1. INTERNATIONAL AIRCRAFT MATERIALS FIRE TEST WORKING GROUP ROUND ROBIN TEST III DATA ANALYSIS Khang D. Tran, Ph.D., Sr. Scientist The Mexmil Company, Santa Ana, California, USA. June 2001 The International Aircraft Materials Fire Test Working Group Meeting Sedgefield, Stockton-on-Tees, Cleveland, United Kingdom, 27-28 June 2001

2. ACKNOWLEDGEMENTS The Mexmil Company Mr. David Indyke, Materials Technology Manager, Mexmil Co. Mr. Tim Marker, Project Manager, The FAA. Mr. Johns Brook, Director of Research, International Aero Inc.

3. INTRODUCTION Purpose: To analyze data obtained from the burn-through round robin test III for aircraft thermal/acoustical insulation blanket. Date : 09/2000 - 02/2001 Participants: 8 Labs (A, B, C, E, F, G, I, J) Burning configuration Burner type: P (Park) Igniter position: 10-12 o’clock (Lab B at 12:30, Lab C at 10). Turbulator-Nozzle distance: 3.75 in. (Lab C at 3.94) Calorimeter manufacturer: V (Labs A and G: M) Thermocouple manufacturer: (Labs A, G, I, J: X; Labs B, C, E, F: T) Air Velocity Meter: O (Lab C: X) Fuel: Jet A Samples: Fabricated by The Mexmil Company.

4. ROUND ROBIN III MATERIALS Notes th< 0.5 in Blanket Blanket Thickness Blanket Barrier Estimated Approx. Approx. ID Construction per Layer Density Material Failure Fiberglass Barrier (inch) (lb/ft 3 ) Mode and Required Required Time (yds 2 ) (yds 2 ) A 2 Layer Fiberglass 1.0 0.6 N/A Burnthrough 214 0 (30-40 sec) B 2 Layer Fiberglass/ 1.0 0.42 Nextel Exceed Heat 214 107 Nextel Paper Paper Flux Limit (>300) C 2 Layer Orcobloc 1.0 0.6 N/A Burnthrough 0 214 OPF (240) D 1 Layer Fiberglass/ 1.0 0.42 Pre Ox Burnthrough 107 107 1 Layer Pre Ox- 3/16 7 PAN Felt (190) PAN Felt (7641) E 1 Layer Fiberglass/ 1.0/0.15 0.42/15 Basofil/ Burnthrough 107 214 2 Layer Basofil/ Aramid (205) Aramid Felt (4759R) Felt F 1 Layer highloft 0.5 0.83 N/A Exceed Heat 0 107 aramid/inorganic Flux Limit (230)

5. FLAME TEMPERATURE PROFILE ( observed from the burner ) A B C E F G I J Thermocouple Number LAB Temperature T of each thermocouple was averaged over 6 calibrations for all labs. The difference  T is as large as 214 °F between flames from Lab A and Lab E. 1804 TEMPERATURE RANGE PROPOSED 927 1038 1149 °C Temperature °F 1093 982 A E

6. FLAME TEMPERATURE-FRONT HEAT FLUX CORRELATION Heat Flux (Btu/s.ft 2 ) E I G B F J C A Temperature (°F) - In the temperature range of 1800-2000 °F, data from 6/8 labs showed that the front heat flux-flame temperature relationship is linear. - A front heat flux in the (14-16 Btu/s.ft 2 ) range could be generated when the average flame temperature is in the range of (1875-1950 °F). - A 5% variation of flame temperature (T 2 - T 1 )/T 1 yields a heat flux variation (Q 2 - Q 1 )/Q 1 as large as 25%. 2  Q = 0.8 Btu/s.ft 2  Q/Q = 3%

7. Average Intake Air Velocity (ft/min) + Temperature (°F) 2  T = 20 ºF 2  V = 100 ft/min INTAKE AIR VELOCITY & FLAME TEMPERATURE CORRELATION G B C I F J A ELab Intake air velocity as recorded by each lab was plotted in the increasing order (blue dots). The corresponding flame temperature was then correlated (green dots). The graph showed that an increase of in-take air velocity V about 70 ft/min could yield a flame temperature increase of 100 °F. Temperature Intake air velocity

8. 70 95 120 145 170 TEST CELL TEMPERATURE - FLAME TEMPERATURE CORRELATION (1) Calibration/Test Test cell temperature (ºF) Flame temperature (ºF) Flame Temp.(°F) TC Temp. (°F) Data from from one lab showed that the flame temperature varied in accordance with the test cell temperature variation.

9. Test cell temperature (ºF) Flame temperature (ºF) The cell temperature was plotted in the increasing order. The above graph showed that, above 100 °F, the flame temperature increased with increasing cell temperature. This behavior was not obvious if the cell temperature is smaller than 100 °F. Indication is that a burn-through test should begin only when the cell temperature had returned to the ambient temperature. 2  T = 20 ºF TEST CELL TEMPERATURE - FLAME TEMPERATURE CORRELATION (2)

10. Relative Humidity (%) Lab 1800 1850 1900 1950 2000 2050 Flame Temperature(ºF) E B C G J F I A RELATIVE HUMIDITY AND FLAME TEMPERATURE 2  T = 20 °F 2  H= 6 % Flame temperature was plotted along with increasing humidity for each lab. Data does not show convincingly that the relative humidity affects flame temperature. For 6 labs from E to A (EBGJFA), the flame temperature increased with increasing humidity while for 4 labs from C to I the variation is reverse.

11. E I B G F J C A FLAME TEMPERATURE AND BURN-THROUGH TIME CORRELATION (Graph 1) A E C D Lab  t/t ~ 5-10% Burn-through time (s) Average burn-through time for 4 sample materials A, E, C, and D by 8 labs was plotted versus increasing flame temperature. For A and E samples, 6 over 8 labs obtained burn-through time data that is in agreement with temperature variation. For C and D samples, 4-5 labs obtained consistent data.

12. E I B G F J C A Lab Temperature (ºF) 2  T = 20 ºF (  T/T~0.5 %) 2  Q = 0.8 (  Q/Q = 3 %) FLAME TEMPERATURE AND BURN-THROUGH TIME CORRELATION (Graph 2) The flame temperatures for all labs were plotted in the increasing order to correlate with burn-through time data. 6/8 burner set-ups generated flame temperature in a pretty narrow range (1850-1950 °F). CURRENT RANGE

13. EFFECT OF MATERIALS COMPOSITION VARIATION The shape of the heat flux curves Q = k f(  T) is a characteristic of materials composition. The rear heat flux curves of samples A, E, C, and D as independently recorded by Labs F, I, and J were utilized to examine the effect of materials composition variation. Below are heat flux graphs for sample E2 and A2 by labs F, J, and I where: (1) the E2-signals were zero for 50 seconds, then started increasing. Burn- through occurred with a sharper increase of the rear heat flux at around 95- 110 s. (2) the A2-signals increased smoothly from nearly zero to 0.7 in t ~ 20-25 s, indication of a near burn-through.

14. Heat Flux (Btu/ft 2 s) Time (s) (Ja and Jb) SAMPLE E2 - LAB J, burned through at 110 s. (Fa and Fb) SAMPLE E2 - LAB F, burned through at 95 s (Fa) (Ja) (Jb) (Fb) HEAT FLUX CURVES OF SAMPLES E2 BY LABS F, J, AND I

15. SAMPLE E2 - LAB I

16. Data from LAB I SAMPLE A2 Data from LABs F and J (Fa) (Fb) (Ja) (Jb) Burned- through

17. EFFECT OF THE TEST CELL VENTILATION CONDITION A strong vertical ventilation = stronger vertical air flow ---> longer burn-through time. A poor vertical ventilation = slower vertical air flow ---> faster burn through Ventilation alters the turbulent nature of the flame, both spatially and temporally FLAME MODELS UNDER DIFFERENT VENTILATION CONDITION

18. For some kinds of sample blanket the vertical clamps cannot hold specimens tightly on the frame during the fire. The specimen was shrunk and/or blown towards the calorimeters. 1/ a failure due to an exceeding heat flux resulting from a nearer distance between the hot surface of the blanket and the rear calorimeter. 2/ an increase of burn-through time due to a longer distance between the burner cone and the blanket. EFFECT OF MATERIALS SHRINKAGE

19. (1) The temperature-front heat flux relationship by 8 labs indicate that, within a 15% error, flame temperature is directly proportional to the flux. (2) The average flame temperature required for generating a heat flux in the range of 14-16 Btu/s.ft 2 in the range of 1875-1950 °F. (3) A 5% variation of average temperature (T 2 - T 1 )/T 1 could yield a heat flux variation (Q 2 - Q 1 )/Q 1 as large as 25%. This could cause a significant difference in burn- through time. Therefore, the allowable fluctuation of the average temperature should be 50 °F (2.6%) instead of 100 °F (5%) (1900 ± 50 °F instead of 1900 ± 100 °F ). (4) Factors need to be controlled The in-take air velocity, the cell temperature, the test cell ventilation, and the shrinkage of materials. Effect of humidity is not evident. (5) Heat flux curves indicated that materials composition variation in samples may not be a significant factor affecting the burn through results. CONCLUSION