THE first sign of impending disaster was a faint smell of smoke in the cockpit. Initially, the pilots put it down to an air-conditioning problem, but 13 minutes later they realised they needed to make an emergency landing. It was too late. Swissair flight 111 crashed into the Atlantic off Nova Scotia, Canada, on 2 September 1998, seven minutes after recognising the emergency. All 229 passengers and crew perished.
After the crash, it became apparent that a fire had broken out right above the pilots’ heads. Why then was the smoke smell so faint? Where did the smoke go? The nearest airport at Halifax was less than 20 minutes away. So, could the pilots have saved the plane if they had diverted immediately?
It has taken nearly five years to answer these questions. The Transportation Safety Board of Canada, which investigated the disaster, commissioned fire safety researchers at the University of Greenwich in London, to model the cockpit blaze. Their pioneering research, published this week, proves that the smoke was sucked out of the cockpit, concealing the fire. Taken together with the TSB’s report, published at the end of March, the research shows that the plane was doomed from the first whiff of smoke. It also raises wider concerns about the safety of electrical add-ons, such as seatback video, that are now common on many planes.
Advertisement
The flight 111 fire seems to have started in the electrical wiring for the in-flight entertainment system. Some sound insulation in the ceiling above the cockpit caught fire and, as smoke detectors were not required in this part of the plane, the first the crew knew about the problem was when they smelt smoke.
Ed Galea and his colleagues at Greenwich’s fire safety group, used a program they developed, called Smartfire, to model the blaze. Usually this program is used to model fires in buildings. “This is the first time a fire model has been used in an official air crash investigation,” says Galea. The program divides the cockpit and the surrounding ducts and spaces into 250,000 cells and calculates air pressure, smoke concentration, turbulence, air velocity and temperature in each cell for every second of the developing fire.
The TSB checked the results of the model against airflow tests, watching to see the way smoke travelled through the ducts. “These tests verified that the model was giving realistic results,” says Galea. The model showed that though smoke initially entered the cockpit, the normal turbulent flow of air there would have dispersed it, leaving only a faint smell.
Then as the fire took hold in the insulation, it destroyed a silicone cap on an air-conditioning pipe behind the cockpit. Once the cap was gone, the flow of air through the pipe fanned the flames, and drew the smoke away from the cockpit.
“By this time,” says Galea, “it was too late to do anything. The fire was driving itself.” A TSB simulation of the flight path to Halifax showed that the pilots could not have diverted successfully. They would have lost control of the plane at least 5 minutes before landing.
The TSB report also reveals a worrying catalogue of other defects in the aircraft’s electrical wiring. The in-flight entertainment system had been retrofitted to the McDonnell Douglas MD-11 plane. It had been directly wired into the plane’s electrical power circuit, a practice the TSB’s report calls a “latent unsafe condition”.
The report says that a survey by the US Federal Aviation Administration found that 10 per cent of retrofitted equipment had been installed “in such a way that prevented flight crew from removing electrical power from the entertainment system without interfering with essential [flight] systems”.
“Clearly this is a problem,” says Nigel Hughes, a specialist in aircraft electronics and a fellow of the Royal Academy of Engineering in London. “A single failure could result in a hazard to the aircraft.” The FAA and Britain’s Civil Aviation Authority have issued guidelines highlighting the risk.