Fire risk

Fire risk is the potential for a fire to cause unwanted, adverse consequences to human life, health, property or the environment. It brings together three questions: what could happen, how bad it would be, and how likely it is. A fire risk therefore depends both on the consequences of possible fires and on the likelihood that they occur. It describes the uncertainty of a loss, not a loss that is certain to happen. Every decision related to fire safety is in effect a fire risk decision, whether or not it is treated as one.

Fire risk combines event scenarios, likelihood and consequences. Based on Watts and Hall (2002), pp. 5-1–5-3 and Meacham (2002), pp. 5-154–5-156.

Zero risk is not an achievable goal, because no technology is completely reliable and natural events such as lightning and earthquakes can also cause fires. The practical questions are how large the remaining risk is and whether it is acceptable. Accepting a level of risk requires a value judgement, and people make such judgements in different ways. Fire risk is estimated from fire statistics, usage data and laboratory tests, and the result is then judged for its acceptability.

Hazard, consequence and probability

A hazard is a chemical or physical condition that has the potential to cause damage to people, property or the environment. An analysis of what could happen and how bad it would be, without considering how likely it is, is called a fire hazard analysis. A fire risk analysis adds the likelihood. A sequence of events that ends in a fire loss is called a scenario. For a whole building, the fire risk is the full spread of possible events and their consequences, not the outcome of one scenario.

Consequences describe how bad a fire would be if it happened. They range from deaths and damage to property to intangible losses such as business interruption, environmental damage and the destruction of cultural objects. A consequence can be direct, such as damaged property, or indirect, such as a company that is out of business for several days. It can also be objective, such as a replacement cost, or subjective, such as the pain and suffering of an injury. In building fire risk analysis, the main things of value are life safety, property, business continuity, the environment and heritage.

Risk is usually estimated by combining the probability that an event occurs with the consequence of the event if it does occur. The terminology is not consistent: a committee of the Society for Risk Analysis found 17 different definitions of risk. The SFPE Handbook also describes fire risk as the possibility of an unwanted outcome in an uncertain situation, where fire may cause loss or harm to something that is valued. A related term is the fire safety level, which the Australian Fire Safety Verification Method handbook (2020) calls the reciprocal of fire risk: when the risk to occupants falls, the safety level rises.

Frequency and probability

A frequency states how often an event occurs in a period, such as the number of fires per year. On its own it is not a probability, because the number of possible occurrences is also needed. In a hypothetical example, 20 fires a year in toasters that are switched on 20 million times a year give a probability of one in a million per start. Fire incident data therefore have to be combined with usage or exposure data, such as the number of appliances in use or the floor area of a type of building.

There are two views of probability. In the frequency view, probability is the proportion of the time that an event occurs in the long run. In the subjective view, it measures how strongly a person believes that an event will occur. Fires are relatively rare, and buildings and their contents change over time, so past losses are a limited guide to future ones. This is a strong argument for using subjective probabilities, based on current knowledge and judgement, in building fire risk analysis.

The amount of available data differs widely. For US homes there are very many recorded fires with limited detail, while for US nuclear power plants there are far fewer fires but much more detail on each. For a single existing building there are very few fires at all. In fire safety engineering, risk analysis is therefore mostly used for an application or a class of facility with a considerable fire loss history.

Typical values

The frequency of fire depends strongly on the use of a building. In British Standards Institution data from 2001, quoted in the International Fire Engineering Guidelines (2005), a dwelling had about 0.003 fire starts per year and an office about 0.006. Schools and industrial buildings had about 0.04, entertainment venues about 0.12 and hospitals about 0.3 fire starts per building per year. The guidelines warn that such values may have changed with new technologies and materials. Finnish data show that the number of fires per square metre falls as buildings get larger, while the number per building generally rises with floor area.

Risk studies often sort frequencies into broad classes. One example in the SFPE Handbook calls events anticipated if they occur more often than once in 100 years. Events are unlikely between once in 100 and once in 10,000 years, and extremely unlikely between once in 10,000 and once in a million years. Rarer events are beyond extremely unlikely. Frequencies should be counted for a given loss rather than for a single scenario: 30 scenarios that each occur once in 10 million years together occur three times in a million years.

Reliability values are typical inputs as well. In reported US home fires, roughly one-third of smoke alarms were not working, but a general study of home installations found only one-fifth not working. Data from unreported fires suggest that the main reason is that working alarms keep many fires so small that the fire service is never called. Reported US fires suggest that sprinklers do not operate in one-sixth to one-fifth of the fires large enough to activate them, yet their real reliability could be well over 90 %.

Individual and societal risk

Risk to people is often expressed in two complementary ways. Individual risk is the probability of injury or death to a person at a specified location relative to a fire or explosion source. It can be drawn as contour lines that connect points of equal risk, regardless of whether anyone is actually present there. In the Australian Fire Safety Verification Method handbook (2020), individual risk is the frequency at which a person may be expected to suffer a given level of harm from a fire in the building.

Societal risk takes into account the likelihood of fires or explosions that cause multiple casualties. It is often shown as an F-N curve, which plots the frequency of events that exceed a given number of fatalities, N, against that number. Such a curve requires the population at risk, the likelihood that people are present and the reliability of measures such as evacuation to be defined. Barry (2002) describes these measures for the chemical process industries, where most calculations of risk to people are based on fatalities.

Individual and societal risk answer different questions. Based on Barry (2002), pp. 5-210–5-211.

The two measures answer different questions. Individual risk shows how the risk to one person varies from place to place, while societal risk shows how often losses of many lives can occur. Single-value risk indices, such as an annual predicted fatality rate, are easy to understand and useful for comparing design options. A ranking of the events that contribute most to the total helps to focus risk reduction. In Australia, the Fire Safety Verification Method Standard 2022 accepts the deemed-to-satisfy provisions of the National Construction Code as reflecting societal expectations, including on individual and societal risk.

Acceptable risk

Acceptable risk is a value judgement applied to a chosen scale for measuring risk. It needs two prior decisions: how risk is measured, and whose views on acceptability count. For a practising engineer, acceptable risk is usually the risk acceptable to the client, while for an authority it is meant to be acceptable to society or to a particular community. A universally acceptable level of fire risk does not exist, because it depends on the problem and on the person judging it.

People judge the same risk in different ways. Experts mainly weigh the calculated reduction in fire risk, while ordinary people also weigh costs and other needs. Enforcers act on behalf of society, and managers such as builders are most directly affected by the costs. The reduced fire risk of a home sprinkler system, for example, appears to be valued less by the average citizen than the benefit of a new car. Risks that are objectively equal are also perceived differently: people prefer certain over uncertain, familiar over unfamiliar, voluntary over involuntary and ordinary over dramatic risks. Dying in a fire is perceived as worse than dying in a car accident, and most people who die in fires die at home.

Reducing fire risk may even increase other risks, as with the potential damage to the ozone layer from halon extinguishing agents. The goal that comes closest to zero risk is risk as low as is technically possible. In fire protection engineering, this may mean accepting the residual risk after all identified risk reduction measures have been adopted. A complication is that this residual risk may then be perceived as zero, although a non-zero risk remains. In Australia, workplace health and safety legislation also requires safe design, which a code of practice defines as eliminating risks to health and safety or, if this is not reasonably practicable, minimising them.

Building regulations rarely state an acceptable risk in numbers. The Australian Fire Safety Verification Method Standard 2022 notes that the National Construction Code contains no specific safety targets. A design assessed with this method must therefore be at least equivalent to the deemed-to-satisfy provisions. It adds that a full quantitative probabilistic risk assessment would ideally be used, but the tools and data for it are not included. The method instead permits a simple deterministic comparison of available and required escape times, with additional checks.

In the Netherlands, where no risk-based limit values exist, the residual risk can be expressed as an acceptable probability of failure. One way to set it is by comparison with a reference situation that follows the rules. In performance-based design, deciding which consequences matter is often described as setting fire safety goals, fire safety objectives and criteria.

Fire statistics as evidence

Fire incident data show how often fires start and how severe they are. In the United States, the national fire incident reporting system collects roughly a million fires a year, but only about half of the fires reported to fire departments. A separate survey of fire departments is therefore used to scale the data up to national estimates. Fires with an unknown cause must be allocated over the known causes. If they are ignored while the share of unknowns falls, every known cause can appear to grow although the total number of fires stays the same.

Not every fire appears in these statistics. Fires handled by the occupants or by private fire brigades are not reported, and fires in buildings with sprinklers or good detection are more likely to be controlled on site. Anecdotes show the range of fires that can happen but cannot be used to estimate rates or probabilities. Detailed databases that are not statistically representative can also mislead, so analysts should not give up statistical validity for detail.

Data on the reliability of fire protection systems are scarce. Expert estimates tend to rate reliable systems even higher and unreliable systems even lower than they really are. Inspection and testing records usually come from well-maintained properties, so they show better reliability than typical buildings achieve. They can also miss short-lived failures, such as a door blocked open.

Assessing and reducing fire risk

Fire risk analysis is a structured approach to decision making under uncertainty. A general procedure identifies the fire hazards, estimates their consequences and probabilities, identifies options to control them, estimates how each option changes the risk, and selects the protection. Methods of fire risk assessment range from checklists and written recommendations through risk indexing to probabilistic methods. Checklists and written recommendations do not quantify risk: a design passes if it follows them, and the residual risk is never evaluated.

A risk matrix sorts hazards by classes of consequence and likelihood, and each cell can record whether such risks are acceptable or how they should be reduced. Probabilistic methods are the most informative, because their results can be traced back through explicit assumptions, data and relationships. Common tools are event trees, which follow the possible outcomes after an initiating event in time order, and fault trees, which show the combinations of system states that lead to an outcome.

Risk can be reduced at several points in the chain from cause to consequence. The SFPE Handbook describes a causal sequence from an initiating event through its outcome and the exposure of people or property to the final consequences. The sequence can be interrupted at each of these points, by blocking the event, the outcome, the exposure or the consequence. In the chemical process industries, some measures reduce how often incidents occur, such as preventive maintenance, control of ignition sources and operator training. Others reduce how severe they are, such as gas and fire detection, fireproofing of steel and water sprays.

Risk reduction can act at several points in the fire scenario. Based on Meacham (2002), pp. 5-157–5-159 and Barry (2002), p. 5-211.

Measures only reduce risk if they work when they are needed. Typical reliability questions are whether the battery in a smoke alarm works, whether a sprinkler valve is open and whether a fire door is blocked open. A US Nuclear Regulatory Commission guide lists causes of sprinkler failure. Examples are a closed water supply valve, an empty tank and a fire pump that fails to start, for instance after a power failure. Failures can also share a common cause, as when a failure of the fire detection system stops a smoke control system from working as intended.

Uncertainty and variability

Fire risk estimates are always uncertain. Unwanted combustion is perhaps the least predictable common physical phenomenon, and uncertainty analysis should underpin the choice of safety factors and safety margins. This concerns the uncertainty of the engineer about how large a risk is, not the uncertainty of a potential victim about whether a fire loss will occur. A totally objective way to measure fire risk does not exist, because defining the problem, collecting data and combining information all involve subjective judgements.

Uncertainty and variability are different things. It may not be known how many people will be exposed, which is uncertainty, while the people who are exposed differ from each other, which is variability. Scientific uncertainties come from theories and models, from data and inputs, from limits of the calculation, from the level of detail and from how well a model represents reality. Further uncertainty lies in human behaviour, in perceptions of risk and values, and in decision making itself.

Calculations can hide much of this uncertainty. Fire models give single values, such as a detector activation time of 121 s, although the real time may be longer or shorter. One widely used egress model assumes that all occupants are readily mobile and start leaving as soon as they hear an alarm, which experience shows is often not the case. In a full uncertainty analysis, only the inputs that could change the decision on whether a design is acceptable, a change called switchover, need detailed quantitative treatment.

Design practice also handles variability and uncertainty in simpler ways. The Dutch national annex to Eurocode 1 (NEN-EN 1991-1-2/NB), for example, gives both average fire loads and 80 % fractiles for a more conservative approach. An 80 % fractile is a value that the actual fire load does not exceed in 80 % of cases. An escape time analysis keeps a safety margin between the available and the required escape time, because many of the inputs are uncertain. The more reliable the inputs are, the smaller the margin needs to be.

See also

• Event tree analysis
• Fault tree analysis
• Quantitative risk assessment
• Reliability of fire protection systems
• Uncertainty analysis
• Fire safety concepts tree

References

• Australian Building Codes Board. (2020a). Data sheets: Fire Safety Verification Method (Handbook annex, Print version 1.1).
• Australian Building Codes Board. (2020b). Handbook: Fire Safety Verification Method (Print version 1.1).
• Australian Building Codes Board. (2022). Fire Safety Verification Method Standard 2022.
• Barry, T. F. (2002). Quantitative risk assessment in chemical process industries. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), SFPE handbook of fire protection engineering (3rd ed., pp. 5-176–5-213). National Fire Protection Association; Society of Fire Protection Engineers.
• Hall, J. R., Jr., & Ahrens, M. J. (2002). Data for engineering analysis. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), SFPE handbook of fire protection engineering (3rd ed., pp. 5-65–5-78). National Fire Protection Association; Society of Fire Protection Engineers.
• Iqbal, N., & Salley, M. H. (2004). Fire Dynamics Tools (FDTs): Quantitative fire hazard analysis methods for the U.S. Nuclear Regulatory Commission Fire Protection Inspection Program (NUREG-1805). U.S. Nuclear Regulatory Commission, Office of Nuclear Reactor Regulation.
• Meacham, B. J. (2002). Building fire risk analysis. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), SFPE handbook of fire protection engineering (3rd ed., pp. 5-153–5-175). National Fire Protection Association; Society of Fire Protection Engineers.
• National Research Council of Canada, International Code Council, Department of Building and Housing, & Australian Building Codes Board. (2005). International fire engineering guidelines (2005 ed.). Australian Building Codes Board.
• Notarianni, K. A. (2002). Uncertainty. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), SFPE handbook of fire protection engineering (3rd ed., pp. 5-40–5-64). National Fire Protection Association; Society of Fire Protection Engineers.
• Van Rossum, E. (2027). Introduction to Fire Safety Engineering. Fire Safety Academy.
• Watts, J. M., Jr., & Hall, J. R., Jr. (2002). Introduction to fire risk analysis. In P. J. DiNenno, D. Drysdale, C. L. Beyler, W. D. Walton, R. L. P. Custer, J. R. Hall Jr., & J. M. Watts Jr. (Eds.), SFPE handbook of fire protection engineering (3rd ed., pp. 5-1–5-7). National Fire Protection Association; Society of Fire Protection Engineers.