Proceedings of the 5th International Seminar on Fire and Explosion Hazards, Edinburgh, UK, 23-27 April 2007

CHARACTERIZING OF DESIGN FIRES FOR CLOTHING STORES
EHAB ZALOK, AND GEORGE HADJISOPHOCLEOUS Department of Civil and Environmental Engineering Carleton University 1125 Colonel By Drive, Ottawa, Ontario K1S 5B6, Canada ABSTRACT
This paper presents the work carried out at Carleton University to characterize design fires for clothing stores. The work includes a survey of clothing stores to determine fire load and type of combustibles and medium-scale testing of fuel packages designed to represent the fire loads and type of combustibles in these stores. The experiments are conducted in collaboration with the National Research Council of Canada (NRC)’s Fire Research Program using NRC’s full-scale facilities. Fourteen different clothing stores were surveyed and the data collected were analyzed to determine the total fire load (MJ) in each of the stores, the fire load density (MJ/m2), and the contribution of different combustible materials to the total fire load. The survey was conducted in 2003 in the Canadian cities of Ottawa and Gatineau. The analysis of the survey data indicated that three different fuel packages should be used to represent clothing stores. One package is used to represent clothing stores with small floor area, the second to represent stores with wood as the major combustible, and the third to represent stores with cloth as the major combustible. The results from the tests show that clothing stores that have wood as the main combustible have almost the same burning characteristics (HRR, gas temperatures and concentrations, and heat fluxes) as small size clothing stores with similar burning trends during the test. However, the test with clothing stores that have cloth as the main combustible produced twice the max HRR, a 50% increase in gas temperatures, and three times the amount of carbon monoxide and carbon dioxide. All three tests followed a medium t2 fire.

KEYWORDS: design fires, experiments, shopping centres, heat release rate, modeling, CFD, performance-based design INTRODUCTION
The prescriptive–based codes deliver the building requirements as fixed values such as maximum travel distance, minimum fire resistance ratings, detection, alarm suppression, and ventilation. A building that follows these requirements is deemed to be safe and acceptable, while a building that does not is not acceptable. Prescriptive codes, however, allow not fully compliant designs to be accepted through equivalency clauses, which state that a building is acceptable if it is deemed to be as safe as the fully compliant building. Performance–based codes establish acceptable or tolerable level of performance in terms of hazard or risk for variety of health, safety, and public welfare in buildings and leave it to the designer to show that his design meets the established performance level [1]. As a result of this approach, performance-based codes have an advantage over the traditional prescriptive codes. They provide more flexibility in design than prescriptive codes and facilitate innovation in design, both of which may lead to lower construction costs without lowering the level of safety. Over the last twenty years, there has been an increasing effort by building code-writing bodies in many countries to move towards performance-based codes. Building designs that follow the performancebased approach are done using engineering calculations and tools that include both computer models and experiments to demonstrate acceptable performance. In many of these calculations or tests, fires that are representative of those expected in the buildings are used to evaluate building performance. These fires are known as design fires. As the development and implementation of performance-based code throughout the world increases so does the need to identify and characterize design fires for a variety of buildings so that their use is acceptable by all involved in the design and approval process. The burning characteristics of design fires depend on the type, amount, and arrangement of combustible materials, and available ventilation in the room of fire origin. The type and arrangement
328 www.see.ed.ac.uk/feh5

Proceedings of the 5th International Seminar on Fire and Explosion Hazards, Edinburgh, UK, 23-27 April 2007

of combustibles affect the growth characteristics, while the total amount of fuel and ventilation characteristics in the compartment of fire origin govern the intensity and duration of the fire. The key parameters that define a design fire include heat release rate, flame temperature, and production of toxic gases. The complete specification of design fires requires knowledge of all stages of the fire; growth, steady burning (fully developed), and decay[2]. The growth characteristics of the design fire influence the time of detection, as well as the time when conditions in the compartment of fire origin become untenable. The faster the fire is detected the earlier occupants will be notified of the fire and begin to evacuate the building; however the time available to the occupants to evacuate safely will depend on the time when untenable conditions are reached in their compartments or the exit routes. The ability of the compartment barriers to withstand the fire attack and contain the fire, preventing it from spreading to other compartments in the building depends on the intensity and duration of the fire. A fire in a compartment with limited fuel is more likely to be contained than a fire in a compartment with large amounts of fuel. The work presented in this paper is aimed at characterizing design fires for clothing stores. To this end, the work carried out were made to characterize the combustibles in clothing stores by conducting a survey to collect all relevant data. Fourteen different clothing stores were surveyed. The fire load survey of clothing stores was part of the fire load survey conducted on 168 commercial premises. Premises surveyed were the typical stores in shopping centres, such as fast food shops, restaurants, shoe stores, bookstores, liquor stores, etc. The survey was conducted in 2003 in the Canadian cities of Ottawa and Gatineau. The data collected were total fire load (MJ), fire load density (MJ/m2), type of combustibles (wood, plastics, cloth, food, etc.), compartment floor area (m2), and ventilation characteristics (area and availability). To ensure the consistency of fire load data for different premises, the surveys were done by the authors and not through a questionnaire. The results of the survey of clothing stores and their analysis produced three different fuel packages to represent these types of clothing stores. The identified fuel packages were then tested in a medium-scale test facility. The data obtained from these tests are presented and discussed in this paper. The next steps towards the characterization of design fires for clothing stores include large-scale tests of the representative fuel and computer modelling using CFD models, such as the Fire Dynamics Simulator (FDS)[3].

SURVEY OF CLOTHING STORES
Determining fire loads in a building is a tedious task. It involves determining the mass of all the different types of combustibles and their effective calorific values. The mass of an item in a compartment can be determined by weighing it (weighing technique), or by determining its volume and identifying its density (inventory technique). In this survey, both techniques have been employed. The direct-weighing method was used for items that could easily be weighed, such as toys and clothing material. The inventory method was used for items such as large objects and built-in shelves. In this method, dimensions of items were measured and their volume was calculated. The mass was then computed by multiplying the volume by the density of the material. To facilitate the survey process, a combination of the weighing and inventory methods was used, in which some common items were pre-weighed and then the surveyor noted their inventory. To ensure a high quality of the survey data and to avoid inconsistencies that might occur if individuals such as storeowners or managers had to complete questionnaires, one of the authors conducted the survey himself. As shown in Fig. 1, clothing stores formed the main part of the total floor area of the buildings surveyed, followed by restaurants, storage areas, and fast food shops. Bennetts et al have also found that clothing stores are the major occupancy in shopping centres [4]. The total floor area of clothing stores in the Ottawa survey was 30% of the total area of surveyed stores, restaurants form 13%, storage areas 9%, arts and crafts supplies 5%, and fast food outlets 4%.

329

www.see.ed.ac.uk/feh5

Proceedings of the 5th International Seminar on Fire and Explosion Hazards, Edinburgh, UK, 23-27 April 2007
Watches Sales 0.59% Travel Agencies 1.20% Toys Retail 0.57% Tobacco Retail 0.23% Ties’ Shop 0.12% Tanning Salons 0.41% Shoe Repairing 0.33% Shoes Retail 1.61% Printing & Photocopy 1.25% Post Office 0.53% Photo Finishing Consum ers 1.04% Pharm acies/Grocery 2.90% Office 0.94% Mail Room 0.29% Liquor Stores 2.95% Leather Goods Retail 0.87% Art Galleries / Picture Fram ing 0.27% Arts & Crafts Supplies 5.26% Book Retail 0.53% Café 0.96% Cellular phones 0.89% Clinic/Optical 0.58%

Com puter Accessories & Stationary Retail 2.90%

Storage Areas 8.74% Clothing Retail 29.73% Restaurants 12.93%

Conference Room 0.15% Tailors 0.56% Dry Cleaning 0.52% Fabric Shops 0.51% Fast Food Shop 3.90% Fast Food Shops / Grocers Retail 2.95% Florists Retail / Gifts 0.60% Pastry Shops 0.41%

Kitchen 2.69%

Jew ellers Retail 0.37%

Jew ellers Mfrs 0.20%

Hairstylists 2.78%

Grocers Retail 2.43%

Gift Shops 3.27%

Figure 1. Percentage of floor area of different premises to the total floor area of surveyed premises.

Figure 2 shows the frequencies of fire load densities of the 168 surveyed stores. It can be seen that the distribution frequency is positively skewed indicating that on the whole, high values of fire load have low occurrence. The fire load densities have a lognormal distribution with a mean of 750MJ/m2, a maximum of 5305MJ/m2, a minimum of 56MJ/m2 and a standard deviation of 832MJ/m2. The four fire load densities at the extreme right hand of the figure are for a bookstore, storage areas for the bookstore, a shoe store, and a greeting card shop.
30 25 Number of samples 20 15 10 5 0 1200 1600 2000 2400 2800 3200 3600 4000 4400 4800 5200 5600 400 800 0

Fire Load Density (MJ/m 2)

Figure 2. Frequencies of fire load density of the 168 surveyed stores.

To obtain a good sample of the fire load densities for clothing stores 14 different clothing stores surveyed. Stores with activities related to clothing stores, such as tailors and custom cloth design were excluded from the sample. The floor area of the surveyed clothing stores ranged from 29 – 1707 m2, and the total fire load ranges from 6256 – 511413 MJ. As seen in Fig. 3, the fire load density varies from 142 – 755 MJ/m2, with a peak at 400 MJ/m2, 95th percentile of 661 MJ/m2, a mean of 393 (MJ/m2), and a standard deviation of 164 MJ/m2. In comparison with other major groups in the survey clothing stores have one of the lowest fire load densities, see Fig. 4. More detailed results of the survey can be found in [5,6].

330

www.see.ed.ac.uk/feh5

Proceedings of the 5th International Seminar on Fire and Explosion Hazards, Edinburgh, UK, 23-27 April 2007
6 Number of samples 5 4 3 2 1 0 1200 1600 2000 2400 2800 3200 3600 4000 4400 4800 5200 5600 400 800 0

Fire Load Density (MJ/m 2)

Figure 3. Frequencies of fire load density of the 14 surveyed clothing stores.
6000 Fire load density (MJ/m 2) 5000 4000 3000 2000 1000 0 Storage areas 1196 881 526 Fast food shops 661 393 Clothing 582 298 Restaurants 4289

Figure 4. Fire load density of major groups. Range, mean, and 95th percentile [Mean (♦), 95th percentile (-)]

To determine the type of combustibles in clothing stores, further analysis of the data was conducted, which indicated that textiles contribute from 16 – 86% of the total fire load density in clothing stores. In 40% of these stores, the textiles are the main combustible with over 50% of the total fire load density. Wood in tables, shelves, and lining materials is another major combustible in clothing stores with fire load density ranging from 7% – 76% of the total fire load density. In 60% of the stores wood is the main combustible with over 50% of the total fire load density. Plastic material contribution to the fire load density ranges from 0 – 23%, as illustrated in Fig. 5. Any combustibles that did not fit into these three material groups were classified as ‘miscellaneous’.
100 90 80 % Contribution 70 60 50 40 30 20 10 0 Textiles Plastics Wood/Paper Misc.

Figure 5. Combustibles contribution in clothing stores. [Mean (♦), 95th percentile (-)]

Further analysis of the data was performed to determine the impact of store floor area on fire load density. For this analysis stores were grouped in three groups. Group 1 included stores with a floor area less than 100 m2, Group 2 stores with floor area from 100 m2 to 300 m2, and Group 3 stores with floor area greater than 300 m2. Figure 6 depicts the influence of floor area on maximum, minimum, and mean fire load densities. A clear decrease in the maximum, 95th percentile, and mean fire load
331 www.see.ed.ac.uk/feh5

Proceedings of the 5th International Seminar on Fire and Explosion Hazards, Edinburgh, UK, 23-27 April 2007

density is associated with the increase in store floor area. However, the data do not indicate a definite relationship between minimum fire load density and floor area.
800 700 Fire load density (MJ/m 2) 600 500 445 400 388 300 200 100 0 Area < 100 100 < Area < 300 Area > 300 419 333 706

587

Figure 6. Effect of floor surface area (m2) on the fire load density of clothing stores. [95th percentile (-), mean (♦)]

FUEL PACKAGES
Based on the analysis of the effect of floor surface area on fire load density of clothing stores, and also the contribution of different combustible materials to the fire load density it was decided to design three fuel packages to represent different clothing stores, details of which are shown in Table 1.
Table 1. Fire loads (MJ/m2) and contribution of combustible materials (%) to the fire load density of clothing stores Rubber/ leather (%) Remaining Mass (kg) & (% Mass Loss) 7.48 (78%) 12.47 (66%) 3.8 (89%) Wood/ paper (%) Textiles (%) Plastics (%) Mass (kg) Fire load density (MJ/m2)

Test ID

Test Title Clothing Store (small store) Clothing Store (mostly wood) Clothing Store (mostly cloth)

S W C

661.0 661.0 661.0

55.0 23.0 86.0

6.0 1.0 2.0

37.0 76.0 12.0

2.0 0.0 0.0

All fuel packages have the same fire load density of 661 MJ/m2, which is the 95th percentile of the values obtained from the survey of all clothing stores, but with different combustible compositions. The first group, Group S represents the combination of combustibles in small stores, less than 100 m2 in floor area. The second, Group W represents those stores that have high wood content and less clothing content, typically upper-class clothing stores that use substantial amount of wood as a decorative material for entrances, shelving, flooring, and on walls and ceilings. The third, Group C, represents the stores with low wood content and high cloth content. In this kind of store, shelves and hangers are mainly made of steel with very few wooden tables and boxes. The internal store lining materials are mainly non-combustible.

FIRE TESTS SETUP
Three fire experiments were conducted in an ISO-9705 compatible room [3], which was instrumented to measure heat release rate, mass loss, gas temperatures and heat flux. The dimensions of the room are 2.4 m x 3.6 m x 2.4 m high, with a 0.8 m x 2.0 m high doorway in one of the 2.4 m x 2.4 m walls. The door opening is directly under a fume hood, which is connected to an exhaust duct having a diameter of 406 mm. To facilitate calculation of the heat release rate using the oxygen consumption
332 www.see.ed.ac.uk/feh5

Total

34.65 36.28 35.44

Proceedings of the 5th International Seminar on Fire and Explosion Hazards, Edinburgh, UK, 23-27 April 2007

method, measurements of mass flow rate and concentrations of oxygen, carbon dioxide and carbon monoxide were taken in the hood exhaust duct [7]. The smoke density was measured in the duct using a pulsed white light meter. The instrumentation of the test room consisted of a tree of thermocouples located at the middle of the room, and a tree of thermocouples attached to the inner wall close to the doorway, as well as a thermocouple at the top of the doorframe to measure the temperature of gases leaving the room. In addition, one thermocouple was placed on the floor close to a heat flux meter at the middle of the room, and four thermocouples were placed at the ceiling. For measuring mass loss rate, a 1.8m x 1.8m suspended steel platform was constructed in the test room at a height of 200mm above the floor. The platform was connected to four load cells using four cables attached to the corners with a total capacity of up to 1814 kg. Three 1m x 1m fuel package-representing combustibles found in clothing stores were tested. The fuel packages were placed on top of the platform inside the test room. The combustible materials used were wood, cloth, plastic hangers and bags, paper bags, and miscellaneous (rubber and leather). All packages were ignited using a 75 kW propane T-burner running for 4 minutes to simulate an ignition source from a wastepaper basket.

TEST RESULTS AND OBSERVATIONS
This section presents the results of the tests performed using the three different fuel packages. A summary of the test results is presented in Table 2. Also, Fig. 7-9 show the fuel package at ignition time and later in the test for the three tests. To facilitate reading the figures, all curves are labelled with a letter that describes the test. “Test S” represent Small size clothing stores, “Test W” clothing stores with Wood as the major combustible, and “Test C” represents clothing stores with Cloth as the major combustible. Figure 10 depicts the heat release rate (HRR) of the three tests. The heat release rate of the test representing small clothing stores reached 720 kW at 4 minutes from ignition, dropped down to 350 kW at 10 minutes, and then decayed. The heat release rate of the test representing stores with mostly wood, reached 733 kW at 3.5 minutes from ignition, dropped down to 250 kW at 6 minutes, and then decayed. The heat release rate of the test representing stores with mostly cloth, had a maximum heat release rate of 1528 kW at 5 minutes from ignition, dropped down to 200 kW at 12 minutes, and then decayed. It can be seen that clothing stores that have wood as the main combustible (Test W) had almost the same maximum heat release rate as small clothing stores (Test S), and both had almost the same trend during the test. However, it can be seen that clothing stores that have cloth as the main combustible (Test C), produced a heat release rate which is twice the value of the maximum heat release rate of the other clothing stores, Test S and Test W. All three tests followed a medium t2 fire. Test S and Test W followed the t2 fire curve up to their peak (about 3.5 minutes) and the HRR of Test C followed the t2 fire curve up to its peak (4.5 min). Concentration profiles of oxygen O2, carbon monoxide CO, and carbon dioxide CO2 for the three tests are shown in Fig. 11 and 12. During Test S, O2 concentration reached a minimum of 19.5%, and CO and CO2 concentrations reached a maximum of 0.0116% (116 ppm) and 1.33% respectively. During Test W, O2 concentration reached a minimum of 19.4%, and CO and CO2 concentrations reached a maximum of 0.0145% (145 ppm) and 1.47% respectively. During Test C, O2 concentration reached a minimum of 16.7%, and CO and CO2 concentrations reached a maximum of 0.0369% (369 ppm) and 4.08% respectively. It can be seen that clothing stores that have wood as the main combustible (Test W) had almost the same O2, CO, and CO2 concentrations as small clothing stores (Test S), and both had almost the same characteristics during the test. However, it can be seen that when comparing clothing stores that have cloth as the major combustible (Test C) with the first two tests (Test S and Test W) that in Test C 16% more oxygen was consumed, and there was a yield of three times the amount of carbon monoxide and carbon dioxide produced by Test S or Test W.

333

www.see.ed.ac.uk/feh5

Proceedings of the 5th International Seminar on Fire and Explosion Hazards, Edinburgh, UK, 23-27 April 2007

Figure 7. Photographs show the test progress, Test S.

Figure 8. Photographs show the test progress, Test B.

Figure 9. Photographs show the test progress, Test C.

Gas temperatures were measured at the middle of the room, at the ceiling, at the west wall surface, and at the sampling point inside the duct, see Fig 13. For Test S, the maximum temperatures measured were 529, 708, 344, and 65ºC respectively. For Test W, the maximum temperatures measured were 488, 870, 383, and 75ºC respectively and for Test C, the maximum temperatures measured were 731, 867, 473, and 193ºC respectively. It can be seen that clothing stores that have wood as the main combustible (Test W) had almost the same gas temperatures at various points inside the room as in small clothing stores (Test S), and both fires burned in a similar fashion. However, it can be seen when comparing clothing stores that have cloth as the major combustible (Test C) with the first two tests (Test S and Test W) that Test C has considerably higher temperature at the middle of the room and at the sampling point inside the duct. In the three tests (Test S, W, and C), the maximum heat flux meter recorded value was 3.74, 4.22, and 11.36 (kW/m2) respectively, and the temperature at the thermocouple right beside the heat flux meter was 46, 48, and 142 ºC respectively, see Fig 14.
334 www.see.ed.ac.uk/feh5

Proceedings of the 5th International Seminar on Fire and Explosion Hazards, Edinburgh, UK, 23-27 April 2007

Test C had high optical density values between 240 s and 720 s, which correspond to the high heat release rate during the same period. After the initial 720 s, the fuel package produced the same optical density until 1560 s when the optical density of Test C decreases at a faster rate than Test S or Test W. The optical density for Test S, W, and C reached a maximum of 0.31, 0.38, 1.07 OD/m, respectively.
Table 2. Summary of HRR, and gas concentrations and temperatures HRR Data Peak HRR (kW) Test Title Time1 (s) Test ID Peak Gas Concentrations (%) CO2 (Max) CO (Max) O2 (Min) Maximum Room Temperatures (oC) Ceiling Middle Door 401 424 567 Wall 344 383 473

S W C

Clothing store (small stores) Clothing store (mostly wood) Clothing store (mostly cloth)

720 733 1528
1750

240 210 300

0.0116 0.0145 0.0369

1.33 1.47 4.08
Test S

19.53 19.47 16.70

529 488 731

708 870 867

1400 Heat Release Rate (KW)

Test W Test C

1050

Medium T-sqared fire

700

350

0 0 600 1200 Time (s) 1800 2400

Figure 10 Heat release rate (kW)
0.040% Te st S 0.035% 0.030% CO Concentration 0.025% 0.020% 0.015% 0.010% 0.005% 0.000% 0 600 1200 Tim e (s) 1800 2400 Te st W Te st C

Figure 11 Carbon monoxide concentration (%)

335

www.see.ed.ac.uk/feh5

Proceedings of the 5th International Seminar on Fire and Explosion Hazards, Edinburgh, UK, 23-27 April 2007
4.5% 4.0% 3.5% CO2 Concentration 3.0% 2.5% 2.0% 1.5% 1.0% 0.5% 0.0% 0 600 1200 Time (s) 1800 2400 Test S Test W Test C

Figure 12 Carbon dioxide concentration (%)
800 700 Tree Temperature ( o C) 600 500 400 300 200 100 0 0 600 1200 Time (s) 1800 2400 Te st S Te st W Te st C

Figure 13 Temperature, middle of the room (oC)

12.0 Te st S 10.0 Heat Flux (kW/m 2) 8.0 Te st W Te st C

6.0

4.0 2.0

0.0 0 600 1200 Time (s) 1800 2400

Figure 14 Heat flux (kW/m2)

SUMMARY AND CONCLUSION
The work presented in this paper deals with research work towards the characterization of design fires for clothing stores. The work include a survey of clothing stores to determine fire load and type of combustibles and medium-scale testing of fuel packages designed to represent the fire loads and type of combustibles in these stores. The analysis of the survey data indicated that three different fuel packages should be used to represent clothing stores. One package to represent clothing stores: one package to represent clothing stores with small floor area, the second to represent stores with wood as a major combustible, and the third to represent stores with cloth as a major combustible.

336

www.see.ed.ac.uk/feh5

Proceedings of the 5th International Seminar on Fire and Explosion Hazards, Edinburgh, UK, 23-27 April 2007

The results from the tests show that clothing stores that have wood as the main combustible had almost the same burning characteristics (HRR, gas temperatures and concentrations, and heat fluxes) as small size clothing stores with similar burning trends during the test. However, the test with clothing stores that have cloth as the main combustible produced twice the max HRR, a 50% increase in gas temperatures, and three times the amount of carbon monoxide and carbon dioxide. All three fires followed a medium t2 fire.

REFERENCES
[1] [2] [3] [4] SFPE Engineering guide to Performance–based fire protection, analysis and design of buildings, Society of Fire Protection Engineers and National Fire Protection Association, 2000. Australian Building Codes Board, Fire Safety Engineering Guidelines, Canberra 2001. ISO 9705–1993, Fire tests–full-scale room test for surface products, International Organization for Standardisation, Genève, Switzerland. Bennetts I.D., Culton, M., Dickerson, M.L, Lewins, R., Poh, K.W., Poon, S.L., Ralph, R., Lee, A.C., Beever, P.F., Cooper, R.J., Haggar, P.I., Moore, I.P., Ramsay, G.C. and Timms, G.R., Simulated shopping centre fire tests, BHP research report No. HPR/SM/R//G/062, 1997. Hadjisophocleous, G., Zalok, E., A Survey of Fire Loads in Commercial Premises, 4th International seminar on fire and explosion hazards, N. Ireland, 2003. Hadjisophocleous, G., Zalok, E., Development of Design Fires for Commercial Buildings, Fire Safety Engineering: Issues and Solutions FSE 2004, Sydney, Australia. Janssen, M., Measuring rate of heat release by oxygen consumption, Fire Technology, August 1991, pp. 234-249.

[5] [6] [7]

337

www.see.ed.ac.uk/feh5