Module 2 · Lesson 2.2
The actions that catch steel out
Drifted snow, restrained temperature and crane dynamics — three cases where the obvious calculation is not the governing one.
Why this matters
Three actions cause a disproportionate share of the trouble in steel buildings, and each for a different reason. Snow drift because a short intense patch is worse for a light member than a big uniform load. Temperature because a restrained member develops force rather than movement — and the force does not care how long the member is. Crane loads because a moving machine applies its load dynamically, and the amplification is real rather than a margin.
By the end of this lesson you should be able to
- Explain why drift governs light secondary members
- Compute the force in a restrained member and say what it does NOT depend on
- Distinguish a dynamic amplification from a partial factor
- Recognise that crane horizontal forces act above the shear centre
Drifted snow: a small load in a bad place
Snow does not settle uniformly. Wind moves it, and it piles up wherever the wind is obstructed — against a parapet, in a valley between two roof slopes, or behind a taller adjacent building.
The drifted intensity can be around twice the uniform value, over a length of a few metres.
For a concrete roof that is often unremarkable: the slab is heavy and continuous, and a local patch of extra load spreads out. For a purlin it is close to the worst thing that can happen. Purlins are light, closely spaced, and often continuous over several bays, so a short intense load in one bay produces more moment than the same total load spread across all of them.
The lesson generalises: a light structure is sensitive to the distribution of load, not only to its total. Heavier structures are dominated by their own weight and care less.
Temperature: force without movement
Heat a steel member and it wants to get longer:
ε = α ΔT, with α ≈ 12 × 10⁻⁶ per K
That is about 1.2 mm per metre per 100 K. Let it move and nothing happens structurally — it simply moves, and a movement joint accommodates it.
Restrain it and the strain has nowhere to go, so it becomes stress:
σ = E α ΔT
For a 100 K change that is 210 000 × 12 × 10⁻⁶ × 100 = 252 N/mm² — most of the yield strength of S275, from temperature alone, with no applied load whatever.
Now look at what is not in that expression.
The length does not appear.
A 2 m restrained tie and a 60 m restrained tie of the same section develop exactly the same stress and exactly the same force for the same temperature change. Almost everyone expects the long member to be worse, and it is not.
What length governs is the movement the member would have made if it were free — and that is what a movement joint has to accommodate. So length decides the joint, and the temperature change decides the force.
Worked example
A restrained tie in an exposed structure
Given
- Steel tie, 5000 mm² cross-sectional area
- Temperature range from −10 °C to +50 °C in service
- Erected at 15 °C
- Considered first as fully restrained at both ends
Find
The force from temperature alone, and how it changes with length.
Assumptions
- E = 210 000 N/mm², α = 12 × 10⁻⁶ per K
- Full restraint — the worst case, and the right place to start
Try it
Force, movement, and what each one depends on
Move the LENGTH slider and watch carefully. The movement tracks it exactly. The force does not move at all — which is the single most counter-intuitive fact about restrained thermal effects.
- Free strain α ΔT
- 0.0480 %
- Stress if fully restrained
- 101 N/mm²
- As a fraction of S275 yield
- 37 %
- Force if fully restrained
- 504 kN
- Member axial stiffness EA/L
- 88 kN/mm
- Fraction of force developed
- 53 %
- Force actually developed
- 269 kN
- Free movement over the length
- 5.8 mm
101 N/mm² if fully restrained — 37% of S275 yield before any applied load. Only 53% of the full-restraint force develops, because the restraint yields 1.9 times more than the rigid assumption allows. Thermal force is shared between the member and whatever resists it — which is why full restraint, though the right starting point, usually overstates the answer.
Things worth trying
- Drag the LENGTH slider across its whole range. The force bars do not move. The movement bar tracks it exactly. Length governs movement and has nothing to do with force — this is the fact almost everyone gets wrong.
- Set ΔT to 100 K. The fully restrained stress reaches 252 N/mm², nearly all of S275's yield strength, from temperature alone. Steel develops roughly 2.5 N/mm² per degree when it cannot move, which is Eα written as a rule of thumb.
- Now drop the restraint stiffness towards 5 kN/mm. Most of the force disappears, because the restraint gives way and the member relieves itself. Real restraints are rarely rigid, which is why full-restraint forces are usually a substantial over-estimate — and still the right starting point.
- Increase the area. The STRESS does not change at all — only the force. Thermal stress depends on E, α and ΔT alone, so you cannot reduce it by using a bigger section.
- Set the length to 60 m and the temperature to 60 K. The movement is over 40 mm. That is what a movement joint must accommodate, and no amount of extra steel will reduce it.
Crane actions: amplification is not a safety factor
A crane does not lower its load gently. Hoisting, travelling, braking and skewing all apply load faster than the structure responds quasi-statically, so the runway beam sees more than the static weight.
That amplification is a dynamic effect. It describes what the structure actually experiences, and it belongs to the action — applied before any partial factor, not instead of one.
The distinction matters because the two get conflated. A dynamic factor of 1.3 and a partial factor of 1.5 are not alternatives and do not overlap: the first says the beam really does see 30% more load than the static calculation suggests, and the second allows for the load being larger than assumed. Both apply, and in that order.
The hoisted load takes the larger dynamic factor, because it is what is being accelerated. The crane's own weight is already travelling with the machine.
And the horizontal forces are worse than they look
Transverse forces from acceleration, braking and skewing are modest — perhaps 10% of the vertical load. But they act at rail level, above the beam's shear centre.
So they produce torsion as well as horizontal bending. And the beam is usually an I section, which resists torsion only through the b t³/3 of its individual plates — very badly indeed, as Module 1 showed.
That is why crane beams so often carry a channel capping the top flange or a horizontal girder alongside. It is not to carry the horizontal force, which is small. It is to provide the torsional restraint the section cannot supply on its own.
Practice
A fully restrained steel member of 4000 mm² undergoes a temperature rise of 30 K. What force develops, in kN? Take E = 210 000 N/mm² and α = 12 × 10⁻⁶/K.
Practice
How far would a free 45 m steel member move under a 60 K temperature change, in mm? Take α = 12 × 10⁻⁶/K.
Practice
A crane has a static wheel load of 50 kN from its own weight and 90 kN from the hoisted load. With dynamic factors of 1.1 and 1.3 respectively, what is the dynamic wheel load in kN?
Check yourself
Why does a crane beam so often have a channel capping its top flange?
Summary
- Drifted snow can be twice the uniform value over a few metres, and light members are worst under short intense patches
- σ = E α ΔT — about 2.5 N/mm² per degree in fully restrained steel
- The thermal FORCE does not depend on member length; only the free MOVEMENT does
- So joints fix movement problems, and only restraint, area or temperature range fix force problems
- Real restraints are not rigid, and usually shed a large fraction of the full-restraint force
- Crane dynamic factors are real amplification applied to the action BEFORE any partial factor
- Crane horizontal forces act above the shear centre, so they produce torsion — which is what caps the flange
This is educational material. It uses simplified examples to teach principles, and must not be relied on for real design or safety-critical decisions. Module overview and checkpoint