Article - Issue 17, October/November 2003
Fighting fire with science
Fire safety in the current age has a strong emphasis on prevention; however, it is inevitable that fires will occur and the possible effects can be devastating. Fire safety science uses a variety of computational techniques to allow engineers to predict these effects and make structures and the people within them safer if the unexpected should happen. Richard Chitty explains.
Fire is one of man's oldest technologies as well as one of his oldest enemies. He has learnt to control it for his own benefit, to fashion tools and to provide himself with warmth, light and propulsion. As well as being useful, fires are attractive: a flickering coal fire has a hypnotic quality and many cultures use fires in their celebrations. However, ‘unwanted’ fire is a devastating destroyer of lives and property. In the UK around 650 people die each year in fires and around 25 times that number are injured.1 The financial burden of fire to the UK is estimated to be just under 1% of GDP every year.
Many of the great scientists of the past have investigated fire in one way or another (see, for example, Michael Faraday’s ‘Chemical history of a candle’, 1860).2 However, it was not until the last decade that fire safety science became a recognised scholarly discipline in its own right.
Fire safety science is an extremely broad subject that encompasses aspects of many other disciplines. An understanding of the combustion processes involved in the fire itself involves physics and chemistry, as well as the engineering disciplines covering heat and mass transfer, and fluid dynamics. In addition, to understand the ‘impacts’ of fire on buildings or transport vehicles and on people, the disciplines of structural engineering, behavioural psychology and toxicology are all essential.3
Developing regulations for fire safety
Over the years, usually in response to major disasters, such as the burning of a city (e.g. the Fire of London in 1666), regulations have evolved to control the use of different materials, how buildings should be constructed and when fire protection systems (such as sprinklers) should be installed. Generally, these regulations were prescriptive, stating what would be permitted and what would not. This was supported by testing to identify materials and construction methods for particular applications. Fire ‘engineering’ was understood to be the process of knowing how to interpret and apply the appropriate rules. During times when technology was changing relatively slowly this approach proved to be successful. However, in an environment such as the present, when technology is rapidly changing, a prescriptive approach can inhibit the development of innovative design and application of new materials.
Introducing a performance-based approach
To address this, many countries around the world, including the UK, are now moving from strict prescriptive specifications to functional or semi-functional requirements. These requirements describe the performance that should be achieved by a design, but do not demand that particular methods are used to achieve those requirements. Where a performance-based approach is adopted it is the responsibility of the designer or contractor to show that the design meets the requirements.
In the United Kingdom, the Building Act, passed in 1984, provided a functional basis for regulations in England and Wales and introduced a simplified system with ‘Approved Documents’.4 The Approved Documents are based on the previous prescriptive regulations and illustrate one way of satisfying the requirements. For example, the requirement for section B4 is that:
‘The external walls of a building shall adequately resist the spread of fire over the walls and from one building to another, having regard to the height, use and position of the building’.
This requirement is intended to prevent fire spread from one building to another and ultimately the development of a fire involving a whole city. The Approved Document (and the documents it refers to) provide several methods to calculate the minimum distance between buildings. These depend on the size of windows and use of the building. There are also detailed clauses so that trivial cases, such as a small bathroom window on an otherwise solid wall, do not require detailed analysis. The basis of these methods is a calculation of radiative heat transfer between surfaces. To reduce the complexity of the calculations the methods given in the Approved Document contain many assumptions and simplifications. These usually result in conservative values for the minimum separation distance between buildings (i.e. the minimum distance is over-estimated). However, to satisfy the requirement of the Building Act the architect (or the engineer undertaking the detailed design) can return to ‘first principles’ and use more precise calculation methods. This may result in greater utilisation of the available land without increasing the possibility of fire spread from one building to another.
The fire safety aspects of the Building Regulations are intended to provide a standard of life safety that is acceptable to the public. Protection of the building structure itself may require additional measures and insurers may seek their own higher standards before accepting the insurance risk.
In cases where the variation from the method given in the Approved Document is small professional judgement may be sufficient to establish compliance. Where a more detailed justification is necessary then fire engineering methods can be used. These may be comparative assessments that show the proposed design is, at least, as safe as the prescriptive methods and current practice. Alternatively the assessment may be deterministic, using criteria that are relevant to the nature and use of the building, for example showing that the occupants of a building can be evacuated before the smoke from a credible fire event will reach the evacuation routes. Finally, a probabilistic approach can be used to show that the frequency of a fire-related event is acceptable for a given design (i.e. comparable to the frequency in existing, similar, structures).
A new approach to product testing
Until recently most fire test methods provided classifications or rankings of different materials or assemblies; these are compatible with a prescriptive regulation system. For example, a door can be tested in a furnace (BS476, part 22) and given a classification ‘FD30’ showing it had maintained its integrity (no gaps occur that would allow the spread of fire) for 30 minutes during the test. The prescriptive regulations, in Approved Document B, indicate where a ‘FD30’ door should be installed (e.g. between a dwelling house and a garage). Such a rating is not very useful in a performance-based analysis. In the example of the ‘FD30’ door, the performance is known only for the conditions prescribed in the test, which may not be representative of those that occur in a ‘real’ fire. Quantitative data that would allow the prediction of behaviour under other exposure conditions are not provided by the standard test. To overcome this, new test methods are being developed to replace the existing tests which assess the ‘reaction to fire’ and ‘fire resistance’ characteristics of a material or product at a more fundamental level. These test methods measure fire-related parameters such as rate of heat release and smoke production which can be used as input to numerical simulations instead of classifications or rankings which cannot (Figure 1).
Performing numerical simulations of fires in buildings presents a significant challenge. At present the movement of smoke and heat transfer into structures can be modelled reasonably reliably. Smoke movement can be predicted using a simple approach. An initial qualitative assessment is made of the anticipated behaviour (smoke from the fire rises to the ceiling, where it forms a layer, like the filling of an inverted bath) and empirical expressions can be applied, with general conservation principles for heat and mass. This gives spatially averaged values for the temperature and depth of a hot smoke layer. The models can be elaborated to include a more detailed calculation of the plume of gases from the fire impinging on the ceiling (a ceiling jet), the flow of gases under balconies and against walls and other features, if they are relevant to a particular scenario. There may also be prediction of gas concentrations and visibility. This is referred to as ‘zone modelling’ by fire engineers and forms the bases of many design methodologies for smoke control systems, for example those used in shopping centres (Figure 2).
Zone modelling provides a practical engineering approach that does not require considerable computational effort. The method includes many assumptions and depends strongly on the initial qualitative assessment of the problem. These tools (computer software, tables and graphs) should not be used as ‘black boxes’ without understanding the inherent assumptions and how they relate to the particular building being examined. The size of a compartment is a particular problem. The simple ‘inverted bath tub’ analogy works well for small ‘domestic’ size compartments, but it is not recommended when the floor area exceeds 2000 m2, as the heat losses to the structure may cause the hot smoke layer to de-stratify. This limit is a ‘reasonable value’ with limited scientific justification. It may be possible to use the method for a larger area if the building is well insulated, but the limit may be too high if the structure has little or no insulation.
Where the assumptions in these simple models are invalid or more detail is required then the computational fluid dynamics (CFD) technique is available. In principle this approach has few assumptions; however, some empirical sub-models are required to represent turbulence, combustion and heat transfer by radiation, etc. In addition, for fire simulations, particular consideration has to be given when specifying the source terms that represent the fire. In many cases a simple approximation using a volumetric heat source and production rate of ‘smoke’ is inadequate. It may not be possible to predict the volume of the flames a priori; the heat release rate and smoke production are dependent on the available oxygen, which in turn is dependent on the ventilation in the room containing the fire. The use of a simple combustion model has been shown to overcome these limitations successfully, but accurate prediction of production of partial products of combustion (e.g. carbon monoxide) and use of ‘real’ materials is currently difficult (Figure 3).
Neither of these approaches addresses the behaviour of the fire itself and both require a prescribed heat release rate. Progress is being made on the prediction of the burning rate of the fire, but this is a research activity and not yet a practical tool. The data from ‘reaction to fire’ test methods currently support the use of these models by providing the heat release rate data for the fire.
Knowing the behaviour of a fire and smoke movement in a building allows estimation of the time occupants have to escape or reach a place of safety before they are exposed to dangerous conditions. This is referred to as the available safe egress time (ASET).
For many applications simple models (zone models) for smoke movement can be used to calculate ASET. These may be built into a Monte Carlo framework to identify the relative hazardousness of different scenarios. Critical scenarios can then be investigated in more detail using the CFD approach.
There is also a need to draw on social sciences to examine the behaviour of people in a fire situation and be able to calculate how long it takes people to respond to a fire alarm and leave a building (Figure 4). This is referred to as the required safe egress time (RSET).
The objective is to ensure for a particular design that RSET is less than ASET so that people can leave the building before they are exposed to smoke and the effects of the fire.
Practical fire safety engineering
Most of the requirements in the Building Regulations state that something should be adequate to achieve a particular purpose. In the case of the prescriptive methods described in the Approved Documents, associated Standards and design guides, the interpretation of ‘adequate’ has evolved from many years of experience and may include significant safety factors. With a performance-based approach there will usually need to be some quantified definition of ‘adequate’ that is acceptable to all the relevant parties (architects, building users, building control officers, fire service, insurers, etc.). In addition they should also be satisfied by the methods used to demonstrate that the criteria have been achieved. A framework for this process and a systematic guide to fire safety engineering design has been developed as a new British Standard (BS7974, 2002).5
The fire modelling team at BRE have been developing these fire modelling tools for over 20 years and a key feature of this development has been validation.6 Large-scale fire experiments have been conducted specifically to provide comprehensive data sets for comparison with the results of numerical simulations. These have included long tunnels and enclosures ranging in size from typical domestic rooms through hospital wards to covered sports stadia. This has provided confidence in the use of these models for design applications, particularly in large spaces of public assembly such as shopping malls, atrium hotels and airport terminal buildings.
These models are now being built into an integrated framework so that designers of buildings can exploit new materials and novel design concepts in buildings with an acceptable level of safety.
Principal Fire Consultant, BRE LTD.
Richard Chitty is a Principal Fire Consultant at the fire group of BRE Ltd. After a period of conducting experimental work and developing instrumentation he has been part of the fire modelling team developing and using a range of fire models ranging from simple calculations to CFD techniques.