THE cinema in Karamay in north west China was packed with school children and their teachers last December when the fire broke out. An electrical short circuit started a small blaze in the ceiling, sending sparks cascading onto the stage. Eyewitnesses say that the stage curtains ignited, creating a ball of flame which engulfed the front rows of the audience. Within moments the entire hall was ablaze. More than 300 people died, most of them children, and at least 200 were injured.
The disaster bore all the hallmarks of a flashover: a small fire that suddenly erupts into a colossal inferno. But scientists modelling the way fires burn now think they understand what makes flashovers happen, and how they can be prevented. In future, architects should be able to use computer simulations based on this knowledge to design rooms that will minimise the risk of a flashover should a fire start.
Flashovers cost lives. The sudden increase that they cause in the size of a fire is the main danger to people in burning buildings, and to firefighters tackling the blaze, according to Dougal Drysdale of the Unit of Fire Safety Engineering at the University of Edinburgh. The most recent figures from the Home Office show that in 1993 fire claimed the lives of 719 people 鈥� 20 per cent less than in 1983. Drysdale attributes this improvement mainly to the use of smoke detectors. However, serious injuries rose by 60 per cent over the same period to 14 620, and Drysdale argues that there is still a need to understand more about how fires spread inside buildings.
Advertisement
Not all fires are difficult to model. For example, a bonfire in an open space will increase in size and temperature more or less steadily, then die away when the fuel is spent. But fires in confined spaces can be far more complex. In 1993, Drysdale and his colleague Alan Beard teamed up with mathematicians Stephen Bishop and Paul Holborn of University College London to model these fires. 鈥淲e wanted to provide a mathematical model that would help architects design new buildings that are intrinsically safe from the possibility of flashover,鈥� says Bishop.
In an enclosed space, a fire can develop in several ways. It may, for example, consume just one object and then die out without doing any more harm. But things become more dangerous when the blaze is powerful enough to heat up surrounding sources of fuel, such as tables, chairs and carpets. This may make them hot enough to catch light, or release flammable gases which mix with smoke and collect near the ceiling. A flashover can then occur if a build-up of heat from the fires causes the mixture of smoke and gas to ignite, increasing the temperature still further and igniting other items in the room. The speed of this process can be startling. In a matter of seconds, the temperature can soar by over 1100 掳C, with catastrophic results. This is what happened in the fire at the Bradford City football stadium in 1985, which left 56 people dead and 200 seriously injured. The heat from a relatively small fire caused bitumen on the roof to boil, releasing highly flammable fumes that collected under the stadium canopy. A flashover occurred when these fumes ignited, sending intense heat down onto the crowded terraces below.
The collaboration between Bishop and Drysdale could help prevent such tragedies happeening again. To construct their model, which uses a branch of mathematics known as non-linear dynamics, they began by dividing a theoretical burning room into three distinct zones 鈥� an upper smoke layer, a lower cooler layer and the region containing the fire. They then considered how certain key factors would vary from moment to moment. On Drysdale鈥檚 advice, they paid particular attention to the changes in the depth and temperature of the smoke layer and the radius of the fire.
The mathematical model gives a remarkably simple picture of how flashover occurs. Plotting the model鈥檚 figures for the temperature of the smoke layer against the radius of the fire produces an S-shaped curve (see Diagram). It can be interpreted like this: as the fire grows, the temperature in the room steadily increases up to the point where the curve folds back on itself. Bishop says this is when a flashover occurs: the temperature of the smoke layer jumps suddenly from a low value to the high one, without any increase in the radius of the fire.
The model predicts that the process can also work backwards as the fire is contracting and the temperature falls. At the point where the curve turns back on itself at the top of the S, the temperature can suddenly drop, and this aspect of the model could be used to answer questions about the best way to tackle fires after they have started. According to Drysdale, this could indicate the size and type of sprinklers or other fire suppression systems best suited to dealing with both small incidents and full-scale flashovers.
Burning by numbers
The model also shows that ventilation has an important influence on whether a flashover occurs. Good ventilation increases the supply of oxygen and therefore the rate of burning, but also helps heat to escape from a burning room. Bishop showed that a small change in the ventilation can radically alter the shape of the curve. This change can be represented in three dimensions with a surface 鈥� called a control surface 鈥� that varies smoothly in one area but folds back on itself in another (see Diagram). The properties of a particular room can be found by taking the appropriate section through this surface.
Last year, the team completed a series of experiments to test their ideas. At the Edinburgh fire safety unit they built 30 model rooms in which they could control the amount of ventilation, and the quantity of fuel they contained. After setting each room alight, they used video monitors and temperature sensors to chart the progress of the fire. Bishop says that the model successfully predicted which fires burn steadily before dying down and which would flashover.
The mathematical model can also be used to predict how changes in the amount of heat lost through the walls can affect the progress of a fire. 鈥淔eatures like the amount of ventilation and the thermal properties of walls can be controlled by architects while they are designing a building,鈥� Bishop says. The model has its limitations, however. For instance, it cannot predict how quickly a flashover will occur. Bishop is hoping to develop a far more advanced and user-friendly version of the model for architects and designers to use. Drysdale believes that models like this will be most useful for designing large public buildings such as airport terminals, in which the architect can control where the furniture goes in each room. 鈥淭he main danger in these buildings is that a fire could break out in a small room and flash over, creating smoke that pours into the large public areas,鈥� says Drysdale.
Paula Beever, who specialises in fire prevention at the engineering consultants Ove Arup & Partners, believes that such a program could be a powerful tool in the way houses are designed too. She points out that domestic blazes are responsible for most fire deaths. 鈥淭he model could help us to see how small changes in the design of the room could produce big improvements in safety.鈥�
It is also possible to cut down the risk of flashover in rooms that have been designed without the benefit of Bishop鈥檚 mathematics. If a single item does not produce enough heat to cause a flashover by itself, then the key is to prevent the blaze spreading to other combustible objects. In the early 1980s, Vyto Babrauskas, a fire safety engineer at the US National Institute of Standards in Engineering in Gaithersburg in Maryland, showed that a fire is unlikely to spread if it is more than a metre away from any other flammable object. This idea is useful in open-plan areas such as an office block atrium, where the furniture can be spread out. The same reasoning leads to the use of low ornamental brick walls to subdivide the floor space and separate flammable items.
Tremendous conclusion
A better understanding of the way fires burn can also help the firefighters who have the difficult job of putting them out. For example, a phenomenon known as backdraught can produce a similar result to a flashover, but has to be tackled in a completely different way. 鈥淭he problem is that there is tremendous confusion amongst firefighters in Britain,鈥� says Richard Chitty of the Fire Research Station in Hertfordshire, who is an expert in fire dynamics.
A backdraught occurs when a fire dies down because it has been starved of oxygen, but flammable gases continue to stream out of hot materials in the room. As the atmosphere cools, air is drawn into the room and this creates an explosive mixture that can be ignited by a single spark or a lingering flame. The resulting fire consumes all the flammable gases within a few seconds, but the heat this produces can easily ignite any remaining material in the room, and may even raise the temperature high enough to cause a flashover. The quick succession of these events is partly to blame for the confusion among firefighters.
Backdraughts can be prevented by ventilating the building to disperse any flammable gases. Conversely, when a flashover is the danger, the ventilation must be reduced to starve the fire of oxygen. Obviously, the wrong decision can be disastrous. 鈥淔irefighters must be trained to be aware of the dangers and recognise the warning signs,鈥� says Chitty.
Most firefighters in Britain do not get any formal training in dealing with fires in enclosed spaces because there is a shortage of training facilities, Chitty says. So they have to find out how to deal with these situations on the job, learning the tricks of the trade from more experienced colleagues. This worked well after the Second World War, when young firefighters learnt from those who put out numerous large fires after bombing raids, says Martin Thomas, an adviser to the Home Office at the Fire Experimental Unit in Morton in Marsh, Gloucestershire. The current generation of firefighters learnt much of their trade from wartime firemen. But he fears that when this generation retires, the experience will be lost. With fewer large fires to practise on, Chitty says the type and amount of training must be reassessed.
Another problem is that training is not coordinated throughout the country. Responsibility for firefighting in Britain is divided among 63 brigades, each responsible for training its own crews, and only the Essex and West Yorkshire services have facilities for training firefighters to deal with fires.
John Taylor is in charge of the West Yorkshire Fire Service Training Centre in Birkenshaw, near Bradford, and has spent the last four years developing a fire training chamber for his crews to practise in. It is modelled on facilities in Sweden, the only country that has a government-coordinated training programme to teach firefighters how to tackle enclosed fires. The chamber consists of a large metal shipping container lined with chipboard to act as fuel. The ventilation can be controlled from the inside or outside by vents in the roof and the sides. Up to six trainees, together with their tutor, enter the chamber, wearing protective clothing and breathing apparatus with built-in radio communication, and carrying firefighting equipment. An experienced firefighter stands at the back of the chamber, ready to defend them from the blaze should anything go wrong.
Flames in the sky
The aim is to teach firefighters to recognise how fires develop in enclosed spaces. For example, the chamber can be used to simulate the conditions that lead to a flashover. This is particularly important because in real fires flashovers usually occur before firecrews arrive, Taylor says. It also allows them to practise techniques for venting flammable gases, to forestall a backdraught. For example, in the technique known as the blowtorch, firefighters create a vent in the roof of a burning building to allow the hot gases to escape into the open air. There they ignite, sending flames roaring into the sky. In a real fire, this reduces the damage inside the building, and anybody trapped there is less likely to be burnt. For the trainees this is also a useful emergency option if the fire inside the chamber threatens to overwhelm them.
鈥淢y biggest problem is convincing firefighters that smoke burns,鈥� says Taylor. Smoke consists mainly of small particles of unburnt carbon, carbon monoxide and other flammable gases. Mixed with oxygen this combination can easily ignite. The amounts of smoke and oxygen in the mixture are crucial: low concentrations of smoke will not catch fire and if the smoke is too thick there will not be enough oxygen for it to burn. The colour of the smoke provides important clues. Black smoke is thick with carbon, while grey smoke indicates that the mixture contains more air 鈥� a potentially explosive combination.
鈥淣obody leaving the training chamber has any doubt that smoke burns,鈥� Taylor says. His methods are changing the way fires are fought. He points to a fire two years ago in the West Midlands, in which one fireman was killed and another was left with burns covering 45 per cent of his body. The accident happened when they opened a door into a burning room, allowing flames to escape and ignite smoke lining the ceiling of the corridor where they were operating.
Conventional firefighting wisdom holds that firefighters should get straight to the seat of a fire and put it out. Instead, Taylor teaches his students to separate the potentially explosive smoke from sources of ignition by sealing the fire in. The next step is to clear the smoke from the rest of the building, possibly with the help of a large fan.
This improves visibility inside the burning building, and so helps the search for any one trapped inside.
Prompted by incidents such as the West Midlands tragedy, the Home Office this year published a supplement to Manual of Firemanship, the firefighter鈥檚 bible. But Taylor feels that it does not go far enough: 鈥淚t still does not state strongly enough that smoke burns.鈥� In any case, he says, you can鈥檛 fight real fires with books. He is adamant that firefighters need to experience fires in properly constructed simulators.
For most of Britain鈥檚 50 000 firefighters, about of quarter of them part-timers, the chances of getting this training are slim. In March, Taylor demonstrated his simulator to 60 senior firefighters from across the country. For the moment, however, there are no plans to introduce similar training facilities elsewhere.