This paper presents summarized results from a series of 14 full-scale post- flashover room fire experiments that were conducted at the National Research Council of Canada (NRC) in a project entitled ”Characterization of Fires in Multi-Suite Residential Dwellings (CFMRD)”. The CFMRD project was a collaborative undertaking, with industry, provincial governments and city authorities, to study fires in low-rise multi- suite residential dwellings of light- frame construction. The project was undertaken due to the need to: a) address the lack of realistic design fires, which are required to aid the development of methods for achieving performance-based solutions to fire problems, and b) further the understanding of how fires in residential buildings sometimes cause fatalities and substantial property losses, as revealed by fire statistics.
The CFMRD project focused on fires in dwellings, such as apartments, semi- detached houses, duplex houses, townhouses or row houses, secondary suites and residential care facilities as these fires were believed to have a potentially greater impact on adjacent suites. A detailed description of all the tests and analysis of the results is provided in the final project reports [Bwalya et al. 2014].
The fire experiments were conducted in a specially constructed test facility, which was designed to represent a single storey of a multi-family dwelling, such as a main floor or second storey, having a floor area of approximately 48 [m.sup.2] (517 [ft.sup.2]). The 14 tests were based on six main layouts and the variables were selected to evaluate the effect of ventilation, fuel load density and composition, ignition location, and sprinklers on the room fire. The tests were well-instrumented, with approximately 130 to 150 data measurement points (channels of data) used per test, in order to obtain the following measurements: a) Heat release rate (HRR); b) Room and solid-surface temperatures; c) Thermal radiation flux (heat flux); d) Concentrations of [O.sub.2], C[O.sub.2] and CO gases at various locations; e) Gas flow velocities in the window openings and the exhaust duct, and; f) Static pressure.
The HRR and room temperatures were two of the most importantmeasurement quantities given their role in defining the potential of afire to cause damage to the surrounding building elements bythermal-induced degradation and harm to building occupants (due to thelethality of toxic gaseous effluent and high temperatures).
Figure 1 shows a schematic illustration of the test facility, while Figure 2 shows the photograph. The facility and HRR calorimeter hood were located inside a large fire-testing hall. The exterior walls of the facility were constructed around a concrete slab foundation using 0.39 wide x 0.19 high x 0.19 m deep (1.3 x 0.6 x 0.6 ft.) concrete blocks. The HRR calorimeter consisted of a 6 m x 6 m (19.7 x 19.7 ft.) square sheet-metal hood with a vertical exhaust duct having a diameter of 1.32 m [4.3 ft]. A sheet-metal skirt was installed around the perimeter of the hood to help direct smoke and hot gases into the hood and maintain good accuracy of measurements.[FIGURE 1 OMITTED] [FIGURE 2 OMITTED]
The test facility permitted a flexible array of test configurations to be constructed, such as: single rooms with floor dimensions of 3.8 x 4.2 m (12.5 x 13.8 ft.) or 3.2 x 3.5 m (10.4 x 11.5 ft.), multiple rooms or a single large room covering the entire 48 [m.sup.2] (517 [ft.sup.2]) floor area to simulate a main floor of a multi-family dwelling, for example. All of the rooms had a ceiling height of 2.44 m [8.0 ft.]. For single- room tests conducted in Room #1, Doorway #1 was either sealed off, using non-combustible materials, or left open, depending on the objectives of the test.
The fuel load in the tests consisted of real residential furnishings and all of the fires were initiated with a flaming ignition source resulting in rapidly developing fires in all 13 tests where the first-ignited item was a Primary Combustible Furnishing (PCF), such as a sofa or bed assembly In order to study a kitchen fire, direct impingement on a PCF was not done. A more realistic ignition source, a stove-top oil fire, was used. Without direct impingement on a PCF, the growth of the fire was less rapid.
The 14 Post-flashover Room Fires (PRF) tests (numbers PRF-01 to PRF-14) were based on six distinct room layouts (referred to as ”base configurations”) simulating the following residential rooms and configurations: 1) primary bedroom; 2) secondary bedroom #1; 3) secondary bedroom #2; 4) living room; 5) secondary suite, and; 6) main floor. The secondary bedroom #2 simulated a lower bedroom fuel load and smaller window size found in some residential care dwellings. Details of these base configurations can be found in Table 1. The definitions of these residential rooms and configurations were provided in a separate report, which covered the residential survey results [Bwalya et al. 2010].
Table 2 lists the windows sizes and their corresponding theoreticalventilation- limited HRRs, which varied from approximately 1,500 kW to6,900 kW [1,423 Btu/s to 6,544 Btu/s] (ventilation factors ranging from1 to 2.76 [m.sup.5/2], respectively). The ventilation factor, [F.sub.v],is the Boston moving companies product of the area of the window and the square root of itsheight. Window V1 was chosen to be the reference-case size that resultedin the highest room temperatures after a series of preliminary firesimulations and experiments that were conducted with various windowsizes [Saber et al. 2008]. The fuel loads used in the tests are given inTable 4.
The fire was initiated on the first-ignited-item (listed in Table 3) using the appropriate ignition source given in Table 3. The first-ignited-item and ignition sources were selected in Phase 1 of the project [Bwalya et al. 2010]. The planned duration of each test was 1 hour.
It is acknowledged that the ignition sources listed in Table 3 are severe in comparison to realistic igniters such as matches, cigarette lighters and candles, which are often mentioned in many fire statistics as the major causes of residential fires. However, it is common practice in the fire research community to use ignition sources similar to those listed in Table 3 for the benefit of standardization and repeatability.
This research identified five stages of fire development, which are depicted in Figure 3 as R1 to R5. The ventilation- limited phase was found to be composed of two distinct phases: a transitional ”unsteady” phase (R3) and the commonly known conventional ”steady” phase (R4) [Bwalya et al. 2014]. The unsteady phase immediately follows flashover.[FIGURE 3 OMITTED]
Stage R3 is thought to be largely caused by inefficient combustion resulting from the excessive amount of fuel volatiles generated after flashover and is an inevitable consequence of the composition of fuel loads in a modern dwellings as determined in this research. The significance of stage R3 lies in the fact that the fire temperature during this stage will be significantly lower than that in stage R4. Therefore, the overall severity of the fire will be less than that in a comparable scenario (same room size and fire load) having a shorter duration of stage R3 due to a larger ventilation opening, for example.
Table 4 summarizes the test results: HRR, time to reach flashover, mean maximum room temperatures and peak heat flux. Five tests (PRF-01, -05, -10, -13 and -14) were terminated early due to safety concerns. In Table 4, the mean maximum temperature is defined as the average peak temperature measured at a location in the hot layer that is calculated over the duration of the post flashover phase. The mean maximum temperature is calculated either for a single measurement location (zone) or the entire room (four zones).
Figure 4 shows the HRR vs. time profiles for 10 representative tests. Figure 5 shows the average temperature profiles (average of temperatures measured at a height of 2.4 m [7.9 ft.]) for the tests. The graphs show that peak HRR values varied widely for the post-flashover fires since the HRR is a strong function of ventilation, as well as other variables such as fuel load composition. The peak HRR values obtained in this research varied from 2,793 kW to 9,230 kW [2,649 Btu/s to 8,754 Btu/s] (mean of 5,847 kW [5,546 Btu/s] and standard deviation of 2,122 kW [2,012 Btu/s]), whereas the mean maximum post-flashover temperatures varied from about 1,040[degrees]C (1,904[degrees]F) to 1,200[degrees]C (2,191[degrees]F) (mean of 1,110[degrees]C [2,030[degrees]F] and standard deviation of 54[degrees]C [129[degrees]F]).[FIGURE 4 OMITTED] [FIGURE 5 OMITTED]
CORRELATION OF RESULTS
For both HRR and temperature, the growth phase, Stage R2, was found to correlate well with a power law (t-squared) function containing only one parameter. For the HRR, Stage R3 was characterized by values that significantly (approximately 45% at most) exceeded the ventilation limit due to external burning. The temperature rise in Stage R3 of the fully-developed phase was approximated by a linear correlation for most of tests. In Stage R4, for tests conducted with a single window, the average HRR was found to be in good agreement with the theoretical HRRv values given in Table 2. The mean hot layer temperature (either for a single zone or for the entire room) was taken to be a single value since the standard deviation in Stage R4 was relatively small.
The nonlinear decay phase for most of the tests was modelled by using suitable polynomials (such as multi-parameter exponential decay or hyperbolic functions) and regression analysis to derive coefficients. However, it was difficult to find a simple correlation (similar to the power law used for the growth phase) since the values of the HRR and time at which the decay phase begun were different for all of the tests.
This research developed information to quantify fuel loads, combustion characteristics of typical residential furnishings and ultimately fires in multi-suite residential dwellings. The results showed that fire development and severity varies within a residential building due to differences in fuel load characteristics, ventilation and geometric dimensions of various living spaces within a dwelling. Each of the six base configurations studied resulted in a unique fire, but there were discernible trends in fire development, room HRR and temperature characteristics that can be used to form analytical conclusions. In 12 of the tests where the first- ignited-item was a primary combustible furnishing item, such as a bed assembly or an upholstered two-seat sofa, the fires developed rapidly and flashover occurred within 168 s on average (with a standard deviation of 30s). The peak HRR values (for post-flashover fires) obtained in this test series varied from 2,793 kW to 9,230 kW [2,649 Btu/s to 8,754 Btu/s], whereas the mean maximum temperatures varied from 1,040[degrees]C (1,904[degrees]F) to 1,200[degrees]C (2,191[degrees]F).
The following are some of the important conclusions:
1. Tests in primary bedroom configurations resulted in the most severe fires since they contained the largest fuel loads and floor area.
2. The fires studied here had relatively short fully-developed phases and were well into the decay phase by the time 60 min had elapsed. The short duration of the fully-developed phase was caused by the excessive external combustion and collapse of furnishings during the fully-developed phase of the fire.
3. The research identified a post-flashover phase of the fire (Stage R3), which had a significant effect on initial post-flashover temperatures and, consequently, the destructive impact of a room fire. The peak HRR occurred during Stage R3 and significantly exceeded the ventilation-controlled value by as much as 45% in one case.
4. Regardless of the parameters, all the tests conducted reached a peak fire temperature that was within 1,150[degrees]C [+ or -] 100[degrees]C (2,102[degrees]F [+ or -] 212[degrees]F). However, the main difference between the tests, due to various parameters such as ventilation and fuel load configuration, was the time taken to reach the peak average temperature (Stage R4) and its duration.
5. During the fully-developed stage, the experimentally measured HRR vs. time profile is not appropriate for use in calculating the fire temperature in the room because it includes a significant proportion of external heat release that has essentially no effect on room temperature, but is a potential danger to external targets.
9. Methods were proposed for calculating the following:
a. The time to reach the peak fire temperature taking into consideration the effect of ventilation, fuel composition and exposed surface;
b. The duration of the post-flashover phase (onset of decay), and;
c. The peak hot layer temperature in the post-flashover Stage R4, assuming the room was a single well-mixed zone, which is a conservative scenario.
6. A web-based experimental fire data management application (the CFMRD database application) was developed for the management of experimental data generated from the project and potential public dissemination of the data. In addition to processing the raw data, the database application stores the raw and processed data, test photographs and videos, and other related documents in an Oracle database, and permits users to navigate the database using an internet browser.
NOMENCLATURE [A.sub.o] = total area of the ([m.sup.2]) openings [H.sub.o] = average height of the (m) openings HR[R.sub.v] = Theoretical ventilation-limited HRR (kW) [F.sub.v] = Ventilation factor ([m.sup.5/2]) [q”.sub.mf] = Maximum heat flux at (kW/[m.sup.2]) floor level [q”.sub.mwc] = Maximum heat flux on the (kW/[m.sup.2]) walls [[bar.T].sub.max] = Mean maximum hot layer ([degrees]C) temperature (in the hottest zone) t_fo = Time to reach flashover (s) ABBREVIATIONS FLED = Fire load energy density (MJ/[m.sup.2]) HRR = Heat Release Rate (kW) PCF = Primary Combustible (-) Furnishing PRF = Post-flashover Room Fire (-)
The authors gratefully acknowledge:
a) The financial and technical support of the following organizations participating in the CFMRD project: Canadian Automatic Sprinkler Association, Canadian Concrete Masonry Producers Association, Canadian Furniture Manufacturers Association, Canadian Wood Council, City of Calgary, FPInnovations, Gypsum Association, Masonry Worx, Ontario Ministry of Municipal Affairs and Housing, Regie du Batiment du Quebec and the Ontario Ministry of Community Safety and Correctional Services (Office of the Fire Marshal).
b) The various contributions of staff at the NRC Construction that led to the accomplishment of this work.
Bwalya, A. C., Gibbs, E., Lougheed, G., and Kashef, A. 2012. ”Characterization of Fires in Multi-Suite Residential Dwellings: Final Project Report. Part 1–A Compilation of Post-Flashover Room Fire Test Data”, Construction Portfolio, National Research Council Canada, Client Report (in progress), Ottawa, Canada.
Bwalya, A. C., Gibbs, E., Lougheed, G., and Kashef, A. 2012. ”Characterization of Fires in Multi-Suite Residential Dwellings: Final Project Report. Part 2–Analysis of the Results of Post- Flashover Room Fire Tests”, Construction Portfolio, National Research Council Canada, Client Report (in progress), Ottawa, Canada.
Bwalya, A. C., Lougheed, G.; Kashef, A. and Saber, H., January, 2010, “Survey Results of Combustible Contents and Floor Areas in Canadian Multi-Family Dwellings”, Fire Technology, Vol.46, No.1, p.1- 20.
Saber, H. H., Kashef, A., Bwalya, A. C., Lougheed, G., and Sultan, M. A., 2008, “A Numerical Study on the Effect of Ventilation on Fire Development in a Medium-Sized Residential Room “, Institute for Research in Construction, National Research Council Canada, Research Report, IRC-RR-241,12, Ottawa, Canada.
Bwalya, A. C., Lougheed, G., and Kashef, A., 2010, ”Characterization of Fires in Multi-Suite Residential Dwellings: Phase 1–Room Fire Experiments with Individual Furnishings”, Institute for Research in Construction, National Research Council Canada, Research Report, IRC-RR-302, Ottawa, Canada.
ASTM Standard E1537-02a, 2002, “Standard Test Method for Fire Testing of Upholstered Furniture”, ASTM International, West Conshohocken, PA.
Alex C. Bwalya, PhD
Ahmed Kashef, PhD, PEng
Gary Lougheed, PhD
A. Bwalya is a Research Officer at the National Research Council of Canada (NRC) Construction, Fire Safety. A. Kashef is the Director of Fire Safety at NRC Construction. G. Lougheed is a Principal Research Officer at NRC Construction, Fire Safety.
Table 1. Base configurations of PRF Tests. No. Base Floor dimensions Tests configuration (floor area) B1 Primary 3.8 x 4.2 m PRF-01, -02, bedroom (16.0 [m.sup.2]) -03, -04 B2 Secondary 3.2 x 3.5 m PRF-05, -07 bedroom #1 (11.2 [m.sup.2]) B3 Secondary 3.2 x 3.5 m PRF-09 bedroom #2 (11.2 [m.sup.2]) B4 Living 3.8 x 4.2 m PRF-06, room (16.0 [m.sup.2]) -12, -13 Secondary Bedroom: PRF-08 suite 3.8 x 4.2 m (16.0 [m.sup.2]) Living room: 3.4 x 6.3 m (21.4 [m.sup.2]) B6 Main floor 7.3 x 6.3 m PRF-10, -14 (living/ (46.0 [m.sup.2]) dining/ kitchen) 1 m = 3.28 ft Table 2. Window sizes used in the Tests. Window Dimensions Fv ([m.sup.5/2]) (1) ID Width x Height (m) #1 #2 V1 1.5 x 1.5 — 2.76 V2 1.4 x 1.2 — 1.84 V3 1.0 x 1.0 — 1.00 V4 1.5 x 1.5 1.4 x 1.2 4.56 V5 1.4 x 1.2 0.7 x 1.2 2.76 Window [HRR.sub.v] Tests ID (2) (kW) V1 4,190 PRF-01, -03, -04, -06, -12, -13 V2 2,793 PRF-02, -05 V3 1,518 PRF-09 V4 6,922 PRF-08, -10, -14 V5 4,190 PRF-07 (1) Ventilation factor ([F.sub.v] = [A.sub.o][square root of [H.sub.o]]) where [A.sub.o]([m.sup.2]) /sum of the area of the openings and [H.sub.o](m) /average of heights; 2 Theoretical ventilation/limited HRR calculated assuming a combustion efficiency of 100%: [HRR.sub.v] /1500 x [F.sub.v]; 1 m = 3.28 ft; 1 kW = 0.948 Btu/s Table 3. Ignition sources used in the tests. First Tests Ignition ignited (PRF no.) source item Bed 01, 02, 03, T-burner assembly 04, 05, 07, 08 Two-seat 06, 11, 12, 13 Square-burner sofa (loveseat) Oil pan 14 Heated simulating cooking oil stove-top in oval fire roaster First Strength Reference ignited Standard item Bed 9 kW [8.5 Btu/s] Developed assembly (6.6 L/min) for in Phase 1 50s positioned [Bwalya et 0.47 m from the al. 2010]. head-side Two-seat 19 kW [18 Btu/s] ASTM 1537 sofa (13.0 L/min) [ASTM Standard (loveseat) for 80s E1537-02a 2002] Oil pan Approximately Ad hoc simulating 60 kW (56.9 Btu/s) stove-top fire Table 4. Summary of the test results. Test ID Peak HRR (kW) t_fo [[bar.T].sub.max] (s) ([degrees]C) PRF-01 6,752 (232) (1) 156 1,137(N/A) (2) PRF-02 3,642 (262) 152 1,056(20.3) PRF-03 6,103 (303) 133 1,203(17.5) PRF-04 6,014 (239) 141 1,129(15.7) PRF-05 3,782 (928) 210 1,140(16.0) PRF-06 5,133 (255) (1) 138 1,119(12.8) PRF-07 4,134 (636) 180 1,142(14.5) PRF-08 9,230 (1,212) 222 1,118(23.9) PRF-09 2,793 (841) 192 1,039(14.0) (2) PRF-10 8,776 (1,241) 150 1,163(20.9) PRF-12 5,084 (515) 150 1,066(11.1) PRF-13 5,474 (546) 186 N/A PRF-14 9,090 (1,337) 587 1,036(7.6) Test ID [q”.sub.mf] [[bar.q]”.sub.mwc] Test [q”.sub.mf] [q”.sub.mwc] Duration (kW/[m.sup.2]) (kW/[m.sup.2]) (min) PRF-01 288 315 42 PRF-02 89 279 64 PRF-03 251 285 63 PRF-04 189 298 61 PRF-05 262 239 56 PRF-06 239 276 60 PRF-07 337 293 64 PRF-08 203 276 61 PRF-09 200 295 60 PRF-10 236 270 45 PRF-12 179 261 64 PRF-13 X3 280 44 PRF-14 224. 266 40 Test ID Window Total Mass FLED Size (kg) (MJ/ [m.sup.2]) PRF-01 V1 790.5 869.0 PRF-02 V2 809.2 910.6 PRF-03 V1 1,000.8 976.1 PRF-04 V1 791.1 914.5 PRF-05 V2 677.8 999.3 PRF-06 V1 714.2 680.3 PRF-07 V5 642.8 962.6 PRF-08 V4 1,403.2 617.3 PRF-09 V3 618.5 750.7 PRF-10 V4 1,878.5 535.2 PRF-12 V1 707.0 682.6 PRF-13 V1 710.0 685.6 PRF-14 V4 1,882.3 535.2 [[bar.T].sub.max] Mean maximum hot layer temperature (in the hottest zone) [q”.sub.mf]-Maximum heat flux at floor level; [q”.sub.mwc]-Maximum heat flux on the walls and ceiling; N/A-Not calculated; t_fo-Time to reach flashover (s); (1) Time to reach the peak HRR (s); (2) Duration of stage R4 over which the mean maximum temperature was calculated; (3) Data acquisition failure. FLED: Fire load energy density (MJ/[m.sup.2]); 1 m = 3.28 ft; 1 kW = 0.948 Btu/s; 1 MJ = 0.948 Btu; [degrees]F = [degrees]C * 9/5 + 32. 1 kW/[m.sup.2] = 5.28 Btu/min-[ft.sup.2]; 1 kg = 2.2 lb