Fire Safety Journal

The September 11th 2001 impact on the World Trade Centre (WTC) resulted in one of the most significant evacuations of a high-rise building in modern times. The UK High-rise Evacuation Evaluation Database (HEED) study aimed to capture and collate the experiences and behaviours of WTC evacuees in a database, which would facilitate and encourage future research, which in turn would influence the design construction and use of safer built environments. A data elicitation tool designed for the purpose comprised a pre-interview questionnaire followed by a one-to-one interview protocol consisting of free-flow narratives and semi-structured interviews of WTC evacuees. This paper, which is one in a series dealing with issues relating to the successful evacuations of towers 1 and 2, focuses on cue recognition and response patterns within WTC1. Results are presented by vertical floor clusters and include information regarding cues experienced, activities prior and subsequent to occupants first becoming aware that something was wrong, perceived personal risk, time taken to respond and the inter-relationships between them. The results indicate differences in occupant activities across the floor clusters and suggest that these differences can be explained in terms of the perception of risk and the nature and extent of cues received by the participants.

We have tested a linear regression model to identify significant predictors of pre-evacuation delay in a sample of evacuees enrolled in the World Trade Center Evacuation Study. We have found that pre-evacuation delay was greater when there were more environmental cues, more seeking out of information, and more pre-evacuation actions. Additionally, higher perceived risk was predictive of shorter pre-evacuation delay times. These findings are compared and contrasted with an analysis of participants in the National Institute of Standards and Technology investigation of the World Trade Center disaster, recently reported by Kuligowski and Mileti (2009). Both studies reported factors associated with pre-evacuation delay that were similar to those associated with community evacuation. Additionally, we found that greater knowledge and greater emergency preparedness were associated with greater perception of risk. Greater emergency preparedness was negatively related to pre-evacuation delay within World Trade Center Tower I, but within World Trade Center Tower II, the relation between emergency preparedness and pre-evacuation delay was positive. These findings have implications for training of occupants of high-rise buildings.Highlights► Emergency preparedness and knowledge of the building both predicted perceived risk. ► Lower perceived risk was associated with more high-rise pre-evacuation delay. ► Predictors of human behaviors in high-rise and community evacuations are similar.

The increasing accessibility of buildings to people with disabilities requires that buildings are also designed and managed to provide accessible means of escape for all. In so doing, it is important to understand the capabilities of building occupants with disabilities to evacuate and their interaction with others. This is particularly important in high-rise buildings where occupants’ primary means of accessing upper floors, i.e. lifts, may not necessarily be the preferred route in a fire emergency.

The intention of this paper is to demonstrate the feasibility of a performance-based solution for providing adequate life safety levels for the occupants of an industrial warehouse. Focus was given to the evaluation of the performance of a smoke venting system and the exit distribution in an industrial warehouse. Quantitative analyses of the available safe egress time and the required safe egress time were carried out using both a simple calculation and a computer modelling technique. The study included a sensitivity analysis to investigate multiple fire scenarios involving variations in design fires, ventilation conditions and number of exits.Computer simulations of fire growth and smoke spread in a large warehouse with a proportionally large ridge vent revealed somewhat interesting smoke movement patterns. The procedure for determining tenability in a well-ventilated high ceiling space is discussed. The result of the study demonstrated the acceptability of the alternative design solution against the established criterion. The limitations of the modelling technique and future research topics are also discussed.

Highlights are given of forty presentations and the concluding general discussion at the International Life Safety and Egress Seminar held November 23 and 24, 1981, at the University of Maryland. The seminar provided a timely opportunity for research peers, educators, designers, consultants, and standards writers to discuss egress and life safety. Participants, numbering over fifty, came chiefly from the United States, but Australia, Britain, Canada, Japan, Sweden, and New Zealand were also represented.

The meteorologic and topologic causes of the 20 October 1991 Oakland Hills conflagration are described here qualitatively. A 3 GW example fire, 100 m in diameter, is used to show the impact of the 10 m/s wind and strong inversion layer at 600 m that existed on 19 October. It is concluded that the dry, high speed NE wind, coupled with the inversion layer and the local topography channeled the hot products of pyrolysis and combustion, along with flaming debris, through a high fuel load region downwind and downslope of the initial fire, thus causing the unusually rapid initial fire spread and consequent conflagration.

In this article the various developments in the field of fire safety regarding the single European market, foreseen by 1992, are reviewed. After a brief discussion on the prospects and the needs of a European harmonization, the subject is approached from two points of view: formal aspects and technical aspects. Under ‘formal aspects’ the essence and probable impact of some European directives are discussed and a summary is given of the various initiatives which are taken to arrive at harmonized technical specifications. Under ‘technical aspects’ both scientific and practical facets of European harmonization in the field of fire safety are reviewed. The role played by Professor Pettersson in this process, is highlighted.

A cross-sectional study was undertaken in Aotearoa New Zealand to investigate the relationship between socioeconomic deprivation and risk of an unintentional fatal domestic fire incident. Addresses of unintentional fatal domestic fire incidents were geocoded to small area (census meshblock) level and analysed with the New Zealand index of socioeconomic deprivation. Fatal unintentional domestic fire incidents occurred disproportionately in dwellings in the most socioeconomically deprived meshblocks. Annual rates of fatal unintentional fire incidents per 100,000 households in the most deprived decile were significantly higher than rates in the least deprived decile (RR 5.6, 95%CI 1.9–16). Strategies to prevent fire related deaths must overcome barriers to household fire safety in population groups experiencing increased risk, including the socioeconomically deprived, seniors, and ethnic minorities. Specific intervention strategies relevant to risks associated with socioeconomic deprivation include improving quality and affordability of housing; increasing prevalence of installed and functioning smoke detectors; and regulation of specific characteristics of cigarettes to reduce risk of ignition from abandoned heat sources. Substantial progress awaits reduction of the underlying socioeconomic determinants of disadvantage.

Fire safety and protection facilities to satisfy current regulatory requirements are a significant component in the cost of many buildings. With the development of systematic approaches to building fire safety and protection, designs based on a fire safety engineering methodology can offer substantial cost savings while achieving satisfactory levels of life safety. An outline is given of recent initiatives in Australia towards the development of a systematic approach for the design of cost-effective fire safety systems in buildings. This systematic approach is based upon a risk assessment framework. While substantial progress has been achieved, considerable research, development and technology transfer must be undertaken before this systematic approach can be used as a routine procedure for design. However, some important applications, based on such an approach, have recently been undertaken.

London is a large capital city with a population of approximately seven million people. It shares many problems with other large cities around the world, including deaths due to fire. Many of these fire deaths can be linked to social problems such as poor housing, loneliness, illness, etc.Data from the London Fire Brigade Real Fire Library—a unique database of information collected from real fire incidents by dedicated teams of fire investigators operating in the Greater London Area has been used to obtain a range of statistics about fatal fires and fire death victims for the 5-year period from 1996 to 2000. Most deaths occurred in unintentional dwelling fires. The statistical information has therefore been analysed to identify the main factors involved as to why people die in unintentional dwelling fires and see what lessons can be learnt from these deaths.Common risk factors identified in the unintentional dwelling fire deaths investigated include smoking, alcohol, old age, disability, illness, living alone, social deprivation and not having a working smoke alarm fitted. Comparisons are also made with the results found from other studies and measures for preventing unintentional dwelling fire deaths are examined.

The paper presents a stochastic model for the interaction between the spread of untenable conditions and occupant egress. Safety is measured by the expected number of deaths. The building is represented by a network for modelling fire spread and by another network for modelling occupant egress. A major innovation is the introduction of the concept of discrete hazard function. It allows the interaction between the various factors involved in the spread of untenable conditions and occupant egress to be taken into account at the time of their occurrence during simulations. Two small-scale examples are worked out in detail and flowcharts for full-fledged programs are given.

In the case of fires in chemical warehouses very toxic fire effluents may be generated. Knowledge of the nature and the amount of the various combustion products is rather limited, and therefore a study has been initiated. This paper describes results obtained from combustion experiments with selected pesticides, solvents, fertilizers and polymers. The substances are: Lindane, MCPA, dimethoate, azinphos-methyl, parathion-methyl, chlorobenzene, ammonium nitrate, polypropylene, polystyrene, PVC and nylon. The DIN 53 436 furnace has been used for all experiments. The experiments have been carried out under various conditions in order to approach the simulation of different fire types. Inorganic combustion products such as carbon dioxide, carbon monoxide, nitrogen oxides, sulphur dioxide, hydrogen chloride, and hydrogen cyanide have been quantified and organic combustion products have been identified.

The use of sacrificial ablative heat shields has proved effective in protecting spacecraft on earth re-entry. However, the use of ablative coatings in the protection of structural steel appears to be limited. This paper uses a mathematical model to assess the likely effectiveness of using such a coating on steel beam sections in furnace tests. In particular, the important coating properties are identified and quantified in order to maximise the failure time (defined as the time taken for the average steel temperature to reach a preset value such as 823 K). The model produces interesting results and demonstrates that it should be possible to use coatings of moderate thickness (∼10 mm) to attain failure times of the order of 1 h, provided that other critical coating properties such as the coating Biot number, Stefan number and diffusive timescale fall within specific ranges.

When a heat release rate limit for a consumer product is set by a regulatory agency, it is of interest to know whether small excursions above that limit, such as may occur due to production line variability, represent a disproportionate increase in fire hazard. This paper presents a methodology to examine this issue. The heat release rate curve of the object is described by a Gaussian time variation; a perturbation peak, also Gaussian, is added to this main peak. The impacts of the perturbation peak on the build up of hazardous conditions in a room fire (where the object is the only item burning) and on the threat of ignition of secondary items are examined. For the peak heat release rate domain studied here, only the ignition threat is significantly affected by the perturbation peak. The results quantify the trade-off between the height of the perturbation peak and its duration for a fixed percentage of increase in the room area threatened by secondary object ignition. The results show that the increased threat is of the same order as the relative perturbation in heat release rate.

Water droplet evaporation is relevant to fire suppression, industrial cooling processes, and many other technologically important applications. A mathematical model of unsteady evaporation of a water droplet is developed. The droplet is assumed to be spherical and semitransparent to radiation. A radiative transfer model based on geometrical optics theory is used to calculate the local volumetric rate of radiation absorption. Published spectral absorption coefficient data for water are used in performing the calculations. The effects of thermal expansion and temperature-dependent thermophysical properties on the evaporation process are accounted for. The internal circulation in the droplet due to the externally imposed flow is accounted for through an effective thermal conductivity of the water droplet. The model predictions are compared with available computational results and experimental data. The results of calculations show that the absorption of radiation and thermal expansion significantly affect the lifetime of a droplet. The results also reveal that neglect of radiation absorption by a droplet underpredicts the rate of water evaporation and greatly overpredicts the droplet lifetime for droplet diameters greater than .

Quantitative approaches to estimating individual and societal risk in the nuclear industry, the chemical and fuel process industries and in building design are considered. Particular attention is paid to what have been proposed as target levels for acceptability of risk in these fields. Available data on multiple fatality fire occurrence in the United Kingdom, the United States and worldwide are examined. On the basis of this data a total societal fire risk for the U.K. is proposed. This, together with information on the number of premises at risk, allows target probabilities to be defined for premises of different size. These range from about 10−6 per annum, for premises where five or more people may be killed in a fire, to 10−7 to 10−8 per annum for premises where one hundred or more may be killed. These might be used with models of risk analysis aimed at estimating actual probabilities. These target values are compared with those used in other areas. Consideration is also given to the evaluation of the weightings used in points schemes and the development of targets of acceptability where such methods are employed.

Differing perceptions of risk by various stakeholders have long played a role in influencing the development of prescriptive-based building fire safety codes. As performance-based building fire safety codes are developed, differing perceptions of risk will continue to be a significant influence. In this paper, the concepts of revealed preference, risk factors, risk adjustment factors and risk conversion factors are discussed, and two methods to address risk perceptions in a performance-based building fire safety code are introduced. The first method proposes the use of risk factors to classify buildings in terms such as low, medium, and high risk. Each class of building would be assigned a risk adjustment factor. Similar to safety factors, risk adjustment factors would be applied during deterministic building fire safety design to provide an increased level of safety in buildings where the risk perceptions would mandate greater safety. The second method would be used with a probabilistic-based building fire safety design approach, and uses risk factors to develop risk conversion factors (RCFs). In the probabilistic approach, a maximum expected-risk-to-life (ERL) value would be established by the code, with appropriate RCFs being applied to adjust maximum ERL values depending on how the building's fire safety risk is perceived.

Experiments have been performed to study pedestrian flow through bottlenecks under oversaturated conditions. The data underline a new phenomenon in pedestrian flows: when high-density conditions occur upstream of a bottleneck, the maximum capacity of the bottleneck (i.e. the maximum number of pedestrians that can flow through the bottleneck in a given time interval) can drop. This is referred to as “capacity drop” and should be carefully taken into account when dealing with building evacuation philosophy.

During the 15-min period, or thereabouts, between the first observation of the fire and its fatal eruption into the ticket hall, the fire at King's Cross Underground Station was observed by many witnesses. Naturally, their descriptions differed appreciably in points of detail, but, overall, the led to a reasonably consistent picture of the fire's development. This picture provided a vital frame of reference against which scientific description of possible sequences of fire development could be tested, particularly with respect to the last 1 or 2 min prior to fire spread into the ticket hall.This paper presents the frame of reference provided by eyewitnesses accounts and the correlation with the outcome of the scientific investigation.

The objective of this work was to determine the accuracy and limitations of a new version of Fire Dynamics Simulator (FDS), developed by McGrattan et al., on axi-symmetric fire plumes. The current version uses LES for turbulence, a mixture-fraction-based infinitely fast chemistry model for combustion, and a constant radiative loss fraction. These sub-models have been tested for unconfined fires of different sizes, based on a dimensionless heat release rate QD* in the range of 0.1 to 10.0, which covers most natural fire scenarios. No adjustment of constants or algorithms in the model FDS2.0 have been made. An examination of plume theory is made first to find the benchmark correlations. This shows a generalization for a collection of correlations based on theory, and which might be “the best”. Using the characteristic length as the scaling factor, it is found that the optimum resolution of a pool fire simulation is around 0.05. With this resolution, the flame height prediction is found to fit well with flame height correlations. Some other parameters such as temperature and mixture fraction are found to be close to the empirical estimations at flame tips. The Froude number, which describes the relative strength of momentum and buoyancy, falls within the measurement range of many researchers. The simulation also reveals that the temperature near the burner is over-predicted, while the centerline temperature and velocity in the non-combusting region is predicted well.

In this study, the predictions of several computer fire models (CCFM, FAST, FIRST and BRI) are compared with each other and with experimental data from Australia. The experimental results were obtained by Keough (Venting Fires through Roofs, Report no. 344, Commonwealth Experimental Building Station, Australia, 1972) at Darwin airport in 1971 for relatively large fires (4 MW and 36 MW). Each model has its own virtues but no model, as yet, is consistently superior. Further work necessary for improvement of the models is discussed; hopefully, this will lead to their application to building design in the future.

The use of acoustic emission (AE) as an early indicator of structural materials exposed to a flame has been investigated and found to be possible. Piezoelectric transducers have been mounted directly on 0.5 m long, simply supported beams of aluminum, gypsum board, wood and plastic, and have been used to record ultrasonic events resulting from a small flame placed under the beam. The number of AE events in a minute and the cumulative energy released during the heating cycle provide a good measure of the overheated state of some of these materials even before a temperature increase is indicated. The measured signals varied in energy and number with the type of material, the thickness of the specimen and heat flux. Wood was particularly susceptible to acoustic emission, producing more than 1000 events/min in a solid fir board and 30/min in 13 mm thick plywood when the flame exceeded 1 kW. A gypsum board produced 16 events in a minute. An aluminum plate did not respond above the background level (0.3 events/min) even though it reached the highest temperature. The differences in cumulative energy were equally striking, with the plywood being four times more energetic than the gypsum board even though the heating period for the wood was half as long, and 30 times more energetic than the aluminum. Some critical issues which remain to be investigated before this technique can be adapted to practical fire detection are mentioned.

The various mechanisms postulated to control the rate of flame propagation across a liquid fuel surface are discussed in detail. Major consideration is given to the surface tension induced flows which control the propagation at liquid temperatures well below the closed cup flash point and the cause of the flame pulsation noted at liquid temperatures just below the flash point. Some new considerations are given to the concepts and methods of experimentally evaluating flash and fire point phenomena. Analytical developments which correctly predict the very high propagation rates across liquids at temperatures well above their flash points are discussed.

The phenomena observed in experimental studies on the behavior of flames propagating across flammable liquids are summarized in this paper. Since the liquid phase phenomena have been investigated extensively and already presented in several well-known review papers, the present authors have concentrated on the gas phase phenomena. It is shown that the gas motion ahead of a propagating flame is important in the mechanisms of flame propagation across a flammable liquid. The effects of wind on the behavior of a flame propagating across a flammable liquid and the flame behavior near an obstacle on the flammable liquid surface are presented. It is pointed out that a knowledge of the detailed aerodynamic structure in the vicinity of the leading flame edge is necessary to understand the effect of wind on the behavior of a flame propagating across a flammable liquid. Also, typical examples of the gas and flame front motion predicted using an inviscid theory are presented. The results imply that the flame behavior near an obstacle on the flammable liquid surface is closely related to the gas motion ahead of the propagating flame.

Twenty-four young adults were exposed twice to a smoke detector alarm activated at 60 dBA. Unlike previous studies, all subjects were unprepared for the first alarm activation (naive) and the stage of sleep in which the alarms occurred was manipulated, with the alarm being activated twice in either stage 4, stage 2 or REM sleep for each subject. Upon being woken, time estimations, dream reports, alarm interpretations and computer reaction times were collected. Five subjects (20%) did not reliably awaken to the alarms and this was associated with their reported lack of sleep the night before, and unrelated to the stage of sleep or whether it was the first or second (non-naive) alarm presentation. Of the awakening, 87% occurred within 1 min of the alarm and no differences in time to awaken were evident between the first ‘naive’ and second ‘non-naive’ awakening. When awoken by the first alarm, 95% took no action within 2 min and 92% did not correctly interpret the alarm nature of the signal. Dream incorporation was not an important variable. Estimations of time to awaken were highly correlated with actual time to awaken for the ‘non-naive’ condition. Neither time to wake across different sleep stages nor analysis of the reaction time data revealed significant differences.