — four times higher, from end fixity alone, with no change to the material or cross-section.

Competitive stretch

If instead this pinned–pinned column were shortened to $L = 0.3\text{ m}$ , its slenderness ratio drops to $L_e/k = 0.3/0.0125 = 24$ , giving $\sigma_{cr} = \pi^2E/24^2 \approx 3427\text{ MPa}$ — far above $\sigma_{yield} = 250\text{ MPa}$ . Euler's formula is not valid here: this short, stubby column would yield or crush at a much lower load than the (physically meaningless) Euler prediction, so the true failure mode is material yielding, not elastic buckling.

Common traps

Using the actual length $L$ instead of the effective length $L_e$ . End conditions swing $L_e$ from $0.5L$ to $2L$ — always identify the supports before plugging into the formula.
Skipping the $\sigma_{cr}\le\sigma_{yield}$ check. A "critical load" implying a stress above yield is not physically meaningful; a stubby column fails by crushing/yielding, not elastic buckling.
Using the wrong $I$ . Buckling uses the minimum area moment of inertia of the section — the axis the column is least stiff about — never the polar moment $J$ , and never the larger of two unequal $I$ values.
Assuming a stronger material means a stronger column. $P_{cr}$ depends on $EI$ and $L_e$ only; a high-yield-strength but low-stiffness material can still buckle at a low load.

Ask the tutor: if you would like to work through a column-design problem where you must pick a cross-section that satisfies both a strength requirement and a buckling requirement at once.

Why does buckling happen at a much lower load than the material's crushing strength?

Because buckling is a stability failure of the whole column's shape, not a local material failure. Once the axial load reaches $P_{cr}$ , any tiny sideways nudge grows instead of dying out, and the column bows sideways under a bending moment it was never designed to carry axially. The steel itself is nowhere near its compressive strength when this happens; it is the geometry — a long, slender shape — that gives way first. Short, stubby columns do not have this problem, which is exactly why the validity check against yield stress matters.

Why does the effective length change so much between fixed and pinned ends?

The effective length measures how much of the column is genuinely free to bow between points of zero moment (points where the buckled shape has an inflection, like the ends of a pinned–pinned column). A fixed end resists rotation, so the buckled shape near it stays nearly straight for some distance before curving — effectively shortening the free-bowing length. A free end, by contrast, lets the column swing out unrestrained, roughly doubling how far the equivalent pinned length extends. Since $P_{cr}\propto 1/L_e^2$ , this geometric effect swings the critical load by a factor of sixteen between the weakest (fixed–free) and strongest (fixed–fixed) support combinations of the same physical column.

Does a thicker column always resist buckling better than a longer, thinner one of the same area?

Not necessarily in the way you might guess — what matters is $I$ , not just cross-sectional area $A$ . For a given amount of material, spreading it farther from the bending axis (a hollow tube instead of a solid rod, for instance) increases $I$ and hence $P_{cr}$ without adding mass. That is precisely why tubular sections are preferred for compression members in trusses and frames: for the same material used, they resist buckling far better than an equal-area solid rod, because $I$ — not $A$ — sets the Euler load.

Column buckling

The core idea

Worked example

Competitive stretch

Common traps

Before this

Where this leads