The value of an holistic approach to fire safety in buildings has long been recognized in Australia. This may be because even urban dwellers in Australia are well aware of the potential for destruction and the threat from unwanted fires. [As this article is being written, the sun has not come up over Melbourne, Victoria, Australia, this morning because the sky is full of the smoke from bushfires that occurred nearby yesterday. On the same day, due in part to extreme weather conditions over a prolonged period, suburbs of Canberra, the national capital, were attacked by bushfires (wild fires), and over 400 houses were destroyed, four lives were lost, and many people were injured.]

However, it is comparatively recently that building code writers and enforcers have become aware, supportive, and accepting of risk oriented approaches to fire safety. Twenty years ago, if an attempt was made to discuss risk due to fire in buildings with a building code official, it was likely that the attempt would be summarily dismissed with a comment to the effect that there is no risk in buildings built to the (Deemed-To-Satisfy) building code. That is, because the building was built to code requirements, the occupants were totally safe. The fact that lives were lost, people were injured, and property was destroyed in such buildings escaped the attention of many officials or was excused by the assumption that the affected buildings were in some way noncomplying.


Despite this, other engineering-oriented people were beginning to probe the building fire safety problem from a risk-oriented perspective. And these efforts have resulted, among other things, in the development of CESARE Risk (a building fire-risk cost-assessment computer model also sometimes known as Fire Risk) substantially through the financial support of Australian building code authorities along with the adoption by the Australian Building Codes Board (ABCB) and the state Authorities Having Jurisdiction of a performance-and risk-oriented approach to building regulation development and reform.


CESARE Risk was developed based on initial research by Vaughan Beck begun in 1979 to use risk-assessment modeling to develop cost-effective building designs that would achieve acceptable levels of fire safety. 1,2 In 1989, a major project on Fire Safety and Engineering was undertaken at the Warren Centre for Advanced Engineering at the University of Sydney, which built on this modeling. 3 This project involved a large number of Australian participants and a considerable contribution from many invited international experts. Subsequently, the Fire Code Reform Centre (FCRC) was established through the cooperation and financial support of government, industry, and research organizations. The FCRC supported the development of CESARE Risk through a project aimed at the development of alternative fire safety system design solutions for the Building Code of Australia (BCA).


CESARE Risk is a fire-risk cost-assessment computer program that can be used to help designers and regulators make informed decisions on the suitability of various combinations of fire safety system components. CESARE Risk has been developed to enable quantification of the effect on fire safety in buildings of changes in fire safety system designs. CESARE Risk is applicable to apartment buildings, hotel and motel buildings, and aged-care facilities. The model estimates the expected risk of injury and to life, and the fire cost expectation. It is specifically intended to provide a basis for consideration of proposed or potential changes to Deemed-To-Satisfy provisions in building codes.


CESARE Risk consists of several linked computer programs using both deter-ministic and probabilistic calculations to estimate the numbers of building occupants killed and injured, and the extent of damage to a building for each fire scenario considered. Many of the programs are run many times, once for each fire scenario, building up a picture of casualties and damage. The results for each scenario are combined to produce an estimated risk of injury (ERI) and estimated risk to life (ERL) for the building occupants and an estimate of the fire cost expectation over the life of the building (FCE).


The ERI and ERL may be expressed in many forms, but are usually expressed as the expected number of casualties (injuries and fatalities) per 1,000 fires reported to the fire brigade, so that they may be compared directly with fire statistics obtained directly from U.S. and Australian fire data. The FCE may also be compared with fire data, usually on an average-cost-per-fire basis.


In CESARE Risk, a total of 384 fire scenarios are considered:

  • three fire types smoldering (three realizations), flaming (three realizations), and flashover (three realizations without fire spread beyond the room of fire origin [RFO] and three realizations with fire spread beyond the RFO).
  • for each of these, four combinations relating to the ventilation conditions in the RFO the door open and closed, and the window open and closed.
  • for each of these, four further combinations relating to the smoke and fire spread situation in terms of the apartment of fire origin door being open and closed, and the stair doors being open and closed.
  • for each of these, two further combinations for the occupants being awake and asleep.

The three realizations mentioned for each fire type are three levels of maximum burn rate for the smoldering fires and three rates of fire growth within the RFO for the flaming and flashover fires.

In theory, each scenario may have an infinite number of realizations. A simplified approach has been used in some parts of CESARE Risk to account for the many possible variations in some factors: continuous distributions have been replaced by equivalent three-point discrete distributions in the Fire Growth, Occupant Response, and Fire Brigade Intervention models. Thus, CESARE Risk probabilistically models fire growth and smoke propagation by repeatedly using deterministic models and probabilistic.


The model distributes occupants throughout the building in the apartments of non-fire origin (ANFO) in proportion to their percentage of the whole population. The occupant groups (or types) vary in mobility and responsiveness. In addition, the occupants are considered when awake and when asleep.


The response model considers the response of the occupants in the recognition and coping stages of an unintended fire, estimating the times at which occupants will be exposed to the cues of:

  • smoke
  • alarms (seven different types of alarm)
  • warnings by other occupants
  • sound of breaking glass

The probability of recognition of a cue and the probability of action thereafter were obtained from data obtained from research at CESARE. 4 The action can be to do nothing, to evacuate, or to investigate. Two types of occupant response times are calculated by the model:

  • the direct evacuation time
  • the investigation time

These times are dependent on the time to recognize the cues and the time for the occupant to start moving, the latter using a three-point realization. Thus, for each scenario, occupants evacuating are assigned three different possible evacuation times with associated probabilities.


Examination of coroner's records and fire statistics has determined the Apartment of Fire Origin (AFO) is the most critical apartment during a fire with regard to the ERI, ERL, and FCE. 5 Initial attempts to use human behavior data applicable to other apartments, obtained from interviews and examination of actual fires, was unsuccessful in that the delay times and probabilities of action were such that the resulting probability of fatalities was far higher than occurs in reality.


The evacuation model calculates the time for occupants to move from their apartments to the corridor, from the corridor to the stairway, and then downstairs to the building exit; the accumulation of carboxyhaemoglobin (COHb) in the blood; and the exposure of occupants to heat radiation. Critical levels of exposure define when incapacitation and death occur. The toxic gases considered are CO and CO2. Temperature is also used to define an occupant fatality condition.


The evacuation model classifies occupants as either being mobile or nonmobile. Nonmobile occupants are either disabled, trapped, incapacitated, or fatalities. Trapped occupants cannot evacuate by themselves because of smoke conditions and require assistance from the fire brigade.


The CESARE Fire Brigade Intervention Model is a simplified version of the Australasian Fire Authorities Council's (AFAC) Fire Brigade Intervention Model. It is a probabilistic model and takes account of all stages of fire brigade actions in attending the scene, fighting the fire, helping occupants reach the building exits, and rescuing injured occupants. It may be used in CESARE Risk or may be excluded from CESARE Risk runs if required, enabling estimation of the effect of fire brigade activities on the outcomes. The Economic Model is used to estimate the monetary costs and losses associated with fire safety and protection provisions, and fire events in buildings. The monetary components are aggregated into the FCE parameter.


In calculating the expected losses, the probabilities of smoldering, flaming, and fully developed fires are estimated, and the losses owing to fire damage, smoke damage, and water damage for each type of fire are calculated. Results from the overall fire spread model are used to estimate the losses from the estimated spread of fire in fully developed fires. Results from the smoke spread model are used to estimate the smoke damage from smoke spread in fully developed and flaming fires. Spread of fire and smoke to areas outside the AFO is considered for flaming and fully developed fires, whereas only smoke damage in the AFO is considered for smoldering fires. Water damage from fire brigade intervention in fully developed fires and sprinkler activation in flaming fires is also considered. The present value of expected losses is calculated over the whole life of the building.


Capital costs associated with fire protection including both active and passive features are also used in the calculation of the fire expectation cost, as are annual costs for maintenance and inspection.


CESARE Risk has been tested using an extensive range of sensitivity studies 6 in which results from CESARE Risk were compared, where possible, with available fire statistics.


A requirement for proposed code changes in Australia is that they do not increase the risk to building occupants, and it is in assessing this requirement that CESARE Risk is currently being used. For example, CESARE Risk has recently been used to assess the implications in relation to the risk to the occupants of possible changes to the BCA for a range of multistory residential occupancies.


Desirable improvements to CESARE Risk have been identified, and it is hoped that further development will occur in the future.


Ian Thomas is with Victoria University.



  1. Beck, V.R., "Outline of a Stochastic Decision-Making Model for Building Fire Safety and Protection," Fire Safety Journal, Vol. 6, No. 2, pp 105-120, 1983.
  2. Beck, V.R., "Performance-Based Fire Engineering Design and Its Application in Australia," Fire Safety Science-Proceedings of the Fifth International Symposium, pp 23-40, Hasemi, Y. (Editor), International Association for Fire Safety Science, 1997.
  3. Beck, V.R., et al, "Project Report" and "Technical Papers, Books 1 and 2," Fire Safety and Engineering Project, The Warren Centre for Advanced Engineering, The University of Sydney, Sydney, Australia, 1989.
  4. Bruck, D., and Brennan, P., "Recognition of Fire Cues During Sleep," Proceedings of the Second International Symposium on Human Behavior in Fire, pp 123-134, Interscience Communications, London, UK, 2001.
  5. Brennan, P., and Thomas, I.R., "Victims of Fire? Predicting Outcomes in Residential Fires," Proceedings of the Second International Symposium on Human Behavior in Fire, pp 123-134, Interscience Communications, London, UK, 2001.
  6. Thomas, I.R., and Verghese, D., "CESARE Risk: Summary Report," Fire Code Reform Centre and Victoria University of Technology, June, 2001.