Steel Column Strength

Real columns buckle sooner than Euler predicts, and the AISC curve accounts for it.

In Slenderness and the Governing Axis, a W14X48 column stood 30 ft tall, pinned at both ends, with girts bracing its weak axis at midheight. Euler's formula gave it 454 kip, at a stress of 32.2 ksi, safely under the steel's 36 ksi proportional limit. By every test that Lesson had, Euler's formula held.

Make that column of 50 ksi steel and design it by AISC, and its nominal strength is 368 kip, about four-fifths of Euler's load. The resistance factor then takes it to 331 kip. That first cut isn't a safety margin. It's the difference between Euler's perfect column and a real rolled one.

What Euler assumed

Euler's derivation pictured a column that is perfectly straight and carries no stress at all until the load arrives. A rolled steel column is neither.

It's crooked. Mill tolerances let a W-shape come out bowed by an amount e0e_0 of about 1/1000 of its length, so a 30 ft column can be 3/8 in out of straight. A perfect column stays straight right up to PcrP_{cr}. A bowed one bends from the first kip: the load, offset by the bow, puts a moment on the column, the moment bows it further, and the bigger bow adds moment. The bending piles extra compression onto the inside of the bow, so that face yields early.

It's already stressed. A W-shape comes off the rolls red hot and cools unevenly. The flange tips, thin and open to the air, cool and stiffen first. The thick junction where the web meets each flange cools last, and as it shrinks, the stiff tips resist it. Once it's cold, the junction is left in tension and the tips in compression, with no load on the column at all. These are residual stresses.

P e0 Initial bow C C T y tips yield first Residual stress
Two ways a rolled column differs from Euler's

Now load the column. Every fiber carries its residual stress plus the average P/AP/A, so the tips, which started out squeezed, reach FyF_y first, while the average is still well below it. A yielded fiber keeps its load but stops resisting: push it further, and it just gives. And the tips are the part of the section farthest from the web, the part that supplies most of a W-shape's IyI_y. So the column loses its weak-axis stiffness early and buckles below Euler's stress.

That's the catch in the last Lesson's proportional-limit check. A test bar cut from the steel stays linear up to its proportional limit, but the whole rolled section, squeezed at its tips by its own cooling, goes soft sooner. The Handbook's column equations never mention either flaw, but they're calibrated for both.

AISC's column curve

The Handbook's Design of Steel Components section gives AISC's answer under Columns. It starts from Euler's critical stress, renamed the elastic buckling stress:

Fe=π2E(KL/r)2F_e = \frac{\pi^2E}{(KL/r)^2}

with E=29,000E = 29{,}000 ksi, printed in the section's heading. Then it chooses one of two critical stresses by slenderness:

KLr≤4.71EFy:Fcr=[0.658Fy/Fe]Fy(inelastic)KLr>4.71EFy:Fcr=0.877Fe(elastic)\begin{aligned} \frac{KL}{r} &\le 4.71\sqrt{\frac{E}{F_y}}: &\quad F_{cr} &= \left[0.658^{F_y/F_e}\right]F_y &&\text{(inelastic)} \\[4pt] \frac{KL}{r} &> 4.71\sqrt{\frac{E}{F_y}}: &\quad F_{cr} &= 0.877F_e &&\text{(elastic)} \end{aligned}

The Handbook writes the conditions with Lc/rL_c/r, where Lc=KLL_c = KL, the same effective length.

Slender columns buckle elastically. Their buckling stress is so low that even the squeezed tips haven't yielded, so residual stress doesn't matter. Only crookedness does, and AISC charges a flat 12% for it: FcrF_{cr} is 0.877 times Euler's stress.

Stockier columns buckle inelastically, with their tips already yielded. Their equation is FyF_y times a factor, 0.658 raised to Fy/FeF_y/F_e. As a column gets stockier, FeF_e grows, the exponent shrinks toward zero, and the factor climbs toward 1. The stubbiest column doesn't buckle at all; it squashes at FyF_y. That's also how to remember which way up the exponent goes. Write it Fe/FyF_e/F_y by mistake, and a stocky column, with its enormous FeF_e, gets almost no strength.

The switch at 4.71E/Fy4.71\sqrt{E/F_y} falls where Euler's stress has dropped to 0.44Fy0.44F_y. There both equations give about 0.39Fy0.39F_y, so the curve runs on without a jump. For 50 ksi steel:

4.7129,00050=113.44.71\sqrt{\frac{29{,}000}{50}} = 113.4
0 50 100 150 200 113 0 10 20 30 40 50 60 KL/r stress (ksi) Euler, Fe AISC, Fcr Fy = 50 ksi inelastic elastic 32.2 26.1
AISC's critical stress for 50 ksi steel, against Euler's
AISC's curve stays below Euler's everywhere and levels off at FyF_y for the stockiest columns.

The marked column is the Hook's W14X48. Its KL/rKL/r of 94.2 is below 113.4, so it buckles inelastically:

Fcr=[0.65850/32.2](50)=26.1 ksiF_{cr} = \left[0.658^{50/32.2}\right](50) = 26.1\ \text{ksi}

That's 81% of Euler's 32.2 ksi. Times the column's 14.1 in², it's the 368 kip nominal strength from the Hook.

Checkpoint
1 of 2

A W-shape column of 50 ksi steel has a governing KL/rKL/r of 118. Which kind of buckling governs, and which equation gives FcrF_{cr}?

A.
Inelastic: Fcr=[0.658Fy/Fe]FyF_{cr} = \left[0.658^{F_y/F_e}\right]F_y
B.
Elastic: Fcr=FeF_{cr} = F_e
C.
Elastic: Fcr=0.877FeF_{cr} = 0.877F_e
D.
Inelastic: Fcr=0.877FeF_{cr} = 0.877F_e

Design strength, start to finish

Strength is stress times area, and the Handbook's nominal strength uses the gross area: Pn=FcrAgP_n = F_{cr}A_g. Its resistance factor for columns is ϕc=0.90\phi_c = 0.90, so the design strength, checked against the factored load as demand versus capacity, is

ϕcPn=0.90 FcrAg≥Pu\phi_cP_n = 0.90\,F_{cr}A_g \ge P_u

For the Hook's column, 0.90(26.1)(14.1)=3310.90(26.1)(14.1) = 331 kip. The Handbook gives this for members without slender elements, flanges or webs thin enough to ripple locally first. None of Table 1-1's W14, W12, or W10 shapes has one.

Every column problem runs the same chain. Find the effective length about each axis, divide each by its own rr, and keep the larger KL/rKL/r. Find FeF_e, compare KL/rKL/r with the switch, take FcrF_{cr} from the right equation, and multiply by 0.90Ag0.90A_g. The costliest slip is the first one: reaching for rxr_x out of habit, when it's usually the weak axis that governs.

Worked Example A braced W14 column
1 of 3

A W14X61 column of 50 ksi steel is 26 ft tall and pinned at both ends. Girts brace it about its y-axis at midheight. Find its design compressive strength.

Table 4-14: the curve, worked out

The Handbook's Table 4-14, Available Critical Stress ϕcFcr\phi_cF_{cr} for Compression Members, has run the equations for you. It lists ϕcFcr\phi_cF_{cr} in ksi, already multiplied by 0.90, for every whole KL/rKL/r from 1 to 200. Like Table 3-2 for beams, it's for Fy=50F_y = 50 ksi only, so it's no help with A36. And its values are stresses: multiply by AgA_g to get a strength in kips.

Design critical stress ϕcFcr\phi_cF_{cr} for 50 ksi steel, for KL/rKL/r from 81 to 100
KL/rKL/rϕcFcr\phi_cF_{cr} (ksi)KL/rKL/rϕcFcr\phi_cF_{cr} (ksi)
8127.99124.6
8227.59224.2
8327.29323.9
8426.99423.6
8526.59523.3
8626.29622.9
8725.99722.6
8825.59822.3
8925.29922.0
9024.910021.7

Adapted from the FE Reference Handbook, AISC Table 4-14: Available Critical Stress φcFcr for Compression Members

The Hook's column has KL/r=94.2KL/r = 94.2, between the rows for 94 and 95. Interpolating gives 23.5 ksi, and 23.5(14.1)=33123.5(14.1) = 331 kip, just as the equations gave. Rounding up to the 95 row instead gives 23.3 ksi and 329 kip, a hair low, which is the safe side.

Checkpoint

A W12X72 column of 50 ksi steel is 28.5 ft tall, fixed at its base and pinned at its top, with no bracing between. Table 1-1 gives Ag=21.1A_g = 21.1 in², rx=5.31r_x = 5.31 in, and ry=3.04r_y = 3.04 in. Using the recommended KK and Table 4-14, what is its design compressive strength?

kip

Table 4-1: strength by shape and length

Table 4-1, Available Strength in Axial Compression, goes one step further and multiplies by AgA_g too. For fifteen W14, W12, and W10 shapes of 50 ksi steel, it lists ϕcPn\phi_cP_n in kips against the effective length KLKL in feet. Here are its rows from 8 to 16 ft:

Design strength ϕcPn\phi_cP_n (kip) of W-shapes of 50 ksi steel, by effective length about the y-axis
KL (ft)W14X74W14X68W14X61W14X53W14X48W12X58W12X53W12X50W12X45W12X40W10X60W10X54W10X49W10X45W10X39
8878804721585529689628551495439717643585505435
9853781700557504670610526472419698625569483415
10826755677528477649590499448397677607551460395
11797728652497449627569471422375655586533435373
12766700626465420603547443396351631565513410351
13734670599433391578525413370328606543493384328
14701639572401361553501384343304581520471358305
15667608543369332527477354317280555496450332282
16632576515338304500452326291257528472428306260

Adapted from the FE Reference Handbook, AISC Table 4-1: Available Strength in Axial Compression, kips—W shapes

The table's length is in feet, so there's nothing to convert; only KL/rKL/r needs inches. And it's taken, in the table's words, "with respect to least radius of gyration ryr_y": the table assumes the column buckles about its weak axis, so you enter it with KyLyK_yL_y. For an unbraced column, that's always right. With weak-axis bracing, check the strong axis first. If KxLx/rxK_xL_x/r_x is the larger ratio, the table overstates the strength, and you need the equations or Table 4-14 instead.

It agrees with what you've computed. The Hook's W14X48, with KyLy=15K_yL_y = 15 ft, reads 332 kip, and the worked example's W14X61, at 13 ft, reads 599.

To select a column, run across the row for your KLKL. Each depth group runs from heaviest to lightest, so in each one, find the lightest shape that reaches PuP_u. Then compare the three winners' weights. The deepest group doesn't always hold the lightest column.

Worked Example The lightest column for a load
1 of 3

A column of 50 ksi steel, 14 ft tall and pinned at both ends with no bracing between, carries a service dead load of 150 kip and a service live load of 170 kip. Using Table 4-1, select the lightest W-shape.

Recap

  • Real columns are a little crooked, and uneven cooling leaves their flange tips squeezed by residual stress. Bending from the bow and early yielding at the tips both make a rolled column buckle below Euler's load.
  • Fe=π2E/(KL/r)2F_e = \pi^2E/(KL/r)^2. At or below KL/r=4.71E/FyKL/r = 4.71\sqrt{E/F_y}, 113.4 for 50 ksi steel, buckling is inelastic and Fcr=[0.658Fy/Fe]FyF_{cr} = \left[0.658^{F_y/F_e}\right]F_y. Above it, buckling is elastic and Fcr=0.877FeF_{cr} = 0.877F_e.
  • ϕcPn=0.90FcrAg\phi_cP_n = 0.90F_{cr}A_g, with the larger of KxLx/rxK_xL_x/r_x and KyLy/ryK_yL_y/r_y, and lengths in inches.
  • Table 4-14 gives ϕcFcr\phi_cF_{cr} in ksi, for 50 ksi steel, so multiply by AgA_g. Table 4-1 gives ϕcPn\phi_cP_n in kips against KyLyK_yL_y in feet, so check the strong axis when bracing differs. To select, take the lightest adequate shape in each group and compare.