Why Buildings Fail Sideways

Structural loads split into two families that behave completely differently. Gravity (vertical) loads are the building's own weight (dead load) plus everything it carries - people, furniture, stored goods, snow (live load). They act downward, always, and they flow down a comfortable path: slab to beam to column to foundation to soil. Columns are in compression, which is what they are best at, so gravity, for all that it must be respected, is the tame load.

Lateral (horizontal) loads are the wild ones. They act sideways, they reverse direction, and they attack the building in a way its vertical structure is not naturally shaped to resist. The two great sources are wind and earthquakes, and it is worth being precise about how they differ. Wind is an applied pressure: the air pushes on the exposed faces of the building, and the force grows with height (wind is faster higher up) and with the size of the face it presses on - so wind governs tall, slender, lightweight buildings. In India, wind loading is set out in IS 875 (Part 3), which maps basic wind speeds across the country.

An earthquake is different and more insidious. The ground does not push the building; it moves out from under it. The building's mass has inertia - it wants to stay where it is - so as the base is dragged sideways, the upper floors lag, and the building is effectively shaken like someone grabbing its ankles. Because the force comes from the building's own mass being accelerated, heavier buildings attract larger earthquake forces - the exact opposite intuition from gravity, where weight is just something to carry. Seismic demand in India is governed by IS 1893. The single most important habit this module builds is to stop seeing a building as a stack of weights and start seeing it as an object that must also resist being shoved and shaken from the side.

When a lateral force hits a building, the building sways - it leans and deflects sideways at the top, then springs back, then leans the other way. Some sway is unavoidable and perfectly safe; the structural questions are how much and how it is controlled. The two measures that matter are total sway at the top and, more importantly, inter-storey drift - the sideways movement of one floor relative to the floor directly below it.

Drift matters more than total sway because damage happens between floors, not at the roof. If one storey racks sideways relative to the next, everything spanning that gap is distorted: columns bend into an S-shape, masonry infill walls crack diagonally, window glass shears in its frame, cladding panels pop, and partitions split. A building can survive a large total sway if it is spread smoothly over many storeys, but a large drift concentrated in one storey is a warning of local overload and the first step toward collapse. Codes therefore cap inter-storey drift - IS 1893 limits drift to a small fraction of the storey height (of the order of 0.4 percent under the design earthquake) precisely to protect these brittle non-structural elements and to keep the frame within safe deformation.

There is also a comfort dimension that architects forget at their peril. In a tall building, wind sway that is structurally harmless can still make the top floors feel like a ship at sea, and occupants become nauseous and frightened long before anything is in danger. This is why supertall towers are often designed for acceleration and human comfort, not just strength - and why damping devices, which we meet later in this module, exist as much to steady a building for the people inside it as to protect it from breaking.

If there is one lateral failure every architect must understand viscerally, it is the soft storey (also called a weak or open ground storey). It is the single most common cause of building collapse in earthquakes, and it is almost always created by an architectural decision, not an engineering one.

Here is the trap. A residential building has masonry infill walls filling the frames on the upper floors - between and around the columns - and those walls, almost by accident, stiffen and brace the upper storeys enormously. But the ground floor is left open: for parking (the ubiquitous stilt floor in Indian apartments), for shops with large glass fronts, for a lobby, for a showroom. That open storey has only the bare columns to resist sideways force, while every floor above it is stiffened by walls. The result is a building that is rigid on top and flexible at the bottom - a heavy, stiff block balanced on a set of slender, unbraced legs.

When the earthquake comes, almost all of the sideways deformation concentrates in that one soft storey. The columns there are asked to absorb the entire drift of the building; they hinge, crush and fail, and the rigid block above simply drops straight down - the horrifying pancake collapse seen after Bhuj in 2001 and in countless earthquakes worldwide. The upper floors are often found nearly intact, sitting on a metre of what used to be the ground floor. The fixes are known and enforced by IS 1893 and IS 13920: brace the open storey with shear walls or cross-bracing, make the ground-floor columns deliberately stronger and more ductile, and never let an open storey be markedly more flexible than the one above it. An architect who wants open parking and a glass lobby must design that stiffness back in, not wish the problem away.

So far we have imagined a building leaning cleanly sideways, all its floors moving in the same direction together. That clean behaviour depends on a hidden balance: the building's centre of mass (where its weight effectively acts) needs to sit close to its centre of stiffness/resistance (where its bracing, walls and cores effectively resist sideways force). When those two centres coincide, a lateral push makes the building translate - slide sideways evenly. When they are offset, the push also makes the building twist about its stiff side, and that twist is called torsion.

Torsion is dangerous because it is uneven. In a twisting building, the columns and walls furthest from the centre of stiffness are dragged through a much larger sideways movement than the rest, so they are overloaded while others are barely working - the failure concentrates on the flank of the building that can least afford it. Torsion is usually the fingerprint of an irregular plan: an L-shaped or U-shaped footprint, a stiff service core or staircase shoved into one corner, shear walls all bunched on one side of the plan, or a building open on one street frontage and solid on the others. Each of these pulls the centre of stiffness away from the centre of mass.

The architectural cure is mostly about plan discipline, and it is squarely in the architect's hands. Aim for regular, symmetrical plans; distribute the stiff elements - cores, shear walls, braced bays - evenly and toward the perimeter where they resist twist most effectively (a core stuck in one corner is a torsion machine); and break a large irregular footprint into simpler, separately braced blocks with seismic joints between them. IS 1893 formally penalises plan and vertical irregularity, requiring more rigorous analysis and higher forces for irregular buildings - the code's way of telling architects that geometry is a structural decision, not just an aesthetic one.

Two further ideas complete the picture of sideways failure. The first is the P-delta effect, a vicious feedback loop that turns leaning into collapse. When a building sways sideways by a distance (delta), its entire vertical weight (P) is now acting slightly off-centre from the columns beneath it. That off-centre weight creates an extra overturning push, which makes the building lean a little more, which moves the weight further off-centre, which pushes harder still. In a stiff building the effect is minor, but in a flexible or already-damaged one it can run away - each cycle amplifying the last until the structure simply lies over. P-delta is why stiffness, not just strength, is a safety property, and why controlling drift matters so much.

The second is progressive collapse: the domino principle, where the failure of one key element sheds its load onto neighbours that were never designed to take it, so they fail in turn, and the collapse propagates far beyond the original damage. The classic warning is Ronan Point in London (1968), where a gas explosion knocked out one corner panel of a precast tower and an entire corner of the building peeled away, floor after floor, because there was no alternative load path and no ties to hold the structure together. The design answers are redundancy (more than one path for load to reach the ground), continuity and tying (physically stitching the structure together so it hangs together when one part is lost), and robustness (accepting that damage will happen and ensuring it stays local).

These are not abstractions; they are lessons paid for in lives. The soft-storey pancakes of Bhuj (2001) and of the Turkey and Syria earthquakes rewrote seismic codes. Ronan Point rewrote robustness rules worldwide. The collapse of buildings on soft alluvial soil in Mexico City (1985), far from the epicentre, taught the profession about resonance between ground and structure. Every clause in IS 1893, IS 13920 and the seismic parts of the National Building Code is, in effect, a scar - a way a building once failed sideways, written down so it need not fail that way again.