Slenderness and the Governing Axis

A column buckles about whichever axis gives the largest KL/rKL/r.

Before this Lesson

Euler's formula makes a strange promise. Halve a column's length, and its critical load goes up four times. Halve it again, and it's sixteen. Nothing in the formula ever says stop. So try it on a stub: a 1 in square steel bar, 6 in long, between two pins.

Pcr=π2(29,000)(1/12)(6)2=663 kipP_{cr} = \frac{\pi^2(29{,}000)(1/12)}{(6)^2} = 663\ \text{kip}

That's 663 kip on one square inch, a stress of 663 ksi. A36 steel yields at 36 ksi, so the stub squashes at about 36 kip, nowhere near the promise. Somewhere between the strand of spaghetti and this stub, Euler's formula stops telling the truth. To find where, think in stresses instead of loads.

One number for slenderness

Divide the critical load by the column's area, and you get the stress in the column at the moment it buckles:

σcr=PcrA=π2E(KL)2⋅IA\sigma_{cr} = \frac{P_{cr}}{A} = \frac{\pi^2E}{(KL)^2}\cdot\frac{I}{A}

That I/AI/A is the square of the radius of gyration, r=I/Ar = \sqrt{I/A}, so the length and the section merge into a single ratio:

σcr=π2E(KL/r)2\sigma_{cr} = \frac{\pi^2E}{(KL/r)^2}

The Handbook prints this under Columns, in its Mechanics of Materials chapter, as the critical buckling stress for long columns, and it calls Kℓ/rK\ell/r the effective slenderness ratio. As in Euler Buckling, this Lesson writes the length as LL.

Look at what got folded together. LL is how long the column is, KK is how its ends are held, and rr is how far its area spreads from the axis it bends about. All of the column's geometry comes down to one number, KL/rKL/r: its effective length, measured in radii of gyration. Two steel columns with the same KL/rKL/r buckle at the same stress, whether one is a rod you could hold in your hand or a column in a ten-story building. And the stress falls with the square: double KL/rKL/r, and σcr\sigma_{cr} drops to a quarter.

Like II, rr belongs to an axis: the one the column bends about when it bows. For a rectangle, use the side it bows across. A 2 in by 3 in bar bowing across its 2 in side has r=2/12r = 2/\sqrt{12}.

KL/rKL/r has no units, but only if KLKL and rr share them. Lengths usually come in feet and rr in inches, so multiply the feet by 12 first.

Worked Example A pin-ended steel bar
1 of 2

A solid steel bar, 2 in by 3 in, is 8 ft long and pinned at both ends. Find its least radius of gyration, its slenderness ratio, and its critical buckling stress.

Where Euler stops telling the truth

Plot σcr\sigma_{cr} against KL/rKL/r, and the curve climbs without limit as columns get stockier. At KL/r≈21KL/r \approx 21, the stub from the start, it's at 663 ksi.

0 50 100 150 200 89 KL/r 0 20 40 60 80 σcr (ksi) proportional limit σpl = 36 ksi to 663 ksi at KL/r ≈ 21 Long columns Euler's formula holds Intermediate columns Euler overpredicts
Euler's critical stress against slenderness, for steel with a 36 ksi proportional limit
Euler's curve meets the proportional limit at KL/r=89KL/r = 89. To the right it holds; to the left, the dashed part promises stresses the steel can't reach.

The trouble is an assumption Euler's derivation made without saying so: the column is as stiff as EE right up to the moment it buckles. EE is the slope of the straight first part of the stress-strain curve, as the Handbook's Uniaxial Stress-Strain section puts it. That straight part ends at the proportional limit, σpl\sigma_{pl}, the highest stress at which stress is still proportional to strain. Past it, each extra ksi buys more strain than before. The material is softer than EE says, and soon it yields.

So Euler's formula holds only while the critical stress stays at or below the proportional limit. Set the two equal, and solve for the slenderness where they meet:

π2E(KL/r)2=σpl⟹(KLr)min⁡=πEσpl\frac{\pi^2E}{(KL/r)^2} = \sigma_{pl} \quad\Longrightarrow\quad \left(\frac{KL}{r}\right)_{\min} = \pi\sqrt{\frac{E}{\sigma_{pl}}}

For steel with σpl=36\sigma_{pl} = 36 ksi, that's π29,000/36=89\pi\sqrt{29{,}000/36} = 89. Euler's formula is valid for any column at least that slender. To turn the limit into the shortest length, multiply back by rr and divide by KK.

A column more slender than the limit is a long column. It buckles elastically, at Euler's stress, and the material never notices. A stockier one is an intermediate column. On its way to Euler's stress it passes the proportional limit, part of its section starts to yield, and its stiffness drops. It fails below Euler's curve, and never above the yield strength. That's why Euler overpredicts a stocky column: it credits the column with a stiffness it no longer has. The very stockiest, called short columns, skip buckling and simply squash.

The Handbook's Columns section prints its stress formula for long columns without saying where "long" ends, and the Handbook lists no proportional limit: its Table 2 gives only yield and ultimate strengths. So a question that wants this limit states σpl\sigma_{pl}. Use it, not the yield or ultimate strength, which are higher and would call some columns long when they're not.

Worked Example How short a column Euler can handle
1 of 2

A pin-ended steel column has a least moment of inertia of 12.0 in⁴ and an area of 6.0 in². Its steel has a proportional limit of 36 ksi and a yield strength of 50 ksi. Find the smallest slenderness ratio, and the shortest length, for which Euler's formula is valid.

Codes and textbooks draw the line between long and intermediate in different places, so a question that asks you to classify a column will state its transition value. Find KL/rKL/r and compare: above the value is long, at or below it is intermediate.

Checkpoint
1 of 2

Euler's formula gives a stocky aluminum strut a critical stress of 70 ksi. The alloy's proportional limit is 30 ksi. Which statement is true?

A.
It buckles at 70 ksi, because Euler's formula doesn't depend on the material's strength.
B.
It fails at exactly 30 ksi, its proportional limit.
C.
It fails above 70 ksi, because Euler's formula is conservative for stocky columns.
D.
It fails below 70 ksi, because once its stress passes 30 ksi, its stiffness drops below E and part of it yields.

Each axis has its own KL/r

Euler Buckling said a column bows about its weak axis. That's true when its ends are held the same way in every direction and nothing touches it in between. Real columns often get help in one direction only. Horizontal wall members called girts tie into a column's flange and stop it moving along the wall, but not out of it. Sheathing nailed to a stud keeps it from bowing across its thin side.

A brace stops the column moving in one direction, so it shortens the unbraced length for buckling about one axis only. Each axis gets its own effective length and its own slenderness ratio, KxLx/rxK_xL_x/r_x and KyLy/ryK_yL_y/r_y. Both share EE and AA, so the larger ratio gives the lower stress, and that's the one the column reaches first. The column buckles about the axis with the larger KL/r. The Handbook's Columns formulas take one rr and one length, so run them once per axis.

Lx = L Buckling about x Ly = L/2 Ly = L/2 Buckling about y
One column, braced at midheight about y only, buckling about each axis
The brace stops sideways movement in the y view only. There the column bows in two half waves, each L/2L/2 long, while about x it bows over the full height.

Keep each length with its own rr. For a W shape, buckling about x bows the column in the plane of its web, and buckling about y bows it sideways, along its flanges. LxL_x runs between points braced against the first, and LyL_y between points braced against the second. Dividing the braced length by rxr_x, or the full length by ryr_y, mixes the two.

Three column shapes, with their areas and radii of gyration
ShapeAA (in²)rxr_x (in)ryr_y (in)
W14X4814.15.851.91
W12X7923.25.343.05
W10X4914.44.352.54

Adapted from the FE Reference Handbook, Table 1-1: W Shapes Dimensions and Properties

A midheight brace halves LyL_y. Whether the weak axis still governs depends on how rxr_x compares with ryr_y. With the same KK both ways, the strong axis takes over when rxr_x is less than twice ryr_y. Deep, narrow shapes like the W14X48 have rxr_x about three times ryr_y. The squat, heavy shapes made for columns are closer to 1.7.

Worked Example A W column with a midheight brace
1 of 2

A W14X48 column is 30 ft tall and pinned at both ends. Girts brace it against buckling about its y-axis at midheight. Its steel's proportional limit is 36 ksi. Using Table 1-1, find its Euler buckling load.

A steel column this long buckles elastically. For a stockier one, Steel Column Strength takes the same governing KL/rKL/r into AISC's design curve.

Different ends, and the balance point

End conditions can differ by axis too. A strut held at each end by a single pin through a clevis can turn about the pin, but the clevis plates clamp its sides, so it acts pinned for buckling one way and close to fixed the other. Then KxK_x and KyK_y differ, and the same comparison picks the axis.

Bracing can also go too far. Add weak-axis braces, and KyLy/ryK_yL_y/r_y drops until it matches KxLx/rxK_xL_x/r_x. That's the balance point, where both axes are equally critical:

KyLyry=KxLxrx⟹Ly=KxLxKy⋅ryrx\frac{K_yL_y}{r_y} = \frac{K_xL_x}{r_x} \quad\Longrightarrow\quad L_y = \frac{K_xL_x}{K_y}\cdot\frac{r_y}{r_x}

For a rectangle, ry/rxr_y/r_x is just the ratio of its sides, since the 12\sqrt{12} cancels. A pin-ended 2 in by 6 in bar is balanced with braces at its third points. Past the balance point, the strong axis governs, and more weak-axis bracing buys nothing.

Worked Example The bar with clevis ends
1 of 3

The 2 in by 3 in steel bar from before, 8 ft long, is now held at each end by a pin through its 2 in thickness. The pins let it turn when it bows across its 3 in depth, but the clevis plates hold it fixed when it bows across its 2 in thickness. Using the theoretical KK, find its Euler buckling load.

Where a brace does the most good

Every brace is a bet on one axis. It pays only if it shortens the unbraced length that governs, and three rules follow from that:

  • Brace the axis that governs. A brace about the other axis lowers a ratio that wasn't in charge.
  • Split the longest segment, near its middle. Between braces, a column buckles over its longest unbraced piece. A brace that splits a shorter piece, or sits near an end of the longest piece, barely touches the governing length.
  • Stop at the balance point. Once the other axis governs, more bracing on the first buys nothing.
Checkpoint
1 of 2
30 ft Buckling about x 18 ft 12 ft Buckling about y
Where the column is braced about each axis

A W12X79 column is braced as shown. Using Table 1-1, about which axis does it buckle, and with what slenderness ratio?

A.
About y, with KL/r = 70.8
B.
About x, with KL/r = 67.4
C.
About y, with KL/r = 118.0

Recap

  • σcr=π2E/(KL/r)2\sigma_{cr} = \pi^2E/(KL/r)^2, and KL/rKL/r folds length, ends, and shape into one number. Take rr about the axis of bending (the side a rectangle bows across over 12\sqrt{12}, or D/4D/4), with both lengths in inches.
  • Euler holds only while σcr≤σpl\sigma_{cr} \le \sigma_{pl}, so for KL/r≥πE/σplKL/r \ge \pi\sqrt{E/\sigma_{pl}}. Stockier, intermediate columns yield first, and Euler overpredicts them. Classify against the transition value a question states.
  • Bracing shortens one axis only. Compare KxLx/rxK_xL_x/r_x with KyLy/ryK_yL_y/r_y, keeping each length with its own rr; the larger governs, and it isn't always the weak axis.
  • Both axes are equally critical at Ly=(KxLx/Ky)(ry/rx)L_y = (K_xL_x/K_y)(r_y/r_x). Brace the governing axis in its longest segment, and stop at the balance point.