Comparative Evaluation Method for Fire Safety Design of Large

Storage Spaces
1,2 1 3 1 1
ZHENKUN WU , HAIHANG LI , YAPING HE , DECHUANG ZHOU , JIAN WANG
1
State Key Laboratory of Fire Science, University of Science and Technology of China, Jinzhai
Road 96, Hefei, Anhui, China
2
Anhui General Fire Brigade, Hefei 230031, Anhui, China
3
School of Computing, Engineering and Mathematics, University of Western Sydney, Kingswood
Campus, Penrith, New South Wales, Australia
ABSTRACT
The design philosophy “the safety level of a performance-based designed building should not be lower than
that of a prescriptive” with objectives of property protection and life safety was implemented in this paper.
Three countervailing solutions were adopted to reduce fire risk in the 11664 m2 case storage. Besides, a
virtual prescriptive storage with identical length-width ratio and similar ventilation conditions as the
performance-based was designed for comparison. After designing the fire development event tree and
simulating six fire scenarios using FDS, the two objectives were achieved as fire severity and risk of
property loss of the performance-based design storage were found to be no higher than those of the
prescriptive design and ASETs were greater than RSETs for all fire scenarios.

KEYWORDS: comparative evaluation, performance-based design, large storage spaces

INTRODUCTION
Performance-based building codes and corresponding fire engineering design methods [1,2] have been
adopted in many countries in order to better meet the needs for designing modern buildings with the desired
functionality, aesthetics and cost-effective safety measures. When designing a new building, the highest
priority should be given to the safety of building occupants especially for densely populated places [3]. The
performance requirements of most building codes are primarily concerned with life safety of building
occupants. But for sparsely populated buildings, such as storage, this requirement may no longer be
adequate and reducing property losses becomes one of the design objectives.
He and Grubits [4] pointed out that extra measures for reinforcing fire protection may counteract increased
fire risk caused by larger building areas. This situation is acceptable if the fire protection rating is no lower
than the fire severity, that is to say, the safety margin is greater than zero. This method can be used to
assess the life safety of building occupants. But according to the regulations [5] in China, fire protection
rating is often defined as Level Two while fire hazard is often defined as Class C. The fire protection rating
and fire hazard cannot be directly used for comparison because their definitions and units are completely
different [6].
“Build a safe building” may become the design philosophy without specifying detailed rules to achieve
this goal. To be specific, buildings can be designed flexibly as long as the performance requirements of the
building codes are satisfied. One of the challenges of performance-based design is to demonstrate
quantitatively that the performance requirements of the building codes are indeed satisfied. In addition to
this challenge, one also needs to demonstrate that all design objectives are met particularly when property
protection is part of the objectives. Quantifiable performance requirements are difficult to find in building
codes. Fire safety engineers often resort to the equivalence approach in which the prescriptive provisions or
deemed-to-satisfy provisions are used as benchmarks [4]. This approach, also known as the equivalence
approach, is based on the design philosophy that “the safety level of a performance-based design building
should not be lower than that of a prescriptive design”.

OBJECTIVES AND OVERVIEW OF METHODOLOGY

In the performance-based approach, the fire safety designer is free to choose how to comply with the design
criteria, as long as the solutions ensure the same protection level as the prescriptive design. Such solutions
are typically verified and supported by engineering assessment showing that the risk of a performance-

FIRE SAFETY SCIENCE-PROCEEDINGS OF THE ELEVENTH INTERNATIONAL SYMPOSIUM pp. 933-943 933
COPYRIGHT © 2014 INTERNATIONAL ASSOCIATION FOR FIRE SAFETY SCIENCE/ DOI: 10.3801/IAFSS.FSS.11-933
 based design is lower than that associated with the prescriptive code complying design. The main objective
of this paper is to demonstrate through a case study the implementation of the design philosophy that “the
safety level of a performance-based building design should not be lower than that of prescriptive code
complying design”. The design philosophy can be addressed from the perspective of either life safety or
property protection or both. In terms of risk or expected loss, the design philosophy can be stated as “the
risk to life and the expected fire loss of a performance-based designed building should not be greater than
that of prescriptive design”.
The particular example in which this philosophy is implemented is the design of an oversized storage
building. In this study the design objective is to achieve adequate protection for both life and property. The
assessment methods to address the two objectives can be detailed as follows.

Life Safety
For the evaluation of life safety, the widely used timeline approach is adopted. The assessment criterion is
that the available safe egress time (ASET) is greater than the required safe egress time (RSET) [7]:

ASET > RSET (1)

The ASET can be determined using computational simulation techniques and tenability criteria [1,8].

Property Protection
The property protection objective is to minimize the potential loss of property. The assessment of property
employs a risk based approach. During the lifetime of a building, the potential fire hazards may be present
in different forms. In a given fire incident, there may be many possible sequences of events. Each sequence
of events constitutes a scenario. The overall fire risk of property loss in a scenario can be quantitatively
expressed as the product of probability of a fire scenario and its consequence [8]:

R j = Fj × C j (2)

where
j = fire scenario index;
R j = risk of fire scenario j in terms of property loss;

F j = probability of fire scenario j ;

C j = consequence of fire scenario j .

The overall expected property loss, or fire loss, R , is evaluated from

N
R = ∑ Rj (3)
j =1

In terms of the expected fire loss, design philosophy can be expressed as

RPBD ≤ RPr e (4)

where subscript PBD represents performance-based designed storage, subscript Pr e represents prescriptive
designed storage. F j can be determined by event tree analysis [9,10] and C j is the property loss of scenario
j . For simplicity, the costs of protection investment and operation are not considered here.

FIRE SAFETY SCIENCE-PROCEEDINGS OF THE ELEVENTH INTERNATIONAL SYMPOSIUM pp. 933-943 934
COPYRIGHT © 2014 INTERNATIONAL ASSOCIATION FOR FIRE SAFETY SCIENCE/ DOI: 10.3801/IAFSS.FSS.11-933
 APPLICATION OF THE METHODOLOGY
The application of the above outlined methodology is demonstrated through a case study in the following
sections.

Project Overview
The building selected for case study is an outbound storage of a new factory in China. A schematic plan of
the one storey storage building is shown in Fig. 1, where it can be seen that the total floor area is 11664 m2.
This storage is divided into 24 smoke bays each 27 m long and 18 m wide. Every smoke bay is equipped
with a fan of 8.3 m3/s exhaust capacity. The smoke exhaust system is designed to activate after sprinkler
activation by linkages to the sprinkler system and the detection system. The building height is 10 m and the
stacking height of goods is 6 m. According to the fire code of China [5], the classification of fire hazard of
this storage is Class C and the fire resistance rating should be Level Two, and the allowable maximum fire
partition area is 3000 m2 after an automatic sprinkler system has been installed as required. However, this
storage building was intended to be designed as a single fire partition due to the requirement of logistics
supply efficiency. So now the contradiction of the building area appears between the requirement of the
building function and the restrictions of the fire code.

Fig. 1. The geometry of the original performance-based design storage. Square filled areas are refrigerator
stackings while blank areas are passages. Measurements are given in meters.

Countervailing Solutions
Several solutions can be adopted to countervail the increased fire risk caused by the enlarged fire partition
area. Early suppression fast response sprinkler system was introduced to control the development of the
initial fire; fire separation zones were set between the refrigerator stackings to prevent fire spread from one
stacking to another; fire retardant coating was used to strengthen the fireproof property of the steel
structures of the storage building. The details of the above solutions are as follows:
Ÿ The parameters of early suppression fast response sprinkler (ESFR) for this storage were selected on
basis of building height and fire risk. The sprinklers have a K factor of 360 L/min.bar1/2 and have an
activation temperature of 74 °C. They have an RTI-value of 36 ms and the outlet pressure of the
sprinklers is 0.45 MPa. Twelve sprinklers were designed to act together with a duration time of one
hour. The corresponding total water flow rate is 152 L s . The sprinklers were mounted at locations 0.5
m below the pitched roof.
Ÿ Fire separation zone in buildings is a empty area of a certain width and combustibles are strictly
prohibited within this area. So fire separation between the refrigerator stackings is proposed to isolate
combustibles without sacrificing building integrity. In an earlier study [11] it has been determined that
the 6 meters wide transport passages can be set as fire separation zones to prevent fire spread between
stackings if the sprinkler system is effective. In case of the sprinkler system failure, the fire separation

FIRE SAFETY SCIENCE-PROCEEDINGS OF THE ELEVENTH INTERNATIONAL SYMPOSIUM pp. 933-943 935
COPYRIGHT © 2014 INTERNATIONAL ASSOCIATION FOR FIRE SAFETY SCIENCE/ DOI: 10.3801/IAFSS.FSS.11-933
 zone transport passages should be broadened to 11 meters to prevent fire spread. The width of 11 m is
adopted in the alternative solution. The design sketch of broadened fire separation zones is shown in
Fig. 2. Although the increased width of transport passages reduces the storage area, it, on the other
hand, brings the benefit of easier access and more rapid turnover in a busy warehouse.
Ÿ Coat protection level of steel structures of this storage meets the requirements of Type 332 building in
NFPA 5000-2006 [12]. Specifically, the fire-resistance time is as follows: walls ≥ 3 hr, columns ≥ 2
hr, beams ≥ 2 hr, and purlins ≥ 2 hr.

Fig. 2. Modified alternative solution with broadened fire separation zones shown in white bias.
Measurements are given in meters.

Virtual Prescriptive Designed Storage

A virtual storage satisfying the prescriptive fire regulation is used for equivalence analysis. The virtual
prescriptive design storage consists of 6 smoke bays with each bay the same as in performance-based
storage. So the length-width ratios of these two storages are identical and the ventilation conditions are
similar. The building area of the prescriptive design storage covers 2916 m2. Standard sprinklers were
selected and the mechanical smoke exhaust volume is 48.6 m3/s. The mechanical smoke exhaust volume
for the performance-based storage is 194.4 m3/s so that the volume of a unit area is the same as the
prescriptive storage. Two stackings are arranged in the prescriptive storage with a central passage dividing
them, Fig. 3.

Fig. 3.Virtual prescriptive design storage. Square filled areas are refrigerator stackings while blank areas
are passages. Measurements are given in meters.

FIRE SAFETY SCIENCE-PROCEEDINGS OF THE ELEVENTH INTERNATIONAL SYMPOSIUM pp. 933-943 936
COPYRIGHT © 2014 INTERNATIONAL ASSOCIATION FOR FIRE SAFETY SCIENCE/ DOI: 10.3801/IAFSS.FSS.11-933
 Tenability Criteria
In [1] the following evacuation safety criteria are given. Based on these, the time when critical conditions
appear can be found. They are:
Ÿ Gas temperature: the temperature below the safety height (2.6 m) should be lower than 60 °C for safe
evacuation.
Ÿ CO and CO2 concentration: the concentration of CO and CO2 below the safety height (2.6 m) should
be lower than 1400 ppm and 6 % respectively.
Ÿ Visibility: the visibility should be better than 10 m in this huge compartment.

Property Loss Criteria

Fire severity is a measure of fire intensity and here it is represented by the smoke layer height, smoke layer
temperature, and radiation intensity at the top of the stackings. In the assessment of the relative safety level
between prescriptive and performance-based design, these parameters will be compared with the critical
values first and then compared mutually. The critical values [1] of these indexes are:
Ÿ Smoke layer height: a height of 1.6 m + 0.1H from the floor, so the critical value of this storage is 2.6
m.
Ÿ Smoke layer temperature: when the temperature of the smoke layer reaches 600 °C, the combustibles
in the room will ignite causing flashover.
Ÿ Radiation intensity at 6 meters high: the combustibles in this storage are classified as easily ignited
goods thus the critical value of radiation heat flux is 10 kW/m2.

Fire Scenarios
After hazard identification, the combustibles in this storage are refrigerators and their packages that are
piled up stacking by stacking. The main igniting sources are failure of electric trailer and electric forklift as
well as irregular operation. The risk analysis is carried out by quantitatively evaluating a number of fire
scenarios. The evaluation calculates the probabilities and consequences for all scenarios in the event tree.
Fire event tree is a logic diagram illustrating the fire development that is mainly affected by the sprinkler
system, the smoke extraction system and the fire separation zone.
Table 1. Reliabilities of sprinkler and smoke extraction systems.
System Reliability (%)
Sprinkler system 89.9 [13]
Smoke extraction system 97.4 [14]

The reliabilities of the sprinkler and smoke extraction systems have been summarized in Table 1. It is
recognized that uncertainties or variabilities exist in these data [15] and the outcomes of the assessment of
each individual designs will depend on the selection of these input data. However, because of the
comparative nature of the assessment methodology employed in the current study and because the impacts
of these input data on the assessment outcomes of the performance-based design and the prescriptive code
compliant design would be the same or similar, it was hypothesized that the dependence or the sensitivity
of the final comparison to the two design assessments would be low or negligible.
For the reliability of the fire separation zone, no statistical data can be found in the literature. According to
our general knowledge, jump fire and bridging can be contributing factors to the failure of a fire separation
zone. The probability of a jump fire is related to the failure probability of smoke extraction system,
considering that fire brands as the result of carton burning will not be exhausted in case of smoke extraction
system failure. Bridging of fires between stackings can also be caused by the existence of electric trailer or
forklift in the transport passages. In normal operation there is one trailer or forklift in the storage, so the
space existence probability of electric equipments around the initial fire stacking can be assumed as the
passage perimeter of the stacking divided by the total length of the passages (here it is 0.32). According to
the data provided by the operating departments, electric trailer and forklift work 14 hours per day so the
time existence probability of electric equipment is 0.58. Besides, the probability of driving electric

FIRE SAFETY SCIENCE-PROCEEDINGS OF THE ELEVENTH INTERNATIONAL SYMPOSIUM pp. 933-943 937
COPYRIGHT © 2014 INTERNATIONAL ASSOCIATION FOR FIRE SAFETY SCIENCE/ DOI: 10.3801/IAFSS.FSS.11-933
 equipment out of the storage by the driver in case of a fire is assumed to be 0.5. To sum up, the probability
of bridging of the initial fire stacking and its adjacent stackings is 0.32×0.58×0.5=0.0928.
Through the analysis and calculation above, if the smoke extraction system is effective the failure
probability of fire separation zone is 0.0928; if the smoke extraction system is ineffective the failure
probability of fire separation zone is 0.026+0.0928=0.1188. Based on these probabilities, the event tree for
a fire development in the present storage building is drawn in Fig. 4. The fire intensity of each scenario will
be calculated by the Fire Dynamics Simulator 5 [16] and will be compared between the prescriptive and the
performance-bas designed storages. On the other hand, the expected number of damaged refrigerators will
be obtained by multiplying the probability of the specific scenario by its consequence. The total risk of
property loss associated with the building is the summation of all the scenarios in the event tree.

Fig. 4. Event tree for a fire development in the performance-based designed storage.
In this paper, the heat release rate of a design fire in its growing phase was simplified by using a t-squared
fire curve, αt 2 . The fire growth rate can be classified into fast according to the property of the
combustibles, α = 0.047 kW s 2 . The effect of the sprinkler system was modeled by a constant heat release
rate from the time of activation [17] that was calculated using DETACT-T2 [18] from NIST. The fire
source was simplified conservatively located at the floor, i.e., 13.5 m below the sprinklers. If the sprinkler
system is effective, the estimated sprinkler activation times are 6.00 and 5.72 min for the prescriptive and
performance-based designs respectively. The corresponding constant maximum heat release rates are 6.1
MW and 5.5 MW respectively. The difference is due to the different RTI-values. If the sprinkler system is
ineffective and the fire separation zone is effective, the maximum heat release rate of the fire will be 129.6
MW when an entire stacking is ignited. If both the sprinkler system and fire separation zone are ineffective,
the constant maximum heat release rate will be 225 MW and 437 MW for prescriptive and performance-
based design storage respectively. This difference is due to different numbers of refrigerators therein. Six
fire scenarios have been designed to describe the probable scenarios as summarized in Table 2. For the
virtual prescriptive design storage, fire Scenarios 3 and 4 no longer exist because fire separation would be
ineffective if the sprinkler system fails. And the probabilities of Scenario 5 and Scenario 6 are 0.0026 and
0.0984 respectively.
The above maximum heat release rates for scenarios are described below:
Ÿ Scenarios 1 and 2 of performance-based: ambient temperature 20 ℃, RTI = 36 ms , activation
temperature 74 ℃, ceiling height was set at 13.5 m as the highest point of the roof for conservatism,
detector spacing 3 m. The result of activation time was 5.72 min with the maximum heat release rate of
5.5 MW for a fast fire growth rate.
Ÿ Scenarios 1 and 2 of the prescriptive design: RTI = 80 ms with other parameters the same as in the
performance-based. The result of activation time was 6 min with the maximum heat release rate of 6.1
MW for a fast fire growth rate.
FIRE SAFETY SCIENCE-PROCEEDINGS OF THE ELEVENTH INTERNATIONAL SYMPOSIUM pp. 933-943 938
COPYRIGHT © 2014 INTERNATIONAL ASSOCIATION FOR FIRE SAFETY SCIENCE/ DOI: 10.3801/IAFSS.FSS.11-933
 Ÿ Scenarios 3 and 4 of the performance-based design: a group of 9 refrigerators stacked as three stories
was considered to be ignited together at first, and the fire to spread to its adjacent groups laterally with
a time delay. The heat release rate curves for these scenarios were obtained by summing the ignited
groups in each time period with the maximum value of 129.6 MW. The heat release rate curves were
simplified as t-squared fast fires which are conservative choices.
Ÿ Scenarios 5 and 6 of the performance-based and prescriptive designs: the maximum heat release rates
were estimated by multiplying the fire area expansion proportion by the maximum HRR of one
stacking area. The maximum HRR of the performance-based and prescriptive storages were 437 and
225 MW respectively. Here these scenarios were also simplified conservatively as t-squared fast fires.

Table 2. Six design fire scenarios and their constant maximum heat release rates.

Smoke Fire Performance-

Fire Sprinkler Prescriptive
extraction separation based storage
scenario system storage
system zone (Probability)
1 √ × √ 5.5 0.0234 6.1 0.0234
2 √ √ √ MW 0.8756 MW 0.8756
3 × × √ 129.6 0.0023
/
4 × √ √ MW 0.0893
5 × × × 437 0.0003 225 0.0026
6 × √ × MW 0.0091 MW 0.0984
“√” represents effective; “×” represents ineffective.

RESULTS AND DISCUSSION

ASET and RSET were quantitatively evaluated for the current performance-based design storage, as shown
in Table 3. As for Scenarios 3 and 5, visibility was the first parameter that reached its risk threshold for
both scenarios and ASETs were 626 s and 342 s, respectively.
Infrared correlation smoke detector is used as the detection system in this storage, so its alarm time can be
chosen as 120 s [19]. An estimate of time used for reaction and decision is 60 s as workers in the storage
area are awake and familiar with the evacuation path according to categories in [19]. The time used for
movement can be estimated as the value of evacuation distance divided by traveling speed as the storage is
sparsely populated. The farthest evacuation distance is 60 meters, so movement time will be 60 s as
traveling speed was assumed to be 1 m/s.
Having compared ASETs with RSETs, we found that workers in the current storage would evacuate safely
in Scenarios 3 and 5. The ASETs for Scenarios 4 and 6 will be greater than 626 s and 342 s, respectively, as
the smoke extraction system is effective therein. Besides, the evacuation safety will be ensured in Scenarios
1 and 2 because initial fires can be suppressed in those two scenarios.
Table 3. Evacuation analysis.
Fire scenario RSET (s) ASET (s) Conclusion
3 252 626 Safe
5 252 342 Safe

In order to verify the second target, the smoke layer height, smoke layer temperature, and radiation
intensity at 6 meters high were calculated using the Fire Dynamics Simulator (FDS) [16] from NIST. This
software package has undergone some validations for its temperature and radiation heat transfer prediction
capabilities [20]. However, it is noted that the effects of sprinkler activation on smoke layer height are not
simulated in FDS though the effect on the heat release rates of fires are. So the results might not display the
situations in real fires accurately. Scenario 1 has been used as an example of the comparison of fire
intensity parameters between prescriptive and performance-based designed storages. Scenario 1 represents
the fire development with effective sprinkler, ineffective smoke extraction, and effective fire separation

FIRE SAFETY SCIENCE-PROCEEDINGS OF THE ELEVENTH INTERNATIONAL SYMPOSIUM pp. 933-943 939
COPYRIGHT © 2014 INTERNATIONAL ASSOCIATION FOR FIRE SAFETY SCIENCE/ DOI: 10.3801/IAFSS.FSS.11-933
 zone. Fig. 5 (a) shows the time curves of smoke layer height from which we can see that both of them are
higher than the critical safety height. The smoke layer heights of the prescriptive and performance-based
design stabilize at about 5.0 m and 7.4 m respectively. From Fig. 5 (b) the smoke layer temperatures of
these two storages are found to be far below their critical values. The steady temperature values of the
prescriptive and performance-based designs are 86.5 ℃ and 48.2 ℃respectively. Finally, the radiation
intensities at 6 meters high are shown in Fig. 5 (c). The stable results of the prescriptive and performance-
based are also far below their critical value as the former is 1.52 kW/m2 and the latter is 0.75 kW/m2. From
the above, it can be concluded that the safety level of the performance-based design storage is not lower
than that of the prescriptive for Scenario 1.

(a) Smoke layer height

(b) Smoke layer temperature

(c) Radiation intensities at 6 meters high

Fig. 5. Fire severity comparison of Scenario 1.

FIRE SAFETY SCIENCE-PROCEEDINGS OF THE ELEVENTH INTERNATIONAL SYMPOSIUM pp. 933-943 940
COPYRIGHT © 2014 INTERNATIONAL ASSOCIATION FOR FIRE SAFETY SCIENCE/ DOI: 10.3801/IAFSS.FSS.11-933
 The stable values of smoke layer height, smoke layer temperature, and radiation intensity for other
scenarios are shown in Table 4. From this table the safety levels of the performance-based design storage
were also found no lower than those of the prescriptive for Scenarios 2, 3, and 4. For Scenarios 5 and 6, all
of the refrigerator stackings in the building will be ignited if the fire separation zone is ineffective. So in
these two scenarios the risk of property loss of the performance-based storage will be far greater than that
of prescriptive.
Table 4. Fire severity comparison for all scenarios.
Radiation
Smoke layer Smoke layer
Fire intensity at
Storage height temperature Conclusion
scenario 6 meters high
(m) (℃ )
(kW/m2)
Performance-based 7.4 48.2 0.75
1
Prescriptive 5.0 86.5 1.52
The safety level of
Performance-based 7.3 48.4 0.76
2 performance-based
Prescriptive 6.9 78.0 1.29
storage is not lower
Performance-based 6.1 188.0 3.45
3 than that of
Prescriptive 1.9 265.5 5.45 prescriptive
Performance-based 6.5 229.9 3.99
4
Prescriptive 2.0 269.3 5.65
Performance-based Risk of property
5
Prescriptive loss of
Performance-based performance-based
All refrigerator stackings will be ignited
storage will be far
6
Prescriptive greater than that of
prescriptive

After the comparison of the safety levels between the performance-based and prescriptive storages, the risk
of property loss will be compared. A total of about 15000 refrigerators can be piled up in the performance-
based storage with 2500 refrigerators in each stacking. If the early suppression fast response sprinkler
system is effective (Scenarios 1 and 2), the expected damaged refrigerators by the initial fire will be 31
(refrigerators ignited before ESFR sprinkler activated); if the sprinkler system is ineffective and the fire
separation zone is effective (Scenarios 3 and 4), the expected damaged refrigerators after fire spread in a
stacking will be 2500; if the sprinkler system and fire separation zone are both ineffective (Scenarios 5 and
6), the expected damaged refrigerators after fire spread in all stackings will be 15000.
A total of about 3750 refrigerators can be piled up in the prescriptive storage as its building area is a quarter
of the performance-based storage and their stacking layouts are similar. If the standard sprinkler system is
effective, the expected damaged refrigerators by the initial fire will be 34 (refrigerators ignited before
standard sprinkler activated); if the sprinkler system is ineffective, fire will spread in all stackings and the
expected damaged refrigerators will be 3750. Comparing the expected damaged refrigerator numbers in
Table 5, we can conclude that the risk of property losses between the performance-based and the
prescriptive design storages are nearly equal.
Table 5. Expected fire loss.
Expected damaged refrigerator number (probability)
Storage Initial suppressed One stacking All stacking Summation
fire ignited ignited
Performance-
31 (0.899) 2500 (0.0916) 15000 (0.0094) 398
based
Prescriptive 34 (0.899) / 3750 (0.101) 409

FIRE SAFETY SCIENCE-PROCEEDINGS OF THE ELEVENTH INTERNATIONAL SYMPOSIUM pp. 933-943 941
COPYRIGHT © 2014 INTERNATIONAL ASSOCIATION FOR FIRE SAFETY SCIENCE/ DOI: 10.3801/IAFSS.FSS.11-933
 CONCLUSION
The main objective of this paper was to demonstrate through a case study the implementation of the design
philosophy “the safety level of a performance-based designed building should not be lower than that of a
prescriptive”. The safety level can be measured by either risk to life or expected fire loss or both. In terms
of the latter two, the design philosophy can be expressed as “the risk to life and the expected fire loss of a
performance-based building design should not be greater than that of prescriptive design”. This statement
also implies that the prescriptive code requirements and the associated risk are used as benchmark for fire
safety engineering assessment. This approach resolves the issue of setting the quantifiable performance
criteria for fire safety engineering design and assessment.
A large refrigerator storage warehouse was selected for the case study. The performance-based design, or
the alternative solution adopted the removal of the fire rated partitions which are required by the
prescriptive code. An early suppression fast response sprinkler system, fire separation zones, and retardant
coating of the steel elements were also adopted to countervail the increased fire risk caused by the enlarged
space. A virtual prescriptive code compliant storage warehouse was used for comparison. An event tree
analysis was conducted to establish multiple fire scenarios and the associated probabilities. For each
individual scenario, the deterministic timeline analysis was employed to assess occupant evacuation safety.
The fire loss was measured in terms of the number of damaged refrigerators.
The result of the analysis revealed that the risk associated with performance-based design was less than that
associated with the prescriptive code compliant design. It was also revealed that the widened transport
passage was an essential feature of the alternative solution in order to minimize the loss of property, or
building contents. The benefit of the widened passage was also reflected in the improvements in access and
turnover.
In the fire loss or fire cost assessment, the current study did not include the cost of initial investment and
the benefit of the widened transport passage. A full cost-benefit analysis would be desirable for decision
making. The outcome of an absolute risk based fire engineering assessment will depend on the probability
or reliability data of various fire protection devices. Uncertainty or variability exists in these input data.
Even though the impact of this uncertainty on the outcome of two design solutions may be the same or
similar because the same assessment approach is used, it may be necessary to conduct a sensitivity analysis
such that it can be demonstrated that the performance-based design is capable of tolerating some extent of
uncertainty in the reliability of the fire protection devices. The above highlighted issues will be the topics
for future studies.

ACKNOWLEDGEMENTS
This research was supported by the Fundamental Research Funds for the Central Universities
(WK2320000012).

REFERENCES
[1] Hadjisophocleous, G.V. and Benichou, N., (1999) Performance criteria used in fire safety design,
Automation in Construction 8: 489-501, http://dx.doi.org/10.1016/S0926-5805(98)00096-X
[2] Meacham, B., Bowen, R., Traw, J., and Moore, A., (2005) Performance-based building regulation:
current situation and future needs, Building Research & Information 33: 91-106,
http://dx.doi.org/10.1080/0961321042000322780
[3] Croce, P.A., Grosshandler, W.L., Bukowski, R.W., and Gritzo, L.A., (2008) The International
FORUM of Fire Research Directors - A position paper on performance-based design for fire code
applications, Fire Safety Journal 43: 234-236, http://dx.doi.org/10.1016/j.firesaf.2007.12.004
[4] He, Y. and Grubits, S., (2010) A Risk-based Equivalence Approach to Fire Resistance Design for
Buildings, Journal of Fire Protection Engineering 20: 5-26,
http://dx.doi.org/10.1177/1042391509360306
[5] "Code of Design on Building Fire Protection and Prevention GB 50016-2006," Ministry of
Construction of the People's Republic of China, 2006.

FIRE SAFETY SCIENCE-PROCEEDINGS OF THE ELEVENTH INTERNATIONAL SYMPOSIUM pp. 933-943 942
COPYRIGHT © 2014 INTERNATIONAL ASSOCIATION FOR FIRE SAFETY SCIENCE/ DOI: 10.3801/IAFSS.FSS.11-933
 [6] Meacham, B.J., (2004) Understanding Risk: Quantification, Perceptions, and Characterization,
Journal of Fire Protection Engineering 14: 199-227, http://dx.doi.org/10.1177/1042391504042454
[7] Nelson, H.E. and Mowrer, F.W., "Emergency Movement," The SFPE Handbook of Fire
Protection Engineering (3rd ed), DiNenno, P.J. (ed.), National Fire Protection Association, Quincy,
MA 02269, 2002, p. 3/367.
[8] "International Fire Engineering Guidelines," Australian Building Codes Board, 2005.
[9] "ISO/TR-13387, Fire Safety Engineering," International Standards Organisation, 1999.
[10] Hurley, M.J. and Rosenbaum, E.R., "Performance-Based Design," The SFPE Handbook of Fire
Protection Engineering (4th ed), DiNenno, P.J. (ed.), National Fire Protection Association, Quincy,
MA 02269, 2008, p. 3/440-3/455.
[11] Zhang, J.Q., Lu, S.X., Yuan, M., Li, C.H., and Li, Q., (2011) Safe Separation Distance Calculation
Model with Changing Area of Fuel Packages in Large Space, Procedia Engineering 11: 666-674,
http://dx.doi.org/10.1016/j.proeng.2011.04.711
[12] "NFPA 5000 Building Construction and Safety Code, 2006 edition," National Fire Protection
Association, Quincy, MA 02169, 2006.
[13] "Code of design for sprinkler systems GB 50084-2001," Ministry of Construction of the People's
Republic of China, 2001.
[14] Fan, W.C., Sun, J.H., and Lu, S.X., Fire risk assessment methodology, Science Press, Beijing,
China, 2004, p. 272.
[15] Bukowski, R.W., Budnick, E.K., and Schemel, C.F., "Estimates of the Operational Reliability of
Fire Protection Systems," Proceedings of the Third International Conference on Fire Research
and Engineering, Chicago, USA, 1999, pp. 87-98.
[16] McGrattan, K., Klein B., Hostikka, S. and Floyd J., "Fire Dynamics Simulator (Version 5) User's
Guide," National Institute of Standards and Technology Report NIST Special Publication 1019-5,
Washington, USA, 2009.
[17] Evans, D. and Stroup, D., (1986) Methods to calculate the response time of heat and smoke
detectors installed below large unobstructed ceilings, Fire Technology 22: 54-65,
http://dx.doi.org/10.1007/bf01040244
[18] "Fire Modeling Software," National Institute of Standards and Technology, 2001.
[19] Fleischmann, C.M., "Prescribing the Input for the ASET versus RSET Analysis: Is This the Way
Forward for Performance Based Design?," Fire Protection and Life Safety in Building and
Transportation Systems, Santander, Spain, 2009, pp. 1-17.
[20] Mell, W., Maranghides, A., Mcdermott, R., and Manzello, S.L., (2009) Numerical simulation and
experiments of burning douglas fir trees, Combustion and Flame 156: 2023-2041,
http://dx.doi.org/10.1016/j.combustflame.2009.06.015

FIRE SAFETY SCIENCE-PROCEEDINGS OF THE ELEVENTH INTERNATIONAL SYMPOSIUM pp. 933-943 943
COPYRIGHT © 2014 INTERNATIONAL ASSOCIATION FOR FIRE SAFETY SCIENCE/ DOI: 10.3801/IAFSS.FSS.11-933