The King's Cross fire 1987 explained: the trench effect, the flashover and the Fennell investigation

In 1987, King's Cross was the busiest station on the Underground, used by over 250,000 passengers on an average weekday. Five lines met there, on five different levels, joined by a labyrinth of passages, stairs and escalators. At its centre was the tube lines ticket hall.

From it, three escalators, numbered four, five and six, ran down to the Piccadilly line. They were installed in 1939, rising 17.2 metres at an angle of 30 degrees. And they were largely made of wood: wooden steps and risers, plywood skirting boards, and plywood balustrades and decking, finished in varnish.

Escalator fires on the Underground went back decades. On Christmas Eve 1944, the Bakerloo line escalators at Paddington were completely gutted by fire. A review at the time counted 77 escalator fires between 1939 and 1944, and found the wooden M type escalators, including the MH type later used at King's Cross, particularly prone to them.

Most were put down to smokers' materials lighting dirt beneath the steps. After Paddington, cleaning was increased, and by 1948 water fog sprinklers had been fitted beneath more escalators, including the three at King's Cross. Automatic detection was considered as early as 1948, and smoke detectors were trialled from 1954, but they were never adopted widely.

Between 1958 and 1987, London Underground recorded over 400 fires and so-called smoulderings on its escalators.

In the three years before King's Cross, the warnings came closer together. In November 1984, a fire in a contractor's store at Oxford Circus filled the station with smoke. Fourteen people went to hospital, and 720 passengers had to walk through the tunnels to escape.

Smoking was banned on the Underground in February 1985. Then came serious fires on wooden MH escalators at Leicester Square, Green Park, Manor House, Holborn, Green Park again, and Bank. Their internal inquiries kept recommending the same things: heavy cleaning, smoke detectors, water fog controls outside the machine room, fire training for staff, and metal in place of wooden skirting boards.

Replacing the wood was refused on grounds of expense, with one engineer noting that the wood was chosen for its fire resistant properties.

Inside London Underground, these fires had a name: smoulderings. Fennell found that management believed fires were inevitable on the oldest and most extensive underground railway in the world, and treated them as an occupational hazard. With passengers and staff acting as fire detectors, they believed there would always be time to get people out.

The fire brigade disagreed. After a fire at Baker Street in August 1985, its senior officers wrote asking to be called at once to any fire on the Underground, warning that it could save lives. The rule book was not changed.

It still divided fires into those staff could put out with their own equipment, and those that needed the brigade. And because no one had ever died in an escalator fire, everyone was lulled into a false sense of security.

Escalator four had a hidden problem. Under each side of the steps ran a running track, where a gap of about 15 centimetres between the wheels collected grease, dust and fluff. The tracks were almost impossible to reach without dismantling the escalator, and at King's Cross they had probably never been cleaned completely.

About two weeks before the fire, gaps had opened between the steps and the skirting board, and 30 per cent of the metal fire cleats, meant to stop matches falling through, were missing. Tests later showed that one lit match in four, dropped onto that greasy, fluffy mixture, set it alight. A glowing cigarette never did.

The fluff worked like a candle wick.

At about 19:25, a lit match, probably dropped by a passenger leaving the station, fell through the gap on the right-hand side of escalator four, near step 48. It landed in the grease on the running track, and the grease began to burn. Because the escalator was still running, the fire was carried further up.

At about 19:30, a passenger saw smoke and a glow beneath the steps, pressed the emergency stop, and raised the alarm. A member of staff went to look, but he had no fire training, and he told neither the station manager nor the line controller. London Underground had no evacuation plan.

Two police officers in the ticket hall went to help. Their radios didn't work below ground, so one ran up to the street to call for help. The fire brigade was called at 19:34.

For the next ten minutes, the fire grew, and the station carried on around it. The police closed the Piccadilly line escalators and sent passengers out another way: up the Victoria line escalators, and through the tube lines ticket hall. They didn't know the station's layout, and believed it was the quickest and only safe route.

Under the escalators, an inspector saw the flames, but couldn't get close enough to use an extinguisher. He forgot the water fog equipment, sprinklers fitted beneath the escalator in the 1940s, and it was never turned on. Trains kept stopping and letting passengers off.

At 19:42, the first fire engine arrived, from Soho, with Station Officer Colin Townsley in charge. No one from London Underground met the crews, and the fire hydrant in the ticket hall was hidden behind a builder's hoarding.

Station Officer Townsley went to look. The fire was burning about halfway down escalator four, described as about the size of a large cardboard box, with flames licking up the handrail. He sent for firefighters in breathing apparatus and a water jet, and ordered the message: make pumps four, persons reported.

By 19:43, the fire had spread across both balustrades and the steps between them. Then, at about 19:45, it changed completely. In seconds, a jet of flame shot up the escalator trench and into the ticket hall.

It hit the ceiling and rolled across it, and the hall filled with intense heat and thick black smoke. A digital clock at the top of the escalators stopped in the heat. It read 19:45.

From the first alarm to the flashover, not one drop of water had been put on the fire.

Why did a modest escalator fire suddenly become a fireball? At the investigation, the experts disagreed. Some blamed the paint on the shaft's ceiling, where up to 20 coats had built up.

The breakthrough came from a computer. Scientists at Harwell modelled the flow of hot gases in the escalator shaft, and found something nobody expected. Once both balustrades and the steps between them were burning, air could no longer reach the uphill side of the flames.

So instead of rising towards the ceiling, the flames lay down in the trench formed by the steps and balustrades. Lying flat over the steps ahead, they radiated heat onto wood that was not yet burning. That wood gave off flammable gases and caught almost at once, so the burning length grew, the flames grew longer, and the fire raced up the 30 degree slope, faster and faster.

For the technically minded, here is why 30 degrees mattered. On a flat surface, a flame creeps against the air drawn in to feed it. Fire scientists call that counter-current spread, and it is slow.

Tilt the surface past a critical angle and the flame leans forward over the fuel ahead. That is co-current spread, and it is fast. Side walls make it far worse, because no air can be drawn in from the sides.

In laboratory trenches the switch came at about 20 degrees. The escalator was at 30. Harwell's computer model found the hot gases clinging to the floor of the trench, rushing up it at up to 14 and a half metres a second.

Every metre of wood that caught heated the next metre faster.

Higher up the shaft, the flow split in two. One stream rose out of the trench and spiralled up over the facia boards, advertisements and ceiling. It set the old paint alight, and made the thick black smoke that swept into the ticket hall and round the passages.

The other stream stayed in the trench and kept accelerating, until a jet of flame erupted into the ticket hall, hit the ceiling and was carried across it. Because those flames came from burning wood, they made little smoke of their own. Train movements may have played a part.

The air in the shaft was calculated to have switched from flowing down at about 1.75 metres a second to flowing up at about 3.25, just as the fire lay down. But once the trench effect took hold, the trains mattered far less.

The theory was tested. In January 1988, investigators dropped a lit match into an undamaged section of escalator four, with its grease still in place. A glow could be seen from above after 2 minutes 17 seconds, flames down the side of the risers after 3 minutes 45, and flames touching the underside of the decking after about 7 minutes.

Glowing cigarettes dropped the same way never lit it. Then the Health and Safety Executive at Buxton built scale models of the shaft and ticket hall, first at one tenth and then at one third of full size. Fire on the model escalator lay down in the trench, raced up it, and erupted into the model ticket hall, just as the computer had predicted.

It became known as the trench effect, in the words of one adviser, a newly discovered phenomenon, not anticipated even in expert circles.

The flashover killed 31 people, and injured many more. Very few of those in the ticket hall at that moment survived. Among the dead was Station Officer Colin Townsley, in charge of the first fire engine from Soho.

He was found at the foot of the steps to the Pancras Road entrance, beside a badly burned passenger he had been trying to help to safety. Fennell wrote that he died a hero's death, giving his life in an attempt to save another, and he was awarded the George Medal after his death. Fennell also praised Police Constable Hanson, whose presence of mind and courage must have helped many people to escape.

And he saluted everyone, passengers, staff and the emergency services, who helped others that night.

The flashover split the station in two. Crews on the surface believed those below were trapped or dead, while those below had no idea what was happening above. Below ground, a fire officer and a police officer fought the fire with a hose from the bottom of the escalators.

Staff flagged down Victoria line trains that had been told not to stop, and the trains carried passengers away. An injured passenger was carried to the emergency exit towards Pentonville Road, only to find its gates locked. The last of them was not opened until 20:17.

On the surface, the fire brigade had no plans of the station until about an hour after the flashover, and asked a British Rail manager to sketch one. Over 150 firefighters attended. The fire was surrounded at 21:48, and the stop message came at 01:46.

One victim was not identified until 2004.

Five days after the fire, the government appointed Desmond Fennell QC to hold a formal investigation. It sat for 91 days in 1988, and reported that November. Fennell found that the staff had not been adequately trained, there was no plan for evacuating the station, communications equipment was poor or not used, and there was no supervision.

No one person at London Underground was responsible for safety overall. Each director thought safety in his division meant mainly the safety of staff, and the escalators fell between departments. The company had a blind spot: it believed fires were inevitable, and that a fire on a wooden escalator would never endanger passengers.

Fennell said that ignored the unpredictability of fire, and the danger from smoke, which he called almost certainly more deadly than fire.

His central message was about managing safety. No passenger transport system, he wrote, can have a fire policy based on fire precaution. It must be based on fire prevention.

He called for a managed safety programme to find and remove hazards, and a safety audit, so that safety could be measured as strictly as a financial audit measures money. He found the Railway Inspectorate's relationship with London Underground too informal, and the law on fire certification for underground stations uncertain. He made 157 recommendations, 33 of them most important.

The government announced 266 million pounds of Underground safety spending over three years, and the chairman of London Regional Transport and the head of London Underground both resigned.

Change followed quickly. Fennell required the wooden skirting boards, balustrades, decking and advertisement panels on every escalator to be replaced with metal by July 1989. He called for heat detectors in every escalator, linked in time to sprinklers or water fog, for proper cleaning, and for six or more escalators to be renewed every year.

The smoking ban was extended to every part of a station below ground. London Underground's instructions now required staff to call the fire brigade at once to any suspected fire or smoke. Fennell also called for staff training and evacuation plans, police radios that worked with the fire brigade's, station plans kept where the brigade could reach them, and joint exercises between the emergency services.

The law changed too. The Fire Precautions (Sub-surface Railway Stations) Regulations 1989, made under the Fire Precautions Act, gave underground stations their own fire safety rules, covering escape, firefighting equipment, detection and warning, fire resisting construction, materials, and staff training. In England, they were replaced in 2009 by regulations made under the Fire Safety Order, which still apply.

Stations with an enclosed underground platform must be kept clear of combustible rubbish. Walls and ceilings in public areas must be of limited combustibility, with surfaces that resist the spread of flame. Staff need basic fire instruction at least every seven months, with fire drills every six months.

And at least two staff must be on duty whenever the public has access.

The lessons reach far beyond the Underground. Small fires are warnings, not routine: every escalator fire that went out without harm made the next one feel safer. Housekeeping matters: this fire started in dirt that had built up for years where nobody could see it.

Call the fire service straight away, and meet them with someone who knows the building and has its plans. Train staff, and have an evacuation plan that has been practised. Make sure people can communicate, above and below ground.

Never forget smoke: within seconds it filled the ticket hall, and very few people caught there at the flashover survived. And manage safety from the top, with clear responsibility, and measure it as carefully as money. The same ideas run through fire risk assessment today.