After a cyclone, the same sentence appears in every report: the roofs were lightweight, so they were blown away. It is a reasonable sentence and it is wrong in a way that matters, because it points at the material when the failure is in the connections.

A corrugated sheet weighs a few kilograms per square metre. A tiled roof weighs hundreds, a concrete slab more. If the problem were weight, the heavy roofs would be the safe ones. They are not. In a strong cyclone, the heavy roof is the one that kills people when it comes down, and the sheet roof is the one that leaves the walls standing.

Wind lifts, it does not press

Air moving over a roof surface accelerates, and where it accelerates its pressure drops. So the sheet has low pressure above it and, because the building is not sealed, near-normal pressure below it. The net force points up. On the windward slope there is a positive pressure pushing in, but the suction on the flat and leeward areas is stronger, and the design case that governs the fixings is the uplift.

This is why the failure looks the way it does. Roofs do not crumple; they peel. A sheet lifts at one edge, the wind gets under it, and the whole run goes in a few seconds — often taking the purlins with it, and sometimes the top of the wall.

Wind over a pitched roof produces suction above the sheet and pressure below it, so the net force is upward; the roof survives only if the load path from sheet to foundation is continuous NET FORCE IS UPWARD wind suction above low pressure over the sheet pressure below the building is not sealed THE CHAIN, AND ITS CHEAPEST LINK sheet fixing one screw purlin truss wall base The chain is as strong as its weakest ring, and the weakest ring costs a few rupees. Replacing J-bolts with U-bolts, and fixing every second corrugation at the edges, is what the national guidance actually asks for. It is not a stronger roof. It is a roof that is still attached.
A lighter roof is not a weaker roof. It is a roof whose survival depends entirely on the fixings — which are the last thing bought and the first thing skimped.

Why this material, and why it spread

Corrugated galvanised iron was patented in 1829 by Henry Robinson Palmer, an engineer to the London Dock Company, and it was exported to the empire almost immediately. Three properties made it travel: it was light, it stacked flat, and it could be installed by semi-skilled labour. In India it arrived with the railways and the tea plantations, and it replaced thatch and wooden shingles in exactly the places where a permanent roof had previously been expensive.

Read those three properties again, in the order of a cyclone. Light means it needs holding down rather than sitting there under its own weight. Stacking flat means it is laid as many small pieces, so there are many joints, and every joint is a place where wind can get underneath. Installable by semi-skilled labour means the person fixing it may not know what a load path is, and has no reason to.

The same three facts that made it the most successful building material of the nineteenth century are the three that make it fail in a storm. That is not a paradox. It is what happens when a product is optimised for transport and cost, and then asked to perform as a structural system.

1999 and 2013

On 29 October 1999 a cyclone crossed the Odisha coast with winds above 250 kilometres an hour and stalled over the land for about two days. The state's own damage memorandum records 9,885 deaths, 1.66 million houses damaged, and nine million trees uprooted. Several international assessments put the death toll between eight and ten thousand. Whichever figure is used, the order of magnitude is the same.

Fourteen years later, in October 2013, Cyclone Phailin made landfall on the same coast at around 200 kilometres an hour. Deaths were in the tens. Close to a million people were moved inland and into shelters before the storm arrived, in what was then the largest evacuation in the country for twenty-three years.

It is tempting to read that as buildings getting better. The record does not support it. Houses were rebuilt, shelters were built, embankments were strengthened — but the great majority of the difference came from people not being inside the houses when the wind arrived. A warning four days out, a siren network, drills held every year, and a population that had learned to act on the instruction did more than any change to the roof.

That is an uncomfortable conclusion for a piece about architecture, and it is the honest one. We learned to move people. We did not learn to tie the roof down. Both are cheap; only one requires a decision by every household separately, at the end of a build when the money is gone.

What the guidance actually says

The national technical document on cyclonic resistance, prepared by the structural engineering research centre of CSIR and issued through the disaster management authority in 2012, is a modest booklet. Its central sentence is not about materials at all: it says that because a low-rise building is an assemblage of walls, purlins, rafters and cladding, it is essential to ensure a safe load path from the roof to the foundation.

Everything else follows. Fix the sheet at every second corrugation, and more closely at the edges and corners, where suction peaks. Replace J-bolts, which hook over a purlin and straighten under load, with U-bolts that wrap it. Carry the tie from the truss into the wall, and from the wall into the footing. Keep the plan compact and symmetrical, because re-entrant corners concentrate suction. Do not build a large overhang you cannot tie back.

None of that is difficult, and almost none of it is expensive. It fails for a reason that has nothing to do with engineering: the roof is the last thing installed and the first thing value-engineered. A household that has spent its budget on walls and a floor does not want to hear that the screws matter more than the sheets.

The honest version of the lesson

There is a version of this story that is told often and is too neat: traditional builders understood the wind, modern materials forgot, and we should go back. The record does not support that either. Thatch roofs fail in cyclones at least as reliably, and a tiled roof in a 250-kilometre wind is a supply of projectiles.

What the sheet did was make a roof cheap enough for everyone to have one. It did not make that roof self-securing, and the fixings were left to the market, which bought the cheapest. The result is a building culture that can put a weatherproof roof over a family for the price of a week's wages, and cannot keep it there for the one night in ten years that matters.

That is worth looking at on any coastal road: not the sheet, which is fine, but the line of fixings along the eave. If they are spaced wide and hooked rather than bolted, the roof is a temporary structure that has not yet been told.

Sources and checks

  1. National Disaster Management Authority / CSIR-SERC Recommended Practices for Improving Cyclonic Resistance of Dwellings/Low-rise Buildings, August 2012
  2. Indian Standards IS 15498 — guidelines for improving cyclonic resistance of low-rise buildings; IS 875 Part 3 — wind loads
  3. Odisha State Disaster Management Authority Memorandum on damages caused by the super cyclonic storm of 29–30 October 1999
  4. UN ESCAP / OSDMA From Coastal Risk to Resilience — the 1999 to 2013 comparison in Odisha
  5. Royal Society Resilience to extreme weather, 2014 — the Odisha 1999 and 2013 case comparison
  6. Herbert, Gilbert Pioneers of Prefabrication: The British Contribution in the Nineteenth Century. Johns Hopkins, 1978

Last checked 1 November 2026 Yellowsea Huang Corrections