Euler Buckling
A long column doesn't crush; it bows sideways at a load set by stiffness alone.
Before this Lesson
- Composite Sections · Statics
- Strain, Hooke's Law, and Poisson's Ratio · Mechanics of Materials
Stand a strand of dry spaghetti on the table and press down on its top. It doesn't crush. Long before the pasta comes near its strength, the strand bows out sideways, the bow runs away from you, and it snaps. Now break off half and try again. The half strand takes about four times the push, though nothing about the pasta changed but its length.
Why a column bows instead of crushing
Take a perfectly straight column with pinned ends, load it straight down its axis, and nudge its middle sideways. Now the load has a lever arm. It still acts along the line between the pins, but the column sits off that line by a distance , so the load bends it with a moment , the way it's already bowed. The column pushes back with its bending stiffness: a moment of times its curvature, the same that sets how far a beam deflects.
Under a light load, the column's push wins, and it springs back straight. Under a heavy one, the load wins, and the bow grows on its own. In between is one load where they tie, the critical load , where the straight shape stops being the only way the column can stand.
A pin-ended column bows into half a sine wave, and a sine wave's curvature is times its offset at every point. Set the two moments equal:
The cancels. At one load, the column balances bowed by any small amount just as well as straight, so the smallest nudge leaves it bowed. The Handbook's Mechanics of Materials chapter prints that load under Columns as Euler's formula, with an effective-length factor for the end supports:
For pinned ends, . The Handbook writes the length there as , and its steel section writes , as this Lesson does.
Notice what's missing: strength. A36 steel and a steel twice as strong give the same , because structural steels all have ksi. The column fails by bending out of the way, so only stiffness counts: for the material, for the shape, and the length, squared. Halve the length, and goes up four times. That's the spaghetti.
All this holds for a long, slender column, one that buckles while its stress is still below the proportional limit. Where stocky columns that yield first take over is the business of Slenderness and the Governing Axis.
It bows the easy way
Press a plastic ruler on end, and it always bows across its thin side, never across its wide one. A column has an about every axis through its centroid, and it buckles about whichever gives the smallest . With the same end supports in every direction, that's the axis with the smallest , the weak axis. For a rectangle, it's the one that cubes the short side.
A round section is the same in every direction, so any diameter will do. A hollow section is the solid minus the hole, with the fourth powers taken before you subtract. Subtract the diameters first, and a pipe with 0.5 in walls comes out with almost no at all.
From Composite Sections, with the short side of a rectangle, and the hole's sides, and a diameter:
The Handbook's area table prints the circles with radii, as . Substituting turns the 4 into 64. A hollow rectangle isn't in the table, so subtract the hole as its own rectangle.
Putting in the numbers
Euler's formula has no unit conversion built in, so keep one length unit throughout. With in ksi and in in⁴, put the length in inches, and comes out in kips. With in MPa, which is N/mm², and in mm⁴, use millimeters, and it comes out in newtons. A length left in feet makes the answer 144 times too big.
When a question names a material but no specific alloy, take from the Handbook's Table 1 - Typical Material Properties, in the Mechanics of Materials chapter:
| Material | E as printed | E in ksi | E in MPa |
|---|---|---|---|
| Steel | 29.0 Mpsi (200.0 GPa) | 29,000 | 200,000 |
| Aluminum | 10.0 Mpsi (69.0 GPa) | 10,000 | 69,000 |
| Wood (fir) | 1.6 Mpsi (11.0 GPa) | 1,600 | 11,000 |
Adapted from the FE Reference Handbook, Table 1 - Typical Material Properties
End restraint shortens the bow
A pin can't resist turning, so the moment there is zero and the curve has no curvature: each pin is an inflection point. Euler's formula really measures the distance between two inflection points, one half sine wave. Hold the ends differently, and the half wave changes length. That length is the effective length .
- Fixed–fixed. Both ends stay vertical, so the curve reverses near each, with inflection points at the quarter points. The middle half is a pin-ended column long: , and four times the pin-ended load.
- Fixed–pinned. One inflection point falls inside the column, about from the pin: , and about twice the load.
- Fixed–free, a flagpole. Mirror it in the ground, and the flagpole and its reflection make one pin-ended column twice as tall: , and a quarter of the load.
The firmer the ends, the shorter the half wave. is a ruler for end conditions.
The Handbook lists those four theoretical values under Euler's formula. Its Civil Engineering chapter, after the steel beam tables, has AISC's fuller version. It describes each end by whether it can rotate and whether it can translate, and it adds a recommended design value:
| End conditions | Theoretical K | Recommended K |
|---|---|---|
| (a) Fixed–fixed | 0.5 | 0.65 |
| (b) Fixed–pinned | 0.7 | 0.80 |
| (c) Fixed base, top sways without rotating | 1.0 | 1.2 |
| (d) Pinned–pinned | 1.0 | 1.0 |
| (e) Fixed–free | 2.0 | 2.10 |
| (f) Pinned base, top sways without rotating | 2.0 | 2.0 |
Adapted from the FE Reference Handbook, Table C-A-7.1 AISC Approximate Values of Effective Length Factor, K
A real fixed end is never perfectly fixed. A footing or a welded connection lets it turn a little, the inflection points drift apart, and the effective length grows. That's why most recommended values are larger. Use the value the question names. Cases (c) and (f), whose tops sway, come next.
A top that can sway
Picture a one-story frame: two columns with a stiff roof girder welded across their tops, and no bracing or walls. The girder keeps the column tops from rotating, but nothing stops the roof from sliding sideways. So when the columns buckle, their tops move over while staying vertical: rotation fixed, translation free.
With a fixed base, each column bends into an S, with an inflection point at mid-height. Each half is a flagpole tall, so . That's case (c): theoretical , recommended 1.2. It's fixed–fixed with the top let loose, and the sway alone doubles .
With a pinned base, the column is an upside-down flagpole. The girder holds the top vertical, like the ground at a flagpole's base, and the pin has zero moment, like a flagpole's tip. That's case (f), with both ways.
Comparing columns without the whole formula
Euler's formula says the capacity goes as . When two columns share the material and the section, everything but cancels from their ratio:
So a flagpole carries of what the same column carries with pinned ends. To rank several columns, find each and sort: the shortest effective length is the strongest. When only the section changes, the ratio is the ratio of the values. For solid round sections, that's the diameter to the fourth power, so doubling the diameter multiplies the capacity by 16.
Sizing a column with a factor of safety
A designer wants the column to stay well clear of buckling under its working load. A factor of safety against buckling, FS, asks that the critical load be FS times the load the column carries:
The Handbook prints no buckling safety factor, so a question will state it. From there, solve Euler's formula backward for whatever is unknown: the longest length, the smallest , or, for a round section, the smallest diameter from .
The factor goes on the load, never on the length. Length is squared in the formula, so a factor of 4 on the load is only a factor of 2 on the length.