de the first 47 pages for reference to the pictures. · the fire situation of maximum severity and...
TRANSCRIPT
Report No. NA-69-37(RD-69-46) IERIM REPORT
Project No. 430-012-02X
EFFECT OF GROUND CRASH FIRE ONAIRCRAFT FUSELAGE INTEGRITY
DECEMBER 1969
Re producd b4
NATIONAL TECHNICALINFORMATION SERVICE
S~pringfield, VS. 22151
DEPARTMENT OF TRANSPORTATIONlFEDERAL AVIATION ADMINISTRATION
National Aviation Facilities Experimental CenterAtlantic City, New Jersey 08405
I.-7-
_ 3-
faiiis skill an sevie tospotFArsacdvlpetadipe
anstyfilavation prgastruhandlforshexdevelomento and evalration of avommon
concPeS, procedUres, systems mrid equipment.
Copies of this report may be purchased for $3. 00 each fromn the
Clearinghouse for Yederal Scientific and Technical Information(CFSTI), Springfield, Virginia, 2Z151.
UL
-US
QL 3
LoD
z 0-
V Mf. ox
Z~~ ~ IS.I ,-- ico
z 2iui 4,
CDoo
u 13 0
w UU
Ix -l >.
Q - .,
00 E
.2 CL)00 0L C.,o -
G3 C)
cc a
0)0
0000006. 0-
*0.0*.0.
00
. 0 0 . 0
.0. *L CL
00 *
40
00 0000
0 * * EUDo ** .. 4 4
0.aU.
o Co
* 0
0 o-,0 -o0
INTERIM REPORT
EFFECT OF GROUND CRASH FIRE ONAIRCRAFT FUSELAGE INTEGRITY
PROJECT NO. 430-002-02X
REPORT NO. NA-69-37(RD-69-46)
Prepared by:
GEORGE B. GEYER
A. OrL
SYSTE4S RESEARCH AND DEVELOPMENT SERVICE
December 1969
This report is approved for unlimited availability. It does notnecessarily reflect FederAl Aviation Adr!n!_-tr-tfnn policy 4n all
respects and it does not, in itself, constitute a standard,specification, or regulation.
DEPARTMENT OF TRANSPORTATIONFederal Aviation Administration
National Aviation Facilities Experimental CenterAtlantic City, New Jersey 08405
ABSTRACT
X- A mathematical model was formulated which permits a calculationto be made of the time required for damage to occur to the aluminumskin covering an aircraft fuselage when it is exposed to maximu: spillfire conditions, The damage time was defined as the time required forthe aluminwii akin to melt.
The model tqas developed through consideration of the heat transferratrs by convection and radiation across a simplified aircraft fuselageconfiguration. The resulting differential equation was &oived using anumerical technique. The results indicate that the minimum time requiredfor skin damage to occur to the largest commercial aircraft now in serviceis less than 40 seconds. The fuselage damage time predictions, madethrough the use of the mathematical model, correspond closely withmeasurements made on simulated aircraft skin configurations employing a40-.foot, stainless-steel-covered section of a four-engine jet aircraft.fuselage.
iii
TABLE OF CONTENTS
ABSTPACT iii
INTRODU(CfO 101
Background 1
DISCUSSIOL 1
General 1Iosts on SFunsese--Ccvd Fu~oiage 2
Discription of the Tests 2Results 8
Tsts on Altuninum Panels 11Description of 1e5t. 11Results 17
Ti-Ttriparature i itory of Aircraft Skirt Heto:ing 17Deve.lopment og a Matbemat.icA Model 17Verification of the Mathe~atical Model 22
Stain.e a Steel TectR 22Al umInum Panel TeitA 22
S IRY OF BESULTS 31
COA'CLUS IONS 32
RECOMMNDATIONS 33
RZ FEFENCES 34
ACKW.EDGM1ENT 35
APPENDIX I Stainless-Steel-Covered Fuselage Tests, 1-1Thermocoup~e and Radiometer Data (9 pages)
APPENDIX II Critical Phases of An Alumin, m Panel 2-1Test (6 pages)
APPENDIX III Aluminua Panel Tests, Thermocouple and 3-1Radiometer Data (10 pages)
APPENDIX IV Development of the Mathematical Model 4-1(8 pages)
V
LIST OF ILLUSTRATIONS
Figure Page
] Exterior View of the Fuselage Section and the 3Fire Pit Location
2 Schematic Drawing of the Stainless-Steel-Covered 4Fuselage and Location of the Fire Pits(Not to Scale:
3 Closeup View of the Radiometer and Thermocouple 5Inst&i.Iations
4 Xnterior View of the Fuselage Showing the 6Thermocouple and Radiometer Locations
5 Plan View of the Fire Test Site (Not to Scale) 7
6 Skin Temperaturec for 0.031-Inch Stainless Steel 9As a Function of Fire Einposlire Time
7 Effect of Wind on Pool Fires 10
8 Cross Section of the Altrainum Fire Test Panel 12Conf:lguration (Not to Scale)
9 Upper Aluminumi Panel Installation 13
10 Lower Alumintmi Panel Installation 14
11 Exrerior View of the Test Panels in Position 15
i.2 Elevation View of the buseiage Section and Fire 16Test Pit (Not to Scale)
]3 Experimental Skin Ttmperatures for 0.020 Inch 18Aluminum as a Function of Fire Fxposure Time
14 Experimental Skin Temperatures for 0.090 Inch 19AlcAninum as a Function of Fire Exposure Time
15 Fr'agments of Aluminum Panels Retrieved 50 to 2060 Feet Downwind from the Fuselage
16 Simplified Model o' Aircraft Skin Heacins 21(Vol to Scale)
vii
LIST OF ILLUSTRATIONS (continued)
F iguzre Page
17 Skin Temperature for 0.031-Inch Stainless 23Steel as a Function of Fire Exposure Timeas Calculated from the Model
18 Calculated Melting Time for Aluminum Aircraft 24Skins as a Function of the Temperature Risefor Stainless Steel
19 Skin Temperatures for 0.020 Inch Aluminum as 25a Function of Fire Exposure Time
20 Skin Temperatures for 0.090 Inch Aluminum as 26a Function of Fire Exposure Time
21 Melting Time for Differert Thicknesses of Aircraft 27Aluminum as a Function of Fire Exposure Time
22 Minimum Skin Thickness of Some Current 29Commercial Aircraft
23 Minimum Skin Melting Time as a Function of the 30
Gross Weight of the Aircraft
viii
INTRODUCTION
Purpose
The purpose of this investigation was to fornulate a mathematicalmodel which would permit a calculation to be made predicting the timerequired for an aluminum aircraft fuselage to melt when exposed toaircraft fuel fires of maximum sevetity, and also to obtain thermal databy conducting full-scale fire tests on a 40-foot, stainless-steel-covered fuselage section of a four-engine jet aircraft to verify thevalidity of the mathematical model for predicting fuselage fire damagetime.
Background
The incidence of fire following survivable aircraft accidentsfrequently leads to tragic loss of lif. which could largely be preventedby a sufficiently rapid fire suppression response.
In incidents involving commercial aircraft, the large number ofpassengers aboard cannot be effectively evacuated through the fire bycurrently available techniques. However, as long as the aircraftfuselage retains its mechanical integrity following a survivable inci-dent, the passengers are afforded some degree of protection from hightemperatures, limited oxygen supply, and the toxic pyrolysis productsof the cabin appointments.
Commercial airliners are constructed of the thinnest aluminumalloys consistant with structural requirements to effect the greatesteconomy in weight. These alloys melt at temperatures significantlylower than those of the flames from burning hydrocarbon fuels. There-fore, passengers may be exposed to maximum hazard conditions relativelysoon after the incident occurs.
Until the present time, there has been no method available topredict, in a precise manner, either the time available to effect adequatefire suppression and passenger rescue or the time available to the firedepartment to respond to an aircraft accident. Therefore, this studywas undertaken to obtain sufficient data to permit a meaningfulestimation of these critical time parameters to be made.
DISCUSSION
General
The development of a mathematical model was based upon the heattransfer to and from an aircraft fuselage when exposed to two differentfire test environments. The first condition exposed a stainless-stecd-covered fuselage section to narrow rectangular JP-4 fuel fires located
at different distances on the upwind side, while the second concernedthe fire situation of maximum severity and danger to the passengers;namely, the case where a large fire is adjacent to the fuselage.
Tests on the Stainless-Steel-Covered Fuselage
Description of the Tests: The test article comprised a 40-footsection of a four-engine jet aircraft fuselage completely coveredexternally with a 0.5-inch-thick layer of ceramic fiber insulation and0.031-inch, Type 304, stainless steel sheets bolted to the fuselage.This configuration was employed to protect the fuselage from destructionby fire during the teat program.
The irztrumentation of the test fuselage and the pool firelocations relative to the fuselage are presented pictorially in Figure 1and schematically in Figure 2.
The stainless steel panels were numbered consecutively from1 to 10 starting at the rear of the fuselage. All instrtuentatton wasconfined to the upwind side of Panel No. 6. The thermocouple wirespenetrated the fuselage from within at Stations Ta, Tb, Tc, Td, Te, Tf,nnA T_ and w o rtark-welded tn the outnide surface of the steel skin.The four water-cooled, nitrogen-gas-purged radiometers were mountedflush with the stainless steel skin and adjacent to the thermocouples atStations RA, RB, RC, and RD (Figure 2). One thermocouple at Station Thwas extended 30 inches horizontally from the center of the fuselage tomeasure the air/flame temperature (Figure 3).
The upper interior portion of the fuselage is shown inFigure 4. All instrumentation wiring was contained in an undergroundconduit system leading from the center of the fuselage to theinstrumentation trailer as Fhown in Figure 5.
Still and motion pictures were taken of each fire test for timedata analysis and documentation from positions shown schematically inFigure 5.
The fire environment comprised three rectangular pits, 10 feetwide and 30 feet long, located equidistant from the ends and parallel tothe fuselage. Each pit contained sufficient water to produce a levelsurface free from the intrusion of "islands" through the fuel surface.The JP-4 fuel charge to each pit was 0.35 gal/ft 2 .
Four fire tests were performed in the following sequence:Test No. I in Pit A located 20 feet from the fuselage, Test No. 2 inPit B located 10 feet from the fuselage, and Test No. 3 in Pit C whichwas adjacent to the fuselage. The fourth test was conducted employing
2
(4-
Hv
Vv 1N
r4
5HIE* L/ ,,~ r.
V VIP ' , T .i
) P1-4
zQ H
oo
I CPJUU
-44 a4C'
U)U
94
4-F4
0I CD E1C
F-__ _E__4 o4
0.zH
F-44
-. 1
'I' 0
z
('.1H
H0
/ 'I
/ .,-. - _ H
~J~) -~. .- H
.4
oUo
0
IA'-4
H
0
Hp.4
1.:( ENi~i )
-14 X ,4. 1r t- A
_i I( A' 11G'' 1
VIHF. OF Iill. I A '('O- ANk
MIAU x11':11 LoCATrINrI:;
< 4>2
w H)0~--
(f-4
O Q
Uz
0 r04
Pits Nos. B and C simultaneously. This series of tests was desigt.edto obtain thermal data concerning the efftct of fire pit location onthe exposed fuselage.
Results. The thermocoaple and radiometer data obtained for thefour stainless-steel-covered fuselage tests are presented in Appendix I.These data show that the most rapid rise in skin temperature wasobtained in Test No. 2. Data from the other three tests showed a slowertemperature rise in the aircrafL skin which vesulted from the differentfire pit locations and the poor fire coverage caused by variable windconditions on the relatively narrow fires at the time of the tests.
The stainless steel skin temperature rise for Test No. 2 isplotted as a function of time in Figure 6. An examination of theinstrument data showed a delay of approximately 13 seconds from the timeof ignition until the fire built up sufficiently to cover the instru-mented area on the simulated aircraft fuselage. This time delay, due tofire buildup, was used to adjunt the data points as shown by the solidpoints in Figure 6. Each solid point represents the same reading as theopen point at the same temperature, but is has been shifted to the leftbideL U, tie , 13 aMf,., U)'. Therefore-the ... L,- Poir, ta are rapra-
sentative of an aircraft incident in which the fuselage is totallyinvolved in fire with little or no delay in ignition time. This approachwas consistent with the requirements for the development of a mathemat-ical model which would predict the fuselage melting time representativeof the immediate involvement of the fuselage in flames.
During the course of the stainless-steel-covered fuselageexperiments, the effect of wind on free-burning pool fires (Reference 1)was evident and is considered to constitute an important factor intactical aircraft firefighting techniques. The effect of wind is tobend the flame In rhe downwind direction- and the flame anzle is afunction of the wind velocity. The flame angle is defined as the angleof tilt of the flame from the vertical. The flame-trailing effect aroundthe test article is shown for wind velocity of 6 to 8 mi/hr in Figure 7.In this test, the downwind edge of the fire pit was 20 feet from thecenterline of the fuselage. Photographh (c) and (d) of Figure 7 show thelarge increase in the effective width of the fire caused by the flame-trailing phenomenon, and it will be noted that the flames are in actualcontact with the fuselage. The temperature data presented in Appendix I,Test No. 1, Figure 1.5, show that the stainless steel skin temperatureat Station Td reached a maximum of 860°F in approximately 100 secondsafter fuel ignition while the ambient air/flame temperature outside thefuselage rose to 1200OF in 30 seconds after fuel ignition.
OZ
000
0@0o ;U 4
r-3 Co_ _ _ _ _ _ _ _ _ _ _ o
j s~aij~~j-zirij'dadia.uNim
9A
I' A
(a) Fire Pi t Loa o (b) Fucdl Ignition
RR io
Comm.S _ag
(c) Burning Tine 10 Seconds (d) Burninig Time 30 Seconds
FIG. 7 EFFECT' OF WINE) ON POOL FIRES
10
Therefore, it is evident that relatively small fires remote andupwind from the fuselage may inflict serious fire damage to an aircraftfuselage as a result of the flame-trailing effect.
Tests on Aluminum Panels
Descriticn of Tests: The second series of tests employed fourdifferent thickneses of st&ndard aluminun aircraft paneling inserted inthree openings, 3 by 3 feet, cur through the fuselage and steel covering.Two of the four tests employed panels constructed of Alloy 2024-T3conforming to Federal Specification QQ-A-362 (Alelad). One was 0.016 inchthick (Test No. 5), and the second was 0,.040 inch thick (Test No. 6).The other two panels were constructed of Alloy 7075-T6 (Alclad) conform-ing to Federal Specification QQ-A-287 and were 0,020 inch thick(Test No. 7) and 0.090 inch thick (Test No. 8). Each panel was backed bya 2-inch-thick layer cf "AA" fiberglas insulation with a density of0.60 lb/ft backed with a facing of polyvinylchloride. This configura-tion was designed to approximate standard aircraft construction and toprovide all of the essential parameters necessary to verify the validityof the mathematical model. A cross-sectional drawing of the test panelconstruction is presented in Figure 8 and photographs of the instrumentedpanels in Figures 9 and 10. An exterior photograph of the fuselagesection with panels installed for testing is shown in Figure 11.
The fire test environment for the aluminum panel tests utilizeda 2500-ft2 pit located on the upwind sie and adjacent to the fuselage.The simulated spill consisted of 750 gallons of JP-4 fuel floated onwater for leveling purposes. The large pit was designed to providerelatively complete fire envelopment of the fuselage and moximum fireexposure. Photographs of a typical fire test are contained inAppendix II, Figures 2.1, 2.2, 2.3, and 2.4.
The fuel was ignited from the instrment panel inside theinstrument trailer by a high-intensity electric spark generated at thefuel surface and in the center of the upwind side of the pit as shown inFigure 12. After ignition, the fuel was allowed to burn until the skintemperature of any one of the panels reached 12000 F. The fire extinguish-ment operaLion was then started and continued until the fire was extin-guished to prevent the des ruction of the internal structure of thefuselage and instrumentation.
Inside the fuselage at the instant of fuel ignition, two electricclocks were activated which were located in the line of sight from theinbtrumentation camera to the aluminum test parels. One camera waspositioned to photograph and record the burn-through time of the two upperpanels and the second to cover the lower panel. These cameras are shownin Figures 9 and 10.
11
STAINLESS STEEL
" TYPE 304ORIGINAL AIRCRAFT --
SKIN CERAMIC FIBERINSULATION
(0. 5 in THICK)
s- BOLT
POLY VINYLGHLORIDE ALUMINUM TEST PANELSHEET (3 ft x 3 it)
CA BIN 3 inINTERIOR
"AA" FIBERGLASS THERMOCOUPLEINSULATION LOCATO NS(2-in THICK)
SUPPORT STRAP-
ASBESTOS GASKET
FIG. 8 CROSS SECTION OF T1E ALUMINUI FIRE TEST PANEL CONFIGUEATION( NOT TO SCALE )
12
w~fit
so MI
'A -4
110
1id...........
E-4a AA
In
C4
'I-
~~fz 1- 1-1
Ci14
1-4
Results: The results of the tests made on the simulated aluminum
aircraft skins are presented in Figure 13 for a thickness of 0.020 inchand in Figure 14 for a thickness of 0.090 inch. In these graphs, theopen points are the direct temperature measurements, and the solid pointshave been adjusted to allow for the flames to spread across the fire pit
as cescribed for the stainless steel tests. The radiometer and thermo-couple data are presented in Appendix III and photographs representativeof the fire conditions in Appendix II.
During the course of the fire tests involving the 0.016-inch,0.020-inch, and 0.040-inch aluminum panels, it was observed thatrelatively large pieces of the test panels, as well as molten globulesof metal, were ripped from the fuselage and carried from 50 to 60 feetdownwind. Representative fragments of the panels and melted metal areshown in Figure 15. Some of the aluminum fragments showed severe heatcrazing and embrittlement, although the edges displayed clean breaksand no melting was evident. Further observation of this phenomenonindicated that the thermal updraft around the fuselage probably reached25 to 30 mi/hr which was apparently sufficient to rip off the thermallyweakened metal before it could melt completely. However, only a fewpieces of the 0.090-inch aluminum panels were retrieved on the down-wind side of the fuselage, and the larger part of these panels was foundcompletely fused under the fuselage.
Time-Temperature History of Aircraft Skin Heating
Development of a Mathematical Model: The development of a mathe-matical model predicting fuselage fire damage time (melting) was basedupon the quantity of heat transferred to and from an aircraft fuselageduring exposure to fire. Primary concern was given to conditions wherethe fire surrounds the aircraft and the flames impinge directly on thefuselage. This environment most closely approaches the steady-stateconditions necessary for the mathematical treatment of a free-burningpool fire.
Figure 16 shows the simplified model of the aircraft skin backedby a layer of thermal insulation through which the heat balance was made.In the model, heat gain to the aircraft skin is assumed to be by radia-tion and convection from the fire, while heat loss is due to (1) radia-tion, (2) convection, and (3) conduction. The difference between theheat gain and heat loss is accumulated by the skin and causes a rise inits temperature. This relationship may be expressed in general terms asfollows:
Heat Accumulated = Heat Input-Heat Loss
The detailed mathematical treatment of this thermal balance is presentedin Appendix IV.
17
_ _ - 0
0@
[z Z
C4
-4
-4
0 0 0 2
0 00~
00
0 0
C> 0 00
a SaaD[3C-2flJ~V)JadV~J. NDIS
18
HFi
0
00
0@
~z00
0 -.
4 4
I j 00
C*0C 0 C>C0 QCC)C
C:> OD(I0A S:3f UD~ G 2d C IV-8a i~a, um
19
Awl*
t1v
.40k
METAL SKIN
- RADIANT 1LEATING
GONDUGTIVI GOULINCi 4-
INSULATXONL NET CONVECTIVE HEATING
ACCUMULATION RATE -
RADIANT COOLING
FIG. 16 SI24PLIFIED MODEL OF AIRCRAft]: SKIN HEATING (NOT TO SCALE)
21
Verification of the Mathematical Model
Stainless Steel Tests - The results of calculations made employ-ing the mathematical model based on the 0.031-inch-thick stainless steelskin are presented in Figure 17. The solid curve is the result ofcalculations made at specific time intervals until the 8kin temperaturereached approximately 15000 F. The parametric data presented in Table 4-I,Appendix IV, were used to obtain the calculated results. For comparisonof the calculated results with tho experimental data obtained for the0.031-inch stainless steel tests, the curve has been superimposed onFigure 6.
It is noteworthy that these data were taken from Test No, 2in which the most rapid temperature rise was recorded. The slowertemperature rise during the other tests was due to poor fire coverage ofthe fuselage which was caused by adverse wind conditions during thesetests.
Figure 18 shows the calculated fire damage time foraluminum aircraft skins as a function of the temperature rise for stain-lees steel. Curves are shown for skin temperatures of QO0oF and 1200°FeThe two data points shown are adjusted values taken from Figure 17. Itwill be noted that they are in good agreement with the data predicted bythe mathematical model,
Aluminum Panel Tests - The results of the calculations madeemploying the mathematical model for aluminum aircraft skins are shownby the curve in Figure 19 for a thickness of 0.020 inch and by the curvein Figure 20 for a thickness of 0.090 inch. The experimental data fromthe fire testa are shown on each of the figures as points. The openpoints are the actual temperature measurements, and the solid pointuwere adjusted a& described for the stainless steel. tests.
In Figure 21, the aluminum panel thickness is plotted as afunction of time to reach the two tempexature levels which constitutethe boundaries of the melting range. For the aluminum alloys employed inthese tests (2024-T3 and 7075-T6), the beginning temperature was approxi-mately 9001F and the ending temperature approximately 200CF. The cal-culated curves and experimental points show reasonable Wgreement althoughthere are some deviations representative of the G,09O-Inch-thickaluminum. The 0.090-inch-thick aluminum panel shows the widest devi.ationfrom the calculated curve as the temperature approaches the upper limitfor the melting range.
The results of comparisons of stainless steel and alwiinumcalculations and experimental test results indicate that the calculLtionsare adequate for use as a method of eatimrting the time required fordamage to occur to an aircraft fuselage in an accident involving instant
22
_____ _______ ______0
00
o
H0r14 4
n0
0 V
4
[-4P-
UI -
XOU)
0 0
NN
MIMS_ z
23.
1340
00
CU)0 0
F-Q
00
000
1.4
00 v
____ aS3VI _____X3 21_ 92 324U
-oo0 00
oo
C,,
0 z *o
0LO~
0Ci)
000
o
C,
rqrz
0r~l : 41
00'-o
0
- - 0
0 0 0 0 0 00 0 0 0 0
,_ S L3OD3U- ..F.,.VU3AtI, MIMS
25
o
0
0
0 0<
0
z r0 !N
U0
0o00
o o IS
000 0 0 0C 0
* 00 o x 0
10 G IV:A NM20
-d,0
00
0 0
N aN
0E-4
_ _ -4.
1- F4
4-
CA)
0 P
.-LE-
_ - .0savl 04ilod329
27I
extreme fire conditions. The total elapsed time necessary for an aircraftskin to melt can be calculated if the skin thickness is known. However,for most aircraft, the skin thickness varies along the fuselage with thestructural requirements.
Figure 22 shows the minimum aircraft skin thickness as afunction of the gross weight of the aircraft. The curve shows approximatevalues for aircraft of several manufacturers which range from small single-engine aircraft to intercontinental jet aircraft. It should be emphasizedthat Figure 22 gives the minimum skin thickness for a given aircraft grossweight and the maximum skin thickness on the same aircraft may be severaltimes the minimum.
The curve in Figure 23 was developed from data taken fromFigures 21 and 22 and shows the time required for an aircraft skin tomelt as a function of the aircraft gross weight. The procedure was toplot the minimum skin thickness of the aircraft taken from Figure 22 andthe fire damage time as the time required to reach 1200°F from Figure 21.The curve in Figure 23 shows that the aircraft skin melting time variesfrom about 10 seconds for small aircraft to nearly 40 seconds for thelarger aircraft. Those melting times are based on immediate fire involve-ment and a larRe fire so they represent the minimum time available forfire suppression before the fire penetrates the cabin. Should ignitionnot occur immediately or If an appreciable time is required for the fireto build up, this addi.ivnal time would be available for fire suppression.However, neither of these conditions can be relied upon in an aircraftincident. Therefore, fire suppression techniques and equipment should bedesigned for effective operation within the minimum time available ormodifications in aircraft construction should be considered to extend theminimum fire damage time if protection to passengers and crew is to beobtained.
28
U0
-4
4-
000
C-o
(1) C) C:
S3HOI-S~MM')HJMIMSI~fIINo
290
E-4
0H rw
1-4
-4
z- _ _ _ _ _ _ _ _ _ _ _ _ _ _ - -
-fJ
CD~
0 030
SU IARY OF RESULTS
The results obtained from theoretical considerations of heat transferand large-scale fire tests are:
1. The calculated melting times obtained using the mathematicalmodel for four different thicknesses of aircraft aluminum show acceptableagreement with the experimental results obtained from the large-scalefire tests.
2. The melting time of four aluminum panels exposed to severefire conditions varied from approximately 8 seconds for the 0.016-inch-thick aluminum to 38 seconds for the 0.090-inch-thick aluminum.
3. A simulated spill fire 10 feet wide and 40 feet long located10 feet from the stainless-steel-covered fuselage on the upwind sideindicated that 0.031-inch-thick aircraft aluminum would melt inapproximately 25 seconds.
4. The effect of wind on pool fires significantly increased thedestructive range of the fire plume downwind from the actual spill boundaryas a Leult Of ttaiie trailing.
31
CONCLUSIONS
Based on the results of thermal calculations and experiments, it isconcluded that:
1. The mathematical model developed in this report is adequateto predict the melting time of aircraft aluminum paneling vznder severefire conditions.
2. The insulated aluminum fuselage skin of current aircraftprovides low resistance to external fuel fire. The melting time offuselage panels and subsequent fire entry into the cabin interior fromfires of maximum severity is on the order of 10 to 40 seconds dependingon skin thickness.
3. A fuel spill fire remote and on the upwind ride of anaircraft fuselage may inflict severe fire damage as a result of theflame-trailing phenomenon.
32
RECOMMENDATIONS
Based upon full-scale fire tests and mathematical methods, it isrecommended that:
1. The mathematical model and the curves developed in thisreport be employed to estimate the approximate melting time for aircraftaluminum of different thicknesses when exposed to aircraft fuel fires ofmaximum severity.
2. The data and information contained in this report for thetime required for the melting of fuselage skin, under severe fireconditions, be used as the primary criteria for estimating airportfirefighting equipment requirements.
3. Consideration be given to the possibility of extensiveflame spread around an aircraft fuselage as a result of flame trailingunder variable wind conditions.
4. Studies be conducted on means of extending occupant survivaltime by encapsulating the aircraft cabin interior in a flame resistantbarrier.
33
REFERENCES
1. Welker, J. R. and Sliepcevich, C. M., The Effect of Wind on Flames,Technical Report No. 2, NBS Contract No. CST-1142, University ofOklahoma Research Institute, Norman, Oklahoma (November 1965).
2. Mickley, H. S., Sherwood, T. K., and Reed, C. E., pplied Mathematicsin Chemical Engineering, Second Edition, McGraw-H 1 (1957).
3. Copley, J. A., An Analytical Method for Predicting the Temperature-Time History of a Hollow Cylinder Enveloped in Flaeme, TechnicalReport No. 2073, U.S. Naval Weapons Laboratory, Dahlgren, Virginia(December 1966). AD804084.
4. Neill, D. T., Heat Transfer from Uncontrolled Buoyant DiffusionFlames, Ph. D. Thesis, University of Oklahoma, Norman, Oklahoma(1968).
5. Perry, R. H., Chilton, C. H., a 1:K ".irkpatrick, S. D., Editions,Chemical Engineers Handbook, Fourth Edition, McGraw-Hill,New York (1963).
6. Stull, D. C., Edition, JANAF Thermocouple Tables, AF ConttactNo. AF04(611)-7554.
34
.ACKNOWLED)GMENT
Appreciation is expressed to Dr. J. Reed Welker of UniversityEngin~eers, Inc., Norman, Oklahoma, for the mathematical. interpretationof the fuselage fire damage time.
35
APPENDIX I
STAINLESS-STEEL-COVERED FUSELAGE TESTS,
niERMOCOUPLE AND RADIOMETER DATA
1-1
0
0£W Z
000
4< N
00 1-40
000
0~~~ ~ ~ Z</IUXnj J3
1-3 ~ ~ PRCDN ASR N
14
-_
_
I_
_
12 -- RA 1 .6RB
RC0
10 RD
RADIOMETERLOCATIONS
4) _)Y
2-'
TIME AFTER IG NITION -SECONDS
FIG. 1.2 TEST NO. 2 - RADl1METER DATA FOR STAINLESS-STEEL-COVERED FUSELAGE
1-4
0
RARB
RC 0
RD
10H RADIOMETER
LOCATIONS
TIMl)E AFTERt IGNIT'ION -SECONDS
FIG. 1.3 TEST N~o. 3 -RADIWAE:TER DATA FOR S3TAINLESS-STEEL-COVEREDl FUiSELAGE
0AA
'NJ RRB
CRD- _ 0
H RADIOMET ER< LOCATIONS
u 0 0 0 01OU0z 14UT1IE AFTERl IGNITION-SECONDS
F7C. 1.4 TEST NO. 4 - RADIOMETER DATA FOR STAINLESS-STEEL-COVERED FUSELAGE
1-5
'0
LLLJ
I-.-
r re
0
H 002 ~ ~ ~ ~ ~ ~ ~ ~ -S <C1, C idr ,U,
2400 -_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
2000
v) 16uow4
[4r
40To Q'
4002 06 8 0
TIME AFT KR IGNITION-SECONDSb
FIG. 1.6 TEST 1NO. 2 - THERMOCOUPLE DATA FOR STAINLESS- STEEL-COVERED FUSELAGE
1-7
2000 -_ _ _ _ _ _ _ _ _ _ _ _ _ _
Tc
16 00 ( : Td ______
U) g 2
THERMOCOUPLE[' 1200[ LOCATIONS
800
400
TIME AFTER IGNITION -SECONDS
FIG. 1.7 TEST NO. 3 - THERHOCOUPLE DATA FOR STAINLESS-STEEL-COVERED FUSELAGE
2000___ _
To T,
T1.600 T - LTh _ _ _ _ __ _ _ _ ___
THERMOCOUPLE~izoo LOCATIONS
40
02 ?0 4 0 6 0 80 100 120
TIME AFTER IGNITION-SECONDS
FIG. 1.8 TEST NO. 4 - THERMOCOUPLE DATA FOR STAINLESS- STEEL-COVERED FUSELAGE
1-9
CRITICAL PHASEk:S OF AN ALUMINUM PANEL rC.ST
2-1~
CD
I-IT BLAN
I ~V 4.,
S., £ ,
4 ~I J~r,
Co
* z0
'Ii NCo4, 0
2_ hI 'C
Hlii N
I #M#,
CV.,
F' 4, C'.
4' I
sat
2-/i
T -o
ASIlkI44
C
7,
r I/K
H
4)r~ NS...F-i
m
KV (:7/i UAt i 1~#
/
r jjibi/~ 4~,1 / ff7/ 4~st~~ /&\ 4y /AA
a. m 'Nb//i1 Wv
APPENDIX III
ALUMINUM PANFLt rESTS, 'MERMOCOUP1,E, ANT) RADIOMt'rFhR DATAA
3- 1
12'0
R A
RB
RD~ NOT OPERATIONAL
RADIO MET RLOCATIONs
4 -
0 10 20 30 40 50 60 70TIMvE AFTERt 10 NITION -SECONDS
FIG. 3.1 TEST NO. 5 - RADIOMETER DATA FOR ALII4INUM ALLOY 2024-T3ALCLAD (0.O16-~INCtH ThICKNESS)
33PRECEDING PAGE H[ANV
RA10A
R
R D
2~~~ 8 ~~RADIOMETER ____ ____
8
.4
-A\-
0 020 30 4 50 6170TIME. AFTxZrt 101 U :'ON-SF7CONT)S
FIG. 3.2 TEST NO. 6 - PU'JIM'ETER DATA FOR ALIIAIB ALLOY 2024-T3ALCLAD (0.040-1tNCI THICKNESS)
12 0RARB
3.0 RC 0
RDCZ
.4
HL
[ o 'A-_ _ _ __ _ _ _ _
-3-0
o NOT OtAINL
RA
RB
RC
RD
___ AI OMETER __
LOCATIONS
.4
0 10Q 20 30 40 50 60 70TIME AFTER IGN'ITION-SECONDS
FIG. 3.4 TEST NO. 8 - RADICI4ETER DATA FOR ALUMINUk4 ALLOY 7075-T6ALCLAI) (0.090-INC11 THICKNESS)
3-6
2400 _____-____
LEGE ND
A TOP P&'! L SKIN TEMPERATURE0) TOPl PANEL INSULATION TEMPERATURE
21 00 0 CENTU R PANEL SKIN TEMPERATURE
E3 CENTER PANEL INSULATION TEMPERATURE
SCLNTIEH PANEL AIR TEMPERATUREBOTTOM PANEL SKIN TEMPERATURE
* BOTTOM PANEL INSULATION TEMPERATURE-BOTTOM PANEL AIR TEMPERATURE
1800 -NOT OPIERATIONA L______
UV)
10 1500
9~100 _____
0.0
3 00 -_____
0u 10 3U 40 50 60 70 80TIME AFTERl ICr*ITION-SECONLUS
FIG. 3.5 TEST NO. 5 - THERMOCOUFLE DATA FOR ALUMlINtUl ALLOY 2024-T3ALCLAD (O.016-INOI IIIICIWESS)
3-7
F- z1
F-F- U o
0 0
0'
3-8n
0
fA
010
OL/
w~ ww
lu D- . 0-00
0, OO z OOw u LO 0
I- I- U L)U to'
3-9
CL
&
-
otOU < a
,---z z
&- - [
z~ w r'
cl <30O3 <
C___
0 H ''
a S ';t Zd U 0 C- 34 1 Ql I V1173 d IN' 3- 1,I
3-10
APPENDIX IV
DEVEL.OP1*TN OF THE MAT1UMATICAL MODE~L
4-1
DEVELOPMENT OF THE MATHIEATICAL MODEL
The development of the mathematical model was based upon heattransfer to and from the aircraft fuselage under conditions where thefire directly contacts the aircraft.
Figure 4.1 shows a simplified model of the aircraft skin backed bya layer of thermal insulation.
Metal Skin
,.,Radiant Heating
a qr
Conductive Cooling Insulation Ar...Net Convective Heating
k h(Tf - T)k (T - To)
Radiant CoolingAccumulation Rate R
p c" dT CUT
dt
Fig. 4.1 Simplified Model of Aircraft Heating.
In the model, heat gain to the aircraft skin is assumed to be by radiationand convection from the fire. Heat loss from the aircraft skin J. due toradiation, convection, and conduction. The difference betwuen the heatgain and heat loss is accumulated by the skin and raises its tenpe:'ature.The following terms are therefore included in the heat balance.
Radiation heating = oq (I)r
Radiation cooling = EaT 4 (2)
Net convective beating = h(Tf-T) (3)
Conductive cooling = Z (T-To ) (4)
Accumulation rate px (5)
4-3 PRECE S PAGE B LANK
The tcrms in Equations 1 through 5 are defined as follows:
T aircraft kin temperaturLTo = temperature inside insulation layer
Tf flame temperLture
= total absorptance of aircraft skin
qr radiant heat output of fire
f total emittance of aircraft skin
a r Stephan-Boltznan constant
h = convective heat transfer coefficient
k = thermal conductivity of insulationz = thickness of insulation
P - density of aircraft shinc = heat capacity of aircraft skin:z = thickness of aircraft skin
t = time
Since
Input - Output - Accumuiation (6)
Equations I through 5 can be co,bined to obtain
dT 4 kPCx -=aq + h ( Tf -T) - (CT _k (T.T 0) (7)
Equation 7 relates the rate of temperature buildup to the net heet gainedby the aircraft skin. In deriving Equation 7, several assumptions havebeen nade jTI order tn simnl i fu the morlPo - Tho temperat-re rhrn,_ghout rhoaircraft skin was assumed to be uniform because the skin is thin and itsthermal conductivity is high. The properties of the metal were assuiredto be known anci constant over the temperature range in question. Theradiant heat transfer from the flame to the aircraft wns assumtd to beconstant and the convective heat transfer coefficient was asauped to beconstant.
Equation 7 does not account for the amount of energy require4 tomelt the aluminum skin of the aircraft. Since the aluwinut is an alloy,it melts over a temperature range rather than at a particular temperature.If it is assumed that the fraction of aluminum melted over a given melt-ing temperature range is proportional to the fraction of the meltingtemperature range travoised, the heating rate necessary for melting canbe given by
X "If dT (8)
E B
4-4
In Equatirn b
%a = heating rate for melting
AHf = heat of fusion
TB - temperature at beginning of melting
TE = temperature at end of melting
If the energy required for melting is included in the heat trrnsferequation, it becomes
Fpx + P X 6 f4 kL T E - B J t r + h(Tf -1) -UT 4 k (T-T) (9)
Equation 9 can only be used after the initial melting temperature isreached. At temperatures below the initial melting temperature,Equation 7 must be used.
If the skin material does not melt on exposure to fire (for
examplek a stainlesi steel skin), Equation 7 can be used throughout the
heating cycle and can be used to calculate the maximum cemperoture reachedduring fire exposure. The maximumw temperature is calclAnied by setting
the accunulation termi in Equation 7 equal to zero. Thus,
o + h(T~ - T) - T 4 - T - T ) 0Q(10r hTf -max) -frTmax - Y max 0
where T ax is the highest temperature reached. Equation 10 c..vn be olveU
by triaT and error to obtain the maximum temperature.
SOLUTION OF THE MODEL
Both Equation 7 and Equation 9 must be used for calculation of thefailure time for aluminum aircraft skin. Equation 7 applies until thetemperature at which melting begins is reached, and Equation 9 appliesfron the start of melting until melting is couplete. Both equations arenonlinear first order differential equarions, and neither can be solved
analytically. Each requires an initial condition for its solution.
In order to simplify the numericul folution of Equations 7 and 9,they were written in the form
dT.. Al + B.1 T + C, T 4 (11)d t
4-.5
and dT A2 + B2 T + 02 T4d +B2C (12)
Equation 11 corresponds to Equation 7 and Equation 12 corresponds toEquation 9. The constants are given by
AI aqr + hTs + To (13)
pcx
B1 .= (h + k/z) (14)P cx
C1i -U (15)p cx
kaq + hT + - T
A2 = S)
pex + pxAHf(TE -TB)
B2 M - (h + k/z) ....
PCX + PX A11f(TE - T1)
and
C 2 ___2 Hf(18)
pcx + P(T E -. TB)
The initial condition applied to Equation 11 is
T = T0 @ t = 0
since the aluminum is initially at the temperature of the burr.Oundings.The initial temperature for Equution 12 is taken as the initial meitingtemperature at the timt, t, at which the initial melting temperature icreached according to the calculations of Equation 11. Sincc Equation 12
4-6
only applies during the melting period, calculations are stopped whenthe end of the melting range is reached.
If the aircraft skirt is nonmelting material such as stainless steel,only Equation 11 is used, and the calculations are continued until thesteady-state solution is approached.
The solutions to Equations II and 12 were obtained using the Range-Kutta technique, which is explained in standard books; for example,Mickley, Sherwood, and Reed (Ref( :nce 2).
Calculations were made for stainless steel and aluminum aircraftskins using the data in Table 1-fV.
Some of the parameters in Table 1-fV are not well known and mustbe estimated in order for the equations to be solved. The primarymachani.sms for heat transfer within the flame are radiation and convection.Heat trvosfe- by radiation depends not only on the intensity of the sourcebt aL:jo cn the absorptance of the receiver. The radiant output of thefire, qr, was asuned to be equal to 31,000 Btu/hr-ft2 , a value obtainedby Copley in fivue tests uaing JP-4 as the fuel (Reference 3). Since sootdepcsitz r::pidly darl-en the aircraft skin, the ahsrptan,,,w,,to be unity. Likewise, the emittance, E , for the surface was assumed tobe unity. The _onvective heat transfer coefficient, h, was estimated tobe 5 Btu/hr-fc2 . The estimate wcs based on forced convection at gasvelocities of about 20 ft/s, and corresponds quite clotely to recent dataobtained by Neill in direct flamc contact heat transfer measurements(Reference 4). The flame temperature, Tf, was taken to be about 20000F,a value based on optical pyrometer readings on hydrocarbon flames.
It should be pointed out that any parameter dependent on flameproperties is not constant. Fluctuations occur which have periods rang-ing from a fraction of a second to at least- several seconds, depending onthe turbulance of the flame and th2 gross movement of the flame due tothe effects of external factors such as the wind. However, when thethermal sink is large enough, tice small-scale fluctuations, such as thosedue to turbulance, are damped out,
6-7
TABLE 1-IV
NUMER1CAL VALUES USED IN CALCULATIONS OF ALUMINUM4MELTING TIME
Parameters ALUMINUk4 STAINLESS STEEL
Value Ref Value Ref
qr 31,000 Btu 3 31,000 Btuhr- ft2 hr-ft2
k 0.7 Btu - 0F/in 0.7 Btu 2 - 0F/inhr-ft hr-ft
z 0.5 inches 0.5 inches
p 175 lb/ft3 508 lb/ft3
C 0.23 Btu/b-°P 5 0.12 Btu/Ib-°F
AHf 170 Btu/Ib 6 NA
TB 9000F NA
TE 1200'F NA
h 5 Btu 5Btu 2hr- ft2 hr-ft2
T8 80 F 80F *
T 2000'F * 20001? *f
1.0 * 1.0
1.0 * 1.0 *
* See discussion in text of report.
4-8