In 1985, the roof of an indoor swimming pool in Uster, Switzerland collapsed without warning, killing twelve people. The roof had been in service for thirteen years. The steel that held it up hadn’t rusted through, buckled under an overload, or fatigued from repeated flexing. It cracked from the inside, in a way no visual inspection would have caught, because the thing corroding it was the air.
Every indoor pool produces chloramines — compounds formed when chlorine, added to disinfect the water, reacts with sweat, skin cells, and urea from swimmers. Chloramines are what people call "the pool smell." They off-gas from the water surface along with the warm, humid air rising above it, and in a pool hall without aggressive ventilation, that chloride-laden moisture doesn’t stay near the water. It rises, spreads under the roof, and condenses on the coldest surface in the room: the structural steel holding the roof up.
High-strength steel exposed for years to a humid film loaded with chloride ions is vulnerable to stress corrosion cracking — a failure mode where cracks initiate at microscopic pits and grow through the metal under sustained tensile stress, with almost no external sign until the member fails outright. At Uster, the steel tension rods supporting the roof failed this way, one at a time, each failure throwing more load onto its neighbours until the structure gave way. Investigators later found the mechanism was already well understood in materials science; it simply hadn’t been connected to natatorium design at the time. Since 1985, stress corrosion failures of the same type have brought down suspended pool-hall ceilings in Switzerland, the Netherlands, and Russia — over fifty deaths combined.
The physics is now well characterised, and the fix is airflow, not materials alone: keep the humid, chloride-laden air from ever reaching the structural steel. Modern natatorium design pulls exhaust air out at deck level, close to the water, before it can rise and pool against the roof — and keeps relative humidity at the steel below roughly 60%, the threshold ASHRAE guidance treats as the line between a dry structure and a slowly corroding one.
Zephyr’s simulation of a natatorium with this geometry shows both states side by side. Without deck-level extraction, humid air rises off the pool, spreads along the underside of the roof, and pools there — the same convection pattern that fed the Uster failure, holding relative humidity at the steel above 75%. Switch on exhaust intakes at deck level and the same simulation shows that humid layer pulled sideways and out before it ever reaches the steel, with relative humidity at the roof settling under 60% within minutes. The difference between those two states isn’t a stronger roof. It’s where the air is allowed to go — and that’s a design decision made before the building exists, not a repair made after.
Tell us the building and we’ll show you where the air goes.