Building sway
A Socratic walk-through of building sway — reasoned out one step at a time, not lectured.
The question we started with
THE QUESTION #Why do engineers let a tall building sway rather than making it rigid?
Stand on the top floor of a tall tower in a gale and you can sometimes feel it: a slow, unhurried lean and return, taking several seconds each way. The instinctive reading is that something is wrong — that a proper building would not move. But this motion was designed in, calculated, and signed off. Why would anyone choose it?
Reasoning it through
REASONING #Begin with what stops motion. A building is a cantilever stuck in the ground, and stiffness against sideways loading comes from depth of structure. Make the tower taller without widening its base and the tip deflects far more for the same push; recovering the old stiffness means adding enormous amounts of material, which adds weight, which needs more material below it, and so on down. There is a point where the building is mostly structure and barely rentable. Absolute rigidity is not forbidden by physics; it is forbidden by economics.
Then ask a sharper question: what is the sway actually threatening? Not collapse. Strength requirements for wind are usually satisfied long before the building stops moving noticeably — the frame is nowhere near yielding at the amplitudes people complain about. The binding constraint is serviceability: cladding and partitions that must not crack, lift guide rails that must stay aligned, and above all the occupants.
And what do occupants sense? Not displacement — you cannot see the whole building translate, because everything visible moves with you. What the inner ear registers is acceleration: how sharply the motion changes. People start noticing accelerations far below anything of structural consequence, and start feeling unwell well below that again. So comfort criteria for tall buildings are written as peak accelerations for a wind of a given return period, not as "how far it leans". A slow, large sway can be less objectionable than a small, quick one.
That reframes the design problem. If the enemy is acceleration, you have three levers rather than one: reduce the force the wind puts in, absorb energy so the motion decays, or change the frequency so the motion is slower for a given amplitude. Only the last is really "stiffness", and it is the expensive one.
What is putting the energy in? Often not the head-on gusts. As air flows past a bluff shape, vortices peel off alternately from either side, and each departure gives a sideways kick. The shedding rate rises with wind speed, and when it approaches the building's own natural frequency the kicks arrive in step with the swaying — so the across-wind response can exceed the along-wind one. This is why so much tall-building design happens in a wind tunnel, and why towers are tapered, stepped back, notched at the corners, drilled with openings, or twisted. Not styling: a cross-section that keeps changing up the height sheds vortices at different rates at different levels, so the kicks never organise into one coherent push.
Where shaping cannot finish the job, add damping. The most visible device is a tuned mass damper — a large mass on the upper floors, on pendulum cables or bearings, its own natural period tuned close to the building's. Taipei 101's is a steel sphere of about 660 tonnes, hung where visitors can watch it. The counter-intuitive part is that it works by lagging. Tuned near resonance, the mass moves roughly a quarter-cycle out of phase with the floor beneath it, so it travels one way as the building travels the other, and its inertia pushes back against the motion in progress. Dashpots then turn that relative motion into heat. If the mass moved with the building, it would do nothing but make it heavier.
The analogy
THE ANALOGY #Think of carrying a shallow bowl of water across a room. Grip it rigidly and every step you take is transmitted straight in, and it slops. Let your wrist and elbow give — moving slightly out of time with the bowl, absorbing each jolt rather than passing it on — and the surface stays calm. The bowl travels further relative to your shoulder, and spills less.
Your arm is actively controlled by a nervous system reacting to what it sees, whereas a tuned mass damper is passive and dumb, tuned in advance to one frequency band — retune the building by adding mass, or drive it well off that frequency, and the damper's usefulness drops away.
Clarifying the model
THE MODEL #A few refinements connect these steps.
"Flexible" does not mean "weak". A tall building's frame is proportioned for strength first; flexibility is what remains once you decline to spend unlimited material on stiffness. Nor does flexibility help uniformly across hazards: in an earthquake a longer natural period generally reduces the force demand, because the ground shakes hardest at shorter periods, but in wind a more flexible tower responds more dynamically, not less. The two hazards pull in opposite directions and towers in seismic wind-exposed cities are compromises between them.
Damping is also the least certain quantity in the calculation. Tall buildings dissipate very little energy inherently, and the inherent damping ratio of a tall frame is hard to predict before construction and usually measured afterwards. That uncertainty is part of why supplementary dampers are attractive: a known, engineered quantity added to an unknown natural one.
One misconception worth dissolving: the sway is not the building "flexing under strain" in the way a bent paperclip is. It is a structure oscillating well within its elastic range, returning exactly to where it started — closer to a tuning fork than to something being damaged.
A picture of it
THE PICTURE #How to readEach point is a design move, placed by what it acts on and what it costs. Read left to right for mechanism: moves on the right change the wind loading itself by altering the shape presented to the flow, while moves on the left leave the loading alone and deal with the resulting motion. Read up and down for the material bill. The cheap, effective quadrant is bottom-right — corner and taper changes that cost geometry rather than steel — which is why aerodynamic shaping is usually settled first in the wind tunnel, and dampers or stiffening brought in for whatever motion remains.
What became clearer
WHAT CLEARED #Rigidity is not the goal, because the thing controlled is not deflection but acceleration, and the constraint is comfort rather than strength. Once that is clear the design moves fall out: change the shape so the wind stops organising its kicks, and add devices that lag deliberately behind the building's motion so its energy bleeds off as heat. A tower that sways slowly and dies away quickly is a better building than one beaten stiff at enormous cost.
Where to go next
ONWARD #- How wind-tunnel testing on a scale model produces load figures a code cannot supply.
- Why active and semi-active dampers, which sense and respond, are still rare compared with passive ones.
- Resonance itself, and why matching frequencies is such a powerful way to feed energy into a structure.
Key terms
TERMS #| Term | What it means |
|---|---|
| Serviceability limit state | a design requirement about usability and comfort rather than safety, such as sway acceleration or cladding movement. |
| Vortex shedding | alternating vortices peeling off either side of a bluff body, applying a periodic across-wind force. |
| Lock-in | the condition where the shedding frequency approaches the structure's natural frequency and the response grows sharply. |
| Tuned mass damper | a large auxiliary mass tuned near the structure's natural frequency, moving out of phase with it to oppose and dissipate motion. |
| Damping ratio | the fraction of critical damping in a structure, governing how fast free vibration dies away. |
Every term the collection defines is gathered in the glossary.