Computer modelling is sometimes the only way to demonstrate how a building and its occupants will react when a fire breaks out. We report on the Connell Mott MacDonald approach.
You've been obsessively feeding quarters into a slot machine at a packed casino when the fire alarm goes off. Does instinct tell you to abandon your stake and head for the nearest emergency exit, or do you push your luck in the hope of hitting the jackpot? The Grand Casino project in Melbourne is just one example that Stephen Logan, associate at Connell Mott MacDonald, puts forward to demonstrate that when it comes to fire engineering it's not necessarily a case of strictly designing to the Codes of Practice.

"As soon as you depart from a simple building design you've got to throw out the Codes and go back to first principles," reasons Logan. "And the authorities are starting to appreciate that more and more".

Approved document B: Fire safety, amongst others, sets out specific criteria for designing buildings, which is fine for most ordinary structures. "But when you get a non-ordinary building the Codes are a bit of a non-sense," he adds.

In such situations a performance based fire-engineering approach is necessary; assessing what is under threat and proving to the fire officer or building control that the solution will work.

Logan points to a sports stadium as a case in hand. The design incorporates unusually deep steel beams to support the seating tiers. "The Code would have you fire spray them, but think about it. Firstly it's an enormous piece of steel and no practical fire would ever heat it up enough to cause it to fail. Secondly it is in a vast concourse open to the sky, so the heat is dissipated," he explains. "Performance fire safety is about doing away with stupidity and waste of resource for no reason."

Performance based fire engineering relies heavily on computer modelling to provide evidence that the proposed design will work, with 3D computational fluid dynamics models commonly used to predict both temperature and smoke movement, and concentration, under fire scenarios.

Evacuation scenarios
Emergency evacuation is also part of this assessment. Over the last four years Mott MacDonald has been developing Simulation of Transient Evacuation and Pedestrian Movements (STEPS), a modelling program for predicting the movement of people under both normal and emergency conditions.

STEPS was born out of the company's railway division, where evacuation particularly of underground stations was a major issue. "However since 11 September we've had a lot more interest for traditional buildings," says Mott MacDonald's Nicholas Waterson. "People have become more aware of the fact that you really do have to be able to get people out of large buildings in a short time."

For evacuation from stations there are international standards such as NFPA130 and similar Codes which make assumptions about the way people move through spaces. "As stations and other buildings get more and more complex, the weakness of those Codes becomes more evident," explains Waterson. "People have been wanting better, more detailed answers. We started in a very simple way to try and simulate the way people move through buildings and it's developed from there."

STEPS models human behaviour within the building under a specific scenario, allowing evacuation simulations to be run which show the decisions that individuals make as they interact with their environment. These can then be compared with, say, smoke movement.

Individuals are assigned specific characteristics, such as levels of patience, fitness, age, size, gender and familiarity with the surroundings. Relationships can be attributed between individuals, as research has shown that families will rarely evacuate unless they are all together. STEPS also claims to be unique in that it is possible to change conditions in the midst of an evacuation, to simulate, say, smoke blocking an exit.

Specific outputs from running the program include information on evacuation times, flow rates through certain exits or locations, graphs of flow rates for paths, as well as numbers of people in particular zones.

Up until a year ago STEPS was used purely as a consultancy tool, being employed on projects including the new Terminal 5 at London's Heathrow and the International Centre for Life in Newcastle-upon-Tyne. However the company is now making it commercially available. "If you develop a software tool and then sell it, the people who are going to buy it are generally going to be your competitors," say Waterson. "But there are benefits for us, including the fact that it becomes more accepted as a general tool."

Meaningful visualisation of the results allows users, as well as clients and architects, to have more intuitive access to the information generated, allowing direct optimisation of the design. For example, in 3D visualisation the crowd is made up of individual figures and poor access to exits becomes immediately evident as the evacuation proceeds.

"We found over the last four years or so there weren't really the tools available to present results, particularly within buildings. As a response to that we developed our own visualisation tool, Model Viewer." says Waterson.

It is this experience that has also enabled Connell Mott MacDonald to adopt a unified modelling approach in its building services division.

Unified modelling
Over the last three to four months the company has been developing a translator that allows a single model, built say in IES' Model Builder program, to be exported for use in other modelling packages. This means the same model built for thermal and daylight evaluation in IES can also be exported and used for cfd evaluation in Phoenics, or to Fire dynamic simulator to establish what would happen in fire conditions. "We've basically managed to build one model and get it to do everything," says Logan.

Ultimately this should see the costs of modelling brought down and encourage clients to use it more – particularly when there is non-compliance with Codes.

With instances where it can take up to five weeks to build a single model, there are obvious benefits of being able to use it for more than one application. "We see unified modelling as a trend for the future," says Logan. That's probably not a bad bet.