The University of Edinburgh BRE Centre for Fire Safety Engineering One Day Symposium on The Dalmarnock Fire Tests: Experiments & Modelling Test One: The Uncontrolled Compartment Fire Cecilia Abecassis Empis, Adam Cowlard, Stephen Welch & Jose L. Torero 1
Overview Objectives of Test One Major Events Data Processing Characterisation of the Internal Fire Characterisation of the External Fire Summary & Exploitation 2
Objectives of Test One The Uncontrolled Fire Test Stand-Alone Objectives: To conduct a real scale (1:1) fire test with a realistic fire load, allowing it to develop freely to post-flashover conditions To record data from a dense array of various types of fire- and structuralmonitoring sensors, at high frequency, during both pre- and post-flashover conditions To provide data from a realistic fire for field model validation (i.e. CFD, FEM, etc.) Test Contribution to Project Objectives: To provide a control freely developing, underventilated scenario which enables an assessment of the bounds of repeatability of this test by comparison with the more ventilated Test Two fire. 3
Fuel Distribution One Day Symposium on The Dalmarnock Fire Tests: Experiments & Modelling Experimental Layout Objectives Sensor Distribution Aim: to provide realistic fuel loading while concentration of fuel load in the back Aim: corner to makes retrieve experiment data from a robust, dense enhancing array of firerepeatabilty. sensors and structural sensors. 5 Horizontal Laser Smoke Obscuration Sensors Heavily Smoke fuel Detectors loaded corner 3 Vertical Laser Smoke Obscuration Sensors 29 Heat Flux Gauges Structural Monitoring 6 Cameras 4 CCTV Window Sill Thermocouple Trees 20Wastepaper Thermocouple basket Trees (x12 Thermocouples (ignition source) each) 14 Bi-directional Air Velocity Probes 4
Ventilation Objectives Ventilation was set up with the aim of providing conditions that favour development to flashover Sealed off Open Open Smoke allowed to accumulate in the experimental compartment while the open doors provide a reasonable supply of air. Ignition Closed Open 5
Overview Objectives of Test One Major Events Data Processing Characterisation of the Internal Fire Characterisation of the External Fire Summary & Exploitation 6
Major Events (1) Clips of video footage from different camera angles 0 s Ignition 9 s Cushions ignite 275 s Bookcase ignites 300 s Fire engulfs bookcase 315 s Ceiling flames (in corridor) 323 s Simultaneous ignition of paper lamp and paper on table 720 s Kitchen Window breaks 801 s NW window pane breaks 1080 s External flaming 1111 s SW window pane breaks 1140 s Firemen begin to extinguish fire 1320 s Mostly smouldering 19 min Free Developing Fire 7
Major Events (2) View from flat corridor View in main corridor Snapshots from video footage from different camera angles 0 s Ignition 9 s Cushions ignite 275 s Bookcase ignites 300 s Fire engulfs bookcase 315 s Ceiling flames (in corridor) 323 s Simultaneous ignition of paper lamp and paper on table 720 s Kitchen Window breaks 801 s NW window pane breaks 1080 s External flaming 1111 s SW window pane breaks 1140 s Firemen begin to extinguish fire 1320 s Mostly smouldering 19 min Free Developing Fire 8
The Uncontrolled Fire (from outside) 9
Aftermath Fallen FRP Smoke detectors Heat Flux gauge & insulation PMMA Thermocouple and rope wire Desk metal legs Metal Bucket Remnants of the living room/office CCTV (Adpro) Computer parts Magazines 10
Overview Objectives of Test One Major Events Data Processing Characterisation of the Internal Fire Characterisation of the External Fire Summary & Exploitation 11
Sensor Calibration Laboratory calibration of laser smoke obscuration sensors Thermocouple lead tested against thermocouple blocks (at BRE) for readings output at different temperatures during both heating and cooling Thin-skin calorimeter heat flux gauge calibration against radiant panel with calibrated laboratory heat flux meter Bi-directional air velocity probe data processed as per calibration from BRE wind tunnel testing (McCaffrey et al. 1976). 12
Smoke Obscuration Sensors Plan view of a scale representation of the experimental compartment 13
Extinction Coefficients Spatially distributed average extinction coefficients (at different heights) Extinction Coefficient [m-1] 30 25 20 15 10 5 0 Laser 1 (Z=450 mm) Laser 2 (Z=1450 mm) Laser 3 (Z=1950 mm) Laser 4 (Z=2150 mm) Laser 5 (Z=2350 mm) Optically Estimated (Camera, Z=730 mm) Flashover 0 100 200 300 400 Time from Ignition [s] Combined to give an upper- and lower-bound average extinction coefficient of the smoke layer over time. Smoke Layer Height, Z-Axi Optically derived values from camera footage show good agreement at equivalent height Smoke layer height over time 2500 2000 250 o C 1500 90 o C 1000 Thermocouple Derived Camera Footage Derived 500 Smoke Layer Height 0 0 50 100 150 200 250 300 350 400 Time from Ignition (s) 14
Thermocouple to Gas-Phase Temperature Correction (for radiation) Extinction coefficient used for temperature correction Thermocouple to gas-phase temperature correction for radiation as per method described in Welch et al., 2007. Maximum single temperature correction was: 11% (Thermocouple Reading = 726 o C, Correction = 78 o C) Around the seat of the fire at the onset of flashover, as were the few other corrections of similar magnitude. Average temperature correction was of the order of 0.1% Negligible differences in this case, particular after flashover when the smoke layer descended to floor level. 15
Overview Objectives of Test One Major Events Data Processing Characterisation of the Internal Fire Characterisation of the External Fire Summary & Exploitation 16
Av. Temperature Time Curve Good Average indication compartment for broad comparison Temperature-Time with other compartment curve, with indicated fires, but spatial time resolution of major of events. data collected allows for comparison with CFD model output. 900 Temperature ( o C) 750 600 450 300 150 0 Flashover Window Breakage -NW Pane External Flaming Fire Brigade Intervention 0 300 600 900 1200 Time from Ignition (s) 17
Resolution of Temperature Distribution W1 W2 W3 W4 W5 N1 N1 N2 N2 N3 N3 N4 N4 W1 W2 W3 W4 W5 Temperature contour maps created using series of thermocouple trees. 18
Av. Temperature Time Curve (Time Steps) Instead compare spatial temperature distribution at several locations in the compartment at different points in time. Temperature ( o C) 900 750 600 450 300 150 0 Time Step 1 (201s) Time Step 2 (251s) Time Step 3 (351s) Flashover Time Step 4 (421s) Time Step 5 (661s) Time Step 6 (901s) 0 300 600 900 1200 Window Breakage -NW Pane Time from Ignition (s) External Flaming Fire Brigade Intervention 19
Temperature Contour Plots (1) Time Step 1 (201s) N1 N1 o C N 900 750 o Temperature C] [ 600 450 300 150 0 0 300 600 900 1200 Time from Ignition [s] Initial localised sofa fire Time Step 2 (251s) 20
Temperature Contour Plots (2) Time Step 2 (251s) N1 N1 o C N 900 750 o Temperature C] [ 600 450 300 150 0 0 300 600 900 1200 Time Step 1 (201s) Time from Ignition [s] Sofa fire begins to spread, continued formation of smoke layer Time Step 3 (351s) 21
Temperature Contour Plots (3) Time Step 3 (351s) N1 N1 o C N 900 750 o Temperature C] [ 600 450 300 150 0 0 300 600 900 1200 Time Step 2 (251s) Time from Ignition [s] Conditions just after flashover Time Step 4 (421s) 22
Temperature Contour Plots (4) Time Step 4 (421s) N1 N1 o C N 900 750 o Temperature C] [ 600 450 300 150 0 0 300 600 900 1200 Time Step 3 (351s) Time from Ignition [s] Post-flashover steady-state conditions Time Step 5 (661s) 23
Temperature Contour Plots (5) Time Step 5 (661s) N1 N1 N o C 900 750 o Temperature C] [ 600 450 300 150 0 0 300 600 900 1200 Time from Ignition [s] Time Step 4 (421s) Time Step 6 (901s) Period of slight steady temperature rise post steady-state conditions 24
Temperature Contour Plots (6) Time Step 6 (901s) N1 N1 o C N 900 750 o Temperature C] [ 600 450 300 150 0 0 300 600 900 1200 Time from Ignition [s] Peak average compartment temperatures once window broke Time Step 5 (901s) 25
What does the Temperature Data indicate? Significant Temperature gradient throughout the compartment at all time steps Gradient is caused by two main phenomena: local flaming local ventilation where cooler air is entrained (postflashover) Gradients can be used to identify local phenomena whereas an Average Temperature vs.time curve provides very limited information. This is the degree of resolution obtained in CFD model output. 26
Ceiling Heat Flux Gauges Plan view of a scale representation of the experimental compartment 27
Heat Flux to Ceiling (kw/m 2 ) Time Step 1 (201s) Y Coordinates (mm) 2800 1.0 2.0 3.0 2600 1.5 2400 2.5 2200 2.0 2000 1800 1600 1400 2.0 2.0 1200 1000 1000 1500 2000 2500 3000 3500 4000 X Coordinates (m m ) Time Step 2 (251s) Y Coordinates (mm) 2800 3 2600 2400 2200 2000 1800 1600 1400 1200 1 2 4 5 7 6 3 4 5 5 4 3 3 4 1000 1000 1500 2000 2500 3000 3500 4000 X Coordinates (m m ) N Time Step 3 (351s) Y Coordinates (mm) 2800 2600 2400 2200 2000 1800 1600 1400 1200 1000 10 30 40 50 60 70 20 80 90 20 30 40 50 60 70 80 30 40 50 60 70 20 1000 1500 2000 2500 3000 3500 4000 X Coordinates ( mm) Time Step 4 (421s) Y Coordinates (mm) 2800 2600 2400 2200 2000 1800 1600 1400 1200 1000 10 20 30 20 30 4050 90 80 70 100 60 40 50 90 80 70 60 20 30 40 50 1000 1500 2000 2500 3000 3500 4000 X Coordinates (m m ) Similar contour plots taken with heat flux to ceiling and to kitchen partition wall at different characteristic time steps. Time Step 5 (661s) Y Coordinates (mm) 2800 10 2600 2400 2200 2000 1800 1600 1400 1200 1000 20 20 30 40 50 30 40 40 3020 50 60 70 60 50 80 1000 1500 2000 2500 3000 3500 4000 X Coordinates (mm) Time Step 4 (901s) Y Coordinates (mm) 2800 2600 2400 2200 2000 1800 1600 1400 1200 1000 40 6080 60 120 100 80 80 60 40 80 100 1000 1500 2000 2500 3000 3500 4000 X Coordinates (m m) Allows for analysis of fire insult to structure and subsequent structural behaviour. 28
What do the Heat Flux contour plots indicate? Significant gradient in the net incident heat flux to the compartment ceiling (and walls) at all time steps Gradients can also be used to identify local phenomena Shows significant variation in severity of heat flux to structure and non-homogenous distribution of heat flux along single structural members. 29
Bi-directional Air Velocity Probes Plan view of a scale representation of the experimental compartment 30
Heat Release Heat Rate (M (MW) One Day Symposium on The Dalmarnock Fire Tests: Experiments & Modelling Fire Size (Total HRR) 900 Estimate of fire size using the principle of oxygen depletion calorimetry 750 (Huggett, 1980) to determine total Heat Release Rate from Bi-directional air Velocity Probe data. Assuming all O 2 entrained into the compartment is consumed. 10 9 8 7 6 5 4 3 2 1 0 Time Step 2 (251s) Flashover Kitchen Window Breakage Window Breakage NW Pane 0 300 600 900 Time from Time from Ignition (s) (s) Temperature ( o C) 600 450 300 150 Window Breakage SW Pane Flashover KEY Kitchen Window NW Window External Flaming Intervention 0 0 300 600 900 1200 Time from Ignition (s) Comparison of HRR with Average HRR average temperature-time Vent Case 1 curve shows good Vent Case 2 correspondence, in time. Vent Case 3 Vent 3 MW Case 4fire post-flashover which grew to a 5 MW fire once compartment window broke. (Indicative HRR) 31
Overview Objectives of Test One Major Events Data Processing Characterisation of the Internal Fire Characterisation of the External Fire Summary & Exploitation 32
External Thermocouple Trees and Heat Flux Gauges Heat Flux Gauges ExW Rulers 5th Floor ExN 4th Floor Thermocouple Trees ExN Bi-directional Air Velocity Probes 3rd Floor ExW Cameras Plan view of a scale representation of the experimental compartment 33
External Temperature Contour Plots View out of window (ExW-ExW Plane) Wind Time (1116 s OR 18 m 36 s) Wind 12 seconds later Vertical slice planes both taken 1116s (or 18m 36s) into the fire Highlight 3D variation in temperature across spill plume and in time. Time (1128 s OR 18 m 48 s) Central view perpendicular to window (ExN-ExN Plane) 34
Overview Objectives of Test One Major Events Data Processing Characterisation of the Internal Fire Characterisation of the External Fire Summary & Exploitation 35
Summary of Test Outcome Large fire compromising whole compartment (flashover) Allowed to burn for 19 min Fuel controlled, ventilation controlled, changes in ventilation Considerable external spill plume once windows broke. Experiment Low number of erroneous/damaged sensors High density and broad range of sensors allowed for complete characterisation of Test One fire, including: pre- and post-flashover phases development of the internal and external fires. Characterisation First test to have: Full scale building Realistic fire scenario Dense measurements of both fire and structural behaviour Measurements during both the heating and cooling phase. 36
The University of Edinburgh BRE Centre for Fire Safety Engineering One Day Symposium on The Dalmarnock Fire Tests: Experiments & Modelling Test One: The Uncontrolled Fire Questions? Cecilia Abecassis Empis, Adam Cowlard, Stephen Welch & Jose L. Torero 37
References McCaffrey, B.J. & Heskestad G. (1979), A Robust Bidirectional Low-Velocity Probe for Flame and Fire Application, Combustion and Flame 26, 125-127. Welch, S., Jowsey, A., Deeny, S., Morgan, R., Torero, J.L. (2007), BRE Large Compartment Fire Tests Characterising Post- Flashover Fires for Model Validation, Fire Safety Journal, in press. Huggett, C. (1980), Estimation of Rate of Heat Release by Means of Oxygen Consumption Measurements, Fire and Materials, 4, p.61-65. 38
Thermocouple to Gas-Phase Temperature Correction Extinction coefficient used for temperature correction. Thermocouple to gas-phase temperature correction for radiation as per method described in Welch et al., 2007. Maximum single temperature corrections were: Hot Layer: 11% (Tree 5, TC at 2150 mm height at 292 s) Average maximum positive correction per TC: 0.9% Cold Layer: -21% (Tree 5, TC at 450 mm height at 253 s) Average maximum negative correction per TC: -2.3% Both around the seat of the fire at the onset of flashover. Average temperature correction was of the order of 0.1% Negligible differences in this case, particular after flashover when the smoke layer descended to floor level. 39
Temperature Contour Plots Time Step 1 (201s) Time Step 2 (251s) N1 N1 N Time Step 3 (351s) Time Step 5 (661s) Time Step 4 (421s) Time Step 6 (901s) Contour plots highlighting spatial temperature variation on plane N1-N1 at different times throughout the fire. This is the degree of resolution obtained in CFD model output. 40