D-A127 142 CORRELATION OF FLAMMABILITY TESTEDATA ON ANTIMISTINO 1/2FUELS(U FALCON RESEARCH AND DEVELOPMVENTC ENGLEWOODCO LAI4000D ET RI DEC_82_FALCON-TR-3648io

UCASIFIED DOT/FAA/CT-8i/i4 DtFA83-88-C-00 . F/G 21/4 NE hELS mo soon6 so m iIEhhhhhhhhhhhhI

pl .!w qu i n. ! l qi qn P . qn uq i,, - - .- -- - . *-. *. . . . .

'p.2

1 . 1.6-~W 12...0

1111.68

1111.25 1111111.4

MICROCOPY RESOLUTION TEST CHART* NATIONAL BUREAU OF STANDARDS,.1963-A

!4

DOTFA/CT8229Correlation of Flamrnmability TestData on Antimisting Fuels

be Levelle MahoodRobert L. Talley%

Falcon Research & Development Co.109 Inverness Drive East

Act Englewood, Colorado 80112

December 1982 .* ; -:

Final Report A

This document is available to the U.S. publicthrough the National Technical InformationService, Springfield, Virginia 22161.

C-

L-I

-e US DMXort1T*-r? O TronW1%PeOtioA

Fedwai tatiion AdmistrattonTechnicol CenterAtiolair City Airport N J 08A05

0 8 04 KO- 0d1007 6

! L

NOTICE

This documen' ia diseinated under the sponsorship ofthe Department of Transportation in the interest ofinformation exchange. The United States Governmentassumes no liability for the contents or use thereof.

The United States Government does not endorse productsor manufacturers. Trade or manufacturer's nasies appearherein solely because they are considered essential tothe object of this report.

," . . . ..

Technical Report Documentation Page

1. Report No. 2. Government Accession No. 3. Recipent's Cotlog No.

DOT/FAA/CT-8AD2

4. Title &nd Subtitle 5. Report rat

CORRELATION OF FLAMMABILITY TEST DATA December 1O, 2ON ANTIMISTING FUELS 6. P.,fo~ming O~gon ,e,*on Co4e

8. Perfoming O,gonization Report No.* . 7. Autherls)

Levelle Mahood and Robert L. Talley Falcon TR-3640109. Performing Organisetion Home and Address |0. Work Unit No. (TRAIS)

Falcon Research and Development Company109 Inverness Drive East 1). Contract o, Grant No.

Englewood, Colorado 80112 DTFA03-80-C-0006113. Type of Rpo,, and Pe,,od Co.erd

12. Sponsoring Agency Home end Address FINALU.S. Department of Transportation August 1980-August 1981Federal Aviation Administration-Technical CtrAtlantic City Airport, New Jersey 08405 4. Sponsoring AgencyCod.

15. Supplementary Notes

16. Absrc, As a part of a comprehensive FAA program to minimize post-crashfire hazards of jet transport aircraft, a correlation study was con-ducted on flammability test data of neat Jet A fuel, and the same fuel

A with various antimisting additives. The data were from fuli-scaleaircraft ctash tests, large-scale fuel spillage/ignition tests, andseveral small-scale flammability tests. Various rheometric tests werealso considered. The ability of certain antimisting fuels to elimi-nate large fireballs during occupant-survivable aircraft crashes wasamply supported. Large-scale crash simulations were found to behighly developed, and provide essential credibility on a given anti-misting fuel near the end of its development. Small-scale flammabil-ity test rigs used for screening antimisting fuels were found general-ly effective, but with some conflicting data between rigs, and withsome deviations from large-scale results. Methods have been defined

*. which can provide higher confidence and /&1ution for selected test* rigs. Enhanced correlation requires characterization of the fuel

droplets and flows produced by a given test rig, and application ofsemi-empirical liquid breakup and ignition/combustion models. Liquid

breakup correlation is provided in a limited case--for a Newtonianfuel at small-scale.

17. Key Words 18. D,urwti rn StatementIFlammability Ignition Document is available to the U.S.Turbine Fuels Liquid Breakup public through the NationalAntimisting Fuels Test Apparatus Technical Information Service,

Springfield, Virginia 2215.1

vqw 19. Security clemesil. (of thris repel,) 0.Security CI..,.l. (of tis page) 21. No. of Pae. 22. Pr-c.

Unclassified Unclassified

Form DOT F 1700.7 (8-72) Reproduction of completed page authorized

S,

U.1

E E En,- E- - - E

* 4 :*.O 0

* - 00 a

.3 -~-I IC

= 230 IL

PREFACE

This final technical report was prepared under contract numberDTFA03-80-C-00061 with the U.S. Department of Transportation,Federal Aviation Administration Technical Center, Atlantic City,New Jersey. Mr. Thomas Rust, Jr., of the Aircraf- Safety Develop-ment Division, Engine/Fuel Safety Branch (ACT-3- ), was the FAATCProject Manager.

The contractor was Falcon Research and Development Company, a sub-sidiary of Whittaker Corporation. The Contract Project Manager atFalcon's Englewood, Colorado, facility was Mr. Levelle Mahood, whoalso served as principal investigator in the areas of aircraftfuel fires and explosions, and related large-scale testing. Mr.Robert L. Talley of Falcon's Buffalo, New York, facility servedas principal investigator in the areas of liquid breakup, fuelsrheology, and related small-scale testing. The invaluable contri-butions of Dr. William A. Sirignano, who acted as a consultant toFalcon R&D in the area of combustion technology, are gratefullyacknowledged.

The authors are indebted to many key participants in antimistingfuel research and development, who gave so generously of theirtime and insights. These include: (1) FAATC--Mr. Eugene P. Klueg,Mr. Thomas Rust, Jr., Mr. Robert F. Salmon, Dr. Thor I. Eklund, Mr.Augusto Ferrara, and Mr. Thomas M. Gustavino; (2) UK/RAE--Dr. S. P.Wilford, and Dr. E. A. Timby; (3) JPL--Dr. V. Sarohia; (4) NAEC--Mr. John Schaible; and (5) Southwest Research Institute--Dr. R. J.

Mannheimer.

* iThe authors also wish to acknowledge the contributions of othersat Falcon R&D, including A. Stein, R. K. Miller, J. M. Otto, D. M.

Taylor, H. H. Iwata, J. 0. Duggan, and D. R. Ludwick.

L-

L .i

ii

TABLE OF CONTENTS

Page

I. INTRODUCTION 1

Purpose IBackground 1

II. DISCUSSION 3

General 3Air Flow 3Liquid Flow 10Liquid Breakup 10Droplet Dispersion 14Heating and Ignition 17Propagation 18Pass/Fail Criteria 22Sustained Fuel Fire 23References 24

III. ANALYSIS OF FLAMMABILITY TESTS 27

General 27Full-Scale Crash Tests 27Fuel and Air Flow 27Liquid Breakup 33Wing Fuel Spillage Tests 41FAA Wing Fuel Spillage Facility 41NWC Wing Spillage Facility 43RAE Rocket Sled Test Facility 43

RAE Minitrack Test Facility 45JPL Mini-Wing Shear Facility 46FAA Flammability Comparison Test Apparatus 48

Analysis 48SWRI Spinning Disc 50

Analysis 50Correlation of Flammability Tests 2Enhancing Correlations 64

Aerodynamics 66Fuel Jets 66Droplet Diagnostics 66

References 68

IV. ANTIMISTING FUELS RHEOLOGY 70

References 78

V

Paqe

V. LIQUID BREAKUP 81

Rationale 81Liquid Breakup Mechanisms 83

Oscillatory Breakup 83Stripping and Taylor Breakup 92

Antimisting Fuels Breakup 92

Liquid Breakup Model 107References 129

VI. CONCLUSIONS 135

vi

LIST OF ILLUSTRATIONS

Figure Page

*. iI-1 Antimisting Fuel Test Groupings 8

11-2 Free Turbulent Jet 9

11-3 Three-Dimensional Incompressible Wall Jet 11

11-4 Opposed Fuel and Air Jets 13

11-5 Typical Diameter Ranges for Various 15Substances

11-6 Typical Mass Fraction Versus Droplet 16Diameter for a Hydrocarbon with ViscoelasticAdditive

11-7 Droplet Combustion Modes 19

11-8 Variations of Fuel Temperature, Mass Fraction, 21and Droplet Radius With Time for a MonodisperseHydrocarbon Spray

III-1 Distance, Velocity, and Acceleration 29Versus Time, SP-211 Crash Test No. 1

111-2 Distance, Velocity, and Acceleration 30Versus Time, SP-2H Crash Test No. 2

111-3 Droplet Diameter Versus Cumulative Mass 34Percent - Time Increment No. 1

111-4 Droplet Diameter Versus Cumulative Mass 35Percent - Time Increment No. 5

111-5 Droplet Diameter Versus Cumulative Mass 36Percent - Time Increment No. 6

111-6 Droplet Diameter Versus Cumulative Mass 37Percent - Time Increment No. 10

111-7 Droplet Diameter Versus Cumulative Mass 38

Percent - Time Increment No. 11,

High Velocity

vii

List of Illustrations (contd.)

Figure Page

II-8 Droplet Diameter Versus Cumulative Mass 39Percent - Time Increment No. 11,Low Velocity

111-9 Fuel Temperature Versus Fuel-to-Air 62Relative Velocity

IV-i Viscometer/Rheometer Schemes 75(Elongational Flow)

V-I Probable Role of Viscoelasticity in Breakup 90

V-2 Breakup Mechanisms 94

V-3 Approximate Shear and Tensile Viscositiesfor 1.3% AM-I in JP-8 Fuel

V-4 Liquid Breakup in Wing Fuel Spillage Configura- 106-tion

* V-5 Breakup Model Prediction Versus Experimental 114Data

* V-6 Neat Jet A Droplet Distribution Predictions 131

V-7 Dynamic Pressure and Velocity Decay in Spray 117Flammability Test Apparatus

118v-8 Droplet Distribution Decay Predictions for

Neat Jet A, 100 psig

V-9 Droplet Distribution Decay Predictions for 119Neat Jet A, 75 psig

V-10 Droplet Distribution Decay Predictions for 120Neat Jet A, 50 psig

V-lI Shear Viscosity of 0.3% FM-9 in Jet A 122

V-12 General Shear Strain Rate - Droplet Size 123

Behavior During Stripping Breakup of aNewtonian Liquid

- viii

List of Illustrations (contd.)

Figure Page

V-13 Computed Breakup of 0.3% FM-9 in Jet A for 126Small Jet Experiment of Eklund and Cox

,ix

LIST OF TABLES

Table Pige

II-1 Summary of AMF Flammability Test 4Facilities/Apparatus

III-1 Selected Data from SP-2H Crash Test No. 1 31(Neat Jet A Fuel)

111-2 SP-2H Crash Test No. 1, Liquid Breakup 32Model Input Velocities and Droplet DataOutput

III-3 Wing Fuel Spillage Facilities 42

111-4 Fuel Ejection Velocity in Meters/Sec. 44

111-5 Spinning Disc Dynamics 51

111-6 AMF Flammability Test Data Summary 53

111-7 Fuel-to-Air Speed in ft/sec by Facility, 75FM-9 Concentration, and Flammability

V-1 Breakup Model Nomenclature 109

x

l

EXECUTIVE SUMMARY

This flammability correlation analysis compared test data from a large.. number of flammability tests used to determine the quality of antimisting fuel." These included full-scale aircraft crash tests, large-scale fuel spillage/

ignition tests, and small-scale flammability tests. All tests comprised thedischarge of the test fuel using various procedures and with various velocitiesinto an environment which included ignition sources of varying intensities.The tests were conducted initially with Jet-A, or equivalent, to obtain astandard or baseline to which the same fuel with various quantities ofantimisting additives was compared to give an indication of the antimistingquality of the latter. Various rheometric tests were also considered as ameans of determining the effectiveness of the fuel relative to its antimistingproperties.

The ability of certain antimisting fuels to eliminate large fireballs duringoccupant-survivable aircraft crashes was amply supported by the data analyzedduring this study. Large-scale crash simulations were found to be highly

-I developed, and provide essential credibility on a given antimisting fuel nearthe end of its development. Small-scale flammability test rigs used forscreening antimisting fuels were found generally effective, but with someconflicting data between rigs, and with some deviations from large-scaleresults. No reasons could be positively identified for these discrepanciesnotwithstanding the limited employment of a combustion modeling technique.

Enhanced correlation among the tests studied requires characterization of thefuel droplets and flows produced by a given test rig, and application of semi-

empirical liquid breakup and ignition/combustion models. This was notaccomplished to the degree necessary due to the limited scope of this program.Liquid breakup correlation is discussed and a solution for a limited case isprovided as an example of the use of this analytical tool. This case was fora Newtonian fuel (not antimisting fuel) on a small-scale basis.

6,

6

-------'..--

I. INTRODUCTION

PURPOSE.

The primary purpose of this effort was to perform a correlationstudy of antimisting fuel (AMF) flammability test data developed bysix test groups--the Federal Aviation Administration TechnicalCenter (FAATC), the United Kingdom's Royal Aircraft Establishment(RAE), the U.S. Naval Weapons Center (NWC), the Naval Air Engineer-ing Center (NAEC), the Jet Propulsion Laboratory (JPL), and South--west Research Institute (SwRI). This effort included study of theapplicable test facilities and procedures, test data, and possibleimprovements, modifications, and additions to the testing to en-

* hance data repeatability and correlation.

*BACKGROUND.

* The Federal Aviation Administration has been conducting a compre-hensive program to minimize fire hazards of jet transport aircraft

* in impact-survivable accidents by developing antimisting fuels.As implied by their generic name, AMF's inhibit the misting whichoccurs when neat kerosene fuel is released at significant velocity

* relative to the atmosphere. Even at bulk liquid and air tempera-tures well below the fuel's flash point, the finely divided mist isreadily ignitible and propagates flame fronts rapidly. The common-ly resulting "crash fireball" heats and ignites pools of spilledfuel at the crash site. Sustained burning of proximate spilledfuel can lead to rapid destruction of an aircraft, and severely

" reduce occupant survivability. By eliminating the crash fireball,AMF's break this hazardous chain of events.

*In the course of development of AMF's, a broad range of test methods,apparatus, and facilities have been essential. The complex polymerchemistry of typical additives involves both standard and special-ized test apparatus and procedures which address rheology and otherproperties of the doped fuels. Small-scale flammability testshave evolved to expose candidate AMF's to idealized crash condi-tions--including flow, expulsion, breakup, and ignition.

Finally, full-scale crash tests and large-scale flammability testshave been essential for proof-of-concept, phenomenology at adequatescale, and demonstrating full-scale effectiveness of near-term

4candidate AMF's.

This study has been initiated at an opportune time. A significantbody of test data has been developed over the full size spectrum

~1

of AMF flammability testing. A data base of rheometric testing isalso available. Equally important, understanding of non-Newtonianflow, liquid breakup, ignition and combustion phenomena has improvedsignificantly in the past few years.

* This report presents analyses of the various test methods and data," relates physical parameters and phenomena to them, and defines

methods for improving confidence in individual tests and compara-bility between tests.

2

II. DISCUSSION

GENERAL.

Table II-1 describes the test facility or apparatus, performingorganization, and applicable references. Figure II-i relates thevarious tests by type (flammability, rheometry), size, and funda-mental method. Three methods of correlation are shown, but semi-empirical methods predominated. The various tests are selectivelyanalyzed and compared in subsequent sections of this report.However, it is illuminating to discuss their similarities anddifferences qualitatively in the context of physical parameters andphenomena.

All AMF flammability tests involve the following phenomenologicalelements in the sequence presented:

1. Air Flow2. Fuel Flow3. Liquid Breakup

* 4. Droplet Dispersion5. Heating

* 6. Ignition7. Propagation

* 8. Sustained Fuel Fire

AIR FLOW.

In the general crash scenario, absolute air flow velocity can benear zero, due solely to wind. The NAEC Crash Tests, RAE RocketSled Tests, and the SwRI Spinning Disc, Drop Test, and Air Guninvolve (and require) near-quiescent ambient air. The remainderutilize liquid fuel orifices which are fixed in relation to theground, and move the air to simulate motion relative to the atmo-sphere. Free air jets are used by the large-scale FAATC WingSpillage and NWC Wing Spillage facilities and the JPL Mini-Wingapparatus. The FAATC small-scale Flammability Comparison TestApparatus (FCTA) utilizes air flow within a 1-inch I.D. pipe.

Figure 11-2 depicts a free turbulent air jet as described in aero-dynamic literature. 2 4 Note that "potential core" flow is free ofentrainment, and maintains a constant velocity over a cross-sectional area which diminishes to zero at a distance of 6.4nozzle diameters from discharge. The limit of diffusion is definedalong a half-angle of 100 from the axis of the jet, extending outabout 100 nozzle diameters. Beyond the "established flow" limitis the "terminal region", where the velocity is near zero.

3

F14

F. I 0 41i 4J C)

1 41~t MT( 1 0 0 0 0Fz4-4- 0s-I r- I 0 -4-) ()) $4 4 V H to .- 0 0

0 W >U 1-)(1 4-)Q)04V 44 -4 0) 44CX 4 .-4 n o : >

04 0) 4-4 -4 0) () Q)) 4J0) r- 0.

$40r 0 C) fou 41 : tH- 04 4- ) Q) a)i C)(12 0) () 4-4 UCo 0o 04 *4 (11 (d: D: )4 Q - 3 -a) 04> DC-4 r ) 0 W FU r 44-4 0)Q) :J M 0 r- (

A (1 0 > 4J A 54-) 0 Q W 0 0) 4 (Ln0 4-4 4- F: '1:4-)Q0 U) tS04 4 rr. 4-)Cd - 4 4 -H ) 4-) -1

4IJ ) m ro - 0) -4 M (1 >) * Cl ) a) %-4 44-4 rO 00)md 4J W 0X-U > )r .0 3:

W- >d 0) ) 4 c0a) r.- -Ira E-4) ) 4J 0-44-) U) )U 44- 0U l H- 1ON -H U)0) 4-4 4 -4 S-i (0 QMd U 0) 4-)O054M r.44C (0 9 ->44-J0 % 4 -

445- $4 4J 0 1 4 0). * E- (W4) OC) r-

4J M ) V-de040 4-) 0 4 0) a Cd 24a 0 F=Q)4 4C -I( JE M 0 04 a) Q -4) r U - V 4j 10 4 -) Z 4-J) :0

0n 4J(d P4 4-) 4-4 ) -i4 Ir :: -1-)4 Q) a 4-) Q) oa) u 4 .- 4 CO W rk- 0 4 ~4-4 ar,)0 Q)-4 (0 )

>14) Q U)Cfl a) p (av :4-4 E- Q)5 40Z r 44 P0 M~U4J z ~ 4)-ri C -P U 4 4-40 - Q~*d )0 Q) 4 44r IQ

r4 0 :X: 504-1( U)(3 )4(0 0 roc ~Q 4 -4 H )H H NC 4J r 4 5-1rr40 n :s 00) 0( Q 0 LA 4 W

E-4 1:: Ha U)0 vZ'000 v ~ 0- 0) Ur >).. wq0 k 1 p 0

U)0 ( w Q a)i d 4 . us ' Qdw 154V (z -r- t

rd U) a)4- 44 Z a(0z k 9- r. 4-) Q) r.4 m V) 0 (aH- 3 r W -'4)W EO C (D 0 ) *r -10 4-1 U a ) (1J Cd fdu r

rX 4(D )$ 4-10 (1 * r. m 1 0) -i -4 rCCfl)r-.00r44-4 - 0 1 C) ~ a, -( 4 -~~4 0-)4 4-4 SI

CO 0 -ri p 4- ) ' $ 0 t -4 C:4C .0s 0 . '-d 00 C>02)0) - Cd4 -r-( 4 ( 4 O f Cd r .44 4 5CE >1-4Q 0 :0

0: N F: 4W: -PH4~O 1 0 J $4 (a-U 0 0.4O4 r1Q.r ~4 ( 1H D a0) (a d0 -,4 U)0 U2r-I WV4 17% l 4~ (dEA = r

4- i> 4)1 4 R3 4 )r- - Z~ Qrj0)~l 44W ZdW4)0)0 4 4 04

0- 0).D r-I0s Z M 0 04J- 0-HI-T - 0 -P P 54 (UC 0-r-I En ) o $ rdz 1 r En 0)I WT3 U -r r ) 4

>1 r ro r., H kW S U) : Hx S w :s m0 - 0 w H50-P dO~ l0

Uri 0 ( E- .)Cfa--4- M CflP44m C~4 q ~u m 0Q-(d 0 4J Q Q 4- 4- > tP -) U)Q) 0 Z- J-4

0) 4-' 0)I .

E-4 4J u Q)I--; 0)- U4 r 0E-)U

Hi E-4 $ m U2-H 0 -*d0d (a -4C; J --4(12

V. Cd 4J 0- 9-O -PP. ) u m)0) o

Cd a)I > (04JE-

E-4 4J CdU a)-H~- U.U ~l -

5-i E ft C -45- l4H r-4W a)1 U) i 9 ) Q)-

u 0 C4-' * - 14) 0)4. r-

A4 tcW *0) :3a ) V0 ) * r Co.l* Zu r4 E- w 4 z 4 U

z

0U

I U)4 .rO 44 4 1 U)

U) $i . W 0 0 U 4 ty U) U (a

-4i -4 r-4 t () rO H-4 () .0 Q W E -4 E4W"- 4 ~ >1W a)00 0 0 I ty : - W ~4.)0 jX 4 c0 0

4 4 a 10M 4 >U( 0) O0 4 4- V4O u iUr OW 0 4J 4-(a (D Z0 r4LfU 0 rW~. Q 4(D4 -,4 9 0 - "- Q)Q) u0 d 0 -)

04 ~ 4J > -W-)C -40 - -4 44 -4 rOQ C p( P 4E- r CO440) ) ) E 0UI00) o w O4- -4 t~ rno 114w 4 0p

Ft n i( -444(A(() 4 W >0 W.a) -4 (0 u(0 Q) ) - C-0 o'-( 1 0)3 : () 0 a)z 4 II ( a) u 4 4 a)

HY rr 41 i 0) 43:r 4J 4 J0)Eti a~0 -4P *-M 0)0 ) Q W 4 04 ri 0 -Cf Cn-4)4l 4 O J. 0 U) 4 ) J)O4-J r

Q) -4 P Or- l4U 00-H J ) u r 0 c -4 r. O ro () r 401- ::1U~~ . -4 - a 0 0 0 j o 0 17WC.rIp D-1

r-4 0 )a UUVa4)- 4- 00 JaP * -H ( 4 0 0 W C 0r-i -14J~l

$ 4 0fn4- 0~ 9)W-4~ 4J a 1 p0 ( N0c u C' P~.4 W-44-4rOP ,4 U 44 40 *UO r.i ~ 0) > 1W 0 :S >1

CD4 4C (DU) U0 4 U 0-~s4 I-o0J C0 0 *$. ~ QW U U 4c )-4J-ICOU-l - 4Ju' 0 U 0 0U 0W~ 01 -14 w HQ) >

H 0 U) V -0 4 -UP4)0 3.' %4 W P .0wd~ 4*xH) z 4 , W n Ir(4 r. ( r4j40 )4 l 0 U)0) ) fl0.j

E- H l r n-10c - )Q (41)0~ U * f4 -lJ rO 4QJ a)4 XW -4U12 (H 4) CU 4- 1d ~ -(-d -0Pr)j WU l0 4 Jk0 C> 4 Q 0) n04U, 4r 4V Q to 4)r4J~-i- 040) f 4 " 4-) 0 - a)-(

43 04 .,4 9 P00l( (D4 0CO -4 tz 4 0)H -14 0)4I M 0 0 0) 4SMH0 4 i,- >Ot 0Cf 0~0-~ Z) 4-)4 4J 9 tp -4 CO I

r-4~~I 0 .-0:% O MSWC -4 r(C: (1)0 S4 0~ C 0--I:) p0)

r- 4-i 0 ::% -. I4- 4 r.-4 4)0 z x0 (a)Uo-a4C: Z )CO M T$ -4 W 4

fts 0 0 00a)Wr-0 4J r. J :3 U)-d 4 4 -Z M 0 ) 4- 4 41 ir00 -4 A 4 -40 t- 0 0 1ll ) -4 -HrO 4-

r. r 4J r z0 -r~4 w 0 4 0 ) 0 : 4 3 -i :s((0 ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ 4v4--. 0 HrU)OC 44 O00-4U )U 44- )a 44

uoQ 4-) 4 E a 3 4 i44rq~ ~- 0 O 4 C -4 U. $4 ) W4 'O 0r. -)M .0 U 0) '00 rO( - O- H Q) Q 0 - Q ~ U C 4- 4J(0Q)W0

1-4 41

U) $:)r

4-1

H - U~ '-I W4rom Eno u 0 0 H-

H 4 0 * ()1 w m.0 040) 4J r. 0)0o >4 As 4 - n rl z -4E-4 H- ) f Q)) W 4 4-j0 En

HL HO (Ti 0' r -4 $0 ~44 U4

tp to P r> rl W 444( (

U)

cu *n w-

rJ r-4)0 NI=U- .--

4-;4- WH 4~

4J -r *H Qdi W0 C rU .4 .1- Cai C.4-0.)>0 U): J ( >1 -1 444~ *3 -4-4 X Q 0 *

C-a r 4J0 0 :3 WO i 4J r-' i-;X >4- -4 4< F

Q) C.4X-n cio nI- a) ~ ~ Cc) *>0,HUlF U

> Eu .0- 0. 4. r E~I-4 4 H4- :5 -4

0404 0> 0 ~ rn >nO w 0) (L u Ca

04) CN 4 44 a 4 () U U -- *44 ) (

4. -nV C 4 4-4 >i 4 -4 U W. 0 -4-J *i*HV E 0.-'

0) >~r4 -4 04 4J C > T3ar4> a) 4 0 U

0 (2)3~ ' a)0 91 r 4-) 0 Q) 'h a0 44 r-f rr-4 0~4.~ 10 -4 O 0m( JJ) 3ME O

H144 U ) U $-4-H u (0-4N~> 0)0 .V rrH nI U

-. a4( 1X 44 d)>)4 CO (a-- . 14n:~ a)- '.C 4-4W W4 0

s- a i- 0 : - -- O f rI0 - 1 C -4 0 - $4r

a) r .. t~ -d-4 4i)0O.-C-4 tCa: P.4- ~ U CO -1 -4 -r -(d 4J i u . 0d (13 U0 > 0 Q)4) 4 o H

k nOU a a) -H 0 mC 4J0)MH4 I-k W0$4U4Q)E -I: 4 u CaI,4J-) 0 4~4. 0

4-) 10(0 O Z 44-r-4 U - EO 0 ~-H 4 JQ)M -HnI040 n a0

E-4- 0 0 -t a W S:43: 4 -E-O0 WI r-

H44 -) k 4 ) : E 0 J-)- 0ci *HOJ t)- En ) a)

O - 0-4 ) 4 0( 4 -- ( 4-i- 1: -ec U- . -- 4-) CJr- 4-) Q 4

4-) 04 : -4W~> 0 4J 0 H400 44 Q C(1) U 0 '0mQ) 4EkJ1)> u1:% a ro E 4J~

a) 0- 0-

-4i (0 41 4 J N(3 ( : - a -l

a)H 4 01 M0 4-4 *.4 r E 44. - 4 -

4- 0 En r.~ Z 4.b) )E ( ,r Z 4 )(Z U .r44 r:; r-I CU-dn40 (D4I 0 0 PW- ) 4 Q)(

*4 UH H- 0aI 4.) J1 - - 0 '.0 44)0 -

H Q4 r. IHI uE-4 ~ *.4- 0)4 J 0 4~

a) r4 I ) 0Q ci)ci *H (U)4 a) -H~ U 4Jr4H

e- 6 a 1

54-

0ca

p 0 4-J U >i t ID4JIco~~t022 *(0- 0 - O 0 0

(0 0H-- r : U E14) -d)) Q)4-

04$ 0 W W 0) a >104 4 0 r 4 0 W4 0 0E-4 4J M l

0> Q)44 r-4O~'t C) S-4r( O 4 Z r-

N.41 w u -4 > z (o 90)020 U 0 it (1)to0JU-r- r.4 4)

E) 4~ 3: 4 U- Ell-4(0 4 Z0 r-i Q - 4J -0W a) , n : -

4-) -, 0 a)r-E 0 > 4)9 a4.0 U 'O 4--jO0).-Ch4 r -4 0 P U

t44-i4-) t Q -) Q ~-e 'J40- 09 4* M U ) 4-) 0 4 4-4E > C O 4 0 4 j0.4

0) CA *'-4- W 04)0 C> U--o0 4r04 mC En ~ j2 4-rI 1 Ca 0 0 0 JQ

ClD :3 rq'- - 0 ~4- W la 4-' C> S - 4k rU4: 9:~4 (D A .- 4U 0 5

4. 0 4-) I'i0 Fr; rN 4J 0 0 F $0

-4 H k t N . ri z-4 W i a>1-~~~- E- .0ma41I Ui(

4-)0 P4 (D 0 . .o D 0EU4 9 d0 0 ( 0 0 EUU 0 >1

wz 024 0 r-4 U4S0 U

(a 00 >.C0tW0JrX < :3 r-4u C -I

TS C4J 4 : -4 -H > tP 4-) : 0 1.7144 -r( U'Nr-4

0 4- -)t 01 : 0 4E~ W~ -4 4.Jr4 0 i

44 4- 0Jr rIrd4 4 r -0 ~~0 4) 4 4( = )024 4 - ): 2

E-4 u~ 0

HE4 EU3E-

E-4 E4 M24J) U

4-) 4JJE0 0

C4 0 W r.4) 0U ~E4

z -HV0zCL V)z 4

In

U'))

CLi

C/u,

zz

F- HUE-

w )- I w CL-I

LU -4

Lnn

I<

4 04

t--- ~ 1 A ~ A S. ~ a ~ - - ---- - - -- - - - - - - -

z

z 0

00LL U

L&JrJ

0L_: E--

-l __ ____ ____ ____

9U

The NWC report9 provides data on velocity profile decay and turbu-lence within the jet. The data show a significant range of vari-ability of air velocity. Profile data were obtained on the large-scale FAATC Wing Spillage facility from the plane of discharge outto 93 inches. Hence, velocities other than those in the "potential

* I core" were not available to compare with the free air jet depicted.

An interesting variation is the introduction of a "ground plane"to the FAATC Wing Spillage facility. This simulztes a crashscenario with a mid- or low-wing aircraft, and t.e considerableinfluence of "ground effect". Figure 11-3 is an isometric view ofa free air jet from a rectangular orifice 2 5 (as at the FAATC facil-ity), depicting the modified velocity profile due to ground effect,as contrasted to Figure 11-2. No velocity mapping data were

. available to correlate with this generalized flow configuration.

LIQUID FLOW.

In the crash scenario, liquid fuel flow is induced through breachesin fuel tank walls primarily by the relatively high transient acceler-ations and decelerations incident to the impact(s); or, in their ab-sence, by the gravity head of the fuel. In the NAEC Full-Scale CrashTests, breaches are deliberately generated in the leading edges, tops,

i- and bottoms of several wing fuel tanks of the SP-2H aircraft.

Large orifices are typically involved, so that even with a highvolumetric rate of spillage, relatively low stress is placed on the

fuel during transport to and through the openings in the tanks.This is particularly important because AMF's are generally suscep-

V . tible to loss of a degree of their antimisting property with expo-sure to excessive shear. The RAE Rocket Sled Tests share with theNAEC Crash Tests the use of deceleration to induce fuel flow from atank. The FAATC and NWC wing spillage rigs and the FAATC FCTA arecarefully designed to avoid shear stress on the AMF's tested. TheFAATC Wing Spillage rig uses a gas pressure cap in the holding tankto induce fuel flow to the orifice, whereas the NWC rig uses gravityflow from an elevated storage tank. The FCTA uses positive expul-

:O sion from a cylinder by a mechanically-driven piston. The SwRI- -Spinning Disc utilizes acceleration of the liquid through four

radial flow channels connecting a center chamber to orifices on the*outer edge of the spinning disc.

LIQUID BREAKUP.

The crux of the post-crash fire hazard with high flash point turbinefuel, such as Jet A, at nominal ambient temperature is liouid break

10

4

1-4

~z

00

YOH

up, i.e. conversion of the fuel from a relatively low surface area,high volume and low ignitability confiquration to a very hichsurface area and, hence, flammable mist. This conversion is due torelative motion between ejected fuel and the atmosphere. In a

. >crash scenario, the air is at low or zero velocity, whereas thefuel is being ejected from an aircraft which is itself movinci atsignificant speed. The highest relative velocity between the airand the fuel is typically achieved (1) at high aircraft decelera-tion while still at relatively high speed, and (2) with fuelejecting through the forward wall(s) of the tank(s). The maximumrelative velocity is significant in that it places the maximumdemand upon the performance of AMF in suppressinq mist formation.

Fuel is ejected a considerable distance ahead of the deceleratingfuel tanks in both the NAEC Crash and the RAE Rocket Sled tests.The FAATC, NWC and JPL Wing Spillage rigsiimulate this criticalencounter with a free air jet flowing against a directly opposedfuel stream exiting from an orifice in a fixed airfoil's leadingedge. The opposed fuel and air jets are schematically illustrated

* .. by Figure 11-4. It is a highly desirable feature of the FAATC WingSpillage facility that breakup occurs in the "potential core"region, Figure 11-2.

The FAATC FCTA uses a fuel jet opposed to air flow within a l-inchI.D. pipe. In cases where the average plume diameter, Dp, isgreater than one inch (Figure 11-4), impingement on the pipe'sinterior surface will occur.

The SwRI spinning disc ejects the fuel from four peripheral orifices,while rotating at rates as high as 30,000 rpm. Hence, the ejected

- liquid "sees" an initial velocity relative to the air which is thevector sum of its radial and tangential velocities at that point.

Liquid breakup is discussed in detail in subsequent sections of thisreport. Suffice it to say that the localized relative fuel/airvelocity is considered crucial to resulting droplet size spectra,and configurations vary with AMF type from compact (e.g., near-spherical) droplets to long filaments. If a significant change inflammability occurs for a given AMF and test apparatus at the samerelative fuel/air velocity because absolute air and fuel velocitieshave been varied, the possible influence of undefined and/or un-

-- controlled variations in the dynamic conditions must be considered.

Liquid breakup is sensitive to scale. Over the range of fuel jet* diameters of the tests analyzed, significantly different phenomena

are encountered. As detailed later, this precludes unadjusted

. comparisons of independent test variables and results between a

12

p -

LL I .J L LiI--E LL .J =

WW OLLj O'L

-wu <

Q.I LU m L C

L LL. L LiJ Li -XiJ C.LAL = -- r-CI-C

C. 0- WjV)C

Lu C LA (- n LC3 <w - w<<=

w < LLJ -Z < UI

< ~- C)

0 000.. CL. LL

LL-

rI-4

c 0

00

U'-

-14

-7-

I-ir

large diameter fuel jet (e.g., on the order of tens of certi-meters) and a nmall one (e.o., one centimeter or less).

Figure 11-5 displays the diameter ranges for some common sub-stances/configurations, and for hydrocarbors oa interest to AIFstudies. Note that for the three hydrocarbors with viscoelosticadditives, the smallest droplets are about 80 microns in diameter.Although droplet size data are lacking for most AMF flammabilitytests, observed flame propagation indicates that the droplet sizesfor neat Jet A fuel can have a lower size bound approaching orless than 10 microns in diameter, depending on relative fuel/airvelocity and other parameters.

Figure 11-6 illustrates the droplet size distribution for a 0.57mass median diameter (MMD) population of a hydrocarbon withviscoelastic additive shown in the previous figure. Data of this

. type are highly desirable in the analysis of AIIF flammabilitytests.

DROPLET DISPERSION.

Given the products of liquid breakup just discussed, their subse-quent tiansport and dispersion are crucial to the flammability ofthe fuel plume or cloud. For example, dispersion can convert aflammable fuel droplet population to one which will not propagateflame--by reducing the number of droplets (or mass of fuel) perunit volume of air below the Lean Flammability Limit (LFL). Thiswas amply demonstrated during development of the Minitrack testapparatus by the RAE. 2 6 In the course of development of thisapparatus, they found unrealisticly high effectiveness of an AMFat very low additive concentrations--compared to large-scaleRocket Sled Test results. They postulated that dispersion wastoo rapid, so that the mist concentration was below the lowerignition limit by the time ignition sources were encountered.The solution was to add a nozzle 1/2-inch (12.7 mm) long to the5/16 inch (7.9 mm) diameter orifice. This nozzle evidentlv nar-rowed the dispersion field and produced the desired comparabilityof results between the Minitrack and Rocket Test Sled devices.

For thorough AHF flammability test analvsis, the fuel dropletsize data previously discussed must be supplemented with dropletconcentration data. A convenient method is to subdivide the drop-let spectra into groups of approximately monodisperse droplets(e.g., mass median diameter); and specify the droplet number den-

* sity (number of droplets per unit volume) for each group.

Considering the flow fields illustrated by Figures 11-2, 1I-3,and 11-4, it is evident that the droplet number density may beexpected to vary significantly at different positions within a

14

z

V0 0E-

z U

LLU

o -LJ

N

LUJ

0 0

(A 0

0 -

LUJ

LUJ

<- 0

< 0) 0

L) 1: >4

CLL

(A V)

0 V) <L) Lj LL)V) I= -LU

z 0

0 - V0(Ao (Ar

0 a- Z LU

cc 0-

a (AJ LUJ1

~LJ

15

00

L,

CL

CL z

F- <->

0 0

I-- 0~

000<

CC

0

z~ 00

u10

000

CL 4401

-J %0 L C1

C; C C

NOIIV'dJSSVW3Aiiinw0

given fuel mist cloud or plume. Locations nf particular concernare (1) the liquid breakup region, (2) regions of hot surfacesand/or ignition sources, (3) transient flame regions, and(4) fireball and sustained fire regions.

At high relative fuel/air velocities, droplet drag is the domi-nant propulsive force--which varies with droplet shape and size.The direction and magnitude of this force is dependent upon theflow configuration of the proximate air mass. At near-zero localair velocities, gravity emerges as the propulsive force, opposedby droplet drag (i.e., settling). Both of these modes of trans-port are highly relevant to the AMF flammability tests analyzed.

Turbulence in the flow fields must be considered. Turbulenceeffects are evident in the full-scale aircraft crash tests--par-ticularly the wina tip vortices and wakes.

HEATING AND IGNITION.

Heating of the fuel mist must be considered separately from igni-tion, since we are dealing with a turbine fuel, i.e. Jet A, which

* "is typically at temperatures well below its flash point (about100*F) under operational conditions. Hence, localized heatingwithin a mist plume or cloud can result in increased evaporationand flammability. Likewise, extremes of bulk fuel and ambientair temperatures, coupled with high aircraft and ground surfacetemperatures, give a "worst case" environmental scenario.

Simulated ignition sources in full-scale crash and large-scaleAMF flammability tests typically provide many orders of magnitudemore than the few millijoules of energy that are the minimumnecessary to ignite a flammable mist in air. The minimum igni-tion energy for a vapor/air mixture is even less, on the order ofa tenth of a millijoule.

It is quite evident from photographic coverage of the large AMFtests that the shrouded rockets, liquid hydrocarbon-fueledtorches, running turbojet engines, fuel pool fires, etc. provideextensive heating effects--involving temperatures well above thefuel's flash point, but below the minimum necessary for ignition."Marginal" test conditions, involving lower concentrations ofantimisting additive and/or higher relative fuel/air speeds inthe breakup region, often provide a clear differentiation betweenprogressive heating effects and localized iqnition.

The clearest demonstration is provided by the wake flame in theflow field downstream from a hydrocarbon gas-fueled torch. Evenwith mist clouds of low flammability, a lengthening of the wake

17

flame is evident. This is due to direct mixing of fuel dropletsin the plume with the torch's flame. With a more flammable

" .air/mist flow past the torch, the wake flame lengthens andthickens. The enhanced flame is both radiating more heat and

- involving more of the flowing cloud in the convective "heating"* :and "combustion" envelopes. With a progressively more flammable

flowing medium, the wake turbulence limits the maximum continuouslength of the wake flame--with large excursions of non-burningmixture across the axis of the flame. Discrete burning regionsjoin the general flow of the plume, with one of two possibleconsequences: (1) heat loss is greater than heat generated byinterior burning, and the discrete region shrinks and self-extin-guishes; or (2) heat loss is less than heat generated, and the

* fireball grows.

The important point is that the large-scale "ignition sources"provide widely varying combinations of high-temperature, lowvolume ignition sources in conjunction with relatively high-volumemist cloud heating/evaporation effects. It is evident that thepreheating and evaporation effects of the large ignition sourcesare often crucial to the ignition of a plume containing AMFdroplets.

In the small-scale tests, the mix of ignition and cloud heatingeffects also varies widely. From these observations, it isevident that in comparing the results of AMF flammability tests,the varying characteristics of the ignition sources may becomehighly significant.

PROPAGATION.

The known flammability properties of monodisperse hydrocarbondroplets (including neat Jet A and AMF) are significant forillustration. Referring again to Figure 11-5, below a dropletdiameter of 10 microns, flame propagation causes immediatevaporization. Hence, droplets below 10 microns may be consideredflammable vapor, provided that their concentration (e.g., number

*. density--number of droplets per unit volume) is above the LeanFlammability Limit. Homogeneous flame propagation occurs overthis range and up to a droplet size of 20 microns. Above 40microns, flame propagation is droplet-to-droplet. An unexpectedexperimental result is that, for one hydrocarbon tested, theminimum ignition energy in the range of ten to a hundred micronswas less than for a near-stoichiometric fuel vapor/air mixtureunder identical conditions.

The degree of dispersion affects droplet burning as well. Figure11-7 depicts conventional individual droplet burning as "Internal

"- 18

• FLME £ OXlI D I ZER

- *PRODUCT

0 0 r

0 a 0 0f

(D 0 0 00 0

FUEL DROPLET

G 0 OXIDIZER

PRODUCTS

A. INTERNAL GROUP COMBUSTION

FUELDROPLET

DROPLETS

10 abe0

go**". -'. 0P 0

*° °'PRODUCTS DROPLETS

VAPORIZING

(NO FLAME)

B, EXTERNAL GROUP COMBUSTION

FIGURE II-7. DROPLET COMBUSTION MODE.SI 00

Group Combustion"2 7' 28. This mode applies provided that thespray is sufficiently dilute in air, or that droplet dispersionis high enough in the preiginition phase to avoid combustion inhi-bition due to fuel evaporation and vapor diffusion. With "Exter-nal Group Combustion", by comparison, individual droplet burningdoes not occur due to the higher droplet number density andslower dispersion. Such variations in burning rate could 3ignif-icantly influence the combustion observed, and an AMF's resultingflammability (pass/fail) rating.

*Of course, polydisperse droplet populations must be consideredfor the AMF flammability tests. Semi-empirical computer modelshave been developed2 9 which address the transport, evaporation,ignition and combustion of polydisperse hydrocarbon dropletpopulations in flowing air. A typical finding is that a very

* small fraction of the mass of a ploydisperse fuel droplet cloud,*at the small end of the droplet size spectrum, can make the

entire cloud flammable. This indicates the importance of evalu-ating AMF's at maximum relative fuel/air velocities to developconservatively small droplets and potentially flammable conditions.

* Figure 11-8 provides the results of the model previously discussed,and illustrates interdependency of three important variables withtime. A monodisperse hydrocarbon droplet population is exposedto an initial temperature of about 390C. The fuel mass fractionincreases as evaporation occurs over time, while the droplet radiusdiminishes. The heat absorbed by evaporation drops the air temper-ature about 210 C. The fuel mass fraction very closely approaches

* the Lean Flammability Limit.

In the full-scale crash tests with neat Jet A, liquid breakup and* dispersion yield a totally flammable vapor/spray cloud. The

* flame propagates throughout the cloud and even to the very tipsof the forward fuel plumes--located several feet ahead of theleading edges of the wings of the SP-2H. Similar engulfmenit wasobserved with the wing fuel spillage rigs testing neat Jet A.

We have noted in the observation of movies of flame propagationwithin a fuel mist cloud from various test facilities that de-finite spatial limits to the fireball sometimes occur. In somecases, these limits may be due to a low concentration of dropletswhich, if closer together, would propagate flame fronts and burn.This seems particularly likely where the fire propagates freelythrough a portion of the cloud formed at lower velocity, and doesnot propagate through a visible portion of the cloud formed atnear-maximum relative fuel/air velocity.

- 20

4P

c0

Ti I.

*ly c )

a : I ;1 7 1 :...i --r

1-4 C) m 0

00 co l- 0-.7 coii0 E-

in r- m

(EOlx) NOILDVdI3 SSVW I2flA

U') QD Ln C) C: C) II (D tn 00

0Do MflIVUI21dW 12L ':ilAC

21

PASS/FAIL CRITERIA.

It is entirely logical and appropriate that the criteria forflammability performance of Fr antimisting fuel be ba'ed rlncharacteristics related to flame propacation under simulatedcrash conditions. For all scales of flammability testing, themaximum size growth rate and/or radiated flux experienred in agiven test are related to established criteria.

At large scale, the work of Salmon et al.5 r,7,S is of particularsignificance. From movies of fireball growth during the FAATCWing Fuel Spillage Tests, a nominal or "effective" linear fire-ball dimension (e.g., radius) was plotted as a function of time.The slope of a linear fit to plots of fireball radius versus timeyields a single value of slope, which is effectively the nominalflame speed in distance per unit tire (e.g., feet per second).Based on extensive experience with a variety of AMF's, criteriawere developed as follows:

Pass 0-10 fpsMarginal 10-20 fpsFail >20 fps

Fireball growth data have been developed from full-scale crashtest data on the SP-211 by Salmon, and were found to compare wellwith the FAA Wing Fuel Spillage data.

It is particularly advantageous that fireball growth is relatedto flame speed in the cloud--the latter beine a fundamentalparameter in applicable fire modeling. Furthermore, it isevident that fireball growth is being applied in the "terminalregion" (see Figure !T-2) of the FAA Wing Fuel Spillage facil-ity's free air jet. This is arplogous to fireball maximum sizeand radiation measurements made as part of the tests with theFAA's Flammability Comparison Test Apparatus (FCTA). Wake flamecriteria were involved in the NItC and JPL (small-scale) 1W7inqSpillage tests. Tn the former testing, a "rass" was designatedif there was no fire or if only self-extinguishing fireballs

rip occurred, or if the torch flames did not propagate (e.g., detach).A "fail" involved igniting the pool of spilled fuel, and envelop-ing the airfoil in flames. A "marginal" condition consisted ofigniting spilled fuel but not enveloping the ,irfoil in flames.

The JPI0 small-scale winq spillwoe tests involved a limit on the- length of the wake flame during a test (typically, I meter)

Measurements were also made. of the greatest hori7nntal distancetraveled by a fireball after detaching from the wake flame before

22

self-extinguishing or contactina a barrier. The length of thefireball at this maximum distance was also rcaFsured.

It is evident that in both these caseF critical pass/fail criteriaare being applied within the "established flew" region of Fiaure11-2, and may even extend forward into the "transition" reagon(from "potential core" to "established" flow).

It is apparent that the pass/fail criteria discussed so far canapply to one continuum--the free air jet as depicted in Figure11-2. Hence, with adequate aerodynamic characterization andheating/ignition/combustion modeling, the several criteria andphenomena of the various test rigs and facilities might becomprehensively and thoroughly related in quantified terms toadequate engineering accuracy.

In contrast, the SwRI Spinning Disc proTides a rotational ratherthan translational environment. Progressive degrees of mistcloud flammability involve (1) supplementing the torch flameintersected by the radial fuel ejecta, (2) the development of a"fire ring", starting with acute angles and, with increasing flam-mability, developing into a stable, 3600 ring of flame, and (3)propagating back to engulfment of the spinning disc itself.

SUSTAINED FUEL FIRE.

L4Sustained burning of the fuel mist cloud and ignition of spilledfuel pools is the ultimate post--crash fire hazard and failuremode for an antimisting fuel. Pool ignition has been included insome pass/fail test criteria, and extensive burning is encoun-tered with severely failed AMF's (or neat Jet A) in all flammabil-ity test facilities.

Propagation in the mist cloud is a better basis for pass/failcriteria, since the allowable fireball sizes, growth rates,and/or heat fluxes are limited to provide a generous and desir-able margin of safety.

r2

!. 23

REFERENCES

1. Anonymous, Full Scale Aircraft Crash nIests of Anti-M-istinqKerosene (InterimReport), Naval Air Engineering Center, Report

SuNo. NAEC-TR-183, 4 August 1980.

* 2. Klueg, E. P., Large-Scale Aircraft Crash Test of AntimistinUFuel, Federal Aviation Administration Conference Proceedings,"Aircraft Research and Technology for Antimisting Kerosene,"February 18-19, 1981, Report No. FAA-CT-81-181, June 1981.

3. Webster, Harry, FM-9 Anti-Misting Fuel Larqe Scale CrashTest #3, Federal Aviation Administration, Summary of Presentation,6th Meeting - US/UK Technical Committee on Anti-Misting Fuel,March 24, 1980.

4. Salmon, R. F., Wing Spi1age Tests UsingAntimisting Fuel,Federal Aviation Administration, Report No. FAA-CT-81-11,

February 1981.

5. Salmon, R. F., Wing Spillage Test, Federal Aviation Admini-stration Conference Proceedings, Report No. FAA-CT-81-181,

June 1981.

6. Klueg, E.P., FM-9 Anti-Mistin2_Fue_ Win_Sillage TestResults, Federal Aviation Administration Summary of Presentation,6th Meeting - US?UK Technical Committee on Anti-Misting Fuel,March 25, 1980.

7. Salmon, R. F., Flame Propagation Study of Jet A and FM-9Anti-Misting Fuel in a Moving Airstream, Federal Aviation Admini-stration Summary of Presentation, 6th Meeting - US/UK TechnicalCommittee on Anti-Misting Fuel, March 24, 1980.

8. Salmon, R. F. and Klueg, E.P., Test Plan, FM-9 AntimistingFuel Wing Spillage Tests, Federal Aviation Administration,

February 15, 1979 (Revised July 12, 1979).

9. San Miguel, A., Williams, M.D., Antimisting Fuel Spillage/AirShear Tests at Naval Weapons Center, Federal Aviation Administration,Report No. FAA-RD-78-50, March 1978.

10. Miller, R. E., Wilford, S. P., Simulated Crash Fire Tests As aMeans of RatingAirc-raft Safety _Fuels, Royal-AircLaft Establishim/ent,Ministry of Defense (Aviation Supply), Technical Report Number71130, May 1971.

24

L.____- - - -

11. Miller, R. D., Wilford, S.P., Fire Resistance of Antimisting

Kerosine (AMK) Under Simulated Crash Conditions, Royal AircraftEstablishment, Ministry of Defence, 16 October 1978.

12. Anonymous, Election of Fuel From Tank on Rocket Sled,

Appendix 1, Royal Aircraft Establishment, Ministry of Defence(Aviation Supply), Undated Draft.

13. Knight, J., Visit of the Technical Group on Safety Fuels to* the UK, 28-30 June 1978, Royal Aircraft Establishment, Ministry of

Defence, Materials Department, Farnborough, Hants, England,July 1978.

14. Makowski, T. D., Film Analysis of Jet Propulsion Laboratory_Small Scale WinqSpillage Tests, Federal Aviation Administration,Technical Center, April 1980.

15. Sarohia, V., Fundamental Studies of Antimisting Fuels, JetPropulsion Laboratory, California Institute of Technology,Pasadena, California, AIAA/SAF/ASME 17th Joint Propulsion Con-ference, Colorado Springs, Colorado, July 27-29, 1981.

16. Ferrara, A., FlammabilityComparison Test Apparatus,Federal Aviation Administration Technical Center, Atlantic City,N.J. (Draft Report).

17. Ferrara, A., Cavage, W., FlammabilityComparison TestApparatus Operator's Manual, Federal Aviation AdministrationTechnical Center, Atlantic City, N.J. (Draft Manual).

18. Eklund, T. I., Cox, J.C., Flame Propagation ThrouqhjSprasof Antimistinq Fuels, Federal Aviation Administration, NAFEC

Technical Letter Report NA-78-66-LR, November 1978.

19. Eklund, T. I., Neese, W. E., Design of an Apparatus forTesting the Flammability of FuelSpray.s, Federal AviationAdministration, Systems Research and Development Service,Report No. FAA-RD-78-54, May 1978.

20. Eklund, T. I., Experimental Scalinq of Modified Fuel Breakup,Federal Aviation Administration, Systems Research and DevelopmentService, Report No. FAA-RD-77-114, August 1977.

21. Zinn, S. V. Jr., Eklund, 'r. I., Neese, W. E., PhotoqraphicInvestigation of Modified Fuel Breakup and Ignition, FederalAviation Administration, Systems Research and Development Service,Report No. FAA-RD-76-109, September 1976.

25

* •

22. Mannheimer, R. J., Deradation and Characterization of Anti-misting Kerosene (AMK), Federal Aviation Administration Report No.FAA-CT-81-153, Southwest Research Institute, San Antonio, Texas,Interim Report, June 1981.

23. Mannheimer, R.J., Influencu of Surfactants on the FuelHandlinq and Fire Safety_Pr2perties of Antimisting Kerosene (AMK).Federal Aviation Administration Report No. FAA-RD-79-62, SouthwestResearch Institute, San Antonio, Texas, Final Report (Draft Copy),February 1979.

24. Davies, J.T., Turbulence Phenomena, Department of ChemicalEngineering, University of Birmingham, Birmingham, England,Academic Press, New York and London, 1972.

25. Sforza, P.M., and Herbst, G., A Study of Three-Dimensional,Incompressible, Turbulent Wall Jets, Polytechnic Institute ofBrooklyn, Farmingdale, N.Y., AIAA Journal, Vol. 8, No. 2,

- April 1969.

26. Miller, R.E., Wilford, S.P., The Desiqn and Development ofthe Minitrack Test for Safety Fuel Assessment, Royal Aircraft

*Establishment, Ministry of Defence, Technical Report No. 71222,November 1971.

27. Suzuki, T., Chiu, H. H., Multi-Dr2plet Combustion of LiquidPropellants, Department of Aeronautics and Astronautics,New YorkUniversity, Bronx, N.Y.

28. Chiu, H.H., Liu, T. M., Group Combustion of LiquidDroplets,Combustion Science and Technology, Vol. 17, pp 127-142, 1977.

29. Mahood, L., External Fire Model for Fuel System VnlnerabilityAssessment, Air Force Aero Propulsion Laboratory Report AFAPL-TR-"79-2053, Falcon R&D, Denver, Co., July 1980.

2

, 26

III. ANALYSIS OP FLAIMABIIITY TESTS

GENERAL.

This section consists of analyses of the antimisting fuel (A IF)flammability tests, data correlations and urcertainties, andmethods for enhancing data correlations.

FULL-SCALE CRASH TESTS.

The full-scale crash tests performed by the Naval Air EnainceringCenter, Lakehurst, N. J., have provided highly credible data onthe performance of neat Jet A and AMF under "real-world" surviv-able crash conditions. The credibility is achieved primarilybecause:

1. An actual aircraft is employed with very plausible fueltank damage mechanisms.

2. A variety of conservatively severe ignition sources areused.

3. Each test involves high initial speeds and decelerationof the aircraft over a carefully prepared impact course.

.' 4. Good diagnostics on aircraft dynamics allow analysis and

comparison of the tests.

The obvious limitation is the cost and time involved to test eachcandidate AMF.

Results of two crash tests (Nos. 1 and 2) of the SP-2H are pre-sented in reference 1. In addition, we have analyzed data fromtest number 4.2 Data comparisons are presented later in thissection in conjunction with data from other tests. The reader isreferred to the referenced reports if details of the testing aredesired. The remainder of this analysis will focus on specific

* test aspects, relating the test conditions and parameters to(1) the phenomenology discussed in the previous section, and(2) to similar parameters of other AIF flanmability tests.

FUEL AND AIR FLOW. Since the test aircraft were movinq in rela-tion to essentially quiescent atmosphere, the velocity of theaircraft relative to the ground suffices for an absolute value ofrelative air velocity. In reference 1, graphs of aircraft veloc-ity versus position are provided for Tests I and 2 (involvinq,

27

respectively, neat Jet A and Jet A with 0.280 and 0.3% FM-9 ati-mistinq additive).

For our analysis, we obtained oscilloqraph traces of accelerationversus time for these same tests. Ie developed and applied acomputer proqram which accepted data points manually extractedfrom the oscillograph traces and computed and/or plotted (1) dis-tance from an initial point (aircraft runout position "0" infeet), (2) longitudinal velocity in feet per second, and (3) lon-gitudinal acceleration in feet per second per second--all as 7function of time. Figure IIT-I presents these graphs to the sametime scale for SP-2H Test Number 1, which involved neat Jet Afuel. Figure 111-2 presents data on SP--211 Crash Test Number 2(Jet A with 0.28% and 0.3% FM-9 antimisting additive) in the same

* format.

We have been informed that NAEC has demonstrated the feasibilityof processing their raw data tapes to provide such data on subse-quent tests without the manual effort we had to expend in pickingoff the data points.

- The utility of the data in thi. format is high. First, it isgenerally accepted that the maximum relative fuel/air velocityplaces the greatest stress on candidate A4F's. Data for comput-

S.ing the maximum relative fuel/air velocity in a given test are* available from this format. Air velocity vs. time is readily* available, and can be related back to position on the crash site

by reference to the distance/time graph.

Velocities must he considered in conjunction with the decelerationforces acting on the fuel in the wing tanks. The latter deter-mine the instantaneous force acting on the fuel being expelledthrough 6-inch high by 9-inch wide openings punched in the for-

0 ward walls of the twelve wing fuel tanks.

The procedure given in reference 1 using instantaneousacceleration was applied in computing a velocity ef neat Jet Arelative to the aircraft. Table ITI-1 presents average and peakdeceleration magnitudes and time duration needed for computinqforward fuel expulsion velocities. Also included are the air-craft speeds for the correspondinq times, and aircraft positiondata. Table 111-2 presents the fuel expulsion velocities andtotal relative fuol-to-air velocities for five selected timeincrements. The increments were selected (1) as the relativelyhigher values of peak acceleration at hiqh voel-ity, and (2)representative sustained deceleration at lower velocity later inthe crash runout.

28

Z03/i NA LA33

4E-4

-E-H

U

Cz >

0-

4C

C129-

In U - U\ -- - - - - - - - - - - - --\

)J3S/13 'NOIIV MMVJJ

0 U' 0 U% C 0 0U'.C14 - -

- C

CN

CN

IE-

-J 0

00

I-4(

LU uLLJ

0o HH

4 ____ __________0

44r3

* ~-~-- ---- "\

4

F: (4 E 0 (N, C- C)i (N ( C

4-J -) (NJ C1.4 ci -P C1 rl C) C CD

P-4 -4 r-- mN mf .-. ( D CN

4-4 0 -F 4-H c rr C N Ci r- Go rq n CN Ln. r- MN C

u ~ (N (N* 'zv zr m' IC C- -zJ Ln C

4-1 Ln Lnv-inLr c,' ri4 C" - 4 JC

4>i

0 -

CD C

-P

C) CO)C 'C ) IH m ,-i r' CC)

r-4 ~ CD ) CD 0 D 0 0 (

m C) - 0 0 C) '-4 0 0 C -)0. u 14 L ; o C

Q 4r C 1 -

0- ___ C______ _______ ______

IJ~rq

'-II

a)0 -. M-- O

4.3 1

4-)J

E- EnU)

CDHC -r.4 E- U, r- H ' 1 ' P

04 V C

0dv -n C!Q 0

U) m ) 4

04 41 o (U

0) 4 QoE a ~v- .- N L' 00 0')

C/) H- .- I r-1 '1 11 m1HU) U 0 0 0 c C C -

0' N m 1 0.P4- (o 0) 00 0 C 1 C

:I r- (n Iq C) .10 (NI% 4-A EU C 0 C 0 0 -

ra 0 10 - C N C (N N N *r.H co m LI) 1.0 mC m)

~4J '

4-) 0 04J (Ni re) I) r NUa)IC 0 E 44 '-4 -4 (N N 0 U

'-4 -1 - "i (- N m 0S 4.1

H ~ U) L-)- r- r0 0 Q4-3 C3 (N P1 4I

* 4) r.. N-4- ::jH l .

E- (a 1-4CD

0U) a)40C U) 00 C .0r- ,

(U- O1 W OD (N 'T -4 41 (

0N -4 :1 z4.4 ,4 '- -

U W-44-4

N CN

U)) Lo CD

* I CD 4--4-

IH) 41 -C

'-4 (r,- U) m 0 C) 1 -.

U ) UN. U) U ( 4

0 0.- 0I 0 d 04 4-4 if)4 t- 3)--,

CD7 4- 4 r

(U Z) ~ P 4 0 I)4* 1E- -

CD C (N N .- (N H U

(U32

- - - - - -

" LIQUID BREAKUP. The major purpose of the analysis of the crashdynamic data was to establish compatibility between larqe-ncaletesting and liquid breakup analsis and modeling. Liquid breakupis discussed in detail in Section V. This particularly imoor-tant considering the general lack of droplet size an. di-tribu-tion data for large-scale tests.

Referring again to Table 111-2, the input data werp applied tothe Falcon R&D Liquid Breakup Computer Model to obtain the dropletdata presented. Applicable assumptions for this computation were:

Fuel: (Neat Jet A)

Density 0.807 q/cm 3

Surface Tension 30 dvne /cm

Viscosity r L .015 poise

Air:

Density pP 1.293 x g/cmIi -4

Viscosity g = 1.722 x 10 poise

Temperature (Fuel. & Air) T = OC

Orifice Diameter (Avg.) D 7.2 inch = 18.3 cm0

It is important to note that the model is limited in this applica-tion by the necessary assumption of a circular orifice, and byits current omission of the effects of Taylor-modified breakup--which is evidently applicable in these tests. Droplet evaporationis also neglected in this example. Table 111-2 summarizes thedata for primary and final droplet mass median diameters, and the10% and 90% diameters for the six time increments selected. FiguresII-3 through 111-8 present the droplet diameters versus cumula-tive mass percent as computed and graphed by the computer programfor six time increments.

A In all cases, the upper (solid) curve is the "primary" dropletsize spectrum, terminating with the symbol "M." The "final"droplet spectrum is a lower dotted line, terminating in a numberwhich represent-, the time steps taken in secondary breakup beforereaching the "final" distribution. For these runs, steps 1

* through 20 were each increments of 0.06 milliseconds; Fteps 21through 40, 0.30 milliseconds, and steps 41 through 60, 1.50 milli-seconds.

° V '. - . - * , --. ,, _ -. -. . . . ..* . -. . . . ..*

. . .. .....

p q

2

.. . .... ... .... ... .. . ...

E0

...- .... ..0.... ......

~ . ... .. .. .. ... 4 4 4 4 4 .4.444 4. -r4414 1

...... \- 0.."

.. .. . ..z .. ....

.. .... . o

...... ... * I. ...... ** . 4. . . . ~ ........ . . . 4 4 .

00 r r- cc 0 0

0~~~ 00 )

- 34

0r'

* Ln

0')

E-4

ON

In H

z U)

E-4

U)>

N.S**III4....4'.i ... .........- v i i , 'S 0

N 0 N

~SI*4hISS40 * S 0 SCI+ I, S llI ,) I---- L

(WO) HLHWIC lqdOH N4

35z

. ... ..... . .

E-

.. . .... ....a .. ..

........ 0..

c,

cc

0 U)

0

m a'

z C)

E-4

Nt . ... .. .m.Hmm)

-- III4 II O:)

CD~ 0%c

(WO) al3WVG I~qOH004

360

Lol

................ . . . . .....

F--

U))

00 U)

*t.....a . g**g I.\i I a a ~ i i . , , i , , ......... ,a..

.2.-l

U'

.........*,o& %o v, -1 *0 .o aq N -4 0ao* ,* ' *, , ,,, , ., .a 0

0~~~~N cn 0 00C0 0 00 0 ; C C( C

a~~~C a, al,, Ct~ a a I I .

37c

.. . .... .. .... .. . .. . .... ...

ON

. . .. ... . ..

.. ...... . . ....... . .. .. ........ .... ... . ...

>> E--

a ~I g I ,' ~ I ~ , I ~ I I t . ,I e * . a ~E->

o a( 0

-- ~ . .... ... .. .. n .~ C lw .

--1 - ... .. -.. o.0-

.... .........

-4 E-. . . . . .. . ...... r

.--. . .....

.. . .. .. . .l

.. . .. .

............ ...... . ..............................................................

0-f C-- l wi - -- - . 0

384

. . . .....

wa

aa

E-aa HN: : :a' z

4.444~44.4 4. . 4. .4. 4I .44 44 ~ 4I4 4I44 *44 .44 44i ~ q

-4 -U

w.44. 0 44 0 ) 4

Lo4

> C4 0

.x . ...... H O4-4

C'44

-4 01

N.414P444~~...... 44 .4 4. a4*0. . 4 4 44* 4 4' 1 4 4 4

.~~ ..... E-4 0

..... ..... ..... .. . .... .j 0 -

.. . .. . .C

0

O0 Wo 0T4 - W 104r r-4C 0 .0 Q 4-00 0 0 0 0 0 0 0 CC)

0 0 0 0 00 0 0

(111) H3JSZWVIG IdUOHO

4 39

Review of Table III-2 and the six related graphs reveals a con-sistent trend towards larger droplets with diminishing rolatiN,,fuel-to-air velocity. Corresponding aircraft position data indi-cate the locations along the "runout" to which these elevatedfuel-to-air velocities applyv. The earlier time increments repre-sent both the higher velocity and deceleration conditions alongthe entire crash runout. Hence, they represent the most stressfulconditions for the spilled fuel. These position data indicatethe sequences of minimum-sized droplet populations that are beingdeployed in the mist cloud along the crash "runout." The timedata can be applied to fuel discharge data to obtain the quantityof fuel for each increment.

Of course, it is evident that all the droplet populations producedduring this crash with neat Jet A fuel were readily flammable--the port wing is engulfed in flames from early in the crash se-quence (per the photos in reference 1). However, for A4F test-ing, the application of the liquid breakup model could be aninvaluable supplement to data analysis, as follows:

1. The model could be used to establish optimum locationsalong the crash runout to array fuel mist cloud (droplet) diag-nostics.

2. Discrete droplet data from a crash test could be used inconjunction with the more comprehensive model predictions todefine subdivided mist clouds with characteristic droplet popula-tions.

3. Any regions of an AMF mist cloud in which ignition orpropagation are observed in films could be compared with corres-ponding droplet population data. This feature would be invaluablewhen "marginal" conditions are encountered in crash tests, as areencountered in other flammability tests.

A similar analysis cannot be performed on Test Nos. 2 and 4 sincethe breakup model does not currently account for Taylor breakup,and a data base of antimistinq fuel droplet spectra is notavailable for model alignment. However, it is quite feasible toextend the model to AMF's with further development.

Quantitative mist cloud analysis applied to full-scale crash testsis highly desirable for direct comparison to both large- andsmall-scale AMF flammability testing. If scaled rigs show that

6_ critical droplet• populations are developed under similar (orscaled) test conditions, credibility is vastly enhanced. Tndeed,a consistent set of critical droplet populations between various

-* 40

AMF flammability tests would enable urprece-b!vcd corre!ation and

credibility.

WING FUEL SPILLAGE TESTS.

Table 111-3 compares the five wing fuel spillage facilitiesstudied in terms of scale and dynamics. They are discussed inthe following paragraphs.

S-"FAA WING FUEL SPILLAGE FACILITY.

This facility represents a mature stage of AJIF flammability test-ing. The facility employs a 69-inch square free air jet drivenby a turbofan engine. This is sufficient for full-scalerepresentation of a large breach in a wing tank, and high fuelspillage rates. An invaluable advantage is the fixed testarticle, which allows for fixed diagnostics, ignition sources,etc. Steady-state spillage, ignition, and propagation conditionscan be studied, in addition to variable air and fuel flowconditions simulating various stages in F crash. Controllabilityof test variables and conditions is greater than for aircraftcrash testing. The generally non-destructive nature and morerapid turnaround time compared to the crash tests make it aninvaluable tool for studying at full scale the sensitivity ofvariables such as: AMF additive type and concentration, relativeand absolute fuel and air velocities, fuel orifice size and shape,variable (reducing) air speed and fuel discharge, angle-of-attack,proximity of the ground plane, sensitivity to various tempera-tures, and type, location(s), and time variations of ignitionsources. As discussed in the previous section, pass/fail cri-teria have evolved to a measurement of fireball growth rate,which has been correlated to the full-scale SP-2H crash tests.

A similar opportunity exists for correlatior with the full-scalecrash tests in terms of liquid breakup. For example, the sixtime increments of Table ITT-2 represent SP-2H crash condition.<averaged over various time durations. Each such increment couldbe simulated under steady-state conditions with this wing spillagefacility. "Cold" tests could be performed at these conditions,allowing deployment of droplet diagnostics without the complica-tion of fire. Comparing droplet size and number density datafrom full-scale crash tests and FAA wing spillage facility couldprovide a wealth of "hard" data for full correlation and, ifnecessary, auiustments in independent test variables, such as airvelocity, fuel velocity, etc.

.d 41

C) I

o- U) H _j(1) :(-4U U)t *Hy t Cd 11 U d

S U2) cri C) C,) U CrJ4 0 C) m 4 C

xCA ) Q)

) t7) UuC) ~ H *H

S 4J -I-) 1'-1 4-'

E- 4-) 4-) 0 0 4

t-,(0

C))

*u .) *Hu )C30 U 1 -o -4 r~f -4 -4 4-

0) (04 ~ dC:ji 41 4

*H, C fLo C.)

1- 4 -4

-4

HU) P4 U) Vi -

.4- 4Ji

r-4l r4 C,)o .-l* -4 r-4 U4-)

4J 0C fu ( H T. -, 4-I-4

tn H ( o 0 )LH 4 4 I 4

C) -4 i-4 r .4

'- .,AJ W ~ U) (

42

The droplet data so developed would be equall, as important as abasis for evaluating droplet data from small-scale ANF flammabi]-ity test rigs for a like comparison.

NWC WING SPILLAGF FACILITY.

This facility is smaller in size than the FAA wing spi?.]age faril-ity, and may be considered a stage in the development of thelatter. It is more limited in the number and range of independenttest variables that can be simulated. It is not considered agood candidate for further characterization in terms of licuidbreakup or flame ignition and propagation.

RAE ROCKET SLED TEST FACILITY.

As with the full-scale crash tests, no airflow simulation is neededsince, in the "standard" test, the wing tank is first acceleratedand then simultaneously opens its forward-facing slit-type orificeand is controllably decelerated. The fuel is expelled into apre-deployed ignition arrav--as many as 36 one-pint kerosene con-tainers with burning wicks. Alternately for the 'run on" tests,the orifice plug is released upon sled acceleration. After rocketmotor burnout, friction provides low deceleration (e.g., 0.5 to1.0 g) and fuel exits the open slit at very low velocity relativeto the tank. Suspended burners ahead of the sled arrestor cableare provided as ignition sources for fuel released in this manner.

Reference 3 provides the basis for various deceleration rates,and compares deceleration pulses achieved with 1- and 2-rockettest runs versus 50th and 90th percentile crashes.

The deceleration pulses vary greatly in time history, with thefollowing peaks:

95th Percentile Accident = 26g50th Percentile Accident = 14a1-Rocket Test = 5g2-Rocket Test = 12g

Fuel ejection velocities versus time are also compared betweenthe I- and 2-rocket "standard" tests and for the 95th percentileaccident for tanks of three different longitudinal dimensions(1.52, 3.05, and 6.1 meters; i.e. 9, 10, and 20 feet). Tab]e TIT-4presents representative values. It can be noted that both sledtests start with their maximum fuel ejection velocity, and grad-ually diminish by about 20 to 25 percent. In contrast, the crashpulses start at about 40 meters per second at time zero, peak at

44

o' *r LA 14

C.)0

U)

a) LA a)

4-)

- Er

0 4.4

410

U -4 4 L A L

M 0 >rX4 C)

44 --

4J~40

Q) C

E-1 4- C4Q

-)4-) 4-,L

4) 0O CO (1)c

E--4

I 444

about 75 milliseconds, and decay to about 15 meters per second at150 milliseconds.

The sensitivity of the differences between the magnitudes of fuelejection velocities between pulse-type crash conditions and thesustained expulsions of the sled testing is not quantitativelyassessed. This is another case where droplet size spectra andnumber density data from selected sled test runs could be compareddirectly with data from similar runs (but with decelerationpulses) of the full-scale crash tests. Liquid breakup modelingcould also be used to assess the variation in droplet populationswith widely varying relative fuel/air velocities between the giventheoretical and test conditions.

In relation to breakup, the range of heights of the slit typeorifices in the leading edges of the large wing tank appears totranscend ranges of distinctly different phenomena. RAE reportsthat for the larger tank, the 30-inch (762-mm) wide slit can bevaried in height from 1/8 inch (3.2 mm) to 1 inch (25.4 mm),whereas the smaller tank slit is 18 inches (457 mm) wide by 1/2inch (12.7 mm) high. Our analysis indicates that for minimumorifice dimensions above and below about 1/2 inch, different break-up phenomena apply (see Section V, "Liquid Breakup"). This wouldindicate that some scaling may be necessary to correlate dropletdata from slits as small as 1/8 inch versus 1-inch slits.

RAE MINITRACK TEST FACILITY.

The Minitrack test facility4 was developed to provide in a smallscale test an indication of the capability of a given candidateAMF to pass the "standard" (i.e., deceleration) sled test usingone rocket for propulsion. Typically, this gives a sled velocityof 35 meters per second at arrestor contact, with a mean deceler-ation of about 5 g. For the Minitrack with a fuel volume limita-tion of 50 ml, and fuel longitudinal dimension of 3 inches (76.2mm), the corresponding speed selected was 37 meters per second,with a mean deceleration of 30 g (six times greater). it isinteresting to note that the orifice dimension of 5/16 inch (8mm) implies that scaling is according to our liquid breakup-basedlimit of 1/2 inch. Comparing.the'results of five different candi-date AMF's, the RAE found that it took 2.3 to 3.3 times as muchadditive concentration in the large-scale sled test fuel comparedto the Minitrack to get comparable "passing" results.

The necessary modification to the Minitrack to evidently limitspray dispersion (i.e., increase droplet number density) has beenpreviously discussed in Section II. This is another case wheredroplet size spectra data on large- and small-scale AMF

45

flammability facilities would be invauable. It is conceivablethat the Minitrack could be adlusted to provide representativemist clouds in quantitative terms. Droplet measurements on a

- - screening test such as this might also indicate anomalous perform-ance of an AMF with a different sensitivity to the Minitrack'sparticular dynamics.

JPL MINI-WING SHEAR FACILITY.

* This facility utilizes a free air jet 8 inches in diameter atdischarge, capable of a maximum speed of 80 meters/second.Approximately two diameters downstream, centered (elevation view)in the jet, is the mini-wing leading edge, incorporating a slit0.5 cm (0.2 inches) high for fuel release upwind. Maximum fuelejection speed is 15 meters/sec. Located approximately 3.6 airjet diameters downstream (centered in the jet) is en oxy-acetylene flame whose holder and mass flow can be altered to varyignition source size and eneraN, release rate.

* - Thermocouples at four locations downstream (from about 7 to 17

diameters from the free jet orifice) along the jet flow ayismonitor wake flames.

Test parameters are air jet speed, fuel ejection speed and massflow, fuel temperature, and AIIF concentration.

This facility must be considered in the broader context of a multi-faceted program by JPL to advance AMF technology. This includesfundamental study of AMF behavior during breakup (e.g., pendantdroplet testing), development of a somewhat larger small-scalewing shear apparatus, and application of advanced imacing terhni-ques to quantitatively evaluate AIIF mists.

The advanced imaging techniques have already been applied to theJPL Mini-wing testing. Reference 5 provides preliminary data andimages on fuel mists for Jet A neat and with FM-9 AMF in concen-trations of 0.1, 0.2, and 0.3% by weight. This work is partic-ularly relevant to correlations of AMF flammability test rigs,

* since combustion-based correlations may be supplemented by quanti-tative droplet data correlations.

. The data presented in reference 5 are promiring, but need the

.. following features to allow direct correlation with liquid break-up models and/or less sophisticatcd diagnostic!- being recommendedfor other AMF flammability tests:

1. The droplet mist should be characterized in the regionof the ignition rource. As illustrated by the example model

46

I

output, Figures 111-3 through 111-8, breakup may still be pro-gressing--particularly in the evident wake of air and fuel offthe trailing edge of the wing.

2. Data in the form of cumulative mass fraction versus drop-let diameter (Figure TI-6) are also highly desirable for theregion of the ignition source.

3. Droplet number density data b- droplet size range (e.a.,increment of cumulative mass fraction, Figure 11-6) in the reqinnof the ignition source are also highly desirable.

The above may represent the practical and economic limit of datathat may be obtained at other AMF flammabilitv test facilities.

It is also desirable that a generous data base be establishedwhich directly relates droplet characterizations to flammabilityresults over a wide range of independent test variables, includ-ing air and fuel ejection speed, fuel mass flow and temperature,AMF additive concentration, flameholder configuration, and flamemixture flow rate. To facilitate comparisons, the values of theseparameters are needed, in addition to any scaling schemes.

Pass/fail criteria related to the JPL Mini-wing have been pre-viously discussed--based on wake flame measurements by Makowskiof the FAA. Another Pass/Fail method is defined by V. Sarohia(reference 5). The aforementioned thermocouples in the wake flamehave registered a large difference in "reduced temperature," asmuch as an order of magnitude, depending on whether the ratio offuel-to-air mass flow rates was above or below about 0.1. Forexample, increasing the mass flow ratio from 0.08 to 0.11increases the "reduced temperature" at one thermocouple location(about 11 diameters from the air jet's orifice) from about 0.02to about 0.21. This was for 0.1% FM-9 by weight in Jet A. Asso-ciated with these flammability test results are correspondingimages of the two mist clouds. According to Sarohia, this largechange in recorded temperature may be used to define "pass" and"fail" mist configurations.

No data have been located characterizing the free air let util-ized, other than the single velocit- in the "potential core," andrelated air mass flow rates and temperatures. Characterizationof extreme and mean velocities and turbulence intensities aredesirable in the regions(s) of liquid breakup, iqnition/heatinv,and propagation.

d4

F.

FAA FLAMMABILITY COMPARTSON TEST APPARATUS.

The FCTA has been developed to provide a common apparatus forevaluating the flamrability of AMF's at various facilities. Fiveunits have been shipped to various test facilities, with a sixthretained at the FAA Technical Center. The FCTA provid(F F porta-ble, self-contained, semi-automatic test syst-m which requiresminimal test fuel and can be operated indoors with adequate spaceand proper precautions. AIF's can be evaluated at various fuelflow rates and air speeds. The units have been calibrated andcome with an operating manual. The FCTA is invaluable for screen-ing the candidate AMF's anId f(r monit-oring the flammability per-formance of various batches of ATIF at different points in theirstorage life.

Details on the FCTA are presented in reference 6. Flammabilitycriteria have been discussed in Section I!. Briefly, air from apressurized tank is introduced into a 1-inch T.D. mixing pipe ina velocit, range of 40 to 80 meters/sec. Test fuel is expelledfrom a cylinder by a moter-driven piston into a 0.25" thin wallstainless steel tube. This tube penetrates the air pipe andextends along its centerline 4.5 inches before expelling the fuelupstream into the air flow, typically at rates from 8 to 18ml/sec.

The two-phase mixture proceeds downstream where it is expelled* through a 16-inch diffuser whose large end is 4 3/16-inches in

internal diameter. The diffuser protrudes through the right-handside of the FCTA's cabinet. A bucket catches liquid from thediffuser's outlet, and a 4-foot by 8-foot steel pan collects air-carried ejecta. A propane torch is mounted along the axis of thediffuser, pointed downstream, with the flame holder tiD aboutfive inches from the exit plane of the diffuser. Any resultingcombustion is recorded photographically against a background grid,and by a radiometer looking orthogonal to the axis of fiow--20inches to the side and 9 inches away from the right-hard wall ofFCTA's cabinet. Runs are typically performed at various combine-tions of fuel flow rate and air speed to obtain sequentiallv moresevere test conditions and combustion responses. The FCTA is theculmination of development work as given bv references 7, 8, 9,and 10.

ANALYSIS. The FCTA differs from :he wing spillage rigs in that(1) no replica wing ic provided for fuel/spray impingement andpossible reintroduction into the air flow, and (2) the airsupply, liquid jet, plume, and diffuser regions are enclosed.Referring back to Section II, Fiqure TT-4, when the diameter of

* 48

4

the fuel plume exceeds the diameter of the bcnunding pipe, impinge-ment of the entire periphery of the plume occurs. This peripheralimpinqement is obviousl,, fundamentally different from impingementupon an airfoil traversing approximatel? the horizontal diametorof the plume. Considerine the mechanics of liquid breakup in theairstream, it is reasonable 1hat the smallest and most rapidlysubdivided fuel particles would be selectively recombined byimpacting the inner wall of the 1-inch T.D. pipe of the FCTA;whereas tests with an airfoil give impingement across a smallhorizontal "slice" through the plume--involvinq only a smallproportion of the outer plume droplets. in such a case, theperipheral impingement of the FCTA might provide coarser dropletsand, hence, less flammable mixtures than with the same conditionsbut without peripheral impingement.

It is not evident whethpr the range or fuel and air flows of theFCTA will yield conditions of substantial peripheral impingementin some cases, and very low or minimal impincTement in others. insome operations of the FCTA, a substantial proportion of the testfuel load, 20% or greater, has been observed to flow as a liquidfrom the diffuser after a test run. This would indicate, barringsome malfunction of the FCTA, that very substantial impingementis occurring.

If FCTA onerationp do cross over from low to high impingement,the test results may be significantly shifted--since dropletsstripped from liquid fuel in contact with the inner surface ofthe pipe are formed in a very complex boundary layer, whose char-acteristic velocity profiles and turbulence intensities may varyradically from the nominal air flow conditions calibrated andmeasured in the main flow stream. This is further complicated bythe fact that fuel mixing is occuring before turbulent air flowhas been fully established after passing through the sonic "choke"orifice.

It might be argued that the impingement and reformation of muchof the liquid in a given FCTA test might represent the closeproximity of a ground plane to a wing--e.g. with a low-wing air-craft crash scenario. Although it is true that large quantitiesof spray will be intercepted by the ground plane, the factremains that as much as 1800 of the outer plume would not impingeon the ground plane, whereas 360' of the plume impinges with theperipheral confinement of the FCTA.

Regardless of the mechanism whereby droplets are formed, an objec-tive criterion can be based on liquid breakup analysis and model-ing with appropriate droplet size and distribution data. The

49

FCTA appears ideally suited for droplet characterization in theIregion of ignition and fireball propagation. The already exist-.% . ing flammability data base could be rapidly supplemented by "cold"

tests (e.g., no ignition source) involvina relati- clv simple,portable droplet diagnostics. By comparina independent testvariables of the FCTA (e.g., fuel and air velocity, AMF concentra-tion) with droplet populations and flammability results, and h'ycomparing similar data from large scale facilities such as theFAA Wing Spillage Facility, verv firm and credible results shouldensue. The postulated differences in breakup mechanisms discussedabove become moot if relative fuel/air velocities can be foundfor the FCTA which give droplet populations which correspond(with scaling) to those of a large-scale facility.

SWRI SPINNING DISC.

The Southwest Research Institute Spinning Disc facility is uniquein that it utilizes rotation to accelerate liquid Pn into theatmosphere past an ignition source. A disc about 2.5 inches indiameter is mounted horizontally on a high speed (variable)motor's shaft. A cup is centered in the disc, into which the

*test fuel flows from a static tube overhead. The cup connectswith four radial flow channels in angular increments of 900 mea-sured in the plane of the disc--each 2 mm in diameter and 10 mm

- long. The channels terminate at ports in the edge of the disc.About 3.9 inches (10 cm) from the disc's edge is an ignition

• -flame. The disc apparatus has the interesting property that theliquid flow rate through it does not vary appreciably with RPM.Hence, for a typical test a suitable flow rate is established atlow RPM, and then rotational speed is increased to a maximum ofthe order of 30,000 RPM. Very discrete and repeatabl(. flammabil-ity phenomena can ensue. The first stage involves increasinggrowth of the flame envelope about the ianition source. At higherRPM, a remarkably stable planar semi-circle of flame developsabout the torch. At still higher RPM, a full (3600), stable ringof flame typically develops. At a "worst case" condition, theflame propagates inward and engulfs the spinning disc.

ANALYSIS. Of particular concern was the physical comparabilityof liquid breakup b, opposed air and fuel jets in longitudinaltranslation, versus the rotatioral cheme of the spinning disc.The relative fuel-to-air velocity for the spinning disc isreported as the tangential velocity of the disc; that is, theproduct of the disc's angular spep1 ind its radius. Table TTT-5presents tangential velocities for various rotational. speeds.

50

TABLE 111-5. SpinningT Disc Dynamics

Rotational Tangential Radial

Speed Velocity Acceleration(RPM) NOi) Wg

6,000 19.8 1,260

12,000 39.6 5,070

18,000 79.2 11,400

24,000 158.4 20,300

30,000 316.8 31,700

51

This table also presents values of radial accoleration actinq ateach exit orifice in the disc's edge, based on:

2AR = rw

where AR is the radial acceleration, r is the disc's radius, ard wis the rotational speed.

Considering the high radial forces acting on the liquid at theejection ports, and the fact that the liquid has been acted or bya radial force which has increased proportionately with distancefrom the disc's axis, it appears that a significant radial veloc-ity should be vectorially added to the tangential velocity toobtain a more realistic value for the initial fuel-to-air veloc-ity. It is also evident that the liquid is being acted upon with-in the flow channel by lateral Coriolis forces. These are typi-cally expressed as the product of twice the angular velocity andthe localized translational velocity vector. In this case ofsimple rotational motion, the translational velocity is relatedto the applied radial force as previously defined. This indicatesthat the liquid in the flow channels is being subjected to complexand extremely high forces. It seems plausible that at such highforces, unrealistic responses may be expected from viscoelastic

*test fuels in particular. It is possible that for one candidateAMF, the "gelling" experienced with high flow through filters may

* occur within this the disc's flow channels.

In addition, there is no evident mechanism except fluid drag todamp out the high rotational rates imparted to the liquid up toejection. This could significantly influence the shapes of AMFdroplets, in contrast to relatively negligible rotational effectsin full- and large-scale spillage tests.

It is particularly important to directly compare droplet popula-tions formed by the spinning disc with droplet data from largt-scale tests. In this way, it is likely that an empiricallyderived relative fuel/air velocity (related to disc RPM) willprove more relevant than tangential velocity of the disc.

CORRELATION OF FLAMMABILITY TESTS.

The various AMF flammability test facilities were consistent indifferentiating between fuels severely deficient in post-crashfireball reduction, and relatively successful ones. Table II-6provides data on the characterizations of the flammability per-formance of neat Jet A and A&IF for each of the facilities, versusappropriate independent test variables. For ease of comparison,

52

M .in4 K XK K

x 0U. w Kr K K K 9 x K K x K K x x x x K

A.iii to 0 FAKx

44 ~ 4 4) CL C- C-VV V V V

Mni 44 4)I M MV) c

-4 4KO innai iin e4 in in U) U, in m in U) W. (r,

-H 04 U) U) I L i .6 66 4.6 664 46 '64 664 64 46 644. 4.6 M. '6-w4~ Vin.4t6 in i

C40 Ajnw n Ajna 'o a) 0. a. 0. 0 a0 0 0 0. 0. a. 0n 0. 0.to4. in i46 inn 4 44 4J 416 416 46 641 41. 61 641 4 4J . 6 ' 6.

mU -M - > ID ' 16.4 0 Ca0 0 0 0 0 0 0 0 0 0 0 0 0

0 in Nn (Ni inm M m V v -W vvwaU.n U) L f160 IDJJii a~.66 a00 0 0 0 0 0

w.6 Ln won c,4 04 0o 0 0 0 0 0 0 0

to -A...6 44 - x -

0 -M c~~aa aLa a. a. a. a. a. a. a. a. a. a. a. Ln a.L ) wl W r

4 0 0

W LWNN 0 0 00 0 0 0 0 0 0 0 0 0 0 0 0 0in . rq r NC NON e if4 CD04 C4O E EN N C, M 0 .4M 4N EmN N NE EN el Ei N " N C4 mN E4O N ENC4m

16. 1. in

C4 eq C4- 0. C..~~ 4 4 en m4 m4 m4 N6 C4 m4 t 0C4 N4. N NVV VV " V4 N %D " 0 V V V4W V C4 M NV

V.6~ K AKK4K41 4 D -r*~~C C46 4 0 0 ... NNNN EE N N N EN N EN N C\ E N eN

in

u 0 Dr 0 0 0 0 0 0 0 0 0 0 0 0 0 0LM C.44J NN ENNNN m 00mm f n e fNEWNNE0NN~ n m M m m m (N('1 N .eNq im E .V

0 .0 V ;C

.6.) m4 0%4P ENW E mE ON) m EN ENEmN E N E N E N E

~ K -66 I- kD D u 9

41 I0 0 0 0 0 0 044f P e0 0 * W 4~.6 *i* 6 * P * 0 6 *..* W 6.* 0* ~ 6 *.

00 $41) 0 N O ENN- ENN E 4 P 4 P 4 E EN 0 0 0% E l 0

1) 41

A46 ~ 4 4E

CU - 0M NN( n..0. ~. g4 .4A

IA

C4a -to FA to K

Q)

41+ ) a) ) 4) 4) 4) 0

(a' 0) 0. 0. w. W.0 c) 0

04

.,40 44 4. 44 04 W44 b'.

c~0 ". C- . C. . . C0 '

0t 0 0 0 0 0 1 44d 44 1.4

I. 0 0 0 0 0 4 4)

0 Lo '..8 G

.44 04 N

044 0. 0. I. In 0. 0D.In C

-40 0: 0 0O C; ' 4f0Oc4 04 ~ N N N 1 N Nc1444 44 .

) a-W %64.4. rm c;m4

c* 4) 4.,U

-n 4j *n 44 Li 44 'a AU L jLto'- to 4 vi 4 *.4 n * in~ in 4 .n L - n 0 CDb.) '44 1*1 w4* N4 w4 N4 w*.. w N w cv 4 c4 4N N- N4 m4 . 4

0 1 414 m!' If.)iA m m m V V V V VU) . r N N N4 N'4.N c4

0 0 N

14 -4N) N -4

0 0 0 0 0 0 0-40 04I444 4444I44 0 0 0 Aj 41 0' -4 0 4 4j 0'

4 441 4( 41 0'0 0 N N 4j 0. a 0 0

14 4414 m N4 N N N

4- a) 0 :c

4. Z U 0 0 0 0 0 0 n n 0 0 in n 0

40* * * 0 0 0 00 0 0

04 0

0.4. . . . . . 4 . 4 .

Q u(A. 4j W

0' 0 0' 0' 0 0'0' ' 0' 0' '0'0 0' ' 0II ~ ~ ~ ~ 4 C I I~~

am

1. 0 C.x x -

a. laE E' m M x x

00 AD a) W )c. C. C.M z4. 0. J. 0 xK w4 x .A

c0 ) 0 U) w M 4j0 m w W, 3t3t

-W 0. 04 Z.- 144 14. 4 .0 3(D 0 0 w 0 W 0 4.. wU 00 4) 4)cc CDc

in .4 C 4C CO C O C- C 0.4- .

CO~ ~~~~~~~~ 0~4~. c 04U 0- .0 - 4 04 N M N

m m0 0420 0 0 0 00000 -0 0 00 0 '0 c 0 4C ' 0 .

cc Z FZ m. ~ - 0 =0 10 1. m w~ IL .IL

0 m

c 0

0

41

0

.D '44D- koK %D KK NKK K MK K4 C4 Nf DCOOK'' Cv -W -W-. -0 0N v vv N N N C4 -f )VC Z C

UC 00 0 0 00 0 0 0 00 N0 ,O

0 M n C4O

0 4. . N N .N .N .N N M M P M I N N I I

0 04

0 ) .0r-M 0 4

.4. 1FA

.4 00 0 0 0 0 0 00 0 2) 0 22 0 0 0.4a2 NNNN INN INN NIN N00 0 0 00 ~ .~i 0' -M

NINN N NN NIN NINI INNI IN NN N NIN NNN 4 N N

0 0 xu-.- x x x

0 U41 4

4 4 4 6 6 6 6 0 m 0 TO M46).0 w w4w N 44)f 4W O W 4 4) a) 4)

E44 4. 0 u. 46 46 6 64 4 46 'U 46 uNtw a i 4. r r_ r. 6) 6 0 C6 r 0 0

Lao 0 U U U O41.4 0 c. 4) 0 0 0 0 0 0 0 0

" 0 V 0 -44 U. 4 .0-. - .0 -0 4 -. mo .4 -I4J -41 4.4 41 4.4 4 4J 44 44 4 4

466 4 .. A 0

441

0 044 44

0 -4 - 0 %a M0N-(. In m N-

0 ) N- ND Nn Ln a) NMnM -W W NN N N N (N

0 04444

0

U4- V V V V V V 1 V V 0 10 V 10 VVVVVVa V V V a

w 40WC 0% 0 0% 0% 0N 00 7 % , M (1 0% N 0 m 00 0, (' m a

440- - , r- a c C, 0 0 00000 0 0) 04 00 0 L - C nL n I I 0 0W 0C 0 00 0 0 00 0 N c cc

4J

w -w 44 0% 0 %r 4r n w

0~~ v N N 0 w0 r- 4 %Q04 c, ~~ 0 ( ' N 4E 44 1. 0 1(4 1' (4 4 1.4 4(4 or o0 o nN 0 N N w cc0

W4 -WU C-4 N - N N (4 N4 Nr4 N NN N CN N- N4

). 46 46 06 0 0 0 N 0. 0-n 1

v .4 rU m vN NN N N M c N N N

ko 6 - 0 %0 0m 0 w 0o4.4 4 4 14 40 14 1" 1- M40

4N~~ 0 C; 04140 46 '6 4

0 o o a o o 0 0a 0 0 0 a U

0 41- 0 (4 a 0 0 0 84 PC -C -C 4 I-.

0% 0. m ON.q 6 ~ 6~r 0,0 044 6 000 0 0Ai) Ad4. 0 0 4 41 41 441 40 0 104 40 (' 410r-Ir-I 04I0- 1-Or-O

pi .W 11 4 . C.4 .

03

41 .0Q. 4 4 4

-60C4 00 40044444 0

0 4W

a uniform set of units was employed in the matrix for all sourcesof test data. The most important independent test variables aregrouped in the left-hand columns of the matrix, and the flarmnabil-ity results are provided at the extreme right. Implicit in ourmatrix is the working hypothesis that the relative fuel-to-air

*: velocity is fundamental to liquid breakup. Since both temporaland spatial conditions are non-uniform for many of the tests, wesought to define maximum relative speeds for each of the tests.Unfortunately, the maximum absolute velocity of ejection of theliquid fuel was not explicit in much of the data, and analyticalmethods were reported in others. The fuel velocity data presentedin Table 111-6 typically represent conservatively low peak fuelejection velocities in the absence of data to the contrary.

In the first case, the NAEC SP-2H Crash Tests, the analyticalmethods of reference 1 were applied. This particularly involvesusing the assumed coefficients of velocity and contraction, 0.985and 0.62, respectively. This provides a relatively high fuelejection velocity, since a sharp-edged orifice is assumed. It isnoted that with a rounded-edge orifice, the contraction coeffi-cient increases to 1.0--which yields no contraction from the 6" x

* 9" rectangular orifice--and hence a minimum fuel ejection veloc-ity for the same conditions.

In the case of the FAATC Wing Fuel Spillage Tests, a coefficientof 1.0 was assumed in the absence of data on fuel stream contrac-tion. Effort is underway to characterize the liquid fuel jet atthis facility which may resolve this question. For the MC WingSpillage Tests, a sharp-edqed orifice was utilized--so a contrac-tion coefficient of 0.61 was applied to the 6-inch round fuelorifice.

In the remainder of the tests of Table 111-6, coefficients of 1.0were applied in the absence of other data. It is important tonote, however, that in the case of the Southwest Research Insti-tute (SwRI) tests, the tangential speed of the spinning disc atits outer edge is utilized for fuel ejection velocity--i.e., theradial component of fuel ejection velocity is ignored. This is

4 consistent with SwRI's method, and computing +he radial componentwith viscoelastic fluids was not feasible.

The orifice shape effects are considered significant to breakup.The column provided gives the configuration of a single, typicalorifice. Where multiple orifices were used, the number is given

in the same column, or under "results."

574'

The fuel flow rate in the next column applies to the orifice de-fined in the prior column. This fuel flow column may seem redun-dant, since liquid ejection velocity and orifice configurationare given. However, it was noted in several of the tests thatflammability results differed significantly with higher versuslower "loadings" of the airstream by ejected fuel--under other-wise similar conditions.

Table 111-7 provides the most comprehensive data for comparingflammability results at nominal fuel temperatures of seven facil-ities--involving FM-9 AMF, particularly at 0.30 weight percentconcentration. The fuel-to-air velocities for the "pass" condi-tion for the SP-2H crash tests and the FAA Wing Spillage Testsare remarkably close. The progressively increasing speeds for a"pass" condition with increasing concentration are self-consis-tent and reasonable. This consistency is shared b, the NWC WingSpillage Tests, but with a very significant shift upwards in levelof FM-9 concentration for a similarly acceptable limit of flamma-bility performance. The 0.40% concentration in the NWC test hasa "pass limit" velocity 16 ft/sec less than the correspondinglimit on the FAA Wing Spillage facility for 0.30% concentration.At the same 0.40% concentration, the "fail limit" is 57 ft/sechigher on the FAA facility than for the NWC's smaller-scale WingSpillage facility. The two most likely possibilities are thatthe NWC test is more severe (as with a scale effect), or that thedifficulties with AIF sample quality, mixing, and/or handlinggave an effectively inferior performance at the NWC facility.The FAA's Flammability Comparison Test Apparatus in intended topreclude the latter ambiguity in the future.

The UK/RAE 2-rocket sled tests give a "pass-to-fail" thresholdvelocity for the 0.30% FM-9 concentration (246 ft/sec) that cor-responds to the median speed of the FAATC's Wing Spillage Testsfor the "marginal" condition and the same FM-9 concentration.The problem is that the "fail" condition for the "2-rocket" test-ing is at a fuel temperature only 61F greater than the nominal81F tested at FAA which gave a "pass" condition. Furthermore,

. testing at the FAA to 961F gave a "pass limit" of 235 ft/sec, adrop of only 3 ft/sec. This discrepancy is aggravated at higherconcentration--0.40% FM-9 fails in the UK/RAE 2-rocket test at

- 246 ft/sec and 971F, whereas the same concentration passes at theFAA Wing Spillage Facility at 322 ft/sec, a difference of 76ft/sec (31%) for a 17*F differential.

o I The UK/RAE 3-rocket sled tests show a "marginal" condition at 322" ft/sec down to 791F fuel temperature, whereas the FAATC Wing

Spillage tests show a "pass" condition at 322 ft/sec--both for

58

........--....- - - - - - - - - - -

LA 0' ( 0 (~1 0r- " m N U

LA

4-)

dP

- 41i

.o - - - -

41 ) LA) L, Ni r- OD0(a ;:m D - rN 0OCN

*4 C4 N (Ni C4 r41. C; V A

Q)41i

0 0a -

m' cn 0 vL LA 0~Q) 0 V 11k C'4 OD mo (No0 N (4N) ('4 e -I(' CN C14 : C o k r-C mrrN

rL 0; oV A AlI v I - Ao-1U.-

N - D - - 0 -4

rNJ mN OV4 (N

C; CN * V A

m 00 cQ) (N a 0 q~U) 14 4N C14 (N

44

1'-

V 0) r-4 U) C r t ) -4 U) 0) -4 u] 0)(D (n a)- -4~W~ 1) $4 -H u) 4-4- U) 54 *q(d CO 0 (a cOm OS m0 itO (13 (0 (aOS o (

U)

0a1

0- 4) ) C. 40 )D .4 Lo U)) U) 41) X)

cn ) A40 u uUW t3) XC) f)u

O'- Un) 00Z 00aa4 E-1 3: t C

U)) ftu 4 3:U) U))

U) E 44

44_

59

In

r. 0

UU

00

.4-)

r. w

0) 410di m :0 D r -0r

o 00C

CuN

4J9

U) 4-) 0- 0) a) 0- ) 0U) $4(~ -HLr U)N HU 4- j 4-

o~3 C9*LAf3'O ')4 0

a)

>4 U)

4.) 0) $

00

C.) - 41 f - (l 0 H' 3r.

1-4 41r- J

(n - (a E-4--

0 04 rd A r

U) (a~ -1 r4 )44c 1

Cu C C Cu-4-o

E.4 - H 0 - 4 -4H4U) - - U '- ) ' - mu)

*~~~a U) 4 0.*- )- ) H

54)

04

Ll- )

the same 0.40% FM-9 concentration. Relevant temperature versusvelocity curves are provided by Figure 111-9.

It is interesting to note that the temperature sensitivity shown* by the UK/RAE testing works in the other direction, as well. An

FM-9 concentration of 0.30 is sufficient for a "pass" result at322 ft/sec at 68°F per the UK/RAE 3-rocket sled tests, whereaschilling the same concentration fuel to 470 F at the FAA WingSpillage facility gives a "pass limit" speed of 225 ft/sec--adiscrepancy of nearly 100 ft/sec (43%). In fact, chilling the0.30% fuel from 81°F to 470 F at the FAA facility only increasedthe "pass limit" speed by 15 ft/sec. Comparing slopes in Figure111-9, a 3.3°F fuel temperature rise in the UK tests and a 350Frise in the FAA tests both give a "pass limit" velocity reductionof 20 ft/sec.

It is apparent that the FAA Wing Spillage and UK/RAE 2-rocketsled tests agree at 0.30% FM-9 concentration and about 80°F fueltemperature--but diverge very significantly as test fuel tempera-ture either increases or decreases. It is possible that the verydifferent natures of the heating and ignition sources of the twofacilities may account for much of the discrepancy. The UK/RAEfacility applies much more heating to a mist configuration whichrapidly decelerates in the region of the ignition sources, whereasthe FAATC Wing Spillage Facility provides transient heating by awake flame in a relatively high velocity flow field involving asmall portion of the total fuel/air plume.

For the JPL Mini-Wing Spillage Tests, data usable for this compar-ison were available from an FAA report (reference 11). JPL testdata were presented (reference 5) in a scaled format which didnot provide all the necessary parameters of Table 111-7. Themaximum speed presented, 284 ft/sec, is not necessarily a "passlimit", since no failure velocities were available. However, thereported maximum wake flame length of 3.1 ft. is near a limitcited elsewhere1 2 of 1 meter (3.28 ft). Comparing the highestspeed, 284 ft/sec, to the FAATC Wing Spillage Facility's "passlimit" of 238 ft/sec gives an apparent overstatement of 46 ft/sec(19%), using the latter as a standard. This difference may in-crease, depending upon how much additional velocity is necessarywith the JPL Mini-Wing to get a 3.3 ft long wake flame. Scaleeffects may account for much of this discrepancy.

The FAATC's Flammability Comparison Test apparatus (FCTA) util-izes maximum heat output as measured by a radiometer--comparedwith the flame size/speed evaluation schemes of the other facil-ities. As detailed in reference 13, "pass," "marginal," and

61

uj 0

'0 >

w >

U, U) H

0- w1 uw (nz% Loc H

L.. V). I 110> Z/7>-FMt4

OLA.

C) E-I U_0 . ::

C0 U

-E-H

*j

0 0 0 0 0

0 '0 CD C)

62

"fail" limits have been defined in terms of heat output--separ-ated by values of 0.6 and 1.0 BTU/ft2 sec. The 0.6 "pass" to"marginal" threshold corresponds to propagation upstream and intothe FCTA's nozzle as observed with a particular batch ("A") ofAMF.

Examination of the three sets of curves provided shows that.. within the range of velocities tested, about 150 to 250 ft/sec,

each fuel typically reaches a peak value of heat output at agiven flow rate and then diminishes with increasing velocity.This is particularly significant where, for a given fuel flowrate, heat output achieves "marginal" and even "fail" values,then falls back at increasing air velocity to "pass" values ofheat output.

It is then possible to interpret very high values of absolute airvelocity (and relative fuel-to-air velocity) as acceptable condi-tions. The entries in Tables 111-6 and 111-7 for the FCTA illus-trate the problem. A "fail" velocity of 243 ft/sec was encoun-tered at .0042 gal/sec (16 ml/sec) flow, whereas a "marginal"speed of 259 ft/sec was found for the same fuel flow rate. A"pass" velocity at this same flow rate was established at 148ft/sec.

Examination of Table 111-7 shows that the relative fuel-to-airvelocities for the "pass" condition vary significantly with flowrate. For the same fuel, 0.30% FM-9 (Batch A), the "pass limit"velocity at lower flow is 257 ft/sec, whereas at somewhat higherflow it is 163 ft/sec--a difference of 94 ft/sec (58%) based onthe higher flow.

In comparison with the FAATC Wing Spillage Facility, the lowerflow range's results are much more desirable--the "pass limit"exceeds that of the Wing Spillage Facility's by only 19 ft/sec.

It seems quite reasonable that the heat output criterion of theFCTA can be related to the criteria of other AMF flammabilitytest facilities--particularly the FAA's Wing Spillage Facility.

4 For example, the drop in heat output for speeds past the failurelimit may be reasonably attributed to the eometry of the flame,nozzle shielding of the flame from the radiometer, and/or thelimited field of view of the radiometer. Past the failure point("propagation into nozzle"), it is likely that a relativelystable flame front forms within the diffuser at the locationwhere the flame speed equals the flow stream's velocity.Increasing the air velocity may increase the intensity of burningwithin the flame front. With more complete combustion inside the

diffuser, less radiation would be sensed outside by the

~63

* radiometer. This may explain why the heat flux diminishes withincreasing air velocity (at the same fuel flow) after the "fail"condition is reached. Droplet characterization may show that thehigher fuel/air velocities give a more finely divided fuel,eliminating any ambiguity that "pass" conditions are encountered

* at higher velocities than "fail" conditions.

Finally, the quantity of fuel that exits the FCTA's diffuser asflowing liquid after the run may require measurement and appro-priate corrections--since variations in that parameter mean dif-ferences in fuel spray available to burn, and may affect totalheat output in an otherwise unknown way.

The SwRI Spinning Disc is physically the most different from theidealized contra-stream flows of fuel and air. However, theupper range of its pass/fail threshold is only 10 ft/sec lessthan the corresponding value for the FAA Wing Spillage Tests.Figure 111-9 shows liquid fuel temperature sensitivity, in compar-ison with other facilities. Note thz.t the slope of the SpinningDisc curve is quite similar to that of the FAA Wing Spillagefacility. The lower velocity values of the Spinning Disc (about40 ft/sec) compared to the Wing Spillage data should be due inpart to the fact that the radial velocity of fuel ejection fromthe disc is not accounted for.

The literature (references 11 and 13) documents a discrepancybetween flammability results of the SwRI Spinning Disc and theJPL Mini-Wing Facility. The Spinning Disc testing indicates thatamine is not necessary in the carrier fluid, whereas the JPLtests indicate the opposite. We could not resolve the differenceusing existing data, but characterizations of test facility condi-tions and droplet spectra should do much to resolve such dif-ferences.

ENHANCING CORRELATIONS.

Considering AMF flammability tests as a group, there are uncer-tainties associated with both independent test variables (e.g.,relative fuel-to-air velocity) and the flammability-relatedresults (dependent test variables). Particular].y for small-scale

* tests, some scaling of fuel-to-air relative velocity may be.. essential to properly replicate full-scale crash conditions.

This is difficult considering the number, complexity, and inter-' dependence of aerodynamic, liquid-breakup, heating, ignition, and* flame propagation phenomena. The varying pass/fail characteriza-

tions of the combustion phenomena of the several test facilitiesdo not always lend themselves to direct, quantitativecomparisons.

64

* . -r* . . . .,.*

It is essential that another set of quantitative dependent varia-bles be determined for those AMF flammability test facilitipswhich are to be utilized in the future. The additionalcharacterizations should consist of fuel droplet size spectr andnumber densities where ignition and flame propagation occur. Forengineering analysis, compact droplets may be assumed spherical,with appropriate empirical correction for higher surfacearea-to-volume ratios.

Such droplet characterizations would provide hard data with whichto correlate the results of liquid breakup models, and provideessential input to ignition and combustion models.

Of course, the JPL Mini-Wing Shear tests are incorporating highlyadvanced methods to perform fundamental studies of all stages ofliquid breakup. Much more rudimentary "off-the-shelf" dropletdiagnostics are available which could provide the minimum set ofdroplet data necessary for directly comparing two or more AMFflammability facilities. The two most evidently compatiblecandidates are the FAA's large-scale Wing Fuel Spillage facilityand their Flammability Comparison Test Apparatus (FCTA). Bothhave established a significant data base on neat Jet A andcertain AMF's. The FCTA has the additional advantage ofstandardization and deployment to several R&D organizations.They are readily amenable to "cold" tests (i.e., without ignitionsources) which can be in other respects identical to flammabilitytests already conducted or to tests specially performed to con-firm repeatability.

The JPL's fundamental. research should provide, as a subset of itsdata output, droplet size spectra and number density data fullycompatible with data from the aforementioned facilities. The RAEand SwRI test rigs are also amenable to droplet characterizationsand subsequent across-the-board comparisons. "Cold" tests (i.e.,no ignition) do not appear feasible for the NAEC full-scale SP-2HCrash Tests. However, the sirilarity in scale and flammabilitycorrelations with the FAATC Wing Fuel Spillage facility pre-viously discussed relax the need for droplet data from the NAECcrash tests.

As an essential complement to the droplet data acquired at facil-ities of interest, the semi-empirical liquid breakup model shouldbe extended to include (1) Taylor breakup efffcts with large jets,and (2) additional AMF fuels. Such a model, qiven adequate align-ment and verification with AMF tests' independent variables anddroplet spectra, would allow uniquely confident interpolation andextrapolation of data, regardless of the source facility. Thecost of such an effort should be more than compensated for by

65

reducing the need for repeated testing (particularly at largescale), and vastly increasing understanding and confidence in AMFperformance.

AERODYNAMICS. Further aerodynamic characterization appears par-ticularly appropriate for the various Wing Fuel Spillage rigsthat employ free air jets. It is evident that the pass/fail cri-teria of the several air jet facilities apply to various jetregions. Velocity and turbulence intensity data are desirable inthe regions of the ignition source(s) and typical propagation.Anemometers with this capability are commonly available. Suchdata are important for assessing liquid breakup and flammabilityof the resulting mist.

FUEL JETS. Determinations of the contraction coefficients re-lated to the jets of ejected fuel should be made for all AIF flam-mability test facilities utilized in the future.

DROPLET DIAGNOSTICS.

Recent emphasis on atmospheric studies has led to the developmentof particle measuring diagnostics which may prove far more cost-effective than the older, more rudimentary and labor-intensivetechniques. The most attractive systems use lasers for directimaging, or a particle sizing interferometric method for the off-axis detection of scattered light (Particle measuring Systems,Boulder, Colorado; and Spectron Development Laboratories, Inc,Costa Mesa, California, respectively). The latter method pro-vides droplet velocity and turbulence intensity data in additionto droplet sizes and number densities--a desirable but not indis-pensible capability. Diagnostic systems can be purchased whichgreatly facilitate data reduction, at large potential savings inmanpower. Equipment can be purchased outright, leased, or pro-vided for a specific time period fully manned with an experiencedcrew. The latter may be particularly attractive for "one-shot"droplet characterizations of a given AIF flammability test facil-ity.

Laser light detection and ranging (LIDAR) methods of the ComputerGenetics Corporation (CGC) were also evaluated for application toaircraft crash tests and large-scale AMF flammability tests. CGChas applied Brillouin, Rayleigh, Mie, fluorescent and Raman scat-tering techniques to atmospheric, hydrographic, ane pollutantpcrameter measurements. A typical use is to remotely monitormunicipal power plant stack emissions.

66

Based on our specific inquiries, acquiring droplet size distribu-tion data, particularly in the densities encountered with AIIFtesting, is not fully developed and operational. with LIDAR sys-tems at the present time.

6

I. -"

REFERENCES

S-1. Anonymous, Full Scale Aircraft Crash Tests of Anti-MistingKerosene (Interim Report), Naval Air Engineering Center, Report

No. NAEC-TR-183, 4 August 1980.

2. Klueg, E. P., Large-Scale Aircraft Crash Test of AntimistingFuel, Federal Aviation Administration Conference Proceedings,"Aircraft Research and Technology for Antimisting Kerosene",

February 18-19, 1981, Report No. FAA-CT-81-181, June 1981.

3. Miller, R. E., Wilford, S. P., Simulated Crash Fire Tests Asa Means of Rating Aircraft Safety Fuels, Royal Aircraft Establish-ment, Ministry of Defence (Aviation Supply), Technical Report

Number 71130, May 1971.

4. Miller, R. E., Wilford, S. P., The Design and Development ofthe Minitrack Test for Safety Fuel Assessment, Royal Aircraft Es-tablishment, Ministry of Defence, Technical Report No. 71222,November 1971.

5. Sarohia, V., Fundamental Studies of Antimisting Fuels, JetPropulsion Laboratory, California Institute of Technology, Pasadena,California, AIAA/SAE/ASME 17th Joint Propulsion Conference,Colorado Springs, Colorado, July 27-29, 1981.

6. Ferrara, A., Cavage, W., Flammability Comparison Test Appa-ratus Operator's Manual, Federal Aviation Administration TechnicalCenter, Atlantic City, N.J. (Draft Manual).

7. Eklund, T. I., Cox, J. C., Flame Propagation Through Spraysof Antimisting Fuels, Federal Aviation Administration, NAFECTechnical Letter Report NA-78-66-LR, November 1978.

8. Eklund, T. I., Neese, W. E., Design of An Apparatus forTesting the Flammability of Fuel Sprays, Federal Aviation Admini-stration, Systems Research and Development Service, Report No.FAA-RD-78-54, May 1978.

9. Eklund, T. I., Experimental Scaling of Modified Fuel Breakup,Federal Aviation Administration, Systems Research and DevelopmentService, Report No. FAA-RD-77-114, August 1977.

10. Finn, S. V. Jr., Eklund, T. I., Neese, W. E., PhotographicInvestigation of Modified Fuel Breakup and Ignition, FederalAviation Administration, Systems Research and Development Service,Report No. FAA-RD-76-109, September 1976.

68

11. Webster, Harry, FM-9 Anti-Misting Fuel Large Scale Crash Test#3, Federal Aviation Administration, Summary of Presentation, 6thMeeting - US/UK Technical Committee on Anti-Misting Fuel, March 24,1980.

12. Mannheimer, R. J., Degradation and Characterization of Anti-misting Kerosene (AMK), Federal Aviation Administration Report No.FAA-CT-81-153 , Southwest Research Institute, San Antonio, Texas,Interim Report (Draft), September 1980.

13. Ferrara, A., Flammability Comnarison Test Apparattus,Federal Aviation Administration Technical Center, Atlantic City,N.J. (Draft Report).-

".69

IV. ANTIMISTING FUELS RHEOLOGY

The broad objectives in examining the rheological characteristics

of antimisting fuels under this program are to assess existinq

. techniques and characterizations of fluid properties relative to*

* flow (through fuel lines, from crash formed orifices,

etc.), and* breakup associated with crash release.

These features, of course, are in support of an overall program

objective of correlating rheological and flammability test data.

In reviewing existing work on fuel rheology, particular attentionwas given to what is measured versus what is needed to characterizefluid behavior in the indicated situations. These features wereexplored with full recognition that no known single measurementtechnique is apt to be appropriate to yield all the rheologicalparameters of interest, and that it is not practical to specify a

set of parameters appropriate to achieve full generality across theundefined range of rheological behavior of possible antimistir-gfuels. It was considered advantageous, therefore, to concentrateon those rheological features which have been observed for currentor past candidate fuels, and on features indicated as potentiallyimportant to liquid breakup.

Accordingly, our rationale for technique examination has involvedtwo considerations--near term and long term. For near term, moredetailed fluid properties are desirable to better quantify flow and

breakup process performance. For the long term, selected fluidproperties are desired which correlate with overall antimisting

fuel performance and therefore can be used for quaiity control of

the fuel.

The general types of liquid properties which appear to be appropriatefor near term needs include: the shear and elongation (tensile)

stress-strain rate flow curves, the corresponding viscosities, dis-

continuities or regions of rapid change of slope in the flow curves,appropriate measures of fluid elasticity, time dependent effects

*ArtIIt ifi t it I i t en W i i I x I,, tW )w w ii t) v TTI~ il Wi t 1 i -

Sita- ive rtmarks rt,,I iv , t t ',intit , 'ra ' l it mt weIe J(i'(,1 l i t.' '

• ittkle is , parent ly kiw i , lo this t- pi, thi-t m,.nin,; iil Ji:;,-is:-iot , i

here is not possible,.

70

(such as thixotropy), gelation-like effects, strain-rate relaxa-tion times, and surface tension.

Parameters suitable for the long term quality control type needsare not well known, in general, but can likely be established forspecific modified fuels which represent viable candidates. Forexample, an FM-9 fuel might be rheologically characterized by simpletests which confirm that the additive concentrations are withincertain prescribed limits, the lower and upper critical shear ratevalues fall within suitable ranges, the shear and tensile viscosi-ties are suitableto give simple breakup and flow performanceswithin selected limits, and the intentional degradation can beachieved with a selected system by applying an amount of mechanicalpower in a given range.

In view of the rheological uncertainties involved in the flow andbreakup of generic types of antimisting fuels, a pragmatic approachto their rheological characterization is to emphasize techniqueswhich actually involve selected features of fuel flow and breakupunder conditions similar to those expected to be encountered invarious flammability tests. This will permit key behavioral fea-tures to be observed and accounted for, if not fully analyzed.

For flow, the techniques which involve the following are believedto be the most relevant:

- rotational rheometers* capillary rheometers (flow-through tubes)* filter/screen/porous bed-type schemes. open siphon/pendant drop-type schemes.

A number of rotational type instruments, including some which arecommercially available, have been reviewed by Van Wazer, et. al., 1

Walters, 2 and Gaskins and Philippoff. 3

Of this group the rotating disc type unit represented by the BrookfieldModel LV Synchro-Lectric Viscometer with LV adapter represents aconvenient device for low-shear viscosities. This relatively in-expensive unit, indicated as a constant temperature viscosity schemeto be examined under this program, is a commonly used and viableinstrument for determining low shear viscosities of antimistinqfuels. Despite its useful... ss in this role, the overall utilityof the Brookfield unit is somewhat limited in its ability to obtaina wide range of shear rates for a narrow viscosity range.

The group of rotating cone and plate rheometers, such as the WeissenberRheogoniometer,l the Mechanical Spectrometer (Rheometrics) and theRotary Rheomete- (Instron)2, have been used to characterize

71

non-Newtonian liquids with varying degrees of success. These units

are reported to suffer from problems with secondary flow and ejec-

tion of material from the wedge region for certain antimistinqfuels (especially FM-9 fuel). As a result, these units have some-what limited ability to characterize the shear behavior of anti-misting fuels, but do represent viable candidates for this purposein low shear rate regions where such effects are unimportant.

The third group of rotating viscometers is represented by thecoaxial cylinder units such as the Rotovisco (Haake).l While thisunit has less inherent capability for assessing viscometric normalstress effects, it has been used with a certain amount of successin characterizing viscoelastic antimisting fuels. 4 It has alsobeen used previously 5 with other viscoelastic materials and, togetherwith an ELZ Visco-Elastic Attachment (Haake),3 has been used formeasuring the elastic recoverable shear strain.* The proceduresfor reducing these couette flow measurements have been given in somedetail by Krieger and Associates. 6 - 8

The capillary viscometers have also been reviewed by Van Wazer, et.al.,1 Walters, and Gaskins and Philippoff. 3 A number of instrumentsof this type are available commercially for gravity or low pressureinduced flow. The high pressure units which have been developed incertain laboratories are essentially one-of-a-kind units. 3 , 21 Theoperation of this type of instrument is based on flow through tubesand a derivation of the associated fluid analysis and instrumental

*. parameters is given in the literature. 1 , 2, 8, 12 Three primaryrheological parameters can be determined with this type of instrument:

*The use of finite elastic strain concepts to describe fluid response was

proposed in several early theories, 9 - 1 5 following Reiner and Freudenthal'sl(

development of a dynamical theory of strength based on a maximum stored distor-

tional energy. Their approach depends on the idea that the energy used in

deforminq a viscoelastic liquid is in part stored (as elastic or recoverable

* strain energy) and in part dissipated (in viscous flow). This scheme introduces

a rate dependence, since if enerqy is input so slowly that it can aII be dissi-

pated, there can be no accumulation of energy to cause rupture. Although it is

recoqnized that such an approach has sometimes been viewed as intiquatc,

Truesdel ]17 has related it to secnd order fluids, and Petric 1 8 notes a rvivlIof interest in the idea. The recent- experimental results from Vinel e ,dov id

. co-workersl9, 20 show that the u;tite ,f ittairs i ; toit s( Simlple, but the Pe. iner

ido,i ()ltrs a use lil first af,1roimatiOll. It i!; on jec t (tl t hit t l.11 . poL i1 may merit some furt-hecr c(nsiderai .H .is i semi-empirica I scheme tor Ise wit],

antimisting futls, because s its rlative simplicity and physicl aptel. Such

activity should take into ac:count t.co,,o rabl e(biiati n straitl as well

re,- ,'(,,v rabl shear strijn.

72

4 ,shear stress, shear rate, and recoverable shear strain. The highpressure capillary viscometers have been successfully used tocharacterize a number of viscoelastic materials at high shear rates(in the range of 104 to 106 sec- 1 ,5 but such units have apparentlybeen sparingly used with antimistinq fuels. In part this has stemmedfrom the dilatant/gelling behavior of FM-9 fuel which tends to pluqthe capillary tubes.*

A Cannon-Ubbelohde four bulb shear dilution viscometer has been usedsuccessfully by Peng 22 to obtain zero shear viscosities.

The filter, screen and porous bed-type measurement schemes have beenadvanced as simple, field-type rheological tests, largely in connec-tion with intentional fuel degradation work. Much of this workoriginated at the Royal Aircraft Establishment (RAE) and includedstudies using filters made of stainless steel meshes, MilliporePTFE membranes and cured aircraft filter papers. 4 The aim of thisRAE work was to characterize as far as possible the filtration be-havior of FM-9 fuel over a range of levels of degradation. Theyfound, as U.S. workers have subsequently, that in many of the experi-ments a build-up of gel (from FM-9 fuel) occurred on the downstreamside of the filter, and this caused variations of flow rate forpassage of successive batches of fuel. Intermittent dislodgementof the gel also resulted in erratic filtration rates. For single,short-duration runs with a new or cleaned filter, the results areapparently rather consistent. Porous bed tests have been pursuedby Mannheimer. 2 3 , 24

Although there is some attractiveness of such simple tests in theirpossible future role in quality control of antimisting fuels, thereis little justification for extending their results to other situa-tions or configurations where the rheological behavior may havelittle relationship to flow through a filter, screen or porous bed.From a more fundamental point of view, these types of schemes yieldresults which represent a confounding of both shear and elongationalflow phenomena. As long as the results desired pertain to flowthrough such a filter or bed, the results actually achieved may bedirectly interpretable. However, accountinq for the results in

*This shear-indiwi(ed qo1 itinin [henomonrn -hserved with PM-9 1-t(l is ntod

in the, dis(.'1S si,1, t I liquiid brejki ij ' l:(,wh r(. in thi; r ,', t. Wh. n ,il , Vl-tion in;

-xt,,nsive mt,('h,)ni i c' I .t ioti, th 0 1,htetimen, n is e te tii ly - ,'Yible, 1. . ,

•uip,n cessation -t sttaininq the q,I r,-laxs -n rI'(lins its low 'isc')sitv.

natinri. Near miferi, l boijri.'ty ?s, howev,!r, their, h s been ' so m, in-lic'ation tl 't

an iddifionil addit iiv, or solvent m.ay ho nCeel,-t V, 'wort .ni htlild',1'.

73

terms of separate shear and elongational flow features representsa questionable procedure at best, and some alternate technique orconfiguration that could qive separately interpretable results isgenerally preferred.

In this connection another simple test noted as the cup viscositytest, was also indicated as a scheme to be examined under this pro-gram. The cup-type viscometers, of which there are several typesincluding the Demmler, Ford, Zahn and Gardner one-shot cups, are anumber of classes of short-tube or orifice viscometers. The commentsby Van Wazer, et. al. 1 relative to these devices are most appropriate.

"The original design concept of these viscometers were derivedfrom the Hagen-Poiseuille law which states that the efflux

" time of a fixed volume through a capillary is proportional tothe viscosity of the fluid. Unfortunately, the viscometersthat were developed consisted of short capillaries or orifices.Flow in these viscometers does not obey the Hagen-Poiseui.llelaw and efflux times are not in any simple relation to theviscosities...

This type of viscometer, however, is widely used. The popu-larity is due to simplicity and inexpensive operation. Insome industrial fields, this type of measurement has beenapplied to correlate a sample property in terms of the resultsobtained. However, this is true only when variation of thefluid property from sample to sample is small. For a highlynon-Newtonian fluid, these viscometers will definitely fail."

* The open siphon and pendant drop type schemes noted as flow measure-ment techniques were employed by Peng and Landel, 2 5 and Sarohia andLande1 2 6 in connection with evaluating the elcngational viscositiesof antimisting fuels. The basic geometrical configuration for theseschemes (as well as for filtration) are shown schematically in FigureIV-l. These two test techniques are somewhat unique as schemes fordirect evaluation of tensile viscosities, although a third schemebased on converging flow (also noted in Figure IV-l) has also been usedby Metzner and Metzner. 2 7 The latter is not known to have been usedfor antimisting fuels. Each of these have limitations: all givenon-steady state results, tend to be laboratory-type schemes, andrequire some degree of fastidiousness to obtain useful results. On

* the other hand, they probably represent the best techniques availableat present for tensile viscosity determinations of antimistinq fuels.

* Attention will now be turned briefly to rheoloqical techniques andmeasurements which are more closely associated with fuel breakup.As might be expected, information in this area is rather sparse.

The reasons, of course, are obvious since breakup represents a

74

PENDANT DROP FILTRATION/SCREEN/POROUS SOLID

OPEN-SIPHON CONVERGING FLOW(FANO FLOW)

FIGURE IV-1. VISCOMETER/RHEOMETER SCHIEMES (ELONGATIONAL FLOW)

75

discontinuity in flow, and the fluid measurements and techniquesnoted apply basically to flow prior to (or during various intermedi-ate stages of) breakup. Correlations of fluid properties withbreakup are about the best that can be expected at present, andthis topic has been discussed at length throughout the liquidbreakup section of this report. It is perhaps useful to highlight

-some of the results indicated, suggested or implied from thatsection.

The current investigation of antimisting fuels breakup has givensome indication of the rheological features apt to be importantin the breakup process. The tensile viscosity and/or elasticityassociated with an elongational strain rate is of special importin the formation and breakup (or not) of liquid filaments andthreads. Shear viscosity plays a minor role in this action,unless this viscosity is quite large.

On the other hand, shear viscosity appears to play a dominant rolein instability (wave) growth and ligament shedding. The contri-bution of tensile viscosity in this case is believed to be minor.

The formation of gel-like material for particular strain rates,such as with FM-9 fuel, may have a profound influence on breakup.The reversibility of this gel-like nature may be of far greaterimport to the flow through pipes, pumps and fuel filters than to

* - liquid breakup, since the fluid is not expected to relax before*.o- breakup occurs.

These observations seem to suggest that in liquid breakup, aprinciple which may be loosely described as a "principle of leastresistance," seems to govern what happens in response to a given

* driving force. This view also seems to be correlated to someextent with the rheological properties of the antimisting fuel.In particular, whichever viscosity--shear or tensile--is thelesser will determine what happens at that stage of the breakupprocess, provided the configuration permits (or causes) a flowgoverned principally by that parameter to occur.

* The closely related situation which may be envisioned as occur-ring frequently in liquid breakup consists of simultaneous shearand elongational flow. The governing parameters in this case arenot well known. This type of motion borders on the hypothesizedactions of drag reducing fluids in suppressing turbulence in a

* liquid. A laminar counterpart to this turbulent flow action in-volves combined lonqitudinal flow and secondary flow in curvedtubes. The effects of polymer additives in this combined flow casehave been examined by Tsang and James. 2 8

76

Measurement techniques specifically configured to characterize thefluid properties relevant to breakup do not currently exist, andfrom a fundamental standpoint may be unnecessary. For purposesof quality control, however, it is probable that one or moresimplified type field tests which emphasize the breakup features

*' of antimisting fuels should be devised.

One relevant technique which appears to be a test to measure theonset of shear induced dilatancy (or gelation in FM-9 fuel) isthe so-called "Inertial Rheometer" devised by Gustavino and co-workers. 29 ,3 0 This scheme has rather surprising sensitivity,but needs further characterization in terms of particular experi-mental variables, pump pressures and flow parameters. It maybe that the secondary flow phenomena explored by Tsang and James2 8

may cause some elongational flow effects when coiled tubes areemployed, but the bulk of the phenomena are believed to be inducedby shear flow.

Another type scheme which is merely suggested at this point, asit has not been explored experimentally under the present effort,is a scheme based on breakup, but intended to serve as a routinecheck (or characterization) of elongational (and perhaps dila-tional) flow. This technique is basically a pendant drop typeconfiguration, with supplementary laminar airflow to cause dropseparation and filament rupture (or not), if necessary. Thepoint is to explore the diameter changes of droplets, or dropletson the threads which are produced for a variation of elongationalviscosity. If the droplet diameter change is sufficiently sensi-tive to changes in elongational viscosity, then with proper cali-bration, a measurement of droplet size can be used to indicateelongational viscosity of the fuel. Calibration of this schememay involve measurements similar to those made earlier by Sarohiaand Landel.

2 6

Finally, it should be noted that one or more particular rheometrictechniques cannot be recommended to the exclusion of others, sinceeach m.-y have a range over which the values it indicates areperfectly valid. However, it does appear that the potentialcapabilities of rheometers such as the Rotovisco (with elasticrecoil attachment) and the high pressure capillary units have notbeen fully exploited in antimisting fuels evaluations.

77

REFERENCES

1. J. R. Van Wazer, J. W. Lyons, K. Y. Kim, and R. E. Colwell,Viscosity and Flow Measurement, Interscience, New York (1963).

2. K. Walters, Rheometry, Chapman and Hall, London (1975).

3. F. H. Gaskins and W. Philippoff, "Instrumentation for theRheological Investigation of Viscoelastic Materials," J. ppl.Polymer Sci., 2(5), Pages 143-154 (1959).

4. J. K. Knight, R. E. Miller, and S. P. Wilford, Rheology ofAntimisting Safety Fuels, RAE Memorandum (19 June 1978).

5. W. H. Andersen, N. A. Louis, and G. Ialongo, Investigation- -of the Aerodynamic Breakup of Viscoelastic Liquids. Phase I--

Subsonic Dissemination, Contractor Report ARCSL-CR-77037 (August1977).

* 6. I. M. Krieger and S. H. Maron, "Direct Determination of theFlow Curves of Non-Newtonian Fluids," J. Appl. Phys., 23(9),Pages 147-149 (1952).

7. I. M. Krieger and H. Elrod, "Direct Determination of theFlow Curves of Non-Newtonian Fluids. II. Shearing Rate in theConcentric Cylinder Viscometer," J. Appl. Phys., 24(2), Pages134-136 (1953).

8. I. M. Krieger and S. H. Maron, "Direct Determination of theFlow Curves of Non-Newtonian Fluids. III. Standardized Treatment

of Viscometric Data," J. Appl. Phys., 25(1), Pages 72-75 (1954).

9. M. Mooney, "Secondary Stresses in Viscoelastic Flow,"J. Colloid Sci., 6, Pages 96-107 (1951).

*.•. 10. M. Mooney, "A Test of the Theory of Secondary Viscoelastic

Stress," J. Apl Phys. 24(6), Pages 675-678 (1953).

11. W. Philippoff, F. H. Gaskins and J. G. Brodnyan, "Flow

Birefringence and Stress. V. Correlation of Recoverable ShearStrains with Other Rheological Properties of Polymer Solutions,"J. Appl. Ph s., 28(10), Pages 1118-1123 (1957).

* 12. W. Philippoff and F. Ii. Gaskins, "The Capillary Experimentin Rheology," Trans Soc. Rheol., 2, Pages 263-284 (1958).

78

13. T. Kotaka, M. Kurata, and M. Tamura, "Normal Stress Effect inPolymer Solutions," J. A]ppl Phl.s., 30(11), Pages 1705-1712 (1959).

14. A. S. Lodge, Elastic Liquids, Academic Press, New York (1964)

15. R. B. Bird, H. R. Warner, Jr., and D. C. Evans, "KineticTheory and Rheology of Dumbell Suspensions with Brownian Motion,"Adv._Polymer Sci., 8, Pages 1-90 (1971).

16. M. Reiner and A. Freudenthal, "Failure of a Material ShowingCreep (A Dynamical Theory of Strength)," Proc. 5th Int. Cong.

Appi. Mech. Pages 228-233 (1938).

17. C. Truesdell, "Fluids of the Second Grade Regarded as Fluidsof Convected Elasticity," Phys. Fluids, 8(11), Pages 1936-1938(1965).

18. C. J. S. Petrie, Elongational Flows, Pittman, London (1979).

19. G. V. Vinogradov, A. Ya Malkin, and V. V. Volosevitch, "SomeFundamental Problems in Viscoelastic Behavior of Polymers in Shearand Extension," Appl. Polymer Symp., 27, Pages 47-59 (1977).

20. C. V. Vinogradov, "Ultimate Regimes of Deformation of LinearFlexible Chain Fluid Polymers," Polymer, 18, Pages 1275-1285(1977).

21. F. H. Gaskins and W. Philippoff, "Instrumentation for theRheological Investigation of Viscoelastic Materials," J. Appl.

" Polymer Sci., 2(5), Pages 143-154 (1959).

22. S. T. J. Peng, Rheological Behaviors of AMK, Jet Propulsion

Laboratory, Presentation, March 24-26, 1980.

23. R. J. Mannheimer, Rheology Study of Antimist Fuels, ReportNo. FAA-RD-77-10, Southwest Research Institute for FAA SystemsResearch and Development Service (January 1977).

24. R. J. Mannheimer, Influence of Surfactants on the FuelHandlin9o and Fire Safety Properties of AntimistinKerosene (AMK),Southwest Research Institute for FAA Systems Research and Develop-ment Service (February 1979) (FAA-RD-79-62)

25. S. T. J. Peng and R. F. Landel, "Preliminary Investiqationof Elongational Flow of Dilute Polymer Solutions," J. Appl. Phys.,47(10), Pages 4255-4259 (1976); and Errata, J._Appl. Phj s., 50(7),Page 5883 (1979).

79

26. V. Sarohia and R. F. Landel, Influence of Antimistin_ Polymer

in Aviation Fuel Breakup, Jet Propulsion Laboratory, AIAA/SAE/ASME,16th Joint Propulsion Conference, June 30-July 2, 1980.

27. A. B. Metzner, and A. P. Metzner, "Stress Levels in RapidExtensional Flows of Polymeric Fluids," Rheoloqica Acta, 9(2),

Pages 174-181 (1970).

28. H. Y. Tsang and D. F. James, "Reduction of Secondary Motionin Curved Tubes by Ploymer Additives," J. ofRheology, 24(5),Pages 589-601 (1980).

29. T. M. Gustavion, P. Cox and W. Cavage, A Status Report onthe Investigation and Development of a Laboratory Test to Character-ize Antimisting Behavior in Polymer Modified Jet A Fuel (AMK)Antimistinq Kerosene, FAA Technical Center Memo to W. T. West-

* .field, May 18, 1980.

30. T. M. Gustavino, Measurements of the Rheology of PolymerModified Aviation Kerosene, FAA Technical Center Memo, June 10, 1980

80

V. LIQUID BREAKUP

RATIONALE.

The introduction of antimisting fuel as a means of reducing fuel

fires in aircraft crashes is in recognition of the fact that coarsedroplet fuel sprays represent a smaller fire hazard than finesprays.

The flow and deformation properties of a fuel are known to influ-ence how the fuel breaks up and what the resultant sizes of spraydroplets are in a crash environment as well as in an aircraftengine. Liquid breakup, however, is the process which actuallydetermines the spray droplet sizes. The key concerns in consider-ing fuel breakup in actual or simulated crash test configurations,therefore, are to:

. Establish the nature and final size distributions of drops(or other shapes) in relevant neat and antimisting fuelsprays.

* Identify mechanisms expected to be important to the breakupprocess over the range of small- to large-scale laboratorytests and in full-scale crash scenarios.

• Ascertain the expected dependence of candidate breakupmechanisms on the rheological characteristics of likelyantimisting fuels.

* Determine techniques for incorporating breakup informationto correlate flammability test data, including the extensionof limited test data to other scenarios.

The expected process, mechanisms and results of breakup are ofprincipal concern in this section.

The overall breakup process depends not only on the propertiesof the liquid fuel, but also on the frontal surface of liquidexposed to the airstream and on thj magnitude of the relativeair velocity. This process may change with scale both due to thesize and shape of the liquid surface interacting with the air-stream, and due to different breakup mechanisms being involvedwith these liquid surfaces of different sizes. These types of

Lchanges are likely to occur over the range of small- to large-scale configurations employed to date in flammability testing ofantimisting fuels.

81

4

The breakup mechanisms of importance are expected to include oscil-

latory, stripping and Taylor breakup. Relevant findings reported

in the general spray literature were used, together with limitedresults directly applicable to antimisting fuels, in identifyingthese key mechanisms and their likely role in jet and droplet

breakup. Information stemming from previous FAA and related work

on antimisting fuels was also taken into consideration.

The droplet size distributions and mass concentrations of spraysproduced in various tests can, in principle, be determined experi-mentally. The exclusive use of experimental techniques to char-acterize droplet size information is very inefficient and is often

not feasible. In our opinion a better approach is to incorporateadditional liquid breakup mechanism information into a physicalmodel of the breakup process and align the model with selectedexperimental findings. The model can then be used to extrapolatenecessarily limited test data to a wide range of operating condi-tions and test scales, including the full-scale crash tests.

Physical models which account for complex liquid breakup phenomenaare relatively rare. Under the present effort one existing modelof this type, patterned after the early work of Mayer, 1 waslocated. This model treats the aerodynamic stripping breakup ofliquid injected into a subsonic airstream, accounts for both pri-mary and secondary breakup, and characterizes the complete result-ing size distribution on a number and a mass basis. Althoughthis model was the result of a major developmental effort forother applications, extensions of the model are necessary to make

it fully applicable to antimisting fuel testing and analysis. Themodel is considered to be an excellent candidate for this typeof modification in the future.

To enhance confidence in the utility of this breakup model we have,at company expense, put a computer coded form of it on-line andfound excellent correlation with available data on Newtonianliquids under conditions of interest under this program. Theseresults are included later in the discussion.

Information as to how the breakup process governs droplet sizesin different situations is important both to achieve an under-standing of why the antimisting fuel works as it does, as well as

- - to obtain insight on potential fuel and/or system modificationsfor improved fire resistance. Liquid breakup is recognized asan intermediate step in establishing the correlation of flamma-

bility test data and in conducting exploratory development testsof antimisting fuels. it isl especially significant in the correla-tion of flarn 1iability test data gathered from a range of small tolarge size test configurations because of its impact on results

82

achieved in tests of different scale. Once data correlation iscomplete, it is probable that breakup considerations can be circum-vented and the flammability results can be directly related tothe rheological properties of the treated fuel. In particular, itis anticipated that a quality-control type of test on selected fuelproperties would be adequate to ensure suitable antimisting fuel

*performance over a wide range of full-scale crash scenarios.

The preceding topics form the rationale for the subsequent dis-cussion of liquid breakup as it applies to antimisting fuels.

LIQUID BREAKUP MECHANISMS.

A large number of past studies have been conducted to further theunderstanding of atomization phenomena and the mechanisms ofliquid breakup. A few investigations which have emphasized themore fundamental aspects of aerodynamic breakup of liquids havesucceeded in clarifying many features of the process. Several ofthese features are relevant to spray formation of antimistingfuels.

Until rather recently most of the investigations concentrated onthe breakup of low viscosity Newtonian liquids with small lateraldimensions, such as jets, sheets and individual drops. The princi-ples and controlling physical properties determined in these smallscale cases also apply to larger scale jets, drops and globs ofliquid for suitable ranges of governing parameters. Studiesinvolving the breakup of highly viscous, non-Newtonian liquidsare rather limited, and the corresponding mechanisms are lesswell quantified. Highlights of the relevant findings of thesevarious studies are included in this section to indicate the natureof the breakup mechanisms and parameters that may be involved inproducing antimisting fuel sprays. A more complete summary ofthe general atomization and spray formation literature can befound elsewhere. 2- 5

OSCILLATORY BREAKUP. The breakup of a small, low-speed liquidjet moving into still air was the first problem of this type tobe treated analytically and this was done by Rayleigh. 6 Whilesuch a jet, per se, is of little direct concern here, the considera-tions of small liquid column or filament breakup which are inti-mately related to that of a small jet are quite relevant. Thisis because such filamentL; are often produced as intermediateelements in the rupture of large jets and drops. As such, therupture behavior of these elements affects the final result ofantimisting fuel disruption.

83

What Rayleigh actually analyzed were the oscillations of a cylinder

of incompressible, inviscid fluid of circular cross section under

the action of surface tension. In this case the column of liquid

is inherently unstable. Small initial instabilities arisinq from

unspecified sources are expected to grow exponentially as axially

symmetric waves--which rupture the column when the larqest wave

amplitude equals the column radius. The lenqths of the resultinq

liquid segments are determined by the wavelength, !opt , of the

most rapidly growing disturbance. This wavelength, opt, dependsonly on the diameter Dj of the column, is independent of the other

physical properties of the liquid or surrounding air, and is given

by

A 4.508 D.. (i)

The corresponding diameter d of spherical drops formed by actionof surface tension on each segment of the ruptured column is

d 3D opt 1/3 (2)

While Rayleigh and many subsequent investigators have used thistheory or its extensions to treat jets, Keller, Rubinou and Tu 7

have shown that the theory is not directly applicable to jets.

The difficulty is not due to (slow) motion of the jet, but arises

" " because the analysis assumes that the oscillation grows in ampli-." tude everywhere along the jet, including the region near the

nozzle where the jet originates. In reality, the oscillation is

negligible near the nozzle and grows with increasing distance

from the nozzle along the jet. The first case of oscillation

over the entire length of the column represents an example of

temporal instability, whereas wave growth with length along ajet represents spatial instability. The general topic of spatial

instability analysis is treated by Betchov and Criminale. 8

Keller and coworkers further showed that spatial growth rates ofcertain modes of instability of the jet are related to the temporal

growth rates of corresponding modes of oscillation of the cylinder

for low jet speeds. This explains why good aqreement is nbserved

between the results predicted for an oscillatinq liquid column and

experiments on low speed jets.

* As it turns out, the case of oscillating liquid columns which

Rayleiqh treated is the one chief interest here, rather than the* case of jets. However, the connection between the two and the

fact that part of the available experimental information applies

to jets gives impetus for examininq the results of other

84

AD-A127 142 CORRELATION OF FLAMMABILITY TEST DATA ON ANTIMISTING 2/2

FUELS(U) FALCON RESEARCH AND DEVELOPNENT CO ENGLEWOODCO L NAHOOD ET AL. DEC 82 FALCON-TR-3646i9

UNCLASSIFIED DOT/FAA/CT-81/i4 DTFR3-88-C-BS6i F/G 21/4 NL_msmmmmmmmmmIsonmommmmmmsmmmmmmmmmml

11111w 15.2

11111L_ 1 .6

MIROOP RSOUION TETCHRNANL BUEA OF STADRS16-

investigations carried out with small jets to estimate likelybreakup behavior of columns or filaments.

Subsequent efforts of Rayleigh 6 and the later work of Weber 9 pro-duced more refined estimates of Xopt by taking into account pro-perties of both the liquid and the surrounding air. With low speedjets, however, the air still only has a secondary influence onbreakup.

Weber derived the following expression for the optimum wavelength

op t =/2 D + 3( ) (3)

where n is the shear viscosity, o is the surface tension, and ppis the density of the liquid. The average diameter d of sphericaldrops initially produced by breakup of the cylindrical column isstill given by equation (2).

In recent years the stability of jets in stagnant air has beeninvestigated further by Middleman and coworkersI 0 - 1 2 to more fullyevaluate the influence of the ambient medium, high speed laminarflow, turbulence in the nozzle, and the use of viscoelasticliquids. They found that the time to breakup tb is given by arelationship

D.t In - (4)b 2 6

where: o* is the growth rate of the instability waveD. is the jet diameter, and63 is the initial radial dimension of the disturbance in

the jet.

In these studies the instability was characterized by measurement*of jet breakup length, Lb, which is related to jet velocity Vj and

breakup time tb by Lb = Vj • tb . Phinney's re-examination of thesedata 1 3 suggests that the initial disturbance at the jet exit maydepend on jet velocity.

These findings further indicate that columns (i.e., jets, liga-ments or filaments) of Newtonian liquids tend to disintegratereadily as a result of initial disturbances which grow as station-ary waves, even at very low speeds relative to air. The dropsresultinq from the breakup of such a column depend mainly on thecolumn diameter.

85

Additional investigations of the breakup of low speed jets ofviscoelastic and viscoinelastic liquids were conducted by Goldin,

et al. 1 4 and by Gordon and coworkers. 1 5 Their findinqs indicatethe instability pattern of a jet of viscoelastic fluid may bedramatically different from that for a Newtonian fluid.

For 0.1% polymer,* shear thinning, viscoelastic solutions, exponen-tial wave growths were absent. Large isolated drops appeared atrandom intervals and the jet became a string of drops, irregularlyspaced, connected by thin filaments of liquid. Some filaments sub-sequently broke, and individual drops were then released. Thefilaments generally showed significant longevity and did not alwaysbreak.

Jets of liquids with lesser elastic properties (e.g., 0.05% solu-tion as above) showed an initial region where a disturbance propa-gated as an exponentially growing wave of constant wavelength.

Growth of this wave was stunted before breakup occurred, and astring of regularly spaced small droplets, joined by filamentsof liquid, was again formed. With increasing distance from thenozzle, a breakup region existed where the connecting filamentsdeveloped secondary instabilities and ultimately broke. Thefilaments then either collapsed to form small satellite drops orwere snapped into an adjacent primary drop.

With viscoinelastic liquids, filaments were also observed, butthese filaments ruptured at the filament-droplet joints. Theresulting free ligame-ts each disintegrated into several small

satellite drops.

The unusual longevity of the filaments between droplets was charac-terized as due to stretching or elogational flow which increasedthe elongational (tensile) viscosity of the filament. While thisflow might be induced by relative motion of the droplets to which

P4 the filaments were attached, Gordon, et al., employed jets perturbedby controlled disturbances. This resulted in negligible relativemotion between droplets. Even in this case of uniform dropletspacing, the steady exponential decrease in filament diameterindicated that a low pressure region within each droplet wasdrawing liquid and creating a stretching motion in the filamentjoining the two drops. Extensional strain rates on the order of20 sec-1 were observed.

0

* Onie (f the materials ised was Separan AP-10 (D-,w Chemical Co. , I

parti, la y hydrolyzed l lyacrylamid , cf hiqh mnoIecti],1 " weiqht.

86

Further considerations in oscillatory-type breakup of jets, sheets

and drops take into account the motion of the fluid and varying

degrees of influence of the surroundinq air. As the velocity of

a jet is increased above the negligibly low value used previously,

breakup of the jet into drops becomes partially influenced by

interaction of the jet with air. The air velocity increases over

the wave crests and decreases in the trouqh reqions. Simultan-

eously, the pressure decreases over the crests and increases over

the troughs, in accordance with Bernoulli's theorem.19 This situa-

tion causes the wave amplitude to increase, as the reduced pres-

sure tends to increase the height of the crests and the increased

pressure tends to deepen the troughs. These motions arc opposed

by surface tension forces and tend to be damped by viscous !orces.

The first order interaction of air with the liquid is thus i

force normal to the average velocity of the liquid jet; shca

forces enter as smaller interactions of higher order.2cI 'h ,

result of this interaction arising from an increase in let

velocity is that the optimum wavelength for breakup decrescs Arld

the growth rate of the optimum wavelength disturbance increases.

These effects lead to a decrease in breakup time, breaku[ Jistan ',

and drop size.

As the jet velocity is increased further, the wave motion on the

jet changes from axially symmetric (varicose or dilational) to

asymmetrical (sinuous)* as a result of increased surface inter-

actions with the air.

Oscillary type breakup of tangentially moving sheets and ho1 low

cones of Newtonian liquids, such as may be produced by fan or

swirl type hydraulic nozzles, has also been consideied by several

investigators. 2 1-24 This breakup occurs in a manner similar to

that of a liquid column. Small disturbances in the sheet develop

to form waves which rapidly grow until they reach a critical

amplitude. Tears occur in the wave crests and troughs, and frag-

ments of the sheet corresponding to a half wavelength (of sinuous

waves) are broken off. These fragments contract by surface ten-

sion into unstable cylindrical ligaments. In addition, holes

that often form in the sheet also result in the formation of liga-

ments. These ligaments ring the holes. Some of the ligaments

move with their long axes in the direction of flow, but the major-

ity of them move transversely. The ligaments, themselves, are

unstable, exhibit growing oscillations and disintegrate into

droplets.

* For ularity it is noted that the (Him.triul ly , ,se] side; ot W let

(o r two surf ,i s o)t islh t) osci I ite it] phase to Ilodie 51ww5 w.wof', intw 0

t h''y osiI lato otit 0?ohi to pt (oile'(ii ilt ionS) i wPJvs.

87

Two different maximum growth rate instability waves are involvedin this process; one is of wavelength A on the sheet, and anotheris of wavelength Aopt on the cylindrical ligament. The latter isgiven approximately by the Raleiqh criterion equation (1). A sheet

--of thickness h which breaks into a band of width t. (where is a--constant in the ranqe between 0 and 1) is expected to collapse into'. a cylindrical liqament of diameter

2

The resultant droplet diameter d resulting from the ligament dis-ruption is qiven by

d - 2.1 (hiX) (5)

It will be noted that the droplet diameter can be expresseddirectly in terms of the wavelenath of disturbance on the sheet.Dombrowski and Johns 2 4 have refined this analysis and tested itexperimentally. They express the drop diameter d in terms of thejet diameter Dj by

1/6S[ 1/3 + 1/63. 3n _ _j (6)

4 with 1/5

1/6 [ K~v 7 ~1/31

D. = 0.9614 [ 2 [1 + 2.6 / J (7)"" 3 pg £ 4 +72.6 2 5

and where K is a constant to be determined empirically,* V is therelative air-to-liquid velocity, Pg is the gas density, and n, ;,

and pt are the viscosity, surface tension, and density of theliquid, respectively.

Sheet breakup of non-Newtonian liquids has been investigated#i experimentally by Dombrowski, et al. 2 5 using shear thinninq alumi-

num soap petroleum "gels", and by Ford and Furmidqe 2 6 using shearthinning water-in-oil emulsions. The former 2 5 shows an examplein which a 1/2-inch diameter jet of petrol gel injected into still

". * T|I'llc 1,.r, me -t, - r K i,; r , l etI t he !-,hoo t ~tt ' I t ,I(1-kIl " . I'

-. h ' n ." r i I by h = K,x.

?88

air at about 150 ft/sec first develops surface irregularities,

then surface waves--which increase in extent until they are formedinto a large sheet. Such sheets are very thin and disinteqratewhen they become unstable. While sheets are formed under partic-

" ular conditions, their formation from jets should be recognized as

possible but not commonplace.

Thin sheets were produced in these investigations by forcing

liquid through a fan nozzle where the liquid experienced a high

shear rate. The shear-thinned viscoelastic sheets so formed alldisintegrated into stable ligaments, but the nature of ligamentbreakup was found to depend on the consistency,* nc, of theliquid. At values of nc = 1 the sheet broke up almost spontan-eously into droplets. Increased values of qc :2 gave ligamentsresistant to fracture and with beads formed on them. In the range

C- of 25 < 9c < 100, the ligaments remained unbroken. For values offlc greater than about 100, the ligaments broke up into shreds,i.e., short cylindrical lengths. At high values of nc (near 1600),

the shreds were formed directly from the sheet. Fraser2 8 furthernotes that sheets of dilatant (shear thickening) viscoinelasticfluids,** consisting of high particulate concentration slurries,break down close to the fan nozzle orifice through the formationof holes. The associated ligaments formed are larger, as are theresulting drops, than observed with Newtonian liquids because thesheet has not expanded so far.

The preceding studies indicate that the major effects of elonga-tional flow and fluid elasticity in the breakup of viscoelastic

filaments are to oppose their breakup and to increase the size ofdroplets or liquid fragments. Figure V-1 indicates the probablerole of viscoelasticity in filament breakup when breakup does occur.

:a.

* Consistency is a viscosity equal to the slope of the shear stress vs.

shear rate curve for the liquid at any given point. 2 7 The values of ric

reported in Reference 25 were in Gardner units, since a Standard Gardner mobil-

ometer was used. The values represent the number of grams correspondinq to a

rate of fall of 0.1 cm/sec of a piunqer which moves down throuqh the liqxuid

in a close (but not tight) fittinq cylinder. These values are approximately

prnoportional to consistency in poise.

* Two types of phenomena are knrnwn to occur with dilatnit materials:

volJmttri dil,.itil'y, which donot (7, -it, incr(,ase inI ttnil v(,lum(, uidcr nlwit

and rheoloqical diLatancy, which refers to an increase in apparent vis(.osity

with ir nresinq shejr rate. This latter property is more commonly :s, ciated

with dilatant fluids, and is the one of main interest here.

89

LIGAMENT BREAKUP

o

FIGURE V-1. PROBABLE ROLE OF VISCOELASTICITY IN BREAKUP

90

,°a.

In closing this section on oscillatory breakup, it is noted thatthe oscillatory breakup of drops has also been treated bothexperimentally and theoretically. 3 0 - 3 4 However, this type breakupis of little interest for antimisting fuels. The scheme accordingto which the breakup of droplets is categorized, on the other hand,is of interest. It will also indicate why oscillatory breakup isessentially negligible in the present context.

A common assumption made in treating the breakup of liquids is thatwhen the maximum aerodynamic force tending to disrupt the liquidexceeds the surface tension force that holds it together, theliquid will burst. This condition is quantified by equating thesetwo forces, which for wind against a spherical surface of diameterDO gives a relationship of the form

1 2SCDPa(U - v) = 4C/D 0 (8)

where CD is the drag coefficient of the drop, u is 'he air velocityand v is the velocity of the liquid at breakup. By rearrangingexpression (8) the breakup criteria can be expressed in terms ofa critical Weber number We by

c2

PaV2D 0 8W e c = - = -- (9 )c (0 CD

for the relative air-to-liquid velocity V.

The assumption on which this condition is based is true only forsmall rates of stress loading and not at high (e.g., shock) loadingrates, since breakup and associated flow of liquids are knownexperimentally to be a time-dependent process. Nevertheless, thecondition has been useful in practice for indicating regions wherecertain types of droplet breakup occur. It should be recognizedthat large liquid viscosity or elasticity can also affect breakup

-* behavior, so that classification based solely on Weber number isknown to be incomplete.

Andersen and Wolfe 34 determined experimentally that a liquid dropcan undergo five separate modes of breakup in an airstream,depending on the conditions:

1. Dumbell oscillation - This action is observed at low values ofWe (>8), just above the critical value required for breakup. Adroplet undergoing dumbell oscillation breaks into two smallerdrops.

,,'. 91

2. Bag breakup - For slightly higher values of We, a drop flattensand its center is then blown out in a concave manner to form athin-walled hollow bag which breaks. With a relatively hiqhviscosity, elastic fluid, the bag typically ruptures and deflateswithout producing small droplets.

3. Stamen breakup - This variation of bag breakup is characterizedby a forward protruding stem, which is formed in the center of thebag prior to breakup.

4. Stripping breakup - At still higher values of We (>100) thebreakup of a flattened drop occurs by continual stripping (shear)of the liquid away into ligaments and droplets by the wind (gasstream). This is the dominant mechanism by which drops arebroken up over a wide range of conditions, as well as by whichjets and bulk liquids are broken up aerodynamically.

5. Taylor breakup - At very high values of We (-104) a drop sud-denly accelerated by a high velocity air stream will flatten aero-dynamically and its frontal surface will quickly become corrugatedby Taylor instability growth. The corrugations rapidly deepenuntil the drops shatter into smaller fragments, which may thencontinue to undergo breakup by surface stripping. This type ofbreakup is important for bulk liquids and large jets which undergosignificant deformation before fragmentation.

STRIPPING AND TAYLOR BREAKUP. As indicated in the previous dropbreakup classification scheme, stripping and Taylor breakup occurat relative air-to-liquid velocities higher than those which causeoscillatory disintegration. These two forms of breakup are groupedtogether here because they compete with each other in the initialbreakup of bulk (large scale) liquid. The conditions of liquiddimensions and relative velocity determine which mechanism is ofdominant importance. The following discussion of these twomechanisms pertains most directly to Newtonian liquids, but thegeneral features and results seem to correlate with experimentsin the very few cases where non-Newtonian liquids have beeninvestigated.

A slug of bulk liquid injected into a low aircraft-speed (60-150knots) counterflowing external airstream undergoes simultaneousbulk deformation and stripping (surface erosion) at its freesurfaces exposed to the air flow. The initial surface strippingof the liquid is slow due to low wind speed and relatively small

* exposed liquid surface area. As a result, deformation of the bulkliquid has time to occur. The deformation results in lateralspreading and qeometrical thinning of the liquid. While the thick-ness of the deformed liquid in the direction of the relative air

~92

flow decreases, Taylor instabilities (waves) grow on the liquid'swindward surface. When the amplitude of these instabilitiesbecomes equal to the thickness of the deformed liquid, the liquidfractures into a number of smaller pieces. This rapidly producesa large liquid surface area. In turn, this increase in areagreatly enhances the overall stripping rate and the pieces continueto break up by surface stripping. In this situation, the Taylorbreakup process dominates the initial disintegration of the liquid,while surface stripping completes the breakup into droplets.

For the case of continuous flow from a large jet into a counter-flowing airstream, as above, a steady-state deformation conditionis reached whereby the jet spreads laterally and its penetrationagainst the airstream is halted at some distance from the oiifice.

°! In this terminal region, the liquid flow is nearly lateral (as itreverses direction relative to the air velocity) and the liquid

* forms a curved sheet convex to the wind with a thickness whichdecreases rapidly with distance from the jet axis. It is in thisregion where Taylor instabilities are expected to grow on thewindward surface and break up the rather thick liquid sheet intosmall pieces. These pieces then continue to break up by surfacestripping, as above.

If the jet is small rather than large, Taylor breakup is neglig-ible, except perhaps at very high air velocities, and the dominantbreakup mechanism is surface stripping.

The stripping and Taylor mechanisms are indicated in a somewhatidealized but illustrative manner in Figure V-2.

Stripping breakup occurs when liquid shear by wind is accompaniedby growth of surface waves which form and shed ligaments. Theseligaments separate from the main liquid mass, undergo growingoscillations due to instability of the column, and break up (ornot) into drops--depending on conditions or liquid properties.If these drops formed by primary breakup are large and the rela-tive velocity is still high enough, the drops may break again.This latter action is called secondary breakup.

As noted earlier, individual drop sizes are generally related to*& specific wavelengths of surface waves, so it is expected that the

size spectrum of drops produced by stripping breakup is determinedby the spectrum of wavelengths excited by the wind. Waves of verysmall wavelength do not develop because of viscous dissipation,and very long wavelengths develop slowly because of inertialeffects. Between the very short and very long wavelengths, thereexists a spectrum of wavelengths that can be excited to appreciable

93

L IGAMENTS

DROP

WIND

AIR 8LQI,

THROUGH

TAYLOR BREAP

PIG(JR. V-2.IB3REAKUP EC.iANTSP

4 S

49

amplitudes during the time o' action of the high velocity gasflow. The waves actually generated show effects of capillarityand wind acceleration. Their growth rate is determined by thebalance of wind energy causing their growth and viscoelasticenergy dissipation in the wave.

The stripping breakup mechanism concept appears to stem from theearly work of Castleman3 5 on the disruption of a high velocityjet. Development of the concept, however, was carried out largelyin connection with the aerodynamic breakup of single drops byseveral investigators; including Lane, 3 6 Hinze, 3 1 ,3 2 Gordon, 37

Hanson and coworkers, 38 Engel, 3 9 Andersen and Wolfe, 34 and Rangerand Nicholls. 4 0 Detailed investigations involving jets werecarried out by Clarke41 and Morrell, 4 2 ,43 Sherman and Shetz, 4 4 andWeiss and Worsham.4 5

The work of Sherman and Schetz 4 4 is of particular interest inthat it confirms by high speed photography the formation ofligaments and ligament shedding under high-speed shear conditions.

The study of Weiss and Worsham 4 5 produced one of the best availablesets of droplet data. They employed hot wax in a hot gas streamto avoid effects of evaporation and devised a scheme to obtainrepresentative samples of droplet mass-size distributions. Theycorrelated their data by means of the following equation for themass-median diameter, dmmd:

( p Dj~i/6 / l ~ l l

d k 5/12 1/3 1 + 103 1/1210)'""~ ~ =.60 ( V)i0

mmd 4/3 Z P.V \g 9~ lgj

where Vj is the jet velocity and Dj is the jet diameter, and allquantities are in cgs units.

Andersen and coworkers 4 6 note that, based on a number of studies,the mass median droplet size produced by small jets is proportionalto the quantity

aD. b

--" Vc

where the experimental index constants vary over the approximateranges: a = 0.2 - 0.45, b 0 - 0.5, and c = 0.75 - 1.33. Values

95

of these indicies (especially b) are uncertain for jets withorifice diameters larger than about 0.5 inches. They also notethat a large amount of experimental work shows that for a widerange of conditions the drop and jet breakup time tb due to surface

stripping are given by

t k(Dj)()l/tb = k (1)

The experimental constant k has been found to range from about 2to 6, with a value near 3.5 often representing a "best" value.

Taylor breakup stems from growth of the instability of the freeboundary surface of a liquid. This instability arises when theliquid is subjected to an acceleration normal to the surface,with the acceleration directed into the liquid. Acceleration inthe opposite direction from the more dense liquid toward theless dense air acts to stabilize the interfacial boundary.*

The classical case of interfacial stability of a semi-infiniteliquid under constant acceleration was analyzed by Taylor4 8 andwas examined experimentally by Lewis. 49 They did not considerliquid sheet fragmentation specifically.

Keller and Kolodner 50 investigated the growth rate of the fastestgrowing Taylor instability induced by a wind impinging normallyon the surface of a nonviscous liquid with finite surface tension.They obtained theoretical expressions for the approximate size ofthe liquid pieces formed by this mechanism, and the time requiredfor this to occur.

Bellman and Pennington 5l also investigated the growth of thefastest growing Taylor instability for the cases of (l) a non-viscous liquid with surface tension, (2) a viscous liquid withoutsurface tension, and (3) a viscous liquid with surface tension.Their results show that the presence of surface tension qenerall%

0

• Another type of phenomenon known as Helmholtz instability 11 o

This type ot instabi I ity diovelops in the suTt-,' cf , 1 i ~lidi 1% h.'

wi nd st ress. Acclrrnq 1(o R4,etiu7 ( V h wIt; ~Ji it t1 l ie dssi( Il I I('Ifl1hiItz Z scllf;u m- inutli sw.di (I im thIc L!t

SI '/08, sli(11't nq f8 I i f hi type o I i TUf;t ib1 i I i ty (boy [I() t fi 'f I .rt r I

has a negligible effect on the wavelength of the fastest qrowinqdisturbance, and that this wavelength is viscosity dependent. Themain effect of viscosity is to retard the growth rate of thedisturbances and to increase the wavelength of the fastest growingdisturbance. This, of course, would result in larger liquidpieces being produced by this breakup mechanism.

Much of what is known about Taylor breakup has resulted from studiesof water droplet breakup conducted by several workers; includingReinecke and Waldman, 5 2 Simpkins and Bales, 5 3 and Harper, et al. 54

These investigations give additional confidence in the validity ofthis mechanism for breakup.

A full theoretical or experimental treatment of Taylor breakupof deformed liquid jets does not seem to be available in theliterature, either for Newtonian or non-Newtonian liquids. Thissituation makes it difficult to know, without further investiga-tion, the combination of conditions of jet size, relative velocityand liquid properties where Taylor breakup becomes important.Preliminary estimates from a brief examination of photographs ofjet breakup in antimisting fuels test configurations under typicaloperating conditions suggests that in these situations Taylorbreakup is unimportant for initial jet diameters less than about0.5 inch, but may begin to become significant for diameters largerthan about 1 inch. This is a topic which deserves further detailedattention in order to better quantify the change in breakupbehavior with test scale.

ANTIMISTING FUELS BREAKUP.

The previous section has concentrated on general breakup mechanisminformation with relatively little input from antimisting fuelsstudies. In the present section the latter is emphasized. Thisseparation has been made purposely in an attempt to clearly indicatehow the rather sparse breakup information on antimisting fuelsdovetails with (and is supported by) the general information.Several of the antimisting fuels that were one-time or currentcandidates are also noted since their breakup behavior is relevant--yet it is not expected to be the same for all such materials.Final droplet size information from sprays of antimisting fuels isnot currently available, except for two tests, so direct correla-tions of this type among the several fuel candidates and testconfigurations are not generally possible at present. Droplet sizedistribution measurements of selected candidate fuels are plannedin connection with investigations being conducted at the JetPropulsion Laboratory.

97

Photographic investigations have provided most of the available

information on breakup mechanisms specifically related to these

fuels. Zinn, Eklund and Neese 5 5 examined the aerodynamic breakupof fuels injected through a 5/16-inch I.D. plain nozzle, costream

into a 110-knot air flow. The nozzle was fed by a 58.6 foot-lonq

fuel hose of 1/2-inch I.D. The fuels used included neat Jet A,and Jet A solutions of 0.2 mass percent Conoco AM-l, 0.4 mass

percent Dow XD8132.01, and 0.7 mdss percent Imperial Chemical

Industries (ICI) FM-4.

While the Jet A fuel, itself, is a Newtonian liquid, the result-

ing fuels modified with AM-l exhibit a shear thinning behavior,56

and with FM-4 and XD8132.01 show a shear thickening or dilatant

nature. The shear thickening behavior of FM-4 differs significantlyfrom that of the Dow material. The former exhibits slight shear

thinning, power law fluid behavior under laminar flow conditionsfor shear strain rates less than about 100 sec -1 , and a mildlydilatant shear stress-shear strain rate (flow curve) nature forlarger values. 5 7 The fuel with Dow additive is essentially

Newtonian _.3low a critical shear rate (about 800 sec-1 for 0.7

weight percent XD8132.01 in Jet A), and tends to form a gel-likesolid above this value. Preliminary results obtained by

Mannheimer 5 6 indicate that a second critical point exists at highshear stress (and slightly higher shear rate) beyond which a near-

Newtonian behavior is again observed.

*The fuels with AM-I and FM-4 additives are known to exhibit

viscoelastic properties57 ,5 8 , and each of the three modified fuelsare known to exhibit high elongational viscosities. 5 9 Theelongational viscosities at given strain rates increase withadditive types in the sequence XD8132.01, FM-4, and AM-l (for

the respective mass concentrations in solution of 0.7%, 0.3% and0.2%). The variations of shear and tensile viscosities with

their corresponding strain rates are shown in Figure V-3 for asolution of 1.3 mass percent AM-I in JP-8 fuel. The tensileviscosity curve in Figure V-3 is dashed to show approximate

behavior over a range of strain rates, including the region

where data (indicated by points) were taken.

The photographs taken by Zinn, et al. at various distances fromthe nozzle indicate that the rheological nature of the modified

fuels has a significant influence on the final fuel particle con-" fiquration. The AM-l fuel is characterized by small liqaments

* near the nozzle, which are drawn out into sheets and into thin

- - droplet-containing threads by the moving air. The FM-4 fuelexhibited somewhat larger ligaments and fewer droplet-ladenfilaments. The XD8132.01 fuel broke up into relatively large,

98

ITENS ILE

j3I

10I

0-

~10- o I 0 0 0

STRAIN RATE -SEC-'

FIGURE V-3. APPROXIMATE SHEAR AND TENSILE VISCOSITIES FOR* 1.3% AM-i IN JP-8 FUEL (ADAPTED FROM REFERENCE 66)

99

irregularly shaped globules. The corresponding flammabilitytests were inconclusive because of uncertainties of fuel spreadingand dilution.

* The long fuel flow hose used on these tests ahead of the nozzle is"expected to have preconditioned the fuel viscosities while achiev-

ing fully developed shear flow. The thread formation observed withAM-4 and FM-4 is consistent with direct ligament formation at thenozzle, followed by inhibited breakup and filament thinning, char-acteristic of viscoelastic liquids with high tensile viscosities,as noted earlier. The large globule formations observed for

*XD8132.01 appear to be the result of direct breakup of liquidligaments, preconditioned by shear strain rates exceeding the firstcritical point and exhibiting a high tensile viscosity.

The subsequent investigations by Eklund6 0 and Eklund and Neese 6l,in connection with developing a small-scale test apparatus tocompare the flammability of modified fuel sprays, also includedtaking photographs of the breakup of antimisting fuels. In thesecases additives were mixed with the neat fuel to form the follow-ing mass percentages in Jet A: 0.4% FM-4 and 0.2% AM-I. Also,0.3% FM-9 in Avtur was tested.

The modified fuels nature indicated earlier is supplemented here

by the shear thickening FM-9 fuel. The rheological properties ofthis fuel are similar to those of the XD8132.01 treated fuel,except that the FM-9 fuel is only slightly viscoelastic under

*particular flow conditions. 6 2 Further, FM-9 exhibits antithixo-tropy,* show concentration-dependent lower and upper critical

- shear rate values,6 3 and exhibits a tensile viscosity of somewhat,. uncertain values. 64 ,6 5** The lower critical shear strain rate at

751F for 0.3 mass percent of FM-9 in Jet A is about 3100 sec-1 .6 6

The second critical shear strain rate for this mate.ial occurs atabout 9100 sec-

1 .6 3

* In these tests the sprays were again directed costream with theair flows, which in this case were 200 mi/h and 285 mi/h. Theobserved breakup of modified fuels was essentially as before,

-1 * Antithixotropy may be of particular importance in liquid breakup cons,-

. , derations since the sheddinq of individual fragments typically occurs in the

millisecond time regime. Primary breakup would be expected to occur before

* aerodynamically induced strain on the liquid would cause significant tensile

viscosity effects to become apparent.

•* The tensile or elonqation viscosity of FM-9 fuels has been a measure-

ment challenge and the results to date show significant disparities.

100

except the photographs showed a significantly greater filamentformation tendency with AM-I fuel than for FM-4 fuel. TheFM-9 treated fuel showed even greater breakup into droplets andregular shaped fuel globs than was apparent earlier with theDow material.

The breakup behavior observed in these tests for FM-9 is interpretedhere as involving an antithixotropic fuel that is not preconditionedto exhibit high tensile viscosity at or near the time of breakup,such as might be expected for a fuel which has experienced largeshear or elongational strain rates via flow through a long pipe ornozzle. The breakup is believed to occur primarily as a result of

. - shear flow-governed instabilities which grow and break in a timeinterval that is short compared to the time it takes the liquid todevelop high tensile viscosity under the flow conditions experiencedduring breakup. Further, the breakup is believed to occur primarilyin the low shear viscosity region where the shear strain rate iswell above the second critical point.

In this region, the range of wavelengths of wind induced instabili-* . ties which can grow is restricted, as are the resultant drop sizes

that are produced when these disturbances break. These size re-strictions arise from the combined effect of available wind drivingforce and the second critical shear strain rate. As the maximumshear strain rate which can be produced by the wind approaches thesecond critical point from above, the lower and upper drop sizelimits approach one another. The drop size which these limitsapproach is typically rather large. Breakup of this treated fuelfor maximum shear strain rates below the second critical point occursin a related manner, but the results are different. In this regionthe drop size range is expected to be rather wide, compared to thatwhich can be generated above the second critical point.

These breakup results are not obvious, and have been determinedusing the breakup model (noted later) in an application involvingFM-9 treated fuel.

The ignition and flame propagation features of these modifiedfuels clearly showed that the AM-I fuel strands ignited rathereasily, and sprays of this fuel possessed a burning tail. BothFM-4 and FM-9 fuels showed good resistance to ignition.

A significant breakup finding embedded in these flammability6 results is that the mere suppression of fuel droplet formation is

not by itself adequate to indicate good resistance to ignition andburning of a fuel. These findings were confirmed by additionaltest work carried out by Eklund and Cox.6 7 This feature was

101

apparently overlooked by Hoyt and coworkers 68 in their very recent

effort to correlate fuel drag reduction properties with jet breakup.

On a strictly physical basis, the reasons seem apparent as to why

a fuel which has formed droplet chains joined by filaments poten-tially represents poor resistance to ignition and flame propaga-tion. Drawinq on the earlier discussion of oscillatory breakupof viscoelastic jets and filaments, it will be recalled that thefilaments between droplets tend to thin rather significantly, andmaj break due to aerodynamic forces. The filaments or threadsoften remain as filaments without disruption or significantretraction into globs or drops. When the surface area of anoriginal filament is compared with the total area of a line of

drops that would be formed if Rayleigh breakup were to occur, it

is found that the two areas are nearly the same (with that ofthe filament being slightly greater). However, in the dropletchain form, as the filament thins the total surface area of the

*filament-droplet configuration will become appreciably greaterthan that of the separate drops. Further, the original filament

size is contributory, in that shear thinning AM-I fuel is expectedto yield finer original filaments than would be expected for ashear thickening material, such as FM-4 fuel. The increasedsurface area of the thin and thinned filament structures areexpected to exhibit enhanced evaporation, ignitability and, hence,

*flammability.

The investigation by Fau1 6 2 of jet and droplet breakup is alsorelevant. He examined jets of neat Avtur and AM-l, and FM-9 inAvtur--as well as droplets of neat Avtur and FM-4 and FM-9 inAvtur. The mass concentrations of additives used in the fuels were0.2%, 0.3% and 0.4% for AM-1, FM-9 and FM-4, respectively.

The jet results tend to be consistent with previous work andwith explanations advanced earlier in this section. In particular,the suppression of turbulence in AM-I fuel jets is a rather welldocumented action observed in connection with drag reducingviscoelastic fluids, a group to which AM-I fuel belonqs. 6 8 ,6 9

A considerably greater degree of turbulence was observed in FM-9jets, implyinq that the viscoelastic effects in this case werefar less than for AM-I fuel.

For the droplet tests, the fuels were fed to the test regionthrough a 0.4 mm inside diameter capillary, with length-tn-diameterratio of about 750. Using low flow rates, Avtur fuel injected]into still air gave a very short jet followed by a stream of dis-crete spherical droplets formed by Rayleigh-type breakup. At thepoint of separation from the capillary jet, the droplets broke

*102

away cleanly--showing no signs of filament formation. An increasedviscosity fuel sample (presumably viscoinelastic), prepared with75 mass percent liquid paraffin and 25 mass percent Avtur, alsoshowed no filament formation, but some occurrence of smalldroplets near larger ones. The latter were presumably caused by asmall amount of necking between the larger drops, which subsequent-ly broke away to form the satellite droplets.

In the case of FM-4 fuels, a continuous string of drops inter-connected by fuel filaments was observed. This type structure istypical for a viscoelastic fluid as noted earlier.

With FM-9 fuel, droplets were produced which appeared similar tothose of FM-4 fuel in that they were joined by filaments duringthe early part of their fall. As Faul noted, 6 2

"On emergence from the capillary an elongated dropletformed which flowed into more discrete droplets as itmoved downwards, eventually accelerating and forminga filament behind it. As the droplet fell away thefilament stretched until it parted, exhibiting elasticbehavior; the upper part of the broken filament re-coiled into the next droplet forming above it. Some-times the shock of the recoil would be enough to detachthis next droplet from its string prematurely ... This

behavior suggests that FM-9 fuel also has viscoelasticproperties, but unlike FM-4 fuel there is a suddenbreaking of the filament followed by elastic recoil.This suggests that with FM-9 fuel there is some visco-elasticity where the fuel has formed a more rigid shearinduced structure."

The flow of liquid through the long capillary prior to ejectionwould be expected to cause each fuel, but particularly FM-9, toexperience an appreciable shear strain rate and thereby precondi-tion the fuel issuing from the capillary tip. Subsequent shearand/or elongation of the liquid as it fell away from the tipcould then readily force the liquid into the gel-like shearinduced structure noted above.

This situation examined by Faul and another by Sarohia and Landel, 6 4

represent the only known cases where elastic effects of FM-9 fuelhave been clearly observed.

Additional tests conducted by Faul consisted of atomizing Avturand AM-l, FM-4 and FM-9 fuels with an airblast atomizer. Theneat Avtur broke rapidly into small sized drops. The FM-4fuel tested under the same conditions showed lonq strinqs of

103

a

fuel pulled out from the atomizer; these remained intact forabout 25-50 mm, even under relatively high blast pressures. TheAM-I fuel behavior was analogous to that of PM-4, except thefilament formation tendency was even more pronounced. The FM-9fuel behavior was more like Avtur but with the resulting dropsbeing larger and less spherical. Faul commented: 6 2

"Examination of cine-film clearly shows that FM-9 fuelis firstly drawn out of the atomizer jet by the airflow and then torn apart to form lumps of fuel varyingin size and shape, suggesting that under these con-ditions the fuel has a form of shear-induced structure.This behavior is also shown to a certain extent in thework on droplets where the FM-9 fuel droplets suddenlybreak from their associated filaments; but unlike theresults from the droplet work there appears to be noelastic recoil or stringing with the airblast atomizersprays. This suggests that the viscoelastic behaviorof FM-9 fuel only becomes apparent under very specificflow conditions as encountered in droplet formation. "

These observations further support the working hypothesis that FM-9breakup depends significantly on its shear induced nature (butnot to the exclusion of tensile effects).

The investigations of Sarohia and Lande1 6 4 also included jet and

drop breakup of neat Jet A, and solutions of AM-I and FM-9 (withcarrier fluid) in Jet A. These jets were injected into movingairstreams. Their results generally confirm the findings ofFaul. They confirmed for jets that the most suitable (i.e.,fastest growing) wavelength for neat Jet A and FM-9 fuels isapproximately 4 to 5 times the jet diameter, as is expected fromthe Rayleigh criterion. Such wave growth was observed to bealmost completely suppressed for AM-I fuels. Both thread andsheets were observed to form from AM-I fuel jets injected into amoving airstream with velocities of 27 m/s.

Sarohia and Landel also carried out pendant drop tests in which alarge drop was allowed to fall under the action of gravity, whilebeing attached by a filament of the liquid to a wetted capillarytip. The purpose of this test was to determine experimentalvalues of the elogational viscosity of FM-9 fuel, which they did.In the process of making these measurements they also observedfilament rupture and elastic recoil of the broken filament back

i* to the adjacent droplets.

An interestinq feature of this configuration was that the measure-ments were carried out under conditions which might be expected to

104

reasonably simulate those occurring in the actual breakup of fila-ments. Some consternation arises, however, in that filament forma-tion seems to be a rare occurrence in the observed breakup ofFM-9 fuel. Nonetheless, it appears that as a scheme for tensileviscosity measurement, this approach or some adaptation of itrepresents one of the few feasible techniaues known.

The previous investigations essentially exhaust the availableinformation on fundamental breakup studies of antimisting fuels.There are, however, two other qualitative breakup features whichresult from observations of the FAA Wing Fuel Spillage tests and

. the full-scale crash tests which should be noted.

In the case of the Wing Fuel Spillage test configuration located. at the FAA Technical Center, liquid breakup of jets is expected to

occur in a manner indicated schematically in Figure V-4. Especiallynote that Taylor breakup is expected to occur in the leading edgeof the aerodynamically deformed fuel jet. Videotape records ofindividual tests conducted with this test rig show the cell-likestructure expected for Taylor breakup and support the hypothesisthat this breakup mechanism is operative in these tests.

Similarly, videotape/photographic records of the full-scale crashtests that have been conducted at Lakehurst, N.J. indicate ananalogous cell-like structure. This indicates that Taylor breakupis operative in the dispersal of antimisting fuels in full-scalecrashes.

In both of these cases where Taylor breakup cells are identified,it will be recognized that the cell structure is not a long-termsteady-state type phenomenon, but rather tends to fluctuate withtime during actual tests. These fluctuations are characteristicof the phenomenon, and some sort of average over their range(perhaps together with estimates of the extremes) represents areasonable expectation in accounting for their role in antimistingfuels breakup. Accounting for these phenomena quantitativelyrepresents a challenge for future endeavors.

It is relevant to note that of the various other test configura-tions which have been or are being used in connection with anti-misting fuels spray and flammability evaluation, none are known

"- to have been characterized in terms of the fuel breakup achieved.In view of the extreme sensitivity of ignition energy require-ments to spray droplet sizes (discussed elsewhere in this report)which has been discovered recently, it appears that some deqree ofdroplet size characterization deserves further attention. Particu-lar attention to this feature is recommended for the full-scale

105

-4

LUU

LUJ

L)L

'-4

.0 H

0S. 5' 0

0 010

* 0

-J

U-i rJ4~

AA

106

tests, the FAA Wing Fuel Spillage rig, the FAA small-scale flam-mability rig, and the Spinning Disc rig (at Southwest ResearchInstitute).*

From the nature of the phenomena involved, it is apparent thatresults from simple engineering tests of the type represented bythe drop test at Southwest Research Institute can hardly beexpected to yield information that will be useful for extrapola-tion, yet may still be somewhat reliable qualitative indicatorsof fuel flammability performance in the tested configuration.The crudeness of the tests gives little confidence that agreementwhich may be achieved upon extending results to other configura-tions is anything more than fortuitous.

LIQUID BREAKUP MODEL.

The incorporation of relevant breakup mechanisms and phenomenainto an analytical model with the capability to calculate theresultant droplet sizes produced by the aerodynamic breakup ofliquids has been accomplished only in recent years. One existingmodel of this type was located under the present effort. Thismodel treats the aerodynamic stripping breakup of a Newtonian or(selected) non-Newtonian liquid injected into a subsonic airstream,accounts for primary, secondary and higher order breakup, andcharacterizes the complete resulting size distribution on anumber and a mass basis. While moderate extensions of thismodel are necessary to make it fully applicable to antimistingfuel testing and analysis, current information indicates that thismodel is an excellent candidate for this type of modification.More importantly, once such a model is aligned with selectedexperimental findings, it can then be used to extrapolate -imitedtest data to a much wider range of operating conditions and testscales, including full-scale crash tests.

The subject model is an outgrowth of Mayer'sl capillary wavestripping model for estimating both stripping rate and dropletsize. Mayer considered the airflow to induce small capillarywaves (instabilities) on a liquid surface. When such a wave qrowsto an amplitude comparable to its wavelength, X, the wave crestis envisioned as being eroded as a ligament, which then breaks upinto droplets.

* The new test f figuration .it the ,14't Pripuls;i , I, ,t i V tut

fitted with a dr-pltet size chtormini nri 'i. bi i y.

L107

Adelberg7 0 ,7 1 extended this model to incorporate the effects ofaerodynamic forces as well as capillary (surface tension) forces.For an airstream moving parallel to the liquid surface, the aero-dynamic forces operate mostly in a direction normal to the localaverage velocity of the liquid jet, as indicated earlier in con-nection with oscillatory type breakup of jets. The correspondingwind driven waves are referred to as acceleration waves.

The model was further extended by Andersen and coworkers4 6 to

account for primary, secondary and higher order breakup on thefinal droplet size distribution produced by stripping breakupof the liquid. The aerodynamic surface stripping of a liquidproduces particles of various sizes, and the larger particlesundergo further stripping to a stable size during their drag de-celeration. This secondary breakup process also involves momentumbalance between the various size classes. Emphasis here will beplaced on summarizing the model equations for primary breakup;secondary breakup involves the same concepts, so nothing funda-mental is omitted.

The meanings of various symbols used in this model are indicatedin Table V-1.

The mass of liquid stripped aerodynamically from a liquid surfacein a differential wavelength range between N and + d ' per unitarea per unit time is given by

IdX = P Xd (12)

where K2 is a proportionality constant whose value is near unity.The quantity L is the time required for a relative wind to formand break a wave of wavelength X from a liquid surface. It isgiven by

f q1/21 2 (13)

S. .. (+ aX21/2 2'

with

f 2 pu) ,2 ( g (14)

108

TABLE V-I. BREAKUP MODEL NOMENCLATURE

a = A wave acceleration parameter

- -- A, B Wavelength parameters

C Dimensionless constant1

C Constant parameters2

C = Drag coefficient (dimensionless)

d = Liquid droplet diameter

D. = Initial diameter of drop or jet3

e Dimensionless parameter that relates maximum wavelengthto the liquid diameter

E = A wavelength parameter

f = A wind forcing parameter

F = Dimensionless constant that determines the diameter ofa droplet produced by the breaking of a wave with wave-length X

g = Viscous retarding parameter

k = Dimensionless stripping constant

K. = Dimensionless mass stripping constant2

A = Wavelength of a wind-induced wave

X 2 X = Wavelength values over a wavelength increment (X - 1

Also, 1 = Ai, A a1 mn'2 max

X = Maximum wavelength that can be induced by a given windmax

on a liquid surface

X mi n = Minimum wavelength that can be induced by a given wind ona liquid surface

= Mass stripping rate per unit area of liquid (gm/cm - sec)

n = Exponent of viscosity

= Droplet formation rate per unit area of liquid(droplets/cm2 - sec)

t = Time (sec)

t = Liquid breakup time (sec)bu = Air (wind) velocity

v = Liquid velocity

V = Relative velocity between a liquid and an airflow

109

TABLE V-I. BREAKUP MODEL NOMENCLATURE (Continued)

* 3 = Sheltering parameter of a wind-induced wave (dimensionless)~-1)

*i = Shear rate of a liquid while undergoing stripping (sec

n = Shear viscosity of liquid

pg = Gas (air) density

P£ = Liquid density

o = Liquid surface tension

T = Time required for a relative wind to form and break a

wave of wavelength from a liquid surface (sec 1 )

-110

I 22 n

8 i C 2g- (15)

a=C CdP9 2 (16)4r

The drag coefficient Cd is given in terms of the Reynolds number

p VD.Re= g (17)

g

by

-0.84C = 27(Re) for 0.< Re < 80

0.2174= .271(Re) for 80 < Re < 10 (18)

= 2 for 104 < Re.

Integrating expression (12) using (13) gives the was fI;.... M ofliquid droplets (nm/cm 2 - sec) stripped from til%. iiquid surfaceby the aerodynamic growth and breaking of waves with wavelengthsbetween X and X (X > Xi),

1 d 2 ,( 2 1

K2 P£ [ f fE A22= ( 2 (B - A) -a3/2 g in - (19)

where

B + a1/2X + 1/2E =in al2- 2 (20)

A + a1 + 11 1/2

2a J4L

4ob111

and

A = (X1 + aX1 2)1/2 (21)

B = (A + aX22)1/2 (22)2 2

In evaluating expression (19), X1 is to be considered the minimumwavelength wave, Amin, induced on the liquid surface. This isgiven by the smallest positive root of the equation

_ 2 = 2f 2g (23)y 'U (X + aX2 )I/ 2 A2

The maximum wavelength, A is currently given by*-z.. max'

X e .. (24)max 4/3 3

V

The smallest wavelength waves formed by stripping are eroded fromthe liquid surface at a finite shear rate Y. Unfortunately, thevalue of this term is not known a priori, nor is the shear dependentviscosity n(') in the parameter g, for a non-Newtonian liquid. Tocircumvent this situation a critical value of i equal to Tc isevaluated using equation (11) given earlier, i.e.,

;"[~~ \/D 1/2.

.c =t k (25)bV gc b

This permits a solution to be obtained for Amin from expression(23).

By dividing the total wavelength range Amin to Amax into zones,the corresponding mass of droplets in each zone can be obtainedfrom (19). The droplet diameter d corresponding to a wave of

77 wavelength A is given by the simple proportionality

d = FA. (26)

* This expression represents a modification introduced recently to obtain

better aqreement with small jet breakup results.

112

Thus by solution of the preceding equations, the mass-size distri-bution of droplets stripped from the liquid surface can be calcu-lated. Likewise, corresponding expressions exist to enablecalculation of the frequency distribution.

It is apparent that the model is not fully predictive, but is

phenomenological and requires that certain parameters be estab-lished to align predicted results with experimental findings. Onthe basis of previous work with the model, approximate values ofthese parameters are known, so this alignment process is usuallyrather straightforward. The fact that the model is phenomeno-logical in nature greatly simplifies its mathematical details.

Only the most important variables affecting liquid breakup need tobe incorporated in order for the model to be effective in predict-ing breakup results for parameter values a considerable distanceaway from those of the alignment point(s).

A computer coded form of this complete model has been put on-line

(separate from this program) and exploratory runs for Newtonian* liquids have been made to indicate its potential usefulness in

antimisting fuels correlations.

An example of a mass-size distribution obtained experimentally

for hot wax in a hot gas stream by Weiss and Worsham4 5 is shownin Figure V-5. The solid curve is the complete distribution

predicted with the model.

In this example the model was matched to the experimental dataat a mass-median diameter (MMD) of 70 im. The predicted shape ofthe distribution results directly from the built-in uniform

probability density over wavelength for wave formation.

The model was also used to calculate the expected mass-sizedistributions for neat Jet A fuel injected contrastream into afree airflow at different relative air velocities. The parametersadopted in this case apply to the FAA Spray Flammability TestApparatus, but the calculated distributions do not include effectsof spray impaction on the walls of the breakup chamber. Thisimpaction is expected to be a function of droplet size, spray

geometry, and fuel tested. In this case the model was alignedusing a single mass-median diameter calculated using the Weiss and

Worsham expression, equation (10). The calculated MMD's for therelative velocities of 40, 50, 60 and 70 m/sec were 111, 83, 65and 53 lm, respectively. The results are shown in Figure V-6. Thecurves show that the expected mass-size distributions and the

MMD's are both sensitive to relative velocity. Further, there isKi a rather large mass fraction of droplets in small sizes where*ignition sensitivity is the greatest.

F 1131 . ,

140

NEWTONIAN LIQUID

120

100 MODEL

~8oLu i

CL

4-

0

20

1 2 5 10 20 30 40 506070 80 90 95 98 99CUMULATIVE MASS - PERCENT

FIGURE V-5. BREAKUP MODEL PREDICTION VERSUS EXPERIMENTAL DATA

- 114

300

SPRAY FLAMMABILITY200 TEST APPARATUS

~50 m/s

100- 0 m/s

o 870 m/s

~60

u

'" FUEL FLOW: 8 ml/s20

- 1

12 5 10 20 30 40 50 60 70 80 90 95 98 99CUMULATIVE MASS - PERCENT

FIGURE V-6. NEAT JET A DROPLET DISTRIBUTION PREDICTIONS

115

Further calculations were made for neat Jet A in the same configur-

ation to explore the sensitivity of the Spray Flammability TestApparatus to pressure (and hence relative air-to-liquid velocity)decay during the course of single test runs. The basic informa-tion on pitot tube dynamic pressure decays and velocity decaysdetermined experimentally at the FAA Technical Center were suppliedby A. Ferrara. Simplified traces of the dynamic pressure decaysfor three air tank pressures are shown in Figure V-7, together withcorresponding air velocity values determined at the beginning,marked central point, and end of the test runs.

The corresponding mass-size distributions for the beginning, mid-

point, and of each of the test runs are shown in Figures V-8, V-9*and V-10. The relative velocities used in these calculations take

into account the liquid injection velocity of 0.6 m/s. Since theair velocities change only slightly during the short time requiredfor liquid breakup to occur, the instantaneous velocity at eachtime of interest was assumed to be constant for purposes of break-up computations.

The results shown in Figures V-8 through V-10 suggest that the size

distribution changes during a single run may not be very signifi-

cant for neat Jet A. This finding may not apply to tests conducted

with antimisting fuels.

Application of the model in an illustrative sense to a full-scalecrash test has also been made and these results are presentedelsewhere in this report. The reader is cautioned not to placemuch credence in the results of calculations for that case, eventhough secondary breakup is taken into account, since the effectsof Taylor breakup were not included.

The model is used in a final example to account for certain dropletbreakup behavior actually observed in a configuration where spraysof FM-9 fuel were tested. 6 0 ,6 7 While this single example does notconfirm the utility of this model in treating the breakup of a non-Newtonian liquid, we believe it is informative. Application ofthe model to this case has also led to findings which would be verydifficult to ascertain otherwise.

* The tests of interest are those conducted by Eklund6 0 and Eklundand Cox 6 7 in which a small scale test configuration was used toform sprays of 0.3', FM-9 in Jet A. The FM-9 fuel was ejected

4 through a 1/4 inch diameter tube (assumed to be 0.493 cm I.D.)costream into air flows of 200 mph and 285 mph in separate tests.Photographs were taken of the resulting drop sprays approximately11-inches downstream from the liquid injection point.i''

81.3 rn/s 69.5 rn/s 61.6 m/s AIR VELOCITY

"3 GAUGE (TANK)

100 psi9

0 2 4 6 8TIME -SECONDS

64.7 m/s 55.3 rn/s 48.3 rn/s AIR VELOCITY

Wr GAUGE (TANK)

PRESSURE:

TIME -SECONDS

S45.2 m/s 40.2 rn/s 38.7 rn/s AIR VELOCITY

A t GAUGE (TANK)PRESSURE:

50 psig

02 4 6 8TIME -SECONDS

FIGURE V-7. DYNAMIC PRESSURE AND VELOCITY DECAY !NSPRAY FLAMMABILITY TEST APPARATUS

117

9998 95 90 80 70 60 50 40 30 20 10 5 21000 1

SPRAY FLAMMABILITYTEST APPARATUS

RELATIVEAIR VELOCITY

100 - 62.2 m/s70.1 ni/s81.9 m/s

uj

i'--

a_0

0

A P

FUEL FLW:12li/

-J

- O-

- M 1

".--"OPERATING PRESSURE: 100 psiq

FUEL FLOW: 12 mi/s

" '2 5 10 20 30 40 50 60 70 80 90 95 98 99"-" CUMULATIVE MASS - PERCENT

FIGURE V-8. DROPLET DISTRIBUTION DECAY PREDICTIONS

FOR NEAT JET A, 100 PSIG

118L'

-L . .: . ' : -- , . , , . . . . o .. . . . . . .. . . . .... . .. . ... .... ... ... .. . . ... .. . . . . . . . . . .

9 98 95 90 80 70 60 50 40 30 20 10 5 2I I I I I I I 0 I

SPRAY FLAMMABILITYTEST APPARATUS

RELATIVE

AIR VELOCITY

49.0 m/s

55.9 m/s

100 65.3 m/s

.7.J

E-

C±)

i-2- -0

OPERATING PRESSURE: 75 psig

FUEL FLOW: 12 ml/s

2 5 10 20 30 40 50 60 70 80 90 95 98 99CUMULATIVE - PERCENT

FIGURE V-9. DROPLET DISTRIBUTION DECAY PREDICTIONSFOR NEAT JET A, /5 PSIC

119

-4.

100099 98 95 90 80 70 60 50 40 30 20 10 5 2

SPRAY FLAMMABILITYTEST APPARATUS

RELATIVE

AIR VELOCITY

39.3 m/s40.8 m/s45.8 r/s

100!-u* --

, __* . Li

1 0 OPERATING PRESSURE: 50 psig

- FUEL FLOW: 12 ml/s

-I

% .-

* " 1 2 5 10 20 30 40 50 60 70 80 90 95 98 99CUMULATIVE MASS - PERCENT

FIGURE V-10. DROPLET DISTRIBUTION DECAYPREDICTIONS POR NEAT JET A, 50 PS1C

120

For purposes of calculations, the values of parameters assumed andused here are those given in the following table. These variousquantities are defined in Table V-i, given earlier.

ASSUMED PARAMETER VALUES

Temperature (Fuel and Air) T 75°F = 23.9 0 C

-3 3. 1.189 x 10 g/cm.g -6

= 184 x 10 poise

p 3.807 g/cm

=30 dynes/cm

D. = .493 cmJ

C, = 2.0a

The apparent viscosity of the treated fuel is determined using astandard reduction technique2 9 and the shear stress versus shear

strain rate curve through points determined experimentally byMannheimer. 6 3 The resultant shear-strain rate dependent viscosityis shown in Figure V-Il. The lower critical shear strain ratepoint on this curve is taken as 3100 sec-1 (as determined by Peng 6 6 )and the upper or second critical point (determined by Mannheimer 6 3 )is taken as about 8000 sec -1 . For breakup, however, the secondcritical shear rate may lie anywhere in the shear rate range from

the peak of the viscosity curve to the greater rate where the vis-cosity curve again levels out--at about 20,000 sec - I for the curvein Figure V-11. The viscosity in the low shear rate region is takenas 0.0298 poise (from Peng) and the peak viscosity at the secondcritical point is about 0.115 poise. After this critical point theviscosity is 0.050 poise at a shear rate of 20,000 sec-1 . While theresulting viscosity curve of Figure V-il represents a composite of

--" experimental values, it is believed to quantitatively represent themajor shear viscosity features of FM-9 fuel. Its accuracy inrepresenting values of the fuel viscosity that were actually testedby Eklund is unknown. To proceed, it is assumed that this viscosityrepresentation can be applied to Eklund's tests; small discrepancies

r may be attributable to errors in this assumption.

The general nature of the shear strain rate (,) behavior of wind-" induced instabilities of wavelength 1, during stripping breakup of* a Newtonian liquid are shown in Figure V-12. This is a plot of equa-

4 tion (23) for a Newtonian viscosity of 0.0298 poise and for differentvalues of A. Since the time i required for the relative wind to formand break a wave of wavelength X from the surface of the jet is in-versely related to ;, the wavelenqths for which values of : are thelargest break more quickly than near the ]argq, ind sma I 1 wavw, onqt h

121

10,

0

a-

I-

SHEAR

10-2

10 101 102 io0, 1041

STRAIN RATE -SEC-'

FIGURE V-11. SHEAR VISCOSITY OF 0.3-, FM-9 IM JET A

(FROM DATA OF REF. 63)

122

-5 X

* (Ix

w VISCOSITY -cp

00

10 10? 10,

DROP SIZE Jipm

*DROP SIZE IS DIRECTLY PROPORTIONAL TO WAVELENGTH A.

FIGURE V-12. GENERAL SHEAF STRAIN RATE - DROPLET SIZEBEHAVIOR DURING STRIPPING BPEAKUP

6 OF A NEWTONIAN LIOUID

123

extremities of the curve. That is, the central wavelengths breakfirst to form drops. The growth of short wavelenqth waves isretarded by viscous effects, whereas the qrowth of long wavelenqthwaves is slowed by inertial effects.

The line with a scale emanating from the maximum of the curve indi-cates the locus of maxima of the curves corresponding to differentviscosities. Since the shape, scale or orientation of the curvedoes not change appreciably when shifted in the ,-? plane shown,a viscosity scale along the locus of maxima indicates the positionof the maximum of a curve of a given viscosity.

The second line through the maximum of the curve indicates the locusof points of curves corresponding to different relative wind

* velocities. Points along the line for values greater than themaximum of the curve apply to larger wind velocities than that forthe curve shown.

The shear rate-wavelength behavior for the non-Newtonian FM-9 fuel.. involves additional considerations. The curve of Figure V-12 is

based on the assumption that waves of different wavelengths"* grow and break independent of one another. In the case of a

dilatant liquid, in particular, such as FM-9, this independenceassumption needs to be tested. Currently available informationon polymer treated fuels is inadequate to examine this issue,but this topic should be given consideration in future investiga-tions.

The crux of the matter hinges on whether the fastest growing wavesshear the entire surface layer of liquid exposed to the wind andhence increase the viscosity of this entire layer. If this occursthen the shorter wavelength disturbances generated on this surfacemiqht be expected to grow at a rate determined by the peak shear

- - rate induced viscosity. On the long wavelength side of the fastestgrowing wavelength, some intermediate viscosity between that of thesurface layer and that at a greater depth within the liquid maygovern the wave growth.

The other extreme is that independence of qrowth of different wave-lengths is maintained. In this case, the locus of possible shearstates over wavelength can be determined by treatinq each pointbelow the second critical point on the curve of Fiqure V-11 as if itwere on a separate Newtonian viscosity curve at that shear strainrate.

When the wind drivinq force is adequate to shear FM-9 fuel at a ratein excess of the second critical rate, the shear rate-wavelenqth

124

behavior is essentially as if the liquid was near Newtonian. It isapparent, however, that a lower shear rate discontinuity exists atthe value of the second critical point. As the maximum shear rateincreases above this critical value, the peaks of the fixed vis-cosity Newtonian curves follow the upper dashed line. When theshear rate forces breakup to occur in this upper region and thetime available is sufficiently long, all the liquid so sheared willbreak up in that region. That is, part of the liquid does notbreak at some lower shear strain rate below the second criticalvalue, since the breakup time required there exceeds that at thecritical value.

The two shear rate-wavelength curves (expressed in terms of dropsize rather than wavelength by use of relation (26) with F = 0.08)are shown in the upper region in Figure V-13. The lower curve is fora viscosity of 0.0298 poise; the upper (which is viscosity shifted)is a composite curve for a viscosity of 0.050 poise. These curvesare self-consistent with each other, with the viscosity-shear ratevalues of Figure V-l1, and with the breakup model. The location ofthe curves has been chosen so the model results closely representthe lower and upper limits of drop sizes produced in Eklund's tests.The situation obtained is not unique due to uncertainties in thevalues measured from spray photographs. The limits indicated bythe curves and measured from photographs are indicated below:

V = 200 mph V = 285 mphD S iFrom From From Fromi Drop Size Limits

Curves Photos Curves Photos

Lower 86 pm 90 Pm 52 pm 50 pm

Upper 330 pm 350 um 540 pm 500 Pm

As can be seen, the agreement is not perfect, but it is remarkablyclose. It should be noted that this is not a comparison of pre-dicted values with experiment. Rather, it is a match of theory toexperiment in order to account for the experimental results.

There are several features of importance relative to this example.

1. The increase of the large drop size limit with increase ofrelative wind velocity is abnormal. From equations (24) and (26)the model indicates, and experiment confirms, that this upper sizelimit normally decreases with an increase in wind velocity. Here,however, the model itself also accounts for this abnormality; thelatter arises from the discontinuity in liouid viscosity at thesecond critical shear strain rate.

125

V =285 MPH

LI

-104 SECOND CRITICAL

w- SHEAR RATE

zV 200 MPH

FIRST CRITICAL- -SHEAR RATE

102

10 10,DROP SIZE" pm

*DROP SIZE IS DIRECTLY PROPORTIONAL TO WAVELENGTH X.

FIGU RE V-13. COMPUTED BREAKUP OF 0. 3.,. FM-9 TN JET AFOR SMALL JET EXPERIMIENT OF EKLUND AND

COX (REF-. 67)

K 126

2. The model accounts for the breakup of the dilatant FM-9 fuel,with its peaked shear viscosity, using only shear viscosity con-siderations. In particular, the various effects indicated in thisexample and section do not indicate a need to invoke tensile orelongational viscosity to account for the breakup of FM-9 fuel.(Tensile viscosity may still be important to flow and intentionaldegradation aspects of FM-9 fuel, to the breakup of other typesof treated fuels, and to the breakup of FM-9 fuel under prestrainedconditions.)

3. The lower critical shear strain rate in the viscosity-shearrate curve is relatively unimportant insofar as representing alocation where significantly different breakup effects begin tooccur. The second critical shear strain rate, on the other hand,is quite important to breakup. Both actually represent key rheo-logical parameters to be measured for a dilatant-type fuel. Theupper parameter can serve to determine the drop size limits pro-duced during liquid breakup.

4. The shear rate-wavelength behavior in the shear rate regionbelow the second critical point is not well known at present fortreated fuels, in view of uncertainties centering around wavelengthindependence of wave growth. This is a question of both theoreticaland practical interest, as the small size portion of the spray dropspectrum is influenced by this result. This drop size regime, inturn, has a significant influence on spray ignition and flamepropagation. It is expected that breakup phenomena in this regioncan be clarified by use of experimental data on drop size distri-butions, where the associated spray experiments are conducted withthis goal in mind. Such experiments appear to be highly desirablein view of their potential importance to fuel spray combustion.

A final consideration before closing this example is how this ex-planation fits with the expected results obtained with other percent-age loadings of FM-9 in Jet A fuel. Peng 4 6 has determined that thelower critical shear rate decreases with concentration increaseaccording to the sequence: 6000 S- 1 , 3100 S- 1, 2000 S-1 and 700 s-I,corresponding to the respective concentrations 0.2%, 0.3%, 0.4%, and0.8% of FM-9 with carrier fluid in Jet A (at 75*F). While values ofthe second critical shear rate are not known to have been measuredas a function of concentration, it is expected that these valuesshould increase as the concentration increases. Then, for a givenair-to-liquid driving force, a sequence of increasing value of thecritical shear rate would cut the shear rate-wavelength (or dropsize) curve at successively higher values. The corresponding dropletsize spectrum expected for breakup in the upper shear region wouldnarrow, and the overall fire resistance of the fuel should improve.

127

This trend of improved fire resistance with increased value of thesecond critical shear rate has actually been observed and reportedby Mannheimer6 3 . He found that by introducing a suitable surfac-tant, both the lower and upper critical shear rate increased. Theignition resistance of the surfactant treated material was signi-ficantly improved over the 0.3% FM-9 treated Jet A without surfac-tant. The fact that the lower critical shear rate increasedapparently had little effect--as one might now expect.

Further calculations for the purpose of data correlation of anti-misting fuels were not made during this effort. There are severalreasons for this. First, the computer model does not presentlycontain the physical model for Taylor breakup. This omission doesnot influence the results noted above for small jets as the Taylorbreakup mechanism is not operative in these cases. Second, theexisting data base of information on antimisting fuels appearsinadequate regarding drop-size spectra produced in representativetests and weak in consistent rheological properties of the candi-date antimisting fuels of interest. In absence of the latter, theadequacy of the model to account for the rheological behavior ofvarious candidate antimisting fuels is not well known. The com-bination of data indicated could be used to ensure sufficiency ofthe model in terms of both rheology and breakup.

i12

REFERENCES

1. E. Mayer, "Theory of Liquid Atomization in High Velocity GasStreams," ARS Journal, 31, pages 1783-1785 (1961).

2. K. J. DeJuhasz, Spray Literature Abstracts, Volumes I-IV,American Soc. Mechanical Engineers (1959-1969).

3. C. E. Lapple, J.P. Henry, and D. E. Blake, Atomization--ASurvey and Critique of the Literature, Stanford Research InstituteTechnical Report SRI-PAU-4900 on Contract DA-18-035-AMC-122(A)Menlo Park, CA (April 1967).

4. C.C. Miesse, "Correlation of Experimental Data on the Dis-integration of Liquid Jets," Ind. and Engr. Chem., 47(9), pages1690-1701 (1955).

5. C. C. Miesse, "From Liquid Stream to Vapor Trail," Proc. Gas.Dynamics Symposium at Northwestern University, Evanston, IL, pages7-26 (1956).

6. Lord Rayleigh, "On the Instability of Jets," Proc. LondonMath. Society, 10, pages 4-13 (1878); "On the Instability of aCylinder of Viscous Liquid Under Capillary Force," Phil. Mag.,34, pages 145-154 (1982).

7. J.B. Keller. S.I. Rubinow, and Y.O. Tu, "Spatial Instabilityof a Jet," Phys. Fluids, 16(12), pages 2052-2055 (1973).

8. R. Betchov and W. 0. Criminale, Stability of Parallel Flows,Academic Press, New York (1967).

9. C. Weber, Z. Angew, "On the Breakdown of a Fluid Jet," EnglishTranslation, Math. Mech., 11(2), pages 136-154 (1931), Ninth Pro-gress Report, Project MX-833, Section II, University of Colorado,Boulder, CO.

10. R. P. Grant and S. Middleman, "Newtonian Jet Stability,"AIChE Journal, 12(4) pages 669-678 (1966).

11. R. W. Fenn, III and S. Middleman, "Newtonian Jet Stability:The Role of Air Resistance," AIChE Journal, 15(3), pages 379-383(1969).

12. F. W. Kroesser and S. Middleman, "Viscoelastic Jet Stability,"AIChE Journal, 15(3), pages 383-386 (1969).

129

13. R. E. Phinney, "Stability of a Laminar Viscous Jet--The*Influence of the Initial Disturbance Level," AIChE Journal, 18(2),

pages 432-434 (1972).

14. M. Goldin, J. Yerushalmi, R. Pfeffer, and R. Shinnar, "Breakupof a Laminar Capillary Jet of a Viscoelastic Fluid," J. Fluid Mech.,38(4), pages 689-711 (1969).

15. M. Gordon, J. Yerushalmi, and R. Shinnar, "Instability ofJets of Non-Newtonian Fluids," Trans. Soc. Rheol., 17(2), pages303-324 (1973).

16. A. S. Lodge, Elastic Liquids, Academic Press, New York, page117 (1964).

17. K. Walters, Rheometry, Chapman and Hall, London, page 219(1975).

18. W. R. Schowalter, Mechanics of Non-Newtonian Fluids, PergammonPress, New York, page 220 (1978).

19. L. M. Milne-Thomson, Theoretical Hydrodynamics, 3rd Edition,MacMillan Co., New York (1955).

20. B. Kinsman, Wind Waves, Prentice-Hall, Englewood Cliffs, NJ(1965).

21. H. B. Squire, "Investigation of the Instability of a MovingFilm," Brit. J. Appl. Phys., 4, pages 167-169 (1953).

22. W. W. Hagerty and J. F. Shea, "A Study of the Stability of* Plane Fluid Sheets," J. App. Mech., 22, pages 509-514 (1955).

23. J. L. York, H. E. Stubbs, and M. R. Tek, "The Mechanism ofDisintegration of Liquid Sheets," Trans. ASME, 75, pages 1279-1286 (1953).

24. N. Dombrowski and W. R. Johns, "The Aerodynamic Instabilityand Disintegration of Viscous Liquid Sheets," Chem. Eng. Sci., 18,

* pages 203-214 and 470 (1963).

25. N. Dombrowski, P. Eisenklam, and R. P. Fraser, "Flow andDisintegration of Thin Sheets of Viscoelastic Fuels," J. Inst. Fuel,1, pages 399-406 (1957).

26. R. E. Ford and C. G. L. Furmidge, "The Formation of Dropsfrom Viscous Water-in-Oil Emulsions Sprayed Through Fan-Jet Noz-zles," Brit. J. Appl. Phys., 18, pages 491-501 (1967).

130

27. L. E. Nielsen, Polymer Rheology, Dekker, New York (1979).

28. R. P. Fraser, "Liquid Atomization," J.-Roy. Aero. Soc., 65pages 749-755 (1961).

29. A. H. P. Skelland, Non-Newtonian Flow and Heat Transfer, Wiley,New York (1967).

30. A. C. Merrington and E. G. Richardson, "The Breakup ofLiquid Jets," Proc. Phys. Soc., London, 59(1), pages 1-13 (1947).

31. J. 0. Hinze, "Forced Deformations of Viscous Liquid Globules,"Appl. Sci. Res., Al, pages 263-270 (1949).

32. J. 0. Hinze, "Critical Speeds and Sizes of Liquid Globules,"Appl. Sci. Res., Al, pages 273-288 (1949).

33. Lord Rayleigh, Theory of Sound, Volume II, Dover, New York,page 373 (1945).

34. W. H. Andersen and H. E. Wolfe, Kinetics, Mechanism, andResultant Droplet Sizes of the Aerodynamic Breakup of LiquidDrops, Report No. 0395-04(18)SP, Aerojet General Corp., Downey,CA (April 1964).

35. R. A. Castleman, Jr., The Mechanism of Atomization Accompany-ing Solid Injection, NACA Report No. 440 (1932).

36. W. R. Lane, "Shatter of Drops in Streams of Air," Ind. Eng.Chem., 43(6), pages 1312-1317 (1951).

37. G. D. Gordon, "Mechanism and Speed at Breakup of Drops,"J. Appl. Phys., 30, pages 1759-1761 (1959).

38. A. R. Hanson, F. G. Domich, and H. S. Adams, "Shock TubeInvestigation of the Breakup of Drops by Air Blasts," Pys. Fluids,6, pages 1070-1080 (1963).

39. 0. G. Engel, "Fragmentation of Water Drops in the Zone Behind* an Air Shock," J. Res. Nat. Bur. Standards, RP 2843, 60(3), pages

245-280 (1958).

40. A. A. Ranger and J. A. Nicholls, "Aerodynamic Shattering ofLiquid Drops," AIAA Journ., 7, pages 285-290 (1969).

41. B. J. Clark, Breakup of a Liquid Jet in a Transverse Flowof Gas, NASA Technical Note TN-D-2424 (August 1964).

131

:.

42. G. Morrell, Rate of Liquid Jet _rea .y a TransverseShock Wave, NASA Technical Note D-1728 (May 1963).

43. G. Morrell and F. P. Povinelli, Breaku 2 of Various LiquidJets by Shock Waves and Applications to Resonant Combustion,NASA Technical Note TN9-2423 (August 1964).

44. A. Sherman and J. Schetz, "Breakup of Liquid Sheets and Jetsin a Supersonic Airstream," AIAA Journ., 9(4), pages 666-673 (1971).

45. M. A. Weiss and C. H. Worsham, "Atomization in High VelocityAirstreams," ARS Journal, 29, pages 252-259 (1959).

46. W. H. Andersen, N. A. Louie, and G. Ialongo, Investigation* of the Aerodynamic Breakup of Viscoelastic Liquids. Phase I--

Subsonic Dissemination, Contractor Report ARSCSL-CR-77037 (August1977).

47. S. Chandrasekhar, Hydrodynamic and Hydromagnetic Stability,Clarendon Press, Oxford (1961).

48. G. I. Taylor, "The Instability of Liquid Surfaces When Acceler-ated in a Direction Perpendicular to Their Planes," Proc. Roy. Soc.,A201, pages 192-192 (1950).

49. D. J. Lewis, "The Instability of Liquid Surfaces When Acceler-ated in a Direction Perpendicular to Their Planes. II," Proc. Roy.Soc., A202, pages 81-96 (1950).

50. J. B. Keller and I. Kolodner, "Instability of Liquid Surfacesand the Formation of Drops," J. Appl. Phys., 25(7), pages 918-921

(1954).

51. R. Bellman and R. H. Pennington, "Effects of Surface Tensionand Viscosity on Taylor Instability," Quart. J.APE1. Math., 12(2),pages 151-162 (1954).

52. W. G. Reinecke and G. D. Waldman, A Study of Drop BreakupBehind Strong Shocks With Applications to Flight, SAMSO-TR-70-142, AVCO Systems Division (May 1970).

53. P. G. Simpkins and E. L. Bales, "Water-Drop Response to

Sudden Accelerations," J. Fluid Mech., 55(4), pages 629-639,(1972).

54. E. Y. Harper, G. W. Grube, and I-Dee Chang, "On the Breakupof Accelerating Liquid Drops," J. Fluid Mech., 52(3), pages 565-591 (1972).

*; 132

55. S. V. Zinn, Jr., T. I. Eklund. and W. E. Neese, PhotoqraphicInvestigations of Modified Fuel Breakup and Inition, Report No.FAA-RD-76-109, NAFEC, Atlantic City, New Jersey, September 1976.

56. R. J. Mannheimer, Rheology Study of Antimist Fuels, ReportNo. FAA-RD-77-10, Southwest Research Institute for FAA SystemsResearch and Development Service, January 1977.

57. A. F. Taylor, Notes on the Flow of Viscoelastic Fuels ThroughPipes and Orifices, College of Aeronautics, Cranfield Institute ofTechnology, January 1973.

58. S. T. J. Peng and R. F. Landel, "Preliminary Investigationof Elongational Flow of Dilute Polymer Solutions," J. Appl. Phys.,47(10), pages 4255-4259 (1976); and Errata, J. Appl. Phys., 50(7),page 5883 (1979).

59. R. Coomber, I. Faul, and S. P. Wilford, Measurement of theExtensional Viscosity of Safety Fuels by Jet-Thrust and Triple-Jet Techniques, RAE, Technical Report No. 77157, October 1977.

60. T. I. Eklund, Experimental Scaling of Modified Fuel Breakup,Report No. FAA-RD-77-114, NAFEC, Atlantic City, New Jersey, August1977.

61. T. I. Eklund and W. E. Neese, Design of an Apparatus forTesting the Flammability of Fuel Sprays, Report No. FAA-RD-78-54,NAFEC, Atlantic City, New Jersey, May 1978.

62. I. Faul, An Investigation Into Jet Droplet and Spray Behaviorof Safety Fuels, RAE Technical Report No. 78131, November 1978.

63. R. J. Mannheimer, Influence of Surfactants on the FuelHandling and Fire Safety Properties of Antimisting Kerosene (AMK),Southwest Research Institute for FAA Systems Research and Develop-ment Service, February 1979. (FAA-RD-79-62)

64. V. Sarohia and R. F. Landel, Influence of Antimisting Polymerin Aviation Fuel BreakuE, Jet Propulsion Laboratory, AIAA/SAE/ASME,

* 16th Joint Propulsion Conference, June 30-July 2, 1980.

65. S. P. Wilford supplied recent data on elongational viscositiesof FM-9 in Avtur prior to publication.

* 66. S. T. J. Peng, Rheological Behavior of AMK, Jet PropulsionLaboratory, Presentation, March 24-26, 1980.

133

67. T. I. Eklund and J. C. Cox, Flame Propagation Through Sprays

of Antimisting Fuels, Report No. NA-78-66-LR, NAFEC, Atlantic

City, New Jersey, November 1978.

68. J. W. Hoyt, J. J. Taylor, and R. L. Altman, "Drag Reduction--Jet Breakup Correlation with Kerosene-Based Additives," Journal ofRheology, 24(5), pages 685-699 (1980).

69. J. T. Davies, Turbulence Phenomena, Academic Press, New York,Chapter 7, 1972.

70. M. Adelberg, "Breakup Rate and Penetration of a Liquid Jetin a Gas Stream," AIAA Journ. 5, pages 1408-1415 (1967).

71. M. Adelberg, "Mean Drop Size Resulting from the Injection ofa Liquid Jet into a High Speed Gas Stream," AIAA Journ. 6, pages1143-1147 (1968).

.13

.1?

1344 . .. . : . . ' , , , . . .

VI. CONCLUSIONS

1. The ability of certain antimisting fuels (AMF's) to minimizethe post-crash fire hazard is amply supported by accumulated test

*data. Full- and large-scale crash simulations are highly developed,and provide essential credibility and data on a given AMF near theend of its evaluation cycle. However, the unavoidable complexityand expense of large-scale testing precludes its use in the screen-ing of all _AMF formulations and their iterations.

2. Small-scale flammability test rigs developed for screeningAMF's have advantages of simplicity, high data production rate, andrelatively low initial and operating costs. The data base each testrig generates is generally self-consistent and repeatable.

3. Results from the various small-scale flammability test rigsgenerally agree where there are substantial differences betweenfuels; e.g., when comparing (1) neat Jet A with an AMF, or (2) amarginal AMF with a very effective one. This is to be expected,since, in the development of a given fire test rig, intuitive as-pects of the design are often adjusted based on the performance ofneat fuel and "known-effective" AMF's--in effect, using these fuelsas informal standards. The assumption is that wake flames orfireballs larger than those experienced with the "standard" AMFindicate a candidate that will be inferior in a real-world crash(or large-scale test).

4. Small-scale flammability test rigs have given conflicting re-sults when comparing two AMF's lacking a wide differentiation inperformance--one reversing the relative ranking of the other.Similar conflict between a small-scale and a large-scale flamma-bility rig has also been encountered--undermining the credibilityof the small-scale rig involved. It is not economically feasibleto perform large-scale flammability tests every time two small-scale rigs conflict. The conclusion is evident that improvementsin the resolution and consistency of small-scale flammability testrigs are highly desirable to (1) support optimization of AMFformulations, (2) confidently and consistently evaluate AMF perfor-mance versus such factors as additive concentration, storaqe timeand degradation, and (3) minimize the need for large-scale testingfor ranking AMF's.

5. The basic problem in developing small-scale flammability test14 rigs for AMF is the lack of design criteria related to liquid

breakup phenomenology and scaling. It is only recently that pre-liminary understandings and models of liquid breakup have emerged.It is possible that a "dynamic bias" may inadvertently exist in a

135

given small-scale test rig; i.e., its peculiar combination of flow,ejection, and/or breakup dynamics may affect some AMF candidatesmore adversely than others--compared to their large-scale rankings.

6. Various AMF's have significantly different breakup character-istics; e.g., relative sensitivity to shear versus elongationalstrains and stresses. In addition, sensitivity to flow and pres-sure history prior to ejection can vary markedly between AMF's.It is concluded that each AMF must be thoroughly examined as to i.ts

" ~rheological behavior under the flow and breakup conditions imposed- by the flammability test rigs.

7. Based on the work presented in this report, it is now feasibleand most effective to exploit current liquid breakup technology

*- and appropriate particle measuring instrumentation in order to (a)

quantitatively characterize the liquid breakup products of flam-mability test rigs of interest, and (b) relate the dynamic physicalcharacteristics of a given test rig to the phenomena and productsof liquid breakup. It is concluded that a semi-empirical modelwhich addresses both large- and small-scale liquid AMF breakupwould facilitate individual rig analysis and collective correlationof selected test rigs.

8. On the basis of successful liquid breakup characterization,combustion technology and related semi-empirical models can beutilized to quantitatively relate the combustion responses of small-scale flammability test rigs and their related pass/fail criteria tolarge-scale fireball evaluations and criteria.

1.3

,136

N44

24''_f"7~

W w.

I j?.~

'' J

'7.