Publisher’s version / Version de l'éditeur:
Vous avez des questions? Nous pouvons vous aider. Pour communiquer directement avec un auteur, consultez la première page de la revue dans laquelle son article a été publié afin de trouver ses coordonnées. Si vous n’arrivez pas à les repérer, communiquez avec nous à PublicationsArchive-ArchivesPublications@nrc-cnrc.gc.ca.
Questions? Contact the NRC Publications Archive team at
PublicationsArchive-ArchivesPublications@nrc-cnrc.gc.ca. If you wish to email the authors directly, please see the first page of the publication for their contact information.
https://publications-cnrc.canada.ca/fra/droits
L’accès à ce site Web et l’utilisation de son contenu sont assujettis aux conditions présentées dans le site
LISEZ CES CONDITIONS ATTENTIVEMENT AVANT D’UTILISER CE SITE WEB.
Paper (National Research Council of Canada. Institute for Research in
Construction), 1989
READ THESE TERMS AND CONDITIONS CAREFULLY BEFORE USING THIS WEBSITE. https://nrc-publications.canada.ca/eng/copyright
NRC Publications Archive Record / Notice des Archives des publications du CNRC :
https://nrc-publications.canada.ca/eng/view/object/?id=5048b9f4-3263-4d0c-91f7-8b9d4676f405
https://publications-cnrc.canada.ca/fra/voir/objet/?id=5048b9f4-3263-4d0c-91f7-8b9d4676f405
NRC Publications Archive
Archives des publications du CNRC
This publication could be one of several versions: author’s original, accepted manuscript or the publisher’s version. / La version de cette publication peut être l’une des suivantes : la version prépublication de l’auteur, la version acceptée du manuscrit ou la version de l’éditeur.
For the publisher’s version, please access the DOI link below./ Pour consulter la version de l’éditeur, utilisez le lien DOI ci-dessous.
https://doi.org/10.4224/40001427
Access and use of this website and the material on it are subject to the Terms and Conditions set forth at
A risk-cost assessment model for evaluating fire risks and protection
costs in apartment buildings
I - - - -
National Research
Consdl national
B*F
Council Canada
de recherches Canada
I
Institute for
T H 1
Research in
recherche en
lnstitut de
I
N21d
,
Construction
construction
no. 1631
c.
2
BLDG
C,
1A Risk-Cost Assessment Model for
Evaluating Fire Risks and Protection
Costs in Apartment Buildings
by D. Yung and V.R. Beck
Reprinted from The Institution of Engineers, Australia
Proceedings, International Symposium on Fire
Engineering for Building Structures and Safety
Melbourne, Australia, November 14-1 5, 1989
p. 15-19
(IRC Paper No. 1631)
NRC
-
L I B R A R Y
\
Les auteurs pdsentent
un
m a l e stochastique d'Cvaluation des risques-coGts pour les
immeubles d'habitation
B
6tages.
On peut s'en servir pour aterminer les risques et coats
d'incendie relatifs de m6thodes de protection contre le feu autres que celles basks sur les
exigences actuelles &s codes
du
batiment.
The
Institution
of
Engineers.
AustraliaINTERNATIONAL SYMPOSIUM
ON FIRE ENGINEERING
FOR BUILDING STRUCTURES AND SAFETY
MELBOURNE
14
-
15 NOVEMBER 1989
M E INSTITUTION OF ENGINEERS, AUSTRALIA NATIONAL CONFERENCE PUBLICATK)N NO. 8911 6
Immatimal Sympoeium on Fire Engineering for Building Sbuciures and Safety Melbourne, 14- 15 November 1989
A Risk-Cost Assessment Model for
Evaluating Fire Risks and Protection
Costs in Apartment Buildings
D. YUNG
Senior Researcher, Fire Research Section, lnstitute for Research in Construction, National Research Council of Canada, Ottawa, Canada
V.R. BECK
Principal Lecturer, Department of Civil and Building Engineering, Footscray lnstitute of Technology, Melbourne
SUMMARY A stochastic risk-cost assessment model is presented for multi-storey apartment buildings. The model can be used to assess the relative fire risks and fire costs of alternative fire protection designs in comparison with designs based on current building code requirements.
1 INTRODUCTION
Since the development of a built environment, there have been great concerns about building fires and about the severe consequences of such fires. In recognition of this potential danger of fires in buildings, communities have traditionally adopted rules and regulations to control such fires. Building codes usually follow a continual revision process because of the need to address new concerns and to incorporate new technologies. The continual revision of building codes has the tendency to make them conservative, adding excessive burden to construction costs. It is a well recognized fact that the simple desire to have absolute protection of life and property from fires in buildings is realistically unattainable or prohibitively expensive. Similarly, it is also recognized that a minimal investment in fire protection will probably result in risks to life and property that are unacceptable to the community. Between these two extremes there exists a set of cost effective solutions in which an appropriate level of spending on fire safety and protection measures will produce levels of life risk and property loss that will be acceptable to the community. The objective of the present risk-cost assessment model is to enable the application of the model by fire professionals to rationally assess the cost
effectiveness of fire safety and protection provisions in apartment buildings. Rather than assessing the cost effectiveness of a single sub-system, the model estimates the expected risks and costs of a complete fire safety and protection system for the building. The system model described here is based on a similar model that was developed to assess the cost effectiveness of fire safety and protection provisions in Australian office buildings (Beck, 1987; Beck and Poon, 1988). The system model has been modified to incorporate, among other things, the construction and occupation features applicable to an apartment building (Beck and Yung, 1989). In this paper, a brief description of the system model is given and some preliminary results are presented. Many of the sub- models are still in the developmental stage or are being improved to provide a more accurate representation of fire growth and human behaviour. Nevertheless, the present paper gives a general description of the risk-cost assessment model and the preliminary results obtained.
2 SYSTEM MODEL 2.1 Model Description
The basic model has been described in detail in previous papers (Beck, 1987; Beck and Yung 1989). Only a brief description of each of the sub-models, shown in Figure 1, is described here:
(a) Fire Growth and Suppression Model
This model is used to calculate the probability of fire extinguishment, or containment, at various states of fire growth; consideration is given to the effects of intervention by various
suppression techniques.
(b) Fire Growth and Occupant Response Model This model is used to calculate the probab~lities of the occurrence of events at which occupants become aware of, and decide to respond to the presence of fire by attempting egress from the building.
(c) Fire Growth and Evacuation Duration Model This is a deterministic model that is used to calculate the durations that are available for occupants to attempt evacuation following their decision to respond to the presence of fire. (d) Flame Models
Two flame models are used: Boundary Element Models and Flame Spread Model. The
Boundary Element Models are used to calculate the probability of failure of structural and non- structural boundary elements of construction when subjected to fully-developed fires. The Flame Spread Model is used to calculate the probability of flame spread between any two separated volumes in a building. Data for this model is obtained from the Boundary Element Models.
(e) Economic Models
These models are used to calculate the fire-cost expectation. The firecost expectation consists of the summation of the expected monetary loss,
which is the monetary loss caused by fire spread in buildings, and the capital and annual costs associated with passive and active fire protection facilities.
(f) Egress Model
The Egress Model is a deterministic model used to calculate the number of occupants that are safely evacuated from a building, and also the residual population (and distribution thereof) remaining in the building at the occurrence of a critical event. The critical event is defined to be flashover in the apartment of fire origin.
(g) Smoke Spread Models
The Smoke Spread Models are used to calculate the probability of smoke spread from the apartment of fire origin to other volumes in the building. Smoke Spread Models I and I1 are applicable to pre-flashover and post-flashover fires respectively.
(h) Life Risk Models I and II
The Life Risk Models are time-independent stochastic, state-transition models that are used to calculate the probabilities of death (and safety) for the residual population. Life Risk Models I and II are applicable to pre-flashover and post-flashover fires respectively.
(i) Expected Number of Deaths Models These models are used to calculate the expected number of deaths in a building. The models are based on consideration of the calculated probability of death, for pre-flashover and post-flashover fires, and they are applied to the residual population remaining in the building at the occurrence of the critical event. (j) Expected Risk-to-Life Model
This model is used to calculate the expected risk-to-life value for occupants of an apartment building.
The system model is based on the supposition that prior to the occurrence of the critical event, egress is possible. The egress process is evaluated using the Egress Model. The durations available for occupant evacuation are determined from the Fire Growth and Evacuation Duration Model. It was assumed that the time-dependent evacuation process occurs
unimpeded by the effects of fire and smoke. That is, prior to the critical event of flashover in the apartment of fire origin, fire and smoke are contained mainly to the apartment of fire origin. At the occurrence of the critical event, it is assumed that the time-dependent evacuation process ceases, and that the residual population remaining in the building at flashover is then subject to the effects of instantaneous smoke spread, and possible fire spread, throughout the building. Since smoke and fire do not spread instantaneously, the present assumption of instantaneous spread provides a conservative estimate of the risk to life. The probability of death and safety for the residual population remaining at the critical event is calculated using Life Risk Models I and 11.
Figure 1 Risk-cost system model
2.2 Model Limitations
I
The modelling of the complex interaction between fire growth and spread, fire detection and warning, fire suppression and protection techniques, and human behaviour in the present system model is described in Section 2.1. As in many risk assessment models, certain assumptions and approximations were made in the mathematical modelling due to the lack of sufficient data base or information. Because of this, the results obtained from the present system model can be regarded only as estimates of reality. Under such circumstances, the proper use of the system
model
is for code equivalency comparisons of alternative fire protection designs. The process of code equivalency comparison is described in detail in Section 2.3. Until such time as the model is further developed, it is not suitable for risk-cost assessments where the design criterion is defined in terms of achieving an acceptable level of risk on an absolute basis.The system model consists of, as was described in Section 2.1, a series of stochastic, state-transition sub- models and interrelated deterministic sub-models. The structure of the system model is such that it permits easy updating of the model by simply changing individual sub-models should additional information or data become available. It is anticipated that the system model will follow a course of continuous development as undoubtedly more information becomes available for improving individual sub-models. Eventually as many of the uncertainties are removed, the model can be used to assess fire risks and costs on an absolute basis.
2.3 Code Equivalency Comparison The performance of any building design is characterized by the system model in terms of two decision-making parameters:
(1 ) expected risk-to-life, and (2) fire-cost expectation.
The expected risk-to-life is the expected number of deaths over the lifetime of the building divided by the total population of the building and the design life of the building. The fire-cost expectation is the expected direct costs and losses, and includes the capital cost for the provision of active and passive fire safety and protection facilities; the maintenance and inspection costs associated with active fire safety and protection facilities; and the expected monetary losses resulting from fire spread in the building, divided by the total cost of the building. The present approach of separating the life risks and the fire costs avoids the difficulties of assigning monetary value to human life. It also provides a more explicit estimate of the expected number of deaths resulting from fires in apartment buildings and allows the comparison of alternative designs to building code requirements in terms of relative risks and relative costs.
In the assessment of alternative design strategies, the following criterion can be used to identify those that are considered equivalent to, or more cost effective than, the current building code provisions (Beck and Yung, 1 989):
"that for an alternative design to be considered acceptable, the fire-cost expectation shall be minimized subject to the constraint that the expected risk-to-life value shall be similar to, and preferably lower than, the expected value based on current code requirements for a particular apartment building under consideration."
I t is recognized that procedural and non-technical considerations can, and do, have a significant role in the decision-making process. Accordingly, it is intended that while the results developed by the present research will assist in the identification of cost- effective alternative design strategies which are considered equivalent to the code provisions, the results must be regarded only as an aid to the decision-making process.
3 PRESENT DEVELOPMENT
The present system model for apartment buildings is
based
on a similar model developed for Australian office buildings. Although the basic framework of the apartment model has been developed, some of the sub-models are still being revised to provide a better representation of fire and smoke spread characteristics and human behaviour in apartment buildings. The basic approach in the development of these sub- models is to adopt stochastic models that are based, where possible, on-either physical laws ortest data. Only when thereare
no physical laws that can be used or when there are no test data that are available, would the stochastic models be based on subjective input data. As an example of this development, the Smoke Spread Models have been revised recently to be based on expen'mental datarather
than on subjective input data (Beck and Yung, 1989).In addition to revising some of the sub-models for better representation of the fire and smoke spread characteristics and human behaviour, other work is being carried out to see whether some of the sub- models can be simplified. The present system model runs on a main-frame computer which makes it difficuil for some users to ut~lize the system model. The goal ol the simplification is to allow the system model to be able to run on a PC. The ability to run the program on a PC allows wider application and hence greater acceptance of the system model by fire professionals.
4 APARTMENT BUILDINGS CONSIDERED 4.1 Generic Building Configuration
It was decided to simplify the layout of apartment buildings and cons~der common or generic apartment buildings which have two categories of floor plan layouts: (a) level of fire orig~n, and (b) level, other than level of fire origin. On the level of fire origin, rt was funher decided to consider two types of apartments: (a) the apartment of fire origtn and {b) the remaining
aoartments on that level wh~ch are aggregated into a srngle apartment. The assumed floor plan layout of the generic apartment bullding is shown in Figure 2.
4.2 Building Design Features
The system model is used to define the cost- effectiveness of various design features that are present in generic apartment buildings. Design features that can be specified include the building architectural layout, passive fire protection systems, active fire protection systems, fire alarm and detection systems, and occupant loading density. Details of the design features that can be included were described in a recent paper (Beck and Yung, 1989).
REMAJNlNG GROUPED
CORRIDOR
ELEVATOR
A) ASSUMEO PlAN ON LEVEL OF FlRE ORIGIN
CORRIDOR
ELEVATOR 8) ASSUMEO P U N ON L M L S
OTHER THAN L M L OF FlRE ORIGIN Figure 2 Generic apartment building layout
5 RESULTS
The present risk-cast assessment model is still in the developmental stage and is therefore not yet fully
operat~onal. As an example to show its potential
appfication in the assessment of the cost effectiveness of fire safety provisions, a simple case of a six-storey
apartment building is considered. The six-storey
apartment building was assumed to have eight apartments on each floor and two stairwells. (In the present system model, the apartments and stairwells
are grouped, as was discussed in Section 4, into the
simple layout of Figure
2.)
It was also assumed thatthere are 144 people living in the building (an average
of 3 per apartment). For such a bullding, the present
National BuiMing Code of Canada requires certain fire resistance ratings (FUR) for certain building
components, and the installation of a central alarm and detection system, but not spnnklers or stairwell
pressurizat~on. To determine the cost effectiveness of
these various fire protection measures, various cases were considered encompassing all possible
combinations of these four fire protection options plus the option of a higher reliability central alarm system
(higher detection and warning rel~abilities, from
approximately 70% to 95%). The results are tabulated
in Table I.
In Table I, all the results are for the option with no staiwell pressurization. The option using stairwell pressurization was considered but the results are not shown since stairwell pressurization (in such a low-rise building) did not reduce the risk, but increased the
protection cost. In
such
a low-rise building with a lowevacuation time, it is expected that the chance of being
trapped in a stairwell
is
very low. The effect of staiwellpressurization may be more pronounced in buildings having a greater number of storeys.
In Table I, the expected risk-to-life (ERL) is defined as
the expected number of deaths over the lifetime of the
building divided by the population of the building and
the des~gn life of the building. assumed to be 50 years.
The fire cost expectation (FCE) is the expected fire cost
(Section 2.3) divided by the total cost of the building.
Nominal fire resistance ratlngs (FRR) are those that are
indicative of the requirements of the building code. In
Table I. the numbers have been normalized by dividing by ERL =1.83 x 10-4 and FCE
=
4.81 x 10-2, which arethe computed results for the reference building code
option (nominal FRR, no sprinklers but wirh central
alarm).
For the case of nominal FUR in Table I, the option of a central alarm only and no sprinklers is the reference option required by the building code. The Table shows that this reference option will reduce the risk from the
no alarm option but will increase the protection cost.
Compared with this reference option, both the option of sprinklers only and the option of a central alarm and
sprinklers will reduce the
risk
but both incur higherprotection costs. The option of a higher reliability
alarm only is the
most
cost effective. This option willreduce the risk but will not increase the cost compared
to the reference option
(in
the presentsystem
model.the cost for
a
higher reliability central alarm wasassumed to be about the same as that of a regular
central alarm). The option of both a higher reliability
central alarm and sprinklers will reduce the risk further,
when compared with the option of a higher reliabilily
central alarm only. The option of a higher reliability
central alarm system and sprinklers has little eftect
on
the risk to life compared with the option of a central
alarm system and sprinklers. This is because the risk has already been reduced substantially in buildings containing regular central alarms and sprinklers.
TABLE I
NORMALIZED' RESULTS FOR A SIX-STOREY APARTMENT BUILDING WITH VARIOUS COMBINATIONS OF PASSIVE AND ACTIVE FIRE PROTECTION MEASURES
112 x Nominal FRR2 Nominal FRR 1-112 x Nominal FRR
No With N o With No With
Sprinklers Sprinklers Sprinklers Sprinklers Sprinklers Sprinklers
(a) No Alarm
Expected Risk-to-Life (ERL) 1.80 0.84 1.62 0.83 1.62 0.83
Fire Cost Expectation (FCE) 0.78 1.31 0.84 1.42 1.06 1.63
(b) With Central Alarm
Expected Risk-to-Life 1.08 0.60 1 .OO 0.60 1 .OO 0.60
Fire Cost Expectation 0.94 1.47 1 .OO 1.57 1.21 1.79
(c) With Higher Reliability Central Alarm
Expected Risk-to-Life 0.82 0.54 0.77 0.54 0.77 0.54
Fire Cost Expectation 0.94 1.47 1 .OO 1.57 1.21 1.79
Notes:
l ~ h e numbers have been normalized by ERL = 1.83 x 10-4 and FCE = 4.81 x 10-2, which are the computed
resutts for the reference building code option (nominal FRR, no sprinklers but with central alarm). 2~~~ = Fire Resistance Rating. Nominal are those indicative of building code requirements.
It is interesting to note from the present results that the effectiveness of active fire protection measures in reducing fire risk is not as dramatic in apartment buildings as in office buildings. Similar results were found in the three-storey apartment building case which was reported in an earlier paper (Beck and Yung, 1989). The present results show that various active fire protection measures can reduce the risk from no active protection in a six-storey apartment building by a factor ranging from 1.6 to 3.0. A previous study (Beck and Poon, 1988) shows that in an office environment, similar fire protection measures can reduce the risk by one to two orders of magnitude. This apparent difference is the result of the fact that in apartment buildings there i s inherently more
compartmentation, fewer people are exposed to the fire in the zone of fire origin, and people in apartment buikiings can be asleep. This latter case can have a significant effect on the risk to life.
Table I also shows the effects of reducing the fire resistance rating of the building components (112 x nominal FRR) and of increasing the rating (1-112 x nominal FRR). For the case of a lower rating (112 x nominal FRR), the results show that there is a small to negligible increase in risks, and a reduction in costs. This reduction in cost is primarily associated with a reduction in the capital cost of building construction. The increase in risks, however, is minimized with the use of a central alarm system or sprinklers or a combination of the two. It would appear that the effect on ERL of fire resistance rating of building components is not as significant in the buildings studied compared with the effect on
ERL
of a central alarm system or sprinklers since there is a good chance the occupants will either be able to evacuate the building or there will be a good chance the fire will be suppressed. For the case of a higher fire resistance rating (1-112 x nominal FRR), the results show that there is a general increase in the construction costs but that there is no significant improvement in fire safety, with or without active fire protection systems. This indicates that any increase in the nominal fire resistance ratings would not increase the fire safety level. Conversely, a moderate reduction in fire resistance ratings will have little or no reduction on the fire safety level.necessarily involves the introduction of many assumptions, both of a conceptual and numerical nature. The present system model is classified as being intermediate between a simplistic model and an overly complex model. The system model was developed based on available physical relations or experimental data; statistical data were used where appropriate. The structure of the system model. however, is such that the individual sub-models can be easily revised should new information or data become available.
The net effect of the assumptions is expected to give a conservative estimate for the calculated expected risk- to-life value for a particular building design. However, the significance of this is diminished because the decision criterion that was adopted is based on the establishment of equivalency of expected risk-to-life values, rather than isolated consideration of an absolute value for the risk-to-life. When the model is further developed with fewer uncertainties, it can then be used to assess risks and costs of build~ng designs on an absolute basis.
The system model can be used to:
-
Appraise both existing code requirements and proposals to change code requirements, and to investigate whether consistent cost-effective performance is provided by the various code requirements.-
Identify alternative design configurations that give equivalent performance to the ex~sting coderequirements (in terms of ERL values), but at a lower net cost (FCE value); that is, the alternative designs are more cost-effective.
-
Provide a performance-based apprcach to the design for fire. For example, designers could be permitted to use any design configuration, provided the design can be shown to give an ERL value which is not greater than that based on the code- specified design requirements.-
Investigate the cost-effective perforniance of proposed and existing buildings.-
Guide future research efforts into those areas which are identified as having a significant impact on the cost-effective provision of f~re safety and protection.6 CONCLUSIONS 7 ACKNOWLEDGEMENT
The complex interactive process between human behaviour and fire dynamics in apartment buildings is modelled in a series of stochastic state-transition sub- models, and inter-related deterministic sub-models. These sub-models were integrated into a system model which is used to calculate the effects of smoke and fire spread in multi-storey apartment buildings. The effects of fire and smoke spread are calculated in terms of two performance parameters; namely the expected risk-to-life (ERL), and the fire-cost expectation (FCE). The present approach of separating the life risks and fire costs avoids the difficulties of assigning monetary value to human life. Further, the code equivalency approach which has been adopted as the basis for decision-making criterion allows the comparison of alternative designs with building code requirements in terms of relative risks and relative costs. As an example, the
preliminary results for a six-storey apartment building are presented.
The use of mathematical models to calculate the
expected effects of fire and smoke spread in buildings
The authors would like to thank Mr. J. Latour of NRCC for making the computer program work at NRCC and for carrying out the computer runs for the present paper.
8 REFERENCES
BECK, V.R. (1 987). A cost-effective decision-making model for building fire safety and protection.
Fire
Safetv Journd. Vol. 12, No. 2, pp. 121 -1 38. BECK, V.R. and POON, S.L. (1 988). Results from a cost-effective decision-making model for building fire safety and protection. Fire ~ a f e t v Journal. Vol. i 3 , NO. 2-3, pp. 197-21 0.
BECK. V.R. and YUNG, D. (1989). A cost-effective risk-assessment model for evaluating fire safety and protection in Canadian apartment buildings. Proceedinas of the International Fire Protection Enaineerina Institute