Steel Beam Flexural Strength

A steel beam keeps carrying load after it starts to yield, until the whole section has yielded.

Before this Lesson

Put a W18X50 of 50 ksi steel, with a slab on its top flange, under a slowly growing moment. The flexure formula says its top and bottom fibers reach yield when M/SxM/S_x hits 50 ksi, at 370 ft-kip. If the beam were glass, that would be the end of it.

It isn't. The beam keeps taking moment, up to about 421 ft-kip, and even then it doesn't snap. It folds, slowly, like a bent paperclip. Those extra 50 ft-kip come from fibers that have yielded but haven't quit, and the Handbook's steel beam strength counts on every one of them.

Yielding spreads inward

At first yield the stress over the depth is the familiar triangle, with FyF_y at both faces. The moment there is the yield moment:

My=FySxM_y = F_y S_x

For the W18X50, that's 50(88.9)=4,44550(88.9) = 4{,}445 in.-kip, or 370 ft-kip. Now push past it. The strain keeps growing in a straight line, but the stress can't follow: a fiber strained past εy\varepsilon_y stays at FyF_y. So the yielded zones creep inward from both faces, and only a core around the neutral axis is still elastic.

NA Fy Fy compression tension yielded elastic core yielded Fy Fy First yield εmax = εy M = My = Fy Sx Partly plastic εmax = 2εy My < M < Mp Fully plastic εmax ≫ εy M = Mp = Fy Zx
The stress over a beam's depth as the moment grows past first yield

With the faces at 2εy2\varepsilon_y, the elastic core is the middle half of the depth. At 10εy10\varepsilon_y, it's the middle tenth. For 50 ksi steel that's a strain of only 0.017, and structural steel stretches to 0.2 or more before it breaks. So the core shrinks almost to nothing, and the limit, with every fiber at FyF_y, is the plastic moment, MpM_p. Past it, more bending buys no more moment. The section just keeps rotating, like a stiff hinge, and that's the Hook's paperclip.

Two blocks of stress

At MpM_p the stress is two blocks: FyF_y in compression on one side of the axis and FyF_y in tension on the other. The beam carries no axial force, so the push and the pull must balance, and with the same FyF_y on both sides, that means equal areas. So the plastic neutral axis (PNA) splits the section into two equal areas. For a symmetric section like a rectangle or a W-shape, that's mid-depth. For a T-section, it isn't at the centroid.

Each half carries Fy(A/2)F_y(A/2) at its own centroid, so the two forces form a couple, and everything in its moment except FyF_y is geometry. That geometry is the plastic section modulus, ZZ:

Mp=Fy A2(yˉc+yˉt)=FyZM_p = F_y\,\frac{A}{2}\left(\bar{y}_c + \bar{y}_t\right) = F_y Z

Here yˉc\bar{y}_c and yˉt\bar{y}_t run from the PNA to the centroids of the compression and tension halves. For a rectangle bb wide and hh deep, each half is bh/2bh/2, with its centroid h/4h/4 from the axis:

A/2 A/2 h b PNA Fy Fy C = Fy(bh/2) T = Fy(bh/2) h/4 h/4 h/2
A fully yielded rectangle: two equal halves and a couple
Z=bh2(h4+h4)=bh24versusS=bh26Z = \frac{bh}{2}\left(\frac{h}{4} + \frac{h}{4}\right) = \frac{bh^2}{4} \qquad \text{versus} \qquad S = \frac{bh^2}{6}

So a fully yielded rectangle carries 1.5 times its first-yield moment, and that's why ZZ, not SS, sets the plastic moment. SS describes the triangle, where only the extreme fibers reach FyF_y and the fibers near the axis loaf. ZZ describes the two blocks, where every fiber works at FyF_y.

A W-shape gains much less. In Bending Stress, the W18X50's web-flange junction ran at 94% of its face stress, so at first yield the flanges are already nearly yielded through. Full yield adds a lot in the web, but the web is thin and close to the axis.

W18X50 NA Fy 0.94Fy flanges: already near Fy web: gains the most, but it's thin
A W18X50 at first yield, with the fully plastic profile outlined
Not to scale. The flanges are drawn about twice their true thickness.

The Handbook defines ZxZ_x only as the "plastic section modulus about the x-axis," with no formula, and Table 1-1 lists it beside SS under Axis X-X.

The five shapes from Bending Stress, with their plastic moduli
ShapeSxS_x (in³)ZxZ_x (in³)Zx/SxZ_x/S_x
W24X681541771.15
W21X4481.695.41.17
W18X5088.91011.14
W16X3656.564.01.13
W12X4051.557.01.11

Adapted from the FE Reference Handbook, Table 1-1: W Shapes Dimensions and Properties. The last column is computed.

Across all of Table 1-1, ZxZ_x beats SxS_x by 10 to 18 percent, most for deep, light shapes like the W21X44, which carry more of their area in the web.

What can stop a beam short of full yield

MpM_p assumes every fiber gets to yield. But a beam's compression side is made of thin plates being squeezed, and a thin plate squeezed hard enough can buckle before it yields, in two ways.

Local buckling

Stand on an empty soda can, and its thin wall wrinkles instead of shortening evenly. A flange or web too thin for its width can ripple the same way before it yields. A section stocky enough to yield all the way through before it ripples is compact. Most W-shapes are compact in 50 ksi steel, and Table 1-1 flags the few that aren't with its note f, "Shape exceeds compact limit for flexure," such as the W12X65. The Handbook prints no limits, so a question names a compact shape or says so.

Lateral-torsional buckling

Hold a plastic ruler on edge and try to bend it the stiff way. It rolls over sideways and twists instead. A W-shape's compression flange, the top one under positive moment, is a long, slender column being squeezed. The web keeps it from buckling downward, but nothing stops it from bowing sideways, while the stretched bottom flange hangs back. So the section swings sideways and twists: lateral-torsional buckling.

A A top (compression) flange bottom (tension) flange support (braced) Lb load moves sideways twists compression tension Plan (looking down) Section A–A
Lateral-torsional buckling of a beam braced only at its supports

The cure is the one for any column: shorten it. A brace is anything that keeps the compression flange from moving sideways or the section from twisting, and the Handbook's unbraced length, LbL_b, is the length between braces. A slab or metal deck fastened along the top flange braces it everywhere, which is full lateral support. The supports always count, since the Handbook notes that bracing must always be provided there.

Short enough, and the flange yields before it can buckle. The Handbook's rule: when Lb≤LpL_b \le L_p, lateral-torsional buckling doesn't apply. Table 3-2 lists LpL_p, which grows with the shape's weak-axis ryr_y, and it's short: 5.83 ft for the W18X50. That beam spanning 30 ft, braced only at its supports, has LbL_b five times LpL_p and can't get near MpM_p.

Mn unbraced length, Lb Lp Mp = Fy Zx Lb ≤ Lp: full plastic moment Lb > Lp: buckles sideways first
Nominal strength against unbraced length, schematic

Past LpL_p, the Handbook's next equation cuts MnM_n below MpM_p, and this Lesson stops there. So Mn=FyZxM_n = F_yZ_x needs three things: a doubly symmetric I-shape such as a W-shape, bent about its strong axis; a compact section; and Lb≤LpL_b \le L_p, which full lateral support satisfies.

Checkpoint
1 of 3

A W18X50 has Sx=88.9S_x = 88.9 in.³ and Zx=101Z_x = 101 in.³ Why is its plastic moment FyZxF_yZ_x rather than FySxF_yS_x?

A.
ZxZ_x adds the extra strength steel gains as it strain-hardens past FyF_y.
B.
ZxZ_x is SxS_x with the resistance factor built in, so it gives the design value directly.
C.
At MpM_p every fiber is at FyF_y, and ZxZ_x measures the couple of the two equal-area blocks.
D.
ZxZ_x is taken about the weak axis, the direction a beam actually fails in.

Design strength

Now the check from Loads and Load Combinations, demand versus capacity, ϕRn≥Ru\phi R_n \ge R_u, in its bending form: ϕbMn≥Mu\phi_b M_n \ge M_u. The Handbook's Design of Steel Components section gives ϕb=0.90\phi_b = 0.90 for these beams, and its Yielding equation gives Mn=Mp=FyZxM_n = M_p = F_yZ_x. Together:

ϕbMn=0.90 FyZx\phi_b M_n = 0.90\,F_y Z_x

FyF_y in ksi times ZxZ_x in in.³ gives in.-kip, so divide by 12 before comparing with an MuM_u in ft-kip. The grade matters too. Table 3-2's heading assumes Fy=50F_y = 50 ksi, the usual grade for W-shapes, and a question that names A36 means 36 ksi, from the Handbook's Table 2 of material properties.

Table 3-2 also lists ϕbMpx\phi_b M_{px}, the product already worked out: 379 ft-kip for the W18X50. It's a handy check, but only for 50 ksi. Used for A36, it would overstate the strength by 50/36.

Worked Example The design strength of a floor beam
1 of 3

A W21X44 floor beam of 50 ksi steel has a concrete slab fastened to its top flange. Find its design flexural strength, and check it against Mu=340M_u = 340 ft-kip.

Running it backward

Any piece of ϕbMn≥Mu\phi_b M_n \ge M_u can be the unknown. Leave ZxZ_x unknown, and the check becomes a requirement:

Zx,req=MuϕbFyZ_{x,\text{req}} = \frac{M_u}{\phi_b F_y}

with MuM_u in in.-kip. Any compact, braced W-shape with at least that ZxZ_x is strong enough in bending. Picking the lightest one is the job of Selecting Steel Beams.

Worked Example The plastic modulus a beam needs
1 of 3

A braced beam of 50 ksi steel spans 30 ft, simply supported, under a factored uniform load of 3.2 kip/ft. Find MuM_u and the required ZxZ_x.

How much a beam can carry

Run it the other way, with the shape given and the load or the span unknown. Set Mu=ϕbMnM_u = \phi_b M_n in wuL2/8w_uL^2/8 and solve:

wu=8 ϕbMnL2L=8 ϕbMnwuw_u = \frac{8\,\phi_b M_n}{L^2} \qquad L = \sqrt{\frac{8\,\phi_b M_n}{w_u}}

Keep ϕbMn\phi_b M_n in ft-kip and LL in ft, and wuw_u comes out in kip/ft. That wuw_u is everything the beam can hold up, its own weight included. The designation gives the weight: a W18X50 weighs 50 lb/ft, or 0.050 kip/ft. It's dead load, so factored it's 1.2(0.050)=0.0601.2(0.050) = 0.060 kip/ft. When a question asks what the beam can carry in addition to its own weight, subtract that.

Worked Example The largest load on a braced beam
1 of 3

A W18X50 of 50 ksi steel spans 30 ft, simply supported, with full lateral support. Find the largest factored uniform load it can carry in addition to its own weight.

The ASD side

ASD keeps the loads at service level and puts the whole margin on the strength, as Rn/Ω≥RaR_n/\Omega \ge R_a. For these beams, the Handbook prints a single number, Ω=1.67\Omega = 1.67, beside ϕb\phi_b. Call it Ωb\Omega_b, for bending. The Handbook prints no ASD equation and Table 3-2 has no ASD column, so you divide the nominal strength yourself:

MnΩb=FyZx1.67\frac{M_n}{\Omega_b} = \frac{F_yZ_x}{1.67}

For the W18X50, Mn=50(101)/12=420.8M_n = 50(101)/12 = 420.8 ft-kip, so the allowable moment is 420.8/1.67=252420.8/1.67 = 252 ft-kip, against a design strength of 379. ASD isn't stricter. Its 252 is compared with service moments, which carry no load factors. The ratio, 0.90×1.67≈1.50.90 \times 1.67 \approx 1.5, is what 1.2D+1.6L1.2D + 1.6L comes to over D+LD + L when the live load is three times the dead load, and at that mix the two methods agree.

Checkpoint

A W12X40 of 50 ksi steel has full lateral support. What is its ASD allowable moment?

ft-kip

Recap

  • Steel yields without quitting, so a beam's moment grows past first yield, My=FySxM_y = F_yS_x, until every fiber sits at FyF_y: the plastic moment Mp=FyZxM_p = F_yZ_x.
  • The plastic neutral axis splits the section into equal areas, and ZZ measures the couple of the two halves: bh2/4bh^2/4 for a rectangle, 1.5 times SS, and 10 to 18 percent above SxS_x for a W-shape, whose flanges are near FyF_y already.
  • Mn=FyZxM_n = F_yZ_x needs a compact section and Lb≤LpL_b \le L_p, so the compression flange can't buckle sideways and twist the beam. A slab or deck fastened to that flange gives full lateral support.
  • ϕbMn=0.90FyZx≥Mu\phi_b M_n = 0.90F_yZ_x \ge M_u, divided by 12 for ft-kip, with 36 ksi for A36. Backward, Zx,req=Mu/ϕbFyZ_{x,\text{req}} = M_u/\phi_bF_y and wu=8ϕbMn/L2w_u = 8\phi_bM_n/L^2, less the factored self-weight.
  • ASD divides: Mn/ΩbM_n/\Omega_b with Ωb=1.67\Omega_b = 1.67, checked against service moments.