CoLateral
Steel design

Web crippling and when you need a bearing stiffener

A beam that passes bending and shear with room to spare can still fail where it sits on its support. Web bearing is a local check, it is governed by the bearing length rather than the span, and it is the reason a perfectly adequate W-shape sometimes comes back needing a pair of plates welded either side of the web.

5 min read

Two failure modes, one clause

CSA S16-19 Cl. 14.3.2 covers what happens when a concentrated force is delivered to a flange and has to get into the web. There are two ways the web can give up, and they are physically different.

  • Web yielding is a strength limit. The load spreads out through the flange and the fillet into a finite length of web, and that length of web squashes.
  • Web crippling is a stability limit. The web is a thin plate loaded on its edge, and it buckles locally just under the flange before it ever reaches yield.

Slender webs tend to be governed by crippling, stocky ones by yielding. You do not get to pick: both are evaluated and the lower factored resistance Br governs.

Utilization = Pf / Br, where Br is the lesser of the yielding and crippling resistances

The dispersion length is the whole game

Both checks depend on N, the bearing length, because the load is assumed to spread from the bearing surface down through the flange into the web at a fixed slope. S16 expresses that as a dispersion added to N:

Load dispersion into the web
LocationEffective web lengthWhy the difference
Interior loadN + 10tThe load spreads in both directions along the web
End reactionN + 4tThere is no web beyond the end to spread into

Here t is the flange thickness plus fillet dimension the clause defines, not simply the flange plate. The practical consequence is blunt: an end reaction is roughly half as good as the same load applied in the middle of a span. Ends are where bearing problems live.

Where the load comes from

Pf is the factored concentrated force at that point. In practice it is one of three things, and the third is the one people forget.

  1. The end reaction where the beam lands on a column, a wall, or a bearing plate.
  2. A point load delivered by an incoming beam or a column landing on the top flange.
  3. A load applied through a connection that does not engage the web directly, such as a seat, where the flange has to carry it in bending before the web ever sees it.

The reaction case usually pairs with a bearing plate check on the concrete below, because the same force has to be acceptable to both the steel web above and the concrete beneath. Passing one says nothing about the other.

When the screen says you need a stiffener

If Pf exceeds Br, the web on its own is not adequate and the load has to be carried by something else. That something is a bearing stiffener: a pair of plates fitted between the flanges, in contact with the loaded flange, that take the force directly.

Cl. 14.4 then treats that assembly as a short column. Three things get checked.

  • Outstand local buckling. The stiffener plate cannot be so wide relative to its thickness that it buckles on its own. The limit is expressed as a width-to-thickness ratio against 200 divided by the square root of Fy.
  • Contact bearing. The end of the stiffener has to physically bear against the flange over enough area to deliver the load.
  • Cruciform column compression. The stiffener pair plus a participating strip of web is treated as a cross-shaped column between the flanges, with an effective length of about 0.75h.

The participating web strip differs by location for the same reason the dispersion length does. An interior stiffener draws on roughly 25 times the web thickness, an end stiffener on about 12, because at an end there is less web to draw on.

What a screening check leaves out

Web bearing is deliberately a local check, evaluated in isolation. The things it does not consider are worth naming, because a beam can pass bearing and still be in trouble.

  • Interaction with coincident shear and moment at the same section.
  • Web sidesway buckling, where the compression flange is not restrained against lateral movement at the load point.
  • Fatigue, which matters for a crane runway or any repeated load and is not covered by a static resistance check.
  • The weld design attaching the stiffener to the web and flange.

Section properties deserve a word too. Fillet dimensions and web thicknesses vary between production runs and between catalogues. If a bearing check is close, the value of t you assumed is worth verifying against the section you are actually specifying rather than a starter library.

The order to work in

  1. Get Pf at each concentrated load and reaction, factored.
  2. Establish N from the real detail, not an assumed value. A bearing plate, a seat angle, and a column cap all give different N.
  3. Check web yielding and crippling, and note which governs.
  4. If it fails, try increasing N before adding steel.
  5. If N cannot grow, screen a bearing stiffener and confirm the cruciform check controls the plate size.
  6. Detail the weld, the clip, and the fit-up, and have the connection reviewed.

Common questions

What is the difference between web yielding and web crippling?

Web yielding is a strength failure: the effective length of web under the load reaches its yield stress and squashes. Web crippling is a stability failure: the web buckles locally as a thin plate just below the flange, before yielding. Both are checked and the lower resistance governs.

Why is an end reaction weaker than an interior load of the same magnitude?

The load disperses through the flange into the web at a fixed slope. At an interior location it spreads both ways, giving an effective length of N + 10t. At an end there is no web on one side, so the effective length is only N + 4t, close to half.

Can I avoid a bearing stiffener by lengthening the bearing?

Often, yes. The resistance depends directly on the bearing length N, so extending the seat or the bearing plate raises Br without changing the section. It is usually cheaper than adding plates and welds, and it is worth trying before you detail a stiffener.

What effective length is used for the bearing stiffener as a column?

The stiffener pair plus a participating strip of web is treated as a cruciform column between the flanges with an effective length of roughly 0.75h, where h is the clear web depth. The participating web strip is about 25 times the web thickness at an interior location and about 12 at an end.

Does passing web bearing mean the connection is adequate?

No. Web bearing is a local check on the web at the load point. The weld, the bolt group, the supporting element below, coincident shear and moment, web sidesway buckling, and fatigue are all separate, and each has to be satisfied on its own.

Keep the check with the project

CoLateral runs these checks inside a workspace that holds the drawing, the assumption, and the review note that go with them, so the number can still be explained six months later.

Everything here is a design aid. Confirm the governing provisions against the code itself, and have a qualified engineer review the result.