-
LOCATION:HOME >> NEWS >> Industry news
-
Industry news
The Reaction Lands on Six Millimetres of Web
A beam is checked in the middle and fails at its ends
Most of the arithmetic on a roof beam goes into the middle. A designer works out the bending moment at mid-span, picks a section modulus and stops. The ends get a line on the drawing that says bearing 150 mm and nothing else. Yet every newton the beam carries leaves it through those two short seats, and at that point the whole load is passing through the web, a plate between four and six millimetres thick on the Q235 H-beam listing. Cross that plate with a reaction it was never sized for and it folds sideways, not because the section was weak in bending but because a concentrated force has nowhere to spread. A column buckles along its length; a beam cripples at its seat. Only one of them is on the calculation sheet.
The size row is a mill envelope, not a section
The Q235 H-beam page prints web widths up to 900 mm beside web thicknesses of 4.5 to 6 mm, flange thicknesses of 8 to 64 mm, flange widths between 50 and 400 mm and stock lengths of 6 to 12 m, held to a one per cent tolerance. Nothing in that row states a bearing capacity. Pick a section inside it, 400 mm deep, 200 mm flanges, a 6 mm web, 13 mm flanges, and the beam and column duty described on the 50 by 150 page applies: it will take bending all day. What the row will not tell you is what happens at the 50 mm of seat where it sits down.
A welded H section gains no more from the joint.
Follow the reaction down to a bearing length
Take an 8 m simply supported beam under a design load of 12 kN per metre, roof cladding plus services hung off the bottom flange. That is 96 kN on the span and 48 kN at each end. The seat is 50 mm long. Through a 13 mm flange a point load spreads at roughly one to two and a half in each direction, so the length of web actually sharing the reaction is about 50 plus five times 13, or 115 mm. Everything else in the frame has been distributing load over metres; these 115 mm have to collect it again.
Average bearing stress is the wrong check
Divide 48,000 N by 6 mm times 115 mm and the average compressive stress is 70 MPa. Against the yield strength implied by the Q235 grade name that looks comfortable, and moving to Q345 would look more comfortable still. The comparison is meaningless. A 374 mm deep web at 6 mm is a slender plate, not a solid block, and a slender plate in compression leaves its plane long before the material reaches yield. The A36 H-beam page describes the web as thin but adequate for the vertical weight of the structure; that is a statement about global bending, not about 48 kN squeezed into 50 mm.
Treat the web as a strut, not a block
Model the strip under the seat as a short column of width 115 mm, thickness 6 mm and length half the clear web depth, 187 mm. Its weak-axis inertia is 115 times 6 cubed over 12, or 2,070 mm4. With E at 206,000 N/mm2 the Euler load is pi squared times E times I over 187 squared, roughly 120 kN. That is two and a half times the reaction, so the 6 mm web survives.
Now order the other end of the same row, 4.5 mm. Inertia falls to 873 mm4 because it goes as thickness cubed, and the Euler load falls to about 51 kN, barely above the 48 kN applied. A quarter less steel has removed 58 per cent of the seat capacity. Nothing about the grade changed, nothing about the span changed, and the mid-span bending check still passes. This is why a range on a mill sheet is not a specification.
What the grade does not buy
Yield strength does not enter the elastic buckling sum at all. Uprating the steel changes the load at which a stubby strut squashes, not the load at which a slender web bows. Restraining the seat works because it shortens a length rather than raising a strength. The grade table is worth reading for weldability, but it is the wrong page for this number.
A stiffener turns bending back into axial load
Weld a pair of 8 by 70 mm flat bars vertically over the seat, one each side of the web, and the reaction travels down 1,120 mm2 of steel in direct compression. At 235 MPa that is 263 kN, five and a half times the applied 48 kN, for about 3 kg of plate. The welding service and the press brake are where that detail is made; the punching station and the cutting line decide whether it fits the flange and clears the bolts. Pair the bars, keep them in bearing at both ends, and stop the weld short of the flange edge so the toe is not a hard spot in a fatigue detail.
Where the reaction actually comes from
On a portal frame the 48 kN is not a load someone drew. It is the sum of a roof bay: C and Z purlins in Q235B, knee braces, tie tubes and round-bar bracing all delivering into the primary member. The galvanized frame schedule names roof purlin C or Z, wall purlin C or Z, gutters and ordinary or high-strength bolts as separate lines. Each of those lines ends at a seat. The purlin lap, the cleat and the gutter bracket are all concentrated reactions on a thin web, and they are the ones that never get a capacity written beside them.
Cladding carries its own load, not the web's
The corrugated profile takes its stiffness from ridges and valleys acting in its own plane, so it spans between purlins; it is not a load-spreading plate for a beam seat. A single-skin sheet screwed at the crests delivers its tributary load as a line onto the purlin, and the purlin hands it to the beam as a point. The circular pipe used for bracing does the same in reverse. Assume spread at the seat and the check passes; measure the 50 mm and it does not.
What the drawing has to fix
Write the web thickness as one number rather than a range, give the bearing length and the flange thickness that spreads it, and mark every stiffener with its size, its weld and its fit. Add the purlin lap and cleat details on the same sheet, because those are seats too. State the protection separately, a hot-dip galvanized beam, Z275 coil for the cladding, or the processed galvanized parts for the cleats, since a coating is not a bearing stiffener. The light steel keel and the sheared cleats belong on the same order. None of this appears in a section modulus, which is exactly why it has to be on the drawing.