IIS

RI."
AFOR
Final Report

TP 72-1'716

" ..

June 30, 1972

Co
INVESTIGATION OF LOW-VELOCITY DETONATION PHENOMENA IN LIQUID PROPELLANTS, FUELS, AND EXPLOSIVES

By:

D. C. WOOTEN, M COWPERTHWAITE, and D. C. ERLICH

Prepared for: DIRECTOR OF AEROMECHANICS AND ENERGETICS AIR FORCE OFFICE OF SCIENTIFIC RESEARCH ARLINGTON, VIRGINIA 22209 Attention: NAE

CONTRACT F44620-69-C-0079

SRI Project SCU-7771

Approved by:
G R ABRAHAMSON, Director
Poulter Ldboratory
C J COOK, Executive Director Physical Sciences Division
,

D C
-I

~fJI~1f
-SEP1,17

NATIONAL TECHNICAL
INFOPMATION SERVICF

Aoq6

toom"l

f.NAe

~rit,

VOOI.;MEN- CUNMPO
CI& .A e
ti flr

i (d~ lift. lhAA

h f

~'nd

D~A'A IiA r-f~ b'

F. D
,* *AJIC'AI

.C(*

'~

AI'.

)~

'

.1I

'...... I .

i ''.I

MENLO PARK), C!AIJFORN)\1A

ITAVF0M) RE~SEARCH I NS'lUI 94025

~j

11. LAS f !"D

jJ.14VESTX.ATION OF LOWv-VLTLO'CI'Y J)LTONATIC)N PWhNO%,11liA R' LIQ~UID i'R0P];LMANITS,

D~ 5CAI ri' Ti Iv

I'1L

OiZ.(7'y,Ye V., tepCoIf 4,1 ' ilcyi

d whzu)

Scientific

Final,_______

~'DA~VID C WNOUSJE11
S30 June 1972
-

MICHAEL. COw1qPidT11WAIT.!:.

DAVID) C ERLICH
7t., 1or A1. 11-. or PA.GE.f.

3,__ 16
W,. or.IGINAToR'S 0,E Po;4 arls

L88
.ht

7b.A4C.. of

M~.1S,
__

(.0141 VAC I 01 C;(At.T NO

F44620-69-C-0079
b. PRO9711-AlO

SRI. Project SCU-7771I 61102F

this repoat) o()(-u ubrld Luiud
__SIR

i.

d.

6681308
ID$ iiUkTl01f STATFMC.14Tj

F________

r C, tI~

SApprovc'cI for p)ublic: r *Ionase; di.

t$ i=1 T.jr'FI.M-L7(-1t77 7T/. 7~11

vb'inuiiitd,
12, WION'AINtG PAIL.l AflY AC11V11

TCOTHER

f
___________

AF Office of Scientific Research (NAE)

11400 WilIson Boulevard

Ar lington, vriginia 22209

Theoretica' and eprncalstudies were peLrformcd to obtain fundament~3. under.s Landill! Of initiaitiOn cv~cJ Prop1agationo of 3.ow-volocity detonation (LVD) in confinc.d enereeot LMI was modeled as a re-active shock prolm.gati on in a liquid cavit~,tr(d by li~qui cis inLIterActions With thn cou'tai..cr1. Will. RiCct~ion required to support. the wave ill LhIis cavitation modol b, al tiated i~y bubble col.Japso ill thle compressud cavitation field. 2..*;) com-putter stut12i~cs of a shock -propagatini. in a contaiiied liquid ShLowed je~:o wall ave:s produce rcg4,,ons of Uenuson suffi cient sxn cavitrlte Ohe liquid hioeo-- it is 1 by thc, miain shock. Simumlationr of hCwt-sipoL-initiatodc reacti.on lby addri~tion~ of cnarr-v 'chind tile ,;Iick substwiti~atod the hypct.hesis of partleal. energy release in ydtJrorans LIND. 'Met hods of c;-Jculating IAYD paramietersI waxe forrautlateL(d with the ,of Ore Cavl tat-ion modal, and .1 completu. equation of state for nitroniethance war, introducEd into time TJICTR code to permit Jrktonc-tion calculations for incomplete react-icn. Calc~ulations -111LC abo-t. 20 p-r(.enL Of liqu~id is, required to support LVD. D~ynanmic!, of Goi 1apin-jl \'dJr bubbbl.s wete anilyzed to st*udy lio.t-sp,,ot reaction intirlt-Ion arid cooidi Lions fl), LVDT initiation. Initial ra-;:e of tempurature increase in Lbheat ano mmss transfer was founid to be 30 to 20 times less than in bcbbles .:tl oIbtaitmd 1in bubblles withiout tran: fer -aL the 1)ul~ble wall, PreSSUre-tiMe hSO:istori iol.1-irtion of LVD , aoid thv~111 ~ coniminud tihoc'ked imonoripx up'lcnts shoe shock ;t e.Aiyl nilt-rl to and 3.0 rom:/i1sec in 1"Cl70. .1 j yaIoci tic" were aboual. 2 .0

jcompressed

F1) uri473
)M

__147

~~

~icrt'i ~ ~
.',.011ci;

C:

MONANOPR Oil"LLANTS
LOW VELOC'ITY DETONATION

(CAVItATION
1RrACTIVE WIAVE PROPAGATION

EQUATION OF STATE BIUGCONlOT CURVEIs

])EPLACRAT1ION INCO>1PLEDIl' REACTION

BUBB)LE 1)YIAMI.CS -HEAT1 -AND ?-ASS~ TRANSFER SHOCK. liPNITIATION YTTIREiUUU PRESS;URE STRESS G)

UNLSSFE
S(cIt .a--fcto

ABSTRACT

Theoretical and experimental studies were p'i'formed to obtain a fundamental understanding of the initiation an' detonation (LVD) in confined energetic liquids. pi-opagation of low-velocity LVD was modeled as a .:i interactions with the by

reactive shock propagating in a liquid cavita container wall. tation model is field.

The reaction required to st,. :kt the wave in this caviinitiated by bubble collapse in the compressed cavitation

Two-dimensional computer studies of a shock propagating in a liquid contained in a cylindrical steel tube showed that precursor wall waves produce regions of tension sufficient t6 cavitate the liquid before it compressed by the main shock. is

Simulatioai of hot-spot-initiated reaction

by addition of energy behind the shock substantiated the hypothesis of partial energy release in LVD. Methods of calculating LVD parameters

were formulated with the hydrothermodynamics of the cavitation model, and a complete equation of state for nitromethane was introduced into the TIGER code so that detonation calcii.ations could be made for incomplete reaction. liquid is The calculations showed that only about 20 percent of the required to react to support LVD. The dynamics of a collapsing

vapor bubble were analyzed to study the hot-spot initiation of reaction and the conditions for initiation of LVI). The initial rate of temperature

increase in a bubble with heat and mass transfer was shown in general to be 10 to 20 times less than that in a bubble without transfer at the wall. Experimental studies resulted in measurements of pressure in LVD. Pressure-time histories in shocked monopropellants contained in cylindrical

steel tubes were recorded by piezoresistant ytterbium in-material stress gages. Gage records obtained by varying the monopropellant, the wall

iii

thickness-to-diameter ratio of the tube, and the initial shock pressure showed shock decay, velocity detonation. 4- to 7-kbar range, initiation of LVD, and the initiation of high-

The peak pressures recorded by LVD were in the and the LVD velocities of propagation were about

.2.0 mm/Rsec in ethyl nitrate and 3.0 mm/Rsec in FEFO.

iv

I II III

INTR6DUCTION

...

....

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

. .

Page 1 3

ASSESSMENT OF PAST WORK LVD COMPUtR MODEL-

WALL INTERACTION P.RODUCING CAVITATION

.. .....

........

IV -HYDRQTHE RMDYNAMICS OF' THE CAVITATION MODEL OFLVD. XV BUBBLE _DYNAMICS . ..........

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

...

..... ..
.

8 11
12
.14

XTV -EXPERIMENTAL RESULTS -VII

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

BtBLIOGPRAPHY.

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

-REFERENCES-

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

15

APPENDICES ,A THE WAVE STRUCTURE IN SHUCKED, CONFINED LIQUIDS ..... ASSOCIATED-WITH LOW;-WVELOCITY ,DETONATION HYDROTHERMPODYNAMICS OF THE CAVITATION MODEL uF LVD A "UNIVERSAL" HUGONIOT FOR LIQUIDS .... .......
.

A( i)

B C D E.

B( i) .
.

C(i) D(i) E(i)

BUBBLE COLLAPSE LEADING TO HOT-SPOT INITIATION

MEASUREEIT OF PRESSURE IN LOW-VELOCITY DETONATION

=4<4

S4

INTRODUCTION

This report presents the results-of -a three-year theoretical and experimental program whose objective was to obtain -a- fundamental understanding of low-velocity detonation (LVD) fuels, and explosives. energetic liquids. It phenomena -in liquid propellants, highly

Low-velocity detonation occurs in confined,

can be initiated by-much smaller shock impulses

than can high-velocity detonation (HD)),- but can cause considerable damage. Present hydrothermodynamic detonation theory predicts the exis-

tence of only HVD for any particular thermodynamic state of the explosive. The existence of LVD therefore represents an anbmaly heretofore unexplained -indetonation theory. A basic understanding of LVD is of great practical

significance in establishing safety requirements for the handling of liquid fuels, propellants, and explosives.

The following tasks were-accomplished during the research program: (1) A complete literature review and a reevaluation ,of existing LVD data were accomplished to identify the individual mechaThe nisms contributing to LVD initiation and propagation. experimentally observed features of LVD were identified and used to evaluate the various models of LVD that have beenproposed. A model that offered the most plausible explanation of the phenomenon and that was consistent with the experimenta.: observation was adopted as a starting point for the subsequent "work. This model was called the cavitation model of LVD. A preliminary, theoretical analysis of the interaction of the confinement system with the liquid explosive was carried out. The studies established that precursor wall waves capable of causing cavitation in the liquid ahead of the main liquid shock are possible. The physical- and chemical properties of liquids known to undergo LVD, as well as those in which LVD was not observed, were The resuits of thfis task suggested the significance examined. of cavitation in the LV. phenomenon.

(2)

(3)

(4)

Detailed computer calculations utilizing the cavitation model of LVD were carried out to study the cavitation production mechanism, the effects of partial energy release, the initiation ofLVD, and the possible transition of LVD to HVD.

(5)

A hydrothermodynamic treatment of the cavitation model of LVD was undertaken in an attempt to develop a satisfactory method for calculating LID propagation velocity. Interaction of the shock and the .cavitation field was accounted for simply by ordering the time scales for the attainment of mechanical, thermal, and chemical equilibrium. Since results, based on the attainment of chemical equilibrium did not agre6 with experiment, the case of incomplete reaci.-on was considered.

(6)

Characteristic times for bubble c6lialpse,

heat transfer within

a bubble, and heat transfer to the surrounding liquid were compared to assess the relative rates of possible processes occurring in the compressed cavitation field. The initial collapse of a vapor bubble was treated analytically in an attempt to obtain a more detailed account of bubble dynamics and to determine conditions for the onset of reaction and the initiation of LVD. (7) Experiments using metriol trinitrate (TMETN) were carried out to examine the effects of confinement upon LVD initiation and propagation. (8) Experiments were carried out to measure the stress field ahead of and during the passage of the liquid shock. Detonation velocity measurements as well as damage criteria were used to verify the observed LVD phenomenon. As a result of the above studies, been developed that is ties of LVD. Further, a quantitative model of LVD has

consistent with the experimentally observed properexperimental verification of some of the more

important quantitative predictions of the model has been carried out. The results of this work, journals, are presented in intended for publication in technical

the manuscripts attached as Appendices to of references of the work performed given in the Bibliography.

this report, and a complete list

during the present research program is

!/.

Il
II ASSESSMENT OF PAST WORK

For most practical applications, condensed explosives are required

to behave reproducibly, to undergo HVD, and to produce pressures in the

lO0-kbar region.

As a result, investigations of detonation phenomena

Consequently, at the beginning of this

have been focused largely on HVD--rather thtan on LVD. which produces

pressures in t*4e lO-kbar-region.

research program, the conditions- for initiation and propagation of LVD were not as well established as, were those for initiation and propagation of HVD. This disparity in our understanding was accentuated by the fact that previous experimental studies of LVD had produced only limited results, generally of a qualitative nature. For instance, it was

clear that a reactive wave traveling at -about sonic speed in virgin explosive was associated with stable LVD, that LVD could.be initiated more easily than HVD, and that initiation depended on the properties and geometry of the container holding the explosive.
1-3* model for HID. Neither were there criteria for calculating detona-

There was, however, no

physical model for LVD corresponding to the Zeldovich-vonNeumann-Doering

tion velocity from the properties of the explosives corresponding to the Chapman-Jouguet (C-J) hypothesis nor a satisfactory model for treating
the coupling between the container walls and the explosive in the LVD

process.
During the initial stages of this program, a comprehensive survey and evaluation was made of existing experimental data and theoretical

models pertaining to LVD.

The results of this study are given in detail

References -are listed at the end of the report.

in the first annual report.

*

Although rather extensive literature on LVD

was known to exist, it

was recognized at the outset that most past obserAmong the U.S., European, there

vations had been scattered and uncoordinated.

and Soviet investigators who had previously studied LVD phenomena, was little

agreement concerning the validity of much of the-reported work.

Moreover, there had been no very thorough attempts to resolve the conflicts,. Accordingly, the program began with an effort to delineate the really important factors contributing to LVD and to discard those concepts and measurements that have tended to confuse the issue. The objective of

this initial study was to establish the direction to be taken in planning and executing the rest of the investigation. Each of the surprisingly numerous existing conceptual models of LVD initiation-and propagation was first examined against the background of accumulated data. Many concepts were thus shown to be highly improbable The remaining concepts, which seemed

and unworthy of further study.

reasonably consistent with experimental information, were subjected to fairly simple analyses as a further test of their validity and relevance

'*1

for subsequent research in this program. Apart from quantitative disagreements with available data, most models failed to explain certain important and experimentally wellestablished physical characteristics of LVD, of container geometry and material (strength) stability of LVD waves. such as the marked influence on the initiation and

Two conceptual models that had been proposed

seemed qualitatively consistent with these behavior characteristics and therefore deserved careful consideration.

r1

Amster et al.

suggested the possibility of hot-spot initiation

i nduced by a Mach disk (local normal shock) occurring at the center of

the containeT when precursor shocks generated by container wall deflections

converge along the centerline.

Although this model affords a plausible a recent analysis at

explanation of certain behavioral chaiacteristics, Stanford Research Institute (SRI)

indicates that the Mach disk generally

is not strong enough to create the local temperatures needed for initiation. The hot-spot initiation concept was retained in a model first suggested 6 by Watson et al. Their model, which qualitatively fits existing experimental observations, envisioned the appearance of cavitation in the liquid, followed by a shock wave that collapses the bubbles, causing 7 local hot spots that act as initiation centers. Watson had also taken photographs showing cavitation ahead of the reaction zone, and Watson 8 et al. had more recently. reported measurements of elastic precursor wall waves that further substantiate the cavitation model of LVD. Based 4,9 apon the work done at SRI under this program and in light of the properties of LVD observed to date by others, the cavitation model appeared to offi.r the most plausible mechanism for the propagation of LVD. The cavitation model for LVD assumes that container interaction with
-

the energetic liquid produces cavitation ahead of a shock in the liquid. The liquid shock subsequently compresses the bubbles, causing initiation and subsequent burning. The approach taken in this program to develop a

better theoretical understanding of the LVD phenomenon has been to examine

the mechanisms by which cavitation is produced in the liquid and theo-

retically to model the reaction zone in an attempt to predict observed LVD behavior. This theoretical work was supplemented by experiments In the following, the

designed to test the analytical model critically. work on the cavitation production mechanism, LVD,

the theoretical modeling of

and the experimental results are summarized.

, k5

4 . V

III

LVD COMPUTER MODEL

WALL INTERACTION PRODUCING CAVITATION

The validity of the cavitation model of LVD requires that a mechanism exists for the production of cavitation in wave. As a part of the current program, the liquid ahead of the LVD

an analysis of the liquid-wall

interaction was performed to determine whether significant cavitation should indeed uccur. In a simplified preliminary analysis, it was found

that precursor shocks in

the container walls can cause a container wall-

liquid interaction capable of producing a substantial degree of cavitation

4
in the liquid. This analysis was carried further by a computer calcushock code to calculate the stresses steel pipe or tube when the liquid lation using SRI's two-dimensional produced in a liquid contained shocked at one end. across its

in a cylindrical

Large tensile stresses were produced in

entire cross section by the wall interaction.

The results

show that large degrees of cavitation can be produced in the liquid. This cavitation and the resulting possibility of hot-spot initiations are very important, not only for LVD, but for the initiation of HVD in

liquid (and even some solid) explosives. The calculations of stress wave propagation down a liquid-filled cylindrical tube, computer code. impacted at one end, were made using SRI's FIBROUS finite-difference code that can hancle both elastic-

FIBROUS is

a two-dimensional

modeled after the description of Wilkins,

10

plastic and purely hydrodynamic constitutive relations. To study further the cavitation model of LVD, the stress history was

computed for a liquid explosive contained in a cylindrical steel container subsequent to its being impacted at one end. The results of the compu-

tations show that coupling between the container wall and the liquid produces precursor tension zones in the liquid leading to cavitation

ahead of the main liquid shock and that recompression of the cavitated zones by the liquid shock is sufficient to initiate reaction to support

the propagation of the liquid shock. in a 1-inch I.D.

The simulation for ethyl nitrate

by 1/8-inch wall thickness steel pipe resulted in 2.4, and 3.0 mm/fisec, respectively,

detonation velocities of about 1.8, for energy releases of Q

,,

0.4, and 1.0 times the heat of reaction. this was taken as

Since experimental LVD propagates at about 2 mm/Lsec, evidence of partial energy rel.ease in LVD.

The results of this work

showed the details of the cavitation production and recompression and further substantiated the cavitation model of LVD. The computer model of LVD provides a menns for studies of initiation .Ad propagation of LVD in various confinement systems and, in addition,

has provided valuable information about the initiation and propagation of LVD. in The results of the computer modeling of LVD are fully described

Appendix A.

IV

HYDROTHERMODYNAMICS OF TilE CAVITATION MODEL OF LVD

Our cavitation model for LVD is cavitated liquid. Propagation is

a reactive shock propagating in a

thus a very complex phenomenon involving subsequent behavior under a theoretical treatment Such calcuand tile

the formation of a cavitation field and its shock compression. of LVD is

The most important problem in

that of calculating the velocity of propagation.

lations must take account of liquid-container wall interaction,

dynamics and thermodynamics of cavitation bubbles with heat transfer, mass transfer, and chemical reaction. processes was used in the initial LVD velocity. The simplest treatment of these

work to develop a method of calculating

Our approach was motivated by the fact that the classical satisfactory for calculating the detonation velocity in

C-J hypothesis is

gases even though a gaseous detonation wave consists of a complex system of interacting shock waves. In the simplest treatment of LVD, the cavita-

tion field in virgin liquid is interactions, and is

formed quickly through liquid-container conditions for assumed to be in

accounted for only by the initial

shock propagation;

the compressed cavitation field is

mechanical equilibrium andtoforma in a state of complete reaction. It

steady-state reaction zone terminating

was demonstrated that the LVD velocity is

uniquely determined by the shock is repre-

our assumptions.

With our assumption of cavitation,

sented by a Rayleigh line passlnr through the initial cavitatod liquid and not the iW!,.Wn

condition of the With the

condition of the liquid.

assumption of mechanical equilibrium4 in and liquid, zone, " there is

the compressed mixture of bubbles

a well-defined pressure at each point of the reaction state for the burning is the initial shocked con-

and the initial

dition in

tile cavitated liquid.

With the assumption of complete reaction,

the end of the reaction zone lies on the equilibrium products Ilugoniot curve centered on the initial shocked condition of the cavitation liquid.

For the initial,

simplest treatment of tile cavitation model,

LVD

propagation velocity is

determined by the C-J condition on the equilibrium liquid.

products flugoniot curve centered on the cavitated

The TIGER code was used to calculate detonation parameters in ethyl nitrate for different degrees of cavitation, calculations are shown in Table 1. and the results of these

The internal energy of the cavitated

liquid ahead of the shock was assumed to be the same as that of the uncavitated liquid. Calculations with the BKW and virial equations of

state do not agree with experiment; propagation velocities calculated with the ideal gas equation of state agree with experiment, tion pressures lie in is inapplicable. The results of the calculations led to the conclusion that the initial assumptions were too restrictive. A more sophisticated treatment of the Thus, a more but detona-

the kilobar region where this equation of state

compressed cavitation field was required to model LVD. detailed study of bubble coll*-pse,

heat and mass transfer,

and chemical

reaction was undertaken to achieve this objective.

In addition, a

modification of the TIGER code to allow partial energy release was used to examine a less restrictive model and to determine the amount of reaction required to support LVI). The hydrothermodynamic treatment of the the manuscript

cavitation model of LVD is discussed in greater detail in of a technical paper attached as Appendix B.

10

-k u

o.

(D

t,

10

5.4 :3

~C .o D

00

C')

to ) M

tN

to

t'

to M

c'

.- CO ) Ln N

CD

Lo

C N

C NC*

N 0; -

t N

0 C; N

to 0

t N

41

to.
z 4.)
P-*4

"o

"
41

b-

cl -4
u_

0) )

N
00 m

0))
.-

t
H-

00

I
0
_ _ _ _

N
04 _ _ 0~ _

,i0-

0
4J4.)

0 O t C CAO
i)

0 1...4

o C

If 4 N

ii

t )

~ ~~ O

) ..

~C

j~~~~~~~~
-

14

0
N' 0

.. 4

I0.4

4j41 cl 94

Q)

$.4J ::
o. 0):..............*

CO

Nt m.

to M*

N* *

N40 0
*

0
*

0
*

0N
*

o
41 41

>
4.

1 0 -H>

0 0 D -4.

01

-0 0;JQ C;

00

00
0 1.4

~ *~ ~ 0~~~~~~ f) N . ~.

N
I

,4
*

.C
* *

N
,.d

1.NN C.4

4J 0 C4 V4

o0 V

P"

BUBBLE DYNAMICS

In

the initial

study of bubble dynamics,

the characteristic times and heat transfer

for bubble collapse,

heat transfer within a bubble,

to the liquid were compared as a means of assessing the relative rates of relaxation processes occurring in the compressed cavitation field behind the shock. The collapse of a bubble was then treated in more detail in

an attempt to determine conditions for the onset of reaction and the initiation of LVD. The dynamics of the collapse of a vapor bubble were analyzed to study the hot-spot initiation mechanism essential to the cavitation model of LVD. The temperature distributions in a collapsing spherical vapor the surrounding liquid were computed. Gas-phase reaction,

bubble and in

heat and mass transfer at the bubble wall, were included in the computations in

and motion of the bubble wall

order to simulate closely the thermal for hot-spot

history of a collapsing vapor bubble and the criteria initiation.

A vapor bubble was studied since during the very rapid even if dissolved

cavitation that occurs in liquids undergoing LVD, gases such as air are present, there is

not sufficient time for diffusion

into the resulting cavity.

Thus,

cavitated zones under such conditions

would consist of vapor bubbles .ontaining negligible amounts of gas.

"This work
* -

on bubble collapse leading to hot-spot initiation is in Appendix D.

discussed in detail

y'

13

VI

EXPERIMENTAL RESULTS

As a part of the studies of LVD undertaken in

this program,

three

separate sets of experiments were carried out to substantiate the theoretical model and to further delineate the important mechanisms that first experiment was to study the role of confinement, that of the second

set of experiments was to measure the wall stress just ahead of the detonation wave, and in the third set of experiments detailed measurethe liquid ahead of and during the The detailed descripthe

ments were made of the stress in

passage of the leading LVD-associated shock wave.

tion of these experiments and the results obtained are described in first and second annual reports and Appendix E. In summary, trinitrate (TMETN) the first set of experiments were run using metriol

in steel tubes of various diameter and wall thicknesses.

For each configuration LVD velocity was measured down the tube for various input shock strengths. Once the proper initiating shock was determined the wall thickness was varied keeping generally resulted in and stronger con-

for a nominal 1/8-inch wall tube, all other conditions constant.

Weaker confinement

decay of the LVD wave as it fitiement resulted in

propagated down the tube,

transition to 11YD.

The important role of circumas

ferential or hoop stress in predicted by theory, axial slit In made, shock.

the cavitation mechanism,

was confirmed by experiments using a tube with an

down one side. the second set of experiments, a limited number of shots were

again using TMETN,

to measure wall stress ahead of the main L[ND also in

Deformation and ringing of the wall were observed,

agreement with theory.

1,5

The most recent experiments,

designed to measure the pressure in were run using The

the liquid ahead of and during passage of the LVD wave,

several liquids and with 1-inch and 3/4-inch I.D. steel tubes.

measured pressures were in close agreement with the pressures calculated by the computer model. work. In general, the close agreement between experimental results and Appendix E provides a detailed discussion of this

theoretical predictions based upon the cavitation model of LVD further supports the cavitation model and provides a better understc.nihg of the LVD phenomenon. further research. In addition, this capability gives a good basis for
3

16

VII

BIBLIOGRAPHY

1.

D. C. Wooten, Hi. R. Bredfeldt, R. W. Woolfolk, and R. J. Kier, "investigation of Low-Velocity Detonation Phenomena in Liquid Propellants, Fuels, and Explosives," Annual Report, March 1970. AFDSR Scientific Report AFOSR 70-1005TR,

2.

D. C. Wooten and M. Cowperthwaite, "Investigation of Low-Velocity Detonation Phenomena in Liquid Propellants, Fuels and Explosives," Annual Progress Report, AFOSR Contract No. F44620-69-C-0079, April 1971. D. C. IWooten, M. Cowperthwaite, and D. C. Erlich, "Investigation

3.

of Low-Velocity Detonation Phenomena in Liquid Propellants, Fuels, and Explosives," Final Reporti AFOSR Contract No. F44620-69-C-0079. 1t. D. C. Wooten and D. C. Erlich, "The Wave Structure in Shocked,

Confined Explosives Associated with Low-Velocity Detonation," (to be published). 5. D. C. Wooten, "Collapse of a Vapor Bubble Leading to Hot-Spot

Initiation" (to be submitted for publication). 6. Erlich and D. C. Wooten, "Measurement of Pressure in a Liquid Explosive Undergoing Low-Velocity Detonation" (to be submitted D. C. for publication. 7. M. Cowperthwaite, "Hydrothermodynamics of the Cavitation Model of for publication).

Low-Velocity Detonation" (submitted 8.

R. W. Woolfolk, M. Courperthwaite, and R. Shaw, "A 'Universal' Hugoniot for Liquids" (submitted for publication).

17 $h

REFRENCES

1. 20

Ya. B. J,

Zeldovichi: Sov.

Phys.-JETP 10,

5,12 (-1940). 549 (1942).

von Neumann,

Theory -of -Detonatien '5'vos,OSRD No. 4!21 (4643).

~~3.,V.oeig
4.

Ann.. Phy~s..43,

-D-D. C. Wooten, :11. -R. Bredfeldt, R. W. Woolfolk, and R. J. Kier, "Investigation of- Low-Velocity Detonation Phenomena in Liquid Propellants, Fuels, :and ExplosYves," Annual Report, AFOSR Scient-ific Report AFOSR 70-1005T7R, March 1970. A. B. Amster, D. M. McEachern, Jr:, and Z. Pressman, "Detonation of ,Nitromethanes-Tetranitromethline Mixtures: Low and High Velocity Waves," Proceedings, Fourth Symposium (International) on Detonation, -ACR-126; Office -of Naval Research,_ Department of the Navy, Washington,, -D.C.,.p. 126.

5.

6.

-R. IV. Watson, C. R. Summers, F. C. Gibson, and R. IV. Van Dolah, "Det6nations in Liquid -Explosive--The Low-Velocity Regime," Proceedings, Fourth-Symposium (International) on Detonation, ACR-126; Office of Naval Research, Washington, D.C., p. 117. D)epartment of the Navy,

7.

R. IV. Watson', "The Structure of Low Velocity Detonation Waves, Twelfth Sympositim (International) on Combustion, The Combustion Institute, Pittsburgh, 1969, p. 723.

8.

R. W. Watson, J, Ribovich, J. E. Hay, and R. IV. Van Dolah, "The Stability of Low-Velocity Detonation Waves," Fifth Symposium (International) on Detonation, August 18-21, 1970, p. 71 (ONR

Report DR-163). 9. D. C. Wooten and M. Cowperthwaite, "Investigation of Low-Velocity Detonation Phenomena in Liquid Propellants, Fuels, and Explosives,"
Annual Progress Report, 10, AFOSE Contract F44620-69-C-0079, April 1971. M. L. Wilkins, Calculation of Elastic-Plastic Flow, Methods of Computational Physics, Vol. III (Fundamontal Methods in Hydrodynamics),

p. 211, Eds. B. Alder, S. Fernbach, M. Rottenberg, Academic Press, New York (1964). 11. R. l~lckling and M. S. Plesset, "Collapse andI Rebound of a Spherical
Bubble in I'Vater," Phys. Fluids 7, 7 (1964).

19

Appendix A

THE WAVE STRUCTURE IN SHOCKED, CONFINED LIQUIDS ASSOCIATED WITH LOW-VELOCITY DETONATION

THE WAVE STRUCTURE IN SHOCKED, CONFINED LIQUIDS ASSOCIATED WITH LOW-VELOCI-TY DETONATION*

by

David C. Wooten, Ultrasystems, Inc..

Newport Beach, California and David C. Erl:ich Stanford Research -Lnstitute Menlo Park, California

February 1972

This work was supported by the U.S. Research under Contract No.

Air

'6:'.re Office of Scientific

F4462-69-C-0U79 a

This is a preprint of a paper intended fc- publication in scientific journal. Since changes ma,. be made prior to publication, this preprint is distribii(Ated with the understanding that it will not be copied or reproduced without express permnis.ion of the authors.

A-iii

ABSTRACT

To study the cavitation rodel of low-velocity detonation (LVD) , the stress history is computed for a liquid explosive contained in a subsequent to its being impacted at one

cylindrical steel container, end. It is

shown that coupling between the container,wall and- the

liquid produces precursor tension zones in the liquid leading to cavitation ahead of the maiv liquid shock. zones by the liquid shock, Recompression of the cavitated

in accordance with the cavitation LVD model,

is assumed to cause hot-spot initiation of the liquid, and the subsequent energy release in turn supports the liquid shock. The simula-

tion for ethyl nitrate in a 1-inch I.D. by 1/8-inch wall thickness steel pipe resulted in detonation velocities of about 1.8, respectively, reaction. for energy releases of 0.1, 2.4, and 3.0 mm/lsec,

0.4, and 1.0 times the heat of this is

Since experimental LVD propagates at about 2 mm/isec,

taken as evidence of partial energy release in LVD.

The results show

the details of cavitation and recompression and furthe.T substantiate the cavitation model of LVD.

A-v

Introduction and Background Low-velocity detonation (LVD) that is is a low-order detonation phenomenon

observed in liquid explosives under particular etinditions of In 1919 LVD was first observed in however, nitruglycerine by the phenomenon remained

confinement. Stettbacker.

Until the past decade,

relatively obscure and there was little

agreement, among investigators More recently, because of fuels,

concerning the observed properties of LVD.

the increasing uses of highly energetic liquids as propellants,

F-

and explosives--many of which are known t0 be susceptible to LVD--a number of investigations of LV D have -been undertaken. these studies, defined, emerged. Experiments indicate that LVD can occur in almost any highly energetic monopropellant, Typically, LVD propagates at speeds only As a result of

the important characteristics of LVD are becoming well

and plausible explanations of the observed phenomena have

slightly greater thaa the liquid sound speed or in the range of 1.5 to

2.5 mm/tisec.

Stabl-. 'LVr3 iF in general observed only in strongly confined

explosive system, cnd ,only i: the sound speed of the container wall
material exceeds that at the undisturbed liquid. in the very wvak containers and in On the other hand,

containers with wall sound speeds sometimes observed

less than that of the liquid,

an unstable LVD is

which propagates in a pulsating manner with an average speed that is subsonic relative 1-s the liquid. Unstable LVI typically propagates

only a short distance before dying out, Though both tyles of detonation are observed In some liquids, the

observed properties of LVD are distinctly different from those of IIVD.

A -1

In addition, energetic liquids are ,generally miuch- icere sensitive to LVD initiation; LVD may be initiated by shock overpressures on the order of 1 to 5 kbar while shocks of 50 to 100 kbar are us.ially required to initiate HVD. Moreover, above a certain minimum chdvge .dliameter, H41)

propagation characteristics are independent of confineme3,4 whereas LVD is strongly dependent upon both confinement geometry and ihi of the wall material. wave, the llVD wave. properties

Thus, LVD represents a stable, low-order detonation Over the years, a number of models of the LVD

phenomenon have been proposed 'but, until recently, attempts to explain 1 LVD theoretically have met with little success. Stettbacker, in an attempt to explain his early observations, and later Deserhkovich and 2 Andreev. suggested t.aat the two detonation velocities in nigroglycerine were related to two known crystalline forms of the solid. More recently,

in a theoretical paper, Bolkhovitinov3 attempted to explain LVD on the basis that a phase transition from liquid to solid takes place behind 'he initiating shock. Ir. addition, as a part of their study of hot-spot
4

initiation in explosives, Bowden and Gurton

suggested that energy losses

due to lateral expansion in the reaction zone give rise to LVD behavior. These theories, although predicting a low-order detonation, do not account for the experimentally observed characteristics of LVD, particularly

that the sound speed in the container material must be greater than that in the liquid. In contrast to the above phenomenological theories, a model of LVD based on purely thermodynamic theory by Eyring et al. approach taken later b, Evans
6
5

and a similar

utilized a variable-reaction-zone length These models predict the

which depended upon dotonation velocity.

existence oe two detonation velocities but the resulting LVD velocity varies inversely with charge diameter, a fact which is not supported by experimental evidence. An attempt to explain the LVD phenomenon with a

A-2

semiempirical model based onl varying energy release rates behind the initiating shock was given by Schall

who suggested that disturbed gas

bubbles distributed in the liquid might account for the energy release
rates responsible for LVW:. While each thermodynamic model of LV7D predicts

a low-order detonation velocity,

they all fail to explain tile experiespecially that LVI) is ouserved propagates

mentally observed characteristics of LVI),

in many 'liquid compounds with different physical properties,

witi" approximately the same detonation velocity independently of the compound, and is apparently strongly coupled to the confinement system.

Recognition of the close interrelationship between LVD properties, such as the ease of initiation, stability, and detonation speed, and the such as container geometry, strength,

charge confinement characteristics,

ani elastic properties of the wall material, which include detonation wave-wall coupling.

has led to models of LVD One model, for example,

that accounted for the empirical interrelationship between LVD incidence and container wall characteristics was given by Woolfolk and Amster8 who suggested that the container wall shock causes pressure waves in the liquid wnich converge at the center to produce a Mach disk capable of initiating detonation. However, in this model, the wall shock and wave

therefore the Mach disk would continuously outdistance the LVI

unless the LVI) were moving at a speed higher than that observed. Another empirical model which also utilizes coupling between the wall and the detonation and which scems to fit the experimentally observed 9 and independently by facts was recently suggested by Watson et al. 10 In this model, precursor waves in the container Voskoboinikov et al, wall cavitate the liquid ahead of a liquid shock. The cavitation bubbles

are subsequently compressed by the liquid shock causing a hot-spotinitiated reoction, Ind the resulting energy release in turn drives the

A-3

11 shock. This model has been further verified by Watson who has taken

photographs of cavitation ahead of the reaction zone and by Watson 12 et al. who more recently have reported tentative measurements of precursor wall waves capable of producing cavitation ahead of the reaction the existence of individual burning sites rather 11 than a continuous reaction front has been observed by Watson in X-ray 13,14 photographs of the LVD reaction wave, and Gibson et al. have experimentally verified the dependence of bubble of the initiating shock. ignition on the strength z'one. In addition,

The cavitation miodel also explains the rela-

tively long dark region observed between the shock front and the lumi11 nescent reaction zone. Thus, the cavitation model, though it involves an incompletely understood interaction between the container and the liquid explosive to produce cavitation and subsequent hot-spot reaction and as yet unknown criteria for stability, appears to offer the most plausible mechanism for the propagation of 11). Our purpose here is, on the basis of the cavitation model, to

examine cavitation and recompression mechanisms ai. propagation of LVD.

in both the initiation

A computer simulation of the cavitation model

of LVD is

utilized to study the stress history in a system consisting

of a liquid explosive in a cylindrical container subsequent to being shocked at one end. The hot-spot-initiated reaction is zone in the liquid is simulated by recompressed

energy release when a precavitated

above a threshold level by the main liquid shock. clarify the liquid-wall

The results help

interaction mechanism that leads to precursor

cavitation and the subsequent collapse of the cavities by the liquid shock, and further substantiates the cavitation model of LVD.

A-4

Cavitation Model of LVD Experimental sutdies of LVD are usually carried out in an arrange-

ment similar to that used to study the sensitivity of liquid explosives. The liquid to be studied cal tube, which is 's placed in a container,

15

most often a cylindriThe strength

shocked at one -end by an explosi'e donor.

of the initiating shock can be varied by using attenuators (usually

Lucite) of different thicknesses.

For a number of liquids,

initiating

shocks on the order of I to 10 kbar typically produce LVD whereas much stronger shocks, on the order of 100 kbar, directly initiate HIT). Experimenters in the past have studied LVI) cameras, velocity-measuring probes,
wall strain.

phenomena with high-speed

S~8-12

and strain gages to measure

circumferential

When the initiating donor shock wave impacts the e.xp)losive system,
the shocks produced propagate down the sound speed in the liquid, the cylinder from thle donor end. is If

the container wall material

greater than that in

then the wall shock outdistances the liquid shock and causes Calculations discussed

a precursor wave system ahead of the liquid shock.

in the fo),lowing sections show that these precursor waves cause zones of tenstoua capable of cavitating the liquid. The cavitated liquid is sub-

sequently recompressed by the liquid shock leading to the possibility of hot-spot initiation. This mechanism is probably the manner in which L\I)

initiation takes place and under some conditions as, stronger donor shock, could lead to JIMI) initiation.

for example,

for a

The calculations
I kbar, thus

demonstrate that relatively weak donor shocks, on are sufficient to produce significant cavitation

the order of in the liquid,

accounting for the high shock sensitivity of liquid explosives to hotspot. initiation and to [M).

A-5

The cavitation model of LVD requires that precursor wall waves produce tension zones leading to cavitation ahead of the liquid shock. If the wall shock caused by the donor were the only mechanism for cavithen all LVD waves would be inherently unsteady, observed, if since

tating the liquid, in

the experimental conditions where stable LIM is

the wall

shock continually outdistances the LVD waves. deflection due to the LVD wave itself cavitation, is

However,

the wall

capable of producing precursor

then a self-sustaining,, steady LVD is possible. Though 11 Watson et al. report some preliminary experimental evidence that there is a precursor pressure wave attached to the reaction zone, it has not

been established In either case, wall material is it

that such a mechanism leads to precursor cavitation. is clear that when the sound speed in the container the liquid, precursor wall waves

less than that in

caused by either the donor shock or the LVD wave are not possible. Pulsating LVD is sometimes observed when the wall sound speed is

12
only slightly greater than that of the liquid; it may occur when the complex interaction between the donor and the receptor causes hot-spot initiation which starts to accelerate toward a full LVD.
.al

Subsequently,

as the resulting reaction wave approaches tlo

sound speed there is

no mechanism for the generation of precursor cavitation and the reaction (lies out from the lack of initiation centers. down due to decreased energy release rate, distance it, As the reaction zone slows

the wall wave may again out-

once again causing cavitation ahead of the reaction zone. in an oscillatory fashion

The whole process may then repeat itself resulting in pulsatory LVI). Clearly,

whether such an unstable LVD will upon the experimental con-

continue to propagate must depend critically ditions. In fact, unsteady LVI) is

frequently observed to (lie out

altogether.

A-6

Computer Simulation A computer study of cavitation and recompression in an energetic

liquid contained in a. cylindrically symmetric tube impacted at one end was undertaken in of LVD. an attempt to simulate the initiation and propagation contained in a 1-inch I.D. by 1/8-inch wall was chosen for the calculations as being repsetup used to study the shock Of

Ethyl nitrate,

cylindrical steel tube,

resentative of a typical experimental

initiation of LVD in the practical part of the research program.

particular interest in these calculations are the following factors: (1) Does the stress wave traveling down the pipe walls create a region of tension in the liquid sufficient to produce the liquid?

cavitation ahead of the main compression wave in (2) What is

the propagation velocity of LVI) and how does it in the reaction

depend upon the amount of energy released zone? (3) Is a stable LVD wave (one

that propagates at a constant to several

velocity) established at some distance (equal pipe diameters) down the pipe?

Computer simulations were performed using the SRI FIBROUS code. This code is stress wave propaga16 that includes tion code modeled after tile description of Wilkins, a two-dimensional finite difference both elastic-plastic and hydrodynamic constitutive relations. FIBROUS code is grid parallel used in the axially symmetrical geometry, When the

a rectangular

to the axis of the cylinder extending from the axis to specified, as shown in it Figure 1. The code

the edge of the cylinder is treats each ceil'of

this plane grid as if

contained the volume swept

out by the 360-degree

rotation of the grid about the cylindrical axis.

A-7

A similar projectile cylinder impacts the target from the right to begin the computation. (The shock produced in the explosive system by the the experimental

projectile cylinder simulates the shock delivered in setup by the donor-attenuator.) impact,

At successive time increments after the stress

all of the relevant parameters needed to characterize for each grid cell. hydrodynamic

wave propagation are calculated position, energy, velocity,

These include pressure, internal

triaxial stresses,

and specific volume.

Certain specified parameters at designated and these for

cells can be stored by the computer at each time increment,

can be tabulated or plotted at the end of the calculation to give, example, stress and specific volume histories,

or stress-volume loading

and unloading paths. For each cell during each cycle the calculation proceeds basically as follows: the stresses in the adjacent cells during the previous cycle for the cell which, along

are used to compute the current acceleration with its previous velocity and position, is

then used to compute its are obtained the current and the specific volume

current velocity and position.

From the latter

triaxial strains, distortional strain energy,

which--together with the equation of state or constitutive relation of the material stresses, in the cell--are used to compute the current triaxial and the internal energy. used to relate changes in The the internal

the hydrostatic pressure,

AMie-Gruneisen equation of state is energy to changes in successive cycles is

17

the hydrostatic pressure.

The time increment between

chosen so that a stress wave propagates less than

the distance between two cells during that time. The computational grid for the simulation of LVD was set up as in Figure l.* The target grid zcntained 4 rows of liquid cells and one row of steel pipe cells in the radial direction; each row contained 97 cells
* The figures are

shown at the end of this appendix.

A-8

in in

the axial direction.

Hlowever,

only about tile first

half of the target the liquid

the axial dlirection yielded useful stress histories in

cells,

since reflection of the stress wave in tile pipe wall from tile the latter half of the target.

free surface disrupted the stress flow in Therefore

the usable LVD simulation calculations were carried out only The projectile, which contained 5

to about 6 diameters down the tube. rows of cells in in

the radial direction with each row containing 10 cells impacted the target at a velocity of 0.5 mm/1sec.

the axial direction,

The cells for both the target and projectile grids were 3 mm square. An elastic-plastic equation-of-state formulation for Armco iron It 18

was used for the constitutive relations for the steel pipe cells. included a nonlinear hydrostatic pressure-specific volune (P-V)

Ilugoniot

*12
where

curve given by P ji =
-

= is

1,590 t + 5,170 in kbar and V

2 + 51,700

i3

1,

= 0.1274 cm/gm.

A shear The

strength of 7.9 kbar and a shear modulus of 820 kbar were used. equation of state for water was used for the liquid cells,

inasmuch as

the density and shock properties of water are similar to those of the liquid propellants of interest in LVD studies. of the liquid was defined by 2 P It = 25.611 + 61.2ti 3 + 122.7t. in The hydrostatic pressure

was found that when a purely hydrodynamic equation of state (one zero) is used for the liquid, the liquid the

which the shear strength is

cells next to the wall become highly elongated because liquid in center of the tube tends to flow down the tube whereas the liquid adjacent wall (no to the wall, is

prevented from sliding along the inside of the the FIBROUS code). Since a

sliding interfaces are allowed in

A-9

highly elongated cell,

particularly one in which one diagonal is inaccuracies, is

much

larger than the other, may lead to calculational was given a small shear strength of 0.1 kbar. It

the liquid

not expected that The equation of state

this had any significant effect upon the simulation. of C-7 was used for the constitutive It

relations for the projectile since the projectile equation

material.

will not be described here,

of state has no effect upon the stress history for time intervals, relevant to the LVD simulation. The cavitation of the liquid ahead of the liquid stress wave and the subsequent detonation of the cavitated liquid upon arrival of the liquid shock were simulated in sile stress (pressure) the following manner: whenever the tenthis

in a cell of the liquid exceeded 0.02 kbar,

cell was tagged and the tensile stress was reset to 0.02 kbar. further tensile strain was applied to this cell, expand at a constant negative pressure

So when

the cell continued to This approx-

level of 0.02 kbar.

imation to the liquid stress-strain behavior during cavitation was handled with only minor adjustment to the FIBROUS code. more realistic treatment of cavitation that is A

possible but involves

more substantial changes in the FIBROUS code will be discussed below. All of these tagged cells (which had undergone negative stress in

excess of 0.02 kbar and hence cavitation) subsequently experienced recompression due to the arrival of the main compressional wave in liquid. When the pressure in the tagged cells exceeded 0.4 kbar in expected the

compression,

an amount of energy equal to the energy that is

to be released by the hot-spot reaction was added to the internal energy of that cell to simulate the energy release in the LVD reaction zone.

The energy was added in three equal amounts over three calculational cycles (the time between successive cycles in the computation was about

A-10

0.35 lisec) bubbles is

since a detonation due-to hot-spot initiation of collapsing not expected to occur instantaneously throughout one region Since there is uncertainty as to the amount of energy evidence that onlya fraction

of the liquid.

actually released (specifically ,there is of the heat of reaction is

released during LIT)) , calculations were made to be added to each cell,

using three different amounts of energy (EVET),

in order to assess the effect of fractional energy release upon the overpressure and propagation speed of the LVD wave. tion of ethyl nitrate which is as the basis of EDET. approximate'ly 5 x 10 The heat of reacergs/gin, was used

A recompression of cavitated zones to the value

of 0.4 kbar was chosen as the compressioii threshold for the initiation of reaction so that the low-amplitude compressional, waves due to ringing behind ,the initial feature, wall shock would not predetonate the liquid. the liquid, is This

in connection with the stress history in

discussed in more detail below.

Results of LVI) Computations The important events that occur after impact are shown in Figure 2 for three simulations in which the EDET equaled 5 x 108 ergs/gm, 2 x 109 ergs/gm, 0.1, initial 0.4, and 5 x 109 ergs/gm or, respectively approximately The

and 1.0 times the heat of reaction of ethyl nitrate.

stress wave in the pipe wall,

which travels at approximately

5.3 mm/[isec,

quickly outdistances that in the liquid which travels at When the former sufficiently outdistances downt the pipe, the following

approximately 1.8 mm/jisec. the latter,

or at approximately 0.7 I.l).

sequence of events takes place. the pipe wall cause a small

The compressive triaxial stresses in Into the

compressional wave to propagate

liquid toward the center of the pipe and at the same time cause the pipe wall to expand radially. With this increase in radial strain, the hoop

A-1l

SX9

z#
stress in the pipe wall goes into tension and a tensile pulse is into the liquid shortly behind the initial compressive pulse. sent

As this

tensile pulse proceeds toward the center of the pipe, to cavitate, central cell. first

the liquid begins

at the cell near the pipe wall and finally at the stress wave in the liquid

Eventually the main compressive

reaches the cavitated zones and recompresses the liquid to above 0.4 kbar, which triggers the detonation. Since cavitation occurs earlier in the

liquid cells near the edge of the pipe than in and since the main compressional wave in

the cells at the center, of the

the liquid reaches all

cells at one axial position at about the same time, earliest

detonation begins down the

in the cell adjacent to the wall (at about 0.7 I.D.

pipe) and latest at the center cell (at about 1.2 I.D.). is

This result reported the

consistent with the observations of LVD in nitroglycerine 19 that the most intense reaction is

by Sosnova et al.

initiated in

nitroglycerine adjacent to the container wall and that the reaction wave profile is concave with respect to the direction of propagation It should be noted, however, in the

initiating region.

that since the inwardthe tensile stress

moving tensile pulses converge along the center line, is highest there and therefore It is the liquid in

the center might be more

highly cavitated.

thus possible that under some conditions, relatively weak, LVl) might occur

perhaps when the initiating shock is primarily along the central region. Figures 3a and

Jb depict the computed stress histories at the respectively,

central cell and at the cell adjacent to the pipe wall, for various distances down the tube in ergs/gm.

the case for which EDET = 5 x 108 initially a relatively small

For each stress history there is

compressive pulse followed by a region of tension where the tensile stress is about 0.02 hbar. These wall-induced precursor waves are

followed eventually by the main liquid compressive wave which triggers

A-12

the detonation. 2.5 diameters,

Note 'that for distances doown the pipe greater than about there are one or more additional cycles of precursor-

induced compression and tension in the liquid prior to the arrival of the main liquid wave. These waves are due to the further ringing of the compressional cavitation, wave. If the compressional

pipe following the initial pulses following the initial

but -prior to the arrival of

the main liquid compression wave, voids and cause detonation,

were strong enough to collapse the

then the LVD wave would be coupled directly

to the pipe wall wave and would travel at approximately the wave velocity

.in

the pipe rather than at a velocity nearly that of the liquid shock. Since it is known experimentally that this does not happen, detonation

was not triggered until the main liquid shock arrived. threshold of 0.4 kbar was chosoji, which in

A detonation

the cases investigated is the liquid caused by

higher than the magnitude of the pressure in ringing of the pipe wall. Figures 4a and 4b depict, circumferential (hoop) respectively,

the computed axial and

stresses in

the pipe at two distances down the The axial stress history of

tube for the same computation as above. Figure 4a clearly shows the initial

wall shock traveling at a speed of

about 5.3 mmn/fsec followed by a more gradual stress release and ringing. The hoop stress shown in Figure 41b rings back and forth between compipe wave until a strong

pression and tension following the initial tension is

produced by the detonatio.n of the liquid inside the pipe.

Figure 5 shows the stress-strain loading and unloading curves of the liquid cell located at the center of the pipe 1.2 diameters down the pipe from the impacted end. detonation parameters used in the two cases in and detonation. The effects of the cavitation and

the computations can be seen by comparing

which the Liquid does nnd does not undergo cavitation

A-13

71I

9*

The simulations in used resulted in 3 and 4.

which EDET of 2 x 109 and 5 x 10

ergs,'gm were

stress histories similar to those depicted in Figures w-ere that the LVD propagated at respectively, than

The two principal differences

a slightly higher velocity--2.36 and 2.99 mm/isec, the 1.84 mm/Vsec for the run using the lower EDET, induced in

and the pressure Figure 6

the liquid by the LVD was significantly higher.

contains stress histories at the central cell located 1.2 diameters down the pipe for all three simulations using the different values of which no detonation

EDET as well as the stress history for a simulation in was allowed to take place.

The simulation represented by these calculaa 1-inch

tions closely approximates LVD in ethyl nitrate contained in

I.D. by 1/8-inch wall steel pipe.

Since measured LVD velocities for

this system are on the order of 1.9 mm/lisec, the above results indicate
that partial energy release is likely in LVD. This hypothesis is further

substantiated by thermodynamic calculations of LVD which also indicate that partial energy release is likely.

20

An additional computer simulation was made using lead as the pipe

wall material.
higher (2.1

The stress wave velocity in the lead was only slightly

than that in the liquid for the particular geometry The result was that the wall wave

mm/tsec)

and pressures of the simulation.

never quite ran far enough ahead of the liquid wave to cause any zone of tension in the liquid; therefore, no cavitation occurred to initiate

LVD.

Computations using different lead pipe wall thicknesses and

precursor cavi-

stronger initiating shocks could conceivably result in

tation and an oscillating LVD wave.

However,

further computations of

this type were beyond the scope of the current work.

A-14

Model for Cavitating Liquid

-

The model for cavitation used in these computations,

in which the

liquid simply expands at constant pressure when the tensile pressure exceeds 0.02 kbar and then subsequently contracts at constant pressure until the strain corresponds to that at 0.02 kbar in tension, is only a

qualitative approximation of the cavitation behavior of a liquid undergoing tension. The more detailed phenomena of cavitation in a liquid the kilobar stress region have

undergoing dynamic tensile failure in

21
been discussed by Erlich et al. A computer model of the dynamic equation of state of a cavitating liquid was developed which calculated growth and collapse of bubbles in a liquid strained at high rates. model, This

which uses a viscous growth law and calculates the void volume

and resulting stress relaxation for a given tensile strain, could be inserted in the FIBROUS code to describe more quantitatively the cavithe LVI)

tation and subsequent bubble collapse for each liquid cell in simulation. The only unknown parameter in

such a calculation would be void volume, which can

the void nucleation frequency or the initial

either be determined to sufficient accuracy for a given material by experiments of the type described in based on data for other liquids. strain, Reference 21 or by an approximation subjected to tensile

When a liquid is

rapid stress release occurs due to the nucleation and growth of

ation bubbles. negative

Sc" Thus,

rather than using the strain at constant a

stress to simulate the cavitation behavior as was done here,

more realistic model would include dynamic bubble growth and the resulting stress release. been cavitated, In addition, once a zone of the liquid has

a modified equation o2" state to include the dynamic Computations using these improvements

bubble motion could be used.

would give a more realistic picture of the cavitation and void collapse phenomena and could thus greatly improve the cavitation model of LVD.

A-15

Discussion
The axisymmetric computer calculations to simulate LVD in a

cylindrical geometry show clearly the manner in which the container wall-liquid interaction produces zo;,es of tension leading to cavitation of the liquid. Thnsion is produced in the liquid by waves propagating The waves propagate into the

from the container walls into the liquid.

liquid {-rom the container walls much like Mach waves and converge at the tube centerline.

-il

If

one computes the Mach angle,

approximately equal to

/V ) , where V ? 1.8 mm/lsec is the propagation speed of the e w e wall waves and V e 5.3 mm/[.sec is the wave propagation speed in the w liquid, a value of approximately 20 is obtained for the present system. from

sin

(V

This indicates that the time delay for precursor waves to propagate the tube wall to the center cell should be about 5.3 lisec, is

a result that

consistent with that obtained by comparison of the precursor waves Thus, the cavitation produced by the precursor

in Figures 3a and 3b. wall waves is

a complex phenomonoa that clearly cannot be accounted for

by a one-dimensional analysis. The recompression of a cavitated zone by the primary liquid shock is assumed to initiate hot-spot reactions, and the subsequent energy

release--simulated by dumping an amount of energy EDET into the recompressed zone--drives the primary liquid shock. As can be seen by a comparison of fully developed the liquid

Figures 3a and 3b, shock is

once the detonation is Thus,

approximately planar.

even though the liquid next to the center of the liquid, recom-

the wall is

cavitated prior to that in

pression occurs almost simultaneously over the tube cross section once the liquid shock is more than 1 or 2 diameters from the initiation end.

Without detailed calculation of the bubble dynamics and appropriate me, cation of the liquid constitutive equations, the detailed effects

of Jhe time history upon the cavitation field cannot be fully explained.

A-16

During the ini.tiation phase, first is

which occurs approximately within the the liquid shock

two diameters from the donor end of the charge,

close enough bebind the wall shock to be within the Mach wave structhe vicinity 19 the region where initiation was observed by Sosnova et al. Thus, reaction takes place only in

ture causing cavitation. of the wall, Watson et al.

S~11

observed a reaction zone in the region downstream of the generated across the entire tube

initiation region where cavitation is

cross section before arrival of the liquid shock. The ringing of the tube wall behind the initial alternating zones of tension and compression main liquid shock (see about 16
1

wall shock causes

in the liquid ahead of the

Figure 3).

The period of the oscillations is Studies of liquid failure,

isec

for the present configuration. 21

or cavitation at high stress rates,

show that significant bubble growth less than 1 I.sec for is therefore likely

and the accompanying stress release can occur in stresses on the order of those observed here. It

that the ringing stresses produce significant bubble motion perhaps capable of collapsing the cavitation produced by the initial wave. Since experimental observations show that the reaction tension zone is

not directly coupled to and does not move with the initial as would occur if

wall shock,

the ringing compressions caused bubble collapse and the artifice was adopted here that a compressiov This effectively

hot-spot initiation,

greater than 0.4 kbar was required to cause initiation. permitted hot-spot initiation at a precavitated of the main liquiid shock. If, aseems likely,

zone only upon arrival ringing of the liquid

in the tube ahead of the main liquid shock is

strong enough to col tapse then the

the bubbles but not strong enough to cause hot-spot initiation, main liquid shock will travel uncontacted type of LVD, out, liquid. In This is

into alternating regions of cavitated and a ve.y possible explanation for the pulsating this dies

which the detonation travels a short distance, and so forth.

then begins again farther down the tube,

A-17

,~-t-,"y-~,-9-3S-

That the recompression due to wall ringing does not cause initiation may be due to several reasons. First, the threshold stress for not known. It is there-

hot-spot initiation by a given bubble .field is

fore possible that the compression due to ringing is to cause initiation. cavitated liquid is On the other hand, only approximated

not strong enough

the equation of state of the

in the present calculations and a

more detailed model of the cavitation zone might provide a much different stress in the liquid due to the wall ringing. For example, the recom-

pression by the final ringing wave following the first be propagating into a dynamically cavitating liquid.

tension zone would Dispersive effects

on the wave propagation caused by the bubble field could significantly alter the stress pattern from that computed here. The inclusion of a

more detailed equation of state for the cavitated liquid which accounts for the dynamic bubble motions would represent a considCrable extension of the current calculations and is subject for further study. As noted previously, phenomenon, if the LVD is to be truly a steady-state must be currently being considered as a

the wall wave caused by the detonation itself

capable of causing precursor cavitation independently of the wall wave caused by the initial shock. (A quasi-steady state LVq), however, could

occur for a considerable distance (town a pipe if remained in

the cavitation bubbles wall no indi-

the liquid long after the ringing due to the initial In the present calculations there is

shock had! disappeared.) cation that this is distance of

the case. and,

One calculation was extended

to an axial

Z/D = 15

even in

the most energetic case (EDET =

5 x 109 ergs/gm)

in which the shock pressure was in excess of 10 kbar, precursor wall waves capable of causing caviThus, caused

detonation-wave-coupled

tation directly ahead of the liquid shock were not observed. while not conclusive,

the work here indicates that cavitation is

only by the precursor wall wave generated

by the initiating shock.

A-18

Conclusions

The computations described above support the cavitation model for the initiation and propagation of LVD and provide a first step towards a detailed quantitative description of these processes. The results

support the hypothesis that onlv partial energy release occurs in LVD. There is no indication that the wall wave caused by the detonation itself produces zones of tension capable of producing cavitation. However,

further conclusions regarding these results must await additional calculations for different geometries, different materials, and more

accurate modeling of the equation of state of the cavitated liquid.

A-19

REFERENCES

1.

A.

Stettbacker,

"Zuinding und D)etonation," 137 (1919).

Z.

ges.

Schiess-

Sprengstoffe, 2. A.

14,

A. Deserlikovich and K. K. Andreev, Z. ges.,

"Uber Die Eigenschaften der 25,

Nitroglycerine-Isomeren," 353 (1930). 3.

Schiess-Sprengstoffe,

L. G. Bolkhovitinov, "The Low Velocity Detonations of Liquid Explosives," Dok. Acad. Nauk, 130 (5), 1046 (1960). F. P. Bowden and 0. A. Gurton, "Birth and Growth of Explosion in Liquids and Solids Initiated by Impact and Friction," Proc. Roy. Soc. A, 198, 350 (1949). 11. I)uffey, and R. 45, 96 (1949). B. Parlin, "Stability

4.

5.

Il. Eyring, R. E. Powell, G. of Detonation," Chem. Rev.,

6.

Al. W. Evans, "Detonation Sensitivity and Failure Diameter in Homogeneous Condensed Materials," J. Chem. Phys. 36, 193 (1962) . R. Schall, "Die Stabilitaet Langsamer Detonation," 470 (1954). Amster, "Low Velocity Detonations: Some Z. Angew. Phys.

7.

VI Band, 8. R.

10,

W. Woolfolk and A. B.

Experimental Studies and Their Interpretation," Twelfth Symposium (International) on Combustion, The Combustion Institute, Pittsburgh 1969, 9. p. 731.

It. W. Watson, C. R. Summers, F. C. Gibson, and R. W. Van Dolah, "Detonations in Liquid Explosives--The Low Velocity Regime," Proceedings Fourth Symposium (International) Office of Naval Research, ACR-126, p. 117. "Low Velocity Sci. USSR, on Detonation, Washington, D.C. Department of the Navy,

10.

L. N. Voskoboinikov, A. V. Dubovik, and V. K. Bobolev, Detonation in Nitroglycerine," Inst. Chem. Phys. Acad. Trans. from I)okl . Akad. , Nauk, 161 (5), 1152 (1965).

A-20

11.

R. IV.Watson, "The Structure of Low-Velocity Detonation Waves," Twelfth Symposium (International) on Combustion, The Combustion Institute, Pittsburgh, 1969, p. 723. R. W. Watson, J. Ribovich, J. E. Hay, and R. IV. Vant Dolah, "The Stability of Low-Velocity Detonation Waves," Fifth Symposium (International) on Detonation, ONR Report DR 163, 71 (1970). F. C. Gibson, R. IV.Watson, J. E. Hay, C. R. Summers, J. Ribovich, and F. If.Scott, "Sensitivity of Propellant Systems," Quarterly Report, U.S. Dept. of Interior, Bureau of Mines, Pittsburgh, February 2, 1966. F. C. Gibson, R. W. Watson, J. E. flay, C. R. Summers, J. Ribovich, F. If. Scott, "Sensitivity of Propellant Systems," Quarterly Report, U. S. Department of Interior, Bureau of Mines, Pittsburgh, May 19, 1966. "Card Gap Test for Shock Sensitivity of Liquid Monopropellants," Liquid Propellant Test Methods, Liquid Propellant Information Agency, Applied Physics Laboratory, Johns Hopkins University, Silver Spring, Maryland (1960) M. L. Wilkins, "Calculations of Elastic-Plastic Flow," Methods of Computational Physics, Vol. III, Fundamental Methods in Hydrodynamics, p. 211, Eds. B. Alder, S. Fernbeck, M. Rottenberg, Academic Press, New York (1964). M. 11. Rice, R. G. McQueen, and J. M. Walsh, "Compression of Solids by Strong Shock Waves," Solid State Physics Vol. 6 (Academic Press, 1958) p. 41, Eds. F. Seitz and 1). Turnbull. J. WV.Taylor and Al. H. Rice, "El.]astic-Plastic Properties of Iron," J. Appl. Phys. 34, 364 (1963) G. S. Sosnova, I. M. Voskoboinikov, and A. V. Dubovik, "Light Emitted by a D)etonation Front of Low Velocity in Nitroglycerine," Dokl. Akad. Nauk, SSSR, 149, 642 (1963). M. Cowperthwaite, "Hlydrot'.drmodynamics of the Cavltation Model of Low Velocity Detonation," Appendix B of this report. 1). C. Erlich, 1). C. Wooten, and R. C. Crewdson, Failure of Glycerol," J. Appl. Phvs. 42 (1971). "Dynamic Tensile

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

A-21

.,

.PPWALL

I"

I IPE I \1I I. 1

GA-7771-IA

FIGURE A-1

RECTANGULAR GRID REPRESENTATION

OF LIQUID-FILLED PIPE USED IN FIBROUS COMPUTER CODE

A\-22

0 80 70

Z/D -- ratio of axial distance to inside diameter of pipe 6.25 5.0 3.75 2.5 1.25

7.5
co

iT-I

VP~

50

0ompressive 30 Wave in Liquid.

jw"

*.,c"

='r

40 10 = rY- i;;100---At = Center of Pipem mNear Pipe Wall

o0
0 2

1
12 8 10 4 6 DISTANCE FOR IMPACTED END 14 cm

16

18

GA-7771-12

FIGURE A-2

WAVE DIAGRAM SHOWING RELEVANT EVENTS IN LVD COMPUTER SIMULATION

A-23

K'L " '::,

"' ... "

. .. "F..

"" . .

. . "

;.. .. '

""

2.5

2.0

Z/D =3.75
Lu

Z/D II

=.2.25

5.25'
(a)IN THE CENTER OF THE PIPE

L1.0
S~II/ z

II

"00.5
0

>-05

0
CAVITATION ZONES

-3.5
0 10 20

1
30

1
40

1
50 - psec

I 1 -60 70
GA.7771.7

TIME AFTER IMPACT

2.5

2.0

ZID

2.25 ZD
= 3.75 I

Z/D = 5.251
7 (b) ADJACENT TO THE PIPE WALL

0..0I'
0 crr

.5-

IY

0 -0.

-o.5
0

I
10

I
-

.I
50 lsec

I
so 70
GA-717118

40 20 30 TIME AFTER IMPACT

FIGURE A-3

COMPUTED LIQUID PRESSURE HISTORIES FOR RUN IN WHICH EDET = 5 x 108 ergs/gm AT CELL LOCATIONS
A-24

i!

" t/

'

Ln-4 (a) AXIAL

-J 4-6

( i'~
VZ/D =5.25 -8 Z/D = 3.75 Z/D =2.25

-10
0 10 20 30 40
-

_I
50 psec
GA.7771-9

60

70

TIME AFTER IMPACT 10 1I

Z/D = 2.25

6
4
2
0 0

Z/D 3.751
ZID (b) CIRCUMFERENTIAL STRESS
5.25 i L HISTORIES IN THE PIPE WALL AT SEVERAL LOCATIONS FOR THE SIMULATIRN IN WHICH EDET 5 = x 10 erg/gm

"I

-4 1

I S-, I

10

20 30 40 so 60 TIME AFTER IMPACT '$sc

70

FIGURE A-4

..

...

.; 2A-25

AXIAL (a) and CIRCUMFERENTIAL (b) STRESS HISTORIES IN THE PIPE WALL AT SEVERAL LOCATIONS FOR THE SIMULATION IN WHICH EDET =5 x 108 Prn,/nm I I I I-

1,

2.5

2.0

X
I,

SI1.5 -

DETONATION/

L/1
CAVITATION
1 . .0
cn

0.5

With Cavitation and Detonation Without Cavitation or Detonation

0 7

-0.5 0 0.01 0.02 0.03 0.04 BULK STRAIN 0.05 0.06

GA-7771-11

FIGURE A-5

STRESS-STRAIN LOADING AND UNLOADING CURVE FOR THE LIQUID IN THE LVD COMPUTER SIMULATION SHOWING EFFECTS
OF CAVITATION AND DETONATION

MODELS USED

A-26

14

---

No Detonation

12

---------. EDET EDET EDET - _

Heat of Reactiorn 0.1 0.4 cx o of Ethyl Nitrate 1.0

I 10
w

-10 14 18 22 26 30
-

TIME AFTER IMPACT

34 pusec

38

GA-7771-13

FIGURE A-6

PRESSURE HISTORIES OF THE LIQUID CELL

IN THE CENTER OF THE PIPE AT A DISTANCE OF 1.2 INSIDE DIAMETERS FROM THE IMPACTED

END FOR DIFFERENT VALUES OF EDET (EDET

FOR ETHYL NITRATE - 5 X 109 ergs/gm)

A-27

Appendix B

HYDROTHERMODYNAMICS OF THE CAVITATION MODEL FOR LOW VELOCITY DETONATION

,[A

HYDROTHERMODYNAM! ICS OF THE CAVITATION MODEL FOR LOW VELOCITY DETONATION*

By

M. Cowperthwaite Physical Sciences Division Stanford Research Institute Menlo Park, California

This work was supported by the U.S. Air Force Office of Scientific Research under Contract No. F44620-69-C-0079.

B-iii

Introduction
Shock wave studies during the last decade,

1-6

extending earlier

work discussed in the book by Taylor,

have established that many liquid

propellants and explosives exhibit two modes of self-sustaining detonation. These modes have distinct velocities of propagation and have been and low-

characterized accordingly as high-velocity detonation (HVD) velocity detonation (LVD).

The high-velocity mode propagates at approxithe low-velocity

mately6 mm/1sec with a pressure of about 100 kilobars;

mode propagates with a slightly higher velocity than ambient soundspeed, at approximately 2 mm/4sec with a pressure of about 10 kilobars. hydrodynamic theory of detonation, (C-J)8,9 hypothesis, The

together with the Chapman-Jouguet

provides a satisfactory model for calculating HIM

velocities when the equation of state of the detonation products is known, but not for calculating LVD velocities at the present time.

The Cavitation Model for LVD The present work is concerned with the problem of formulating a It is based on

model for calculating the propagation velocity of LND.

the physical model proposed by workers at the Bureau of Mines after an This model was discussed extensive experimental investigation of LVD. 5 LVD by Watson at the Twelfth International Symposium on Combustion. is modeled as a reactive shock propagating in a cavitated liquid. The

cavitation field ahead of the shock is the liquid and container wall; is iaitiated in

produced by interactions between

the reaction required to support the wave This model is called

bubbles compressed by the shock.

the cavitation model of LVD.

B-1

Propagation of LIT is formation of a cavitation compression. wall

thus a very complex phenomenon field and its

involving tile

subsequent behavior under shock

A detailed treatment must account for liquid-container

interactions and also for the dynamics and thermodynamics of a mass transfer, and chemical reaction.

cavitation field with heat transfer, Here, is however, it is

assumed that a detailed account of these processes The pro-

not required to calculate the propagation velocity of LITD.

cesses will be modeled their characteristic dimensional C-J model

simply with well-defined assuihiptions related to This approach is motivaued which is by the onesatisfactory for

times.

for gaseous detonation,

calculating the average velocity of self-sustaining detonation waves without treating the complex system of interacting transverse waves necessary for their propagation. As with the C-J model, the validity

of assumptions and the model based upon them must be determined by tie comparison of calculated LVD propagation velocities with those obtained experimentally.

IlydIrothe rmodynamic Assumptions Our one-dimensional treatment of LVD is based on the following mechanical equilibto be

assumntions ordering the time scales rium, and chemical reaction.

for cavitation,

The cavitation field is assumed

formed quickly ahead of the main shock in compression is

the liquid, and under shock

assumed to attain mechanical equilibrium and form a The implications of the assumptions are as

steady-state burning zone. follows.

The rapid formation of the cavitation field necessitates that

the shock in the liquid be represented by a Rayleigh line8 passing through the initial state of the cavitated liquid and not the initial The assumption about mechanical equilibrium reaction

state of the virgin liquid.

implies that presstwe be well defined at each point of thO

B-2

zone, and specifies the initial state-for burning as the initial shocked state produced in the cavitated liquid. Tne steady-state assumption

leads to the conclusion that the reaction zone lies on the Rayleigh line representing the main shock in the cavitated liquid. An additional required to We

assumption about the termination of the reaction zone is

formulate a model for calculating the propagation velocity of LVD. will first considei

the case when the reaction attains equilibrium at does not.

the termination of the steady state and then the case when it

Complete Equilibrium Considerations based on the properties of Hugoniot curves lead to the conclusion that the cavitation model for LVD gives a unique propagation velocity when the reaction zone ends in a state of complete equilibrium. The Hugoniot for cavitated liquid is SK shown schematically

in } the pressure-specific volume (p-v)

plane as OSlI in Figure 8,9 the 1; centered

detonation branch of the equilibrium products llugoniot curve at 0 is J shown as as OJS. 0 'JH'

and the Rayleigh line tangent at its C-J

point

The point of intersection of the Rayleigh line Oi 0 , with slope i -(D/v.) e, and

passing through the initial condition

OSII represents the initial shocked state propagating at velocity vo. D

produced by a shock

in the cavitated liquid with initial volume the initial shocked state,

Since the initial state for burning is

the final Lurnt state must lie on the deflagration branch of an equilibrium products Ilugoniot centered on 0Cal. We are therefore interested in the

deflagration branches of the family of equilibrium products Hugoniot curves centered on OSII For convenience, we will consider the family of deflagration lHugoniot S with p. > p .
I S

curves centered on OSII above

is shown schematically

A member of
When

this family
,

in Figure 1 as

1 J '10

'.

p. > p)

we will

*Figures

are shown at the end of this appendix. B-3

show that the points of intersection of the Rayleigh

line

Oi

with

the detonation branch of the products Hugoniot curve centered at the

initial conditlior

liLe onl the deflagration branch of the pr~oducts

i. The points of intersection of It is convenient first Oi to i and

flugoniot curve centered at

0 iJl'

are shown in 0,

Figure 1 as 1 and 2. i , and 2

relate the states and 2

by expressing the conditions that 0 and also on the then

lie on the Hugonioi. curves centered at 02i. Let e

Rayleigh line (c, tion p, 8 v)

denote specific internal energy,

states on a tiugonilo

curve are related by the llugoniot equamomentum, and energy across a

expressing the balance of mass,

shock discontinuity. centered at

The tlugonLot equation for a Hlugoniot curve is

O(eopovo)

2(e - eo)=

(p + po)(vo

v)

(1)

and the conditions for from Eq.

to be on

OS1i

and

to be on

OJIl'

follow

(1)

as 2(e*
-

0o)

(p1

+ po)(Vo - v.)

(2)

2(e

eo)

(P 2 + PO)(vo - V2 )

(3)

The condition for these states to be on the Rayleigh 8 from the Rankine-flugoniot jump conditions as

line

Oi'

follows

- I Vo -v 2

= -

i - PO - v.

Tihe combination of Eqs. and vC

(2),

(3),

and (4)

to eliminate

eo,

Po,

leads to the equation

2(02

e)

p)(v

2)

(5)

B -4

relating the states equation

and

2,

and a similar argument leads to the

2(e,

- e.) 1 and 1.

(P 1

+ P.)(v.1 1

v1 )

(6)

relating the states

Equations (5)

and (6)

show that states

2 and I satisfy the Ifugoniot equation

2(e - e.)

(P + pi)(v

v)

(7)

for the deflagration branch of the products Iugoniot at i. Thus- when P > Ps i 0
,

0 JO. I

centered

tOe deflagration branch of the products

Hugoniot centered at Hugoniot centereO. at line connecting at 0. It is 0

and, the detonation branch of the products intersect at the points where the Rayleigh i intersects the detonation branch centered

-and

clear from Figure 1 that the intersection points 1 and 2 pi p approaches > p
s

approach each other as Pi = P."

,

and coincide at iJ'

when

Moreover,

when

the Rayleigh line

tangent to of the

O Jl 0
initial

intersects the condition

p = 0 When p
-

axis at a point to the left < p

S

0,

the deflagration branch of the

Blugoniot centered at

J de s not intersect the detonation branch of i , and its tangent Rayleigh line intersects 0. A unique situa0 and

the -Hugoniot centered at the p = 0

axis at a point lying to the right of pi = PS

tion arises when -i

.for then the IHugoniots centered at OS at

are tangent to the Rayleigh line

J , and the C-J points of

both curves coincide.

This condition is

the only one compatible with

our assumption -of a steady state terminating in complete equilibrium because in this -case the Rayleigh line must be tangent to the deflagration branch of tihe tlugoniot centered on the initial must also pass through the initial shocked state and liquid.

state of the cavitated

B-S

Thus conditions for a steady state define a unique propagation velocity in the simplest case when the reaction attains equilibrium. The propa-

gation velocity for LVD with the reaction attaining equilibrium in the cavitation model is determined by the C-J condition on the detonation branch of the equilibrium Hugoniot centered on the cavitated liquid. To test the model with a realistic equation of state, the TIGER code developed at Stanford Research Institute for the Ballistic Research Laboratories under Contract No. DA-04-200-MAIC-3226(X) , was used to calcu-

late the detonation parameters of ethyl nitrate for different degrees of cavitation. The internal energy of the cavitated liquid ahead of the The

shock was assumed to be the same as that of the uncavitated liquid. results of these calculations are shown in Table 1.

Calculations with

the BKW and the virial equation of state do not agree with experiment, even when the volume of the cavitated liquid greatly exceeds that expected experimentally. Propagation velocities calculated with the ideal gas but the detonation pressures inapplicable.

equation of state agree with experiment,

lie in the kilobar region where this equation of state is

The results of these calculations lead to the conclusion that the assumption of complete reaction is not valid. Attention was therefore

given to a study of the cavitation model when the reaction does not attain equilibrium.

Incomplete Equilibrium We will use a reaction coordinate

X to denote the fraction of

Then

unburnt liquid when the reaction does not attain equilibrium.

X = 0

in the former case when the reaction attains equilibrium and all and

the liquid is consumed,

X= I

when there is no reaction.

Our

treatment of partial equilibrium is

based on the assumption that the

13-6

%4

reaction is

frozen with the pivoducts in chemical equilibrium.

In this

case a value of

defines an equilibrium products flugoniot curve for frozen Hugoniot--centered on the initial conliition And the values of

incomplete reaction--a

of the cavitated liquid.

in the range

0 < X < 1

define a family of frozen Hlugoniot curves lying between the llugoniot curve of the cavitated liquid and the equilibrium products Ilugoniot for complete reaction. i It follows from the previous treatment of complete equilibrium that LVD will be represented by a Rayleigh line tangent to the detonation branch of a frozen Ilugoniot centered on the initial state cavitated liquid. 0 of the

An equation of state for nitromethane was incorporated

into the TIGER code so that detonation parameters could be calculated with a realistic thermodynamic description of the explosive and the products of reaction. of X Detonation parameters were calculated as a function

to determine if LVD could be modeled with the cavitation model.

10

A complete equation of state

of the liquid explosive must be

incorporated into the TIGER code to compute detonation parameters when the reaction does not p:'oceed to equilibrium. plete pressure-volume-temperature A knowledge of the incom-

(p-v-T) equation of state and the variaCP along the atmospheric

tion of specific heat at constant pressure isobar is

re,,uired to introduce a complete equation of state into the The thermodynamic functions required to compute a thermo-

TIGER code.

dynamic state are calculated by integrating thermodynamic identities relationship 11 state for liquids used in the present work was based on the equation of along isotherms from the atmospheric isobar. formulated previously for shock temperature calculations. description of liquids, (Dp/BT)
v

The (p-v-T)

In this

is assumed to be constant and the

specific heat at constant volume temperature, i.e. C (T) . v

Cv is assumed to be a function of Nitromethane was chosen as a typical liquid

13-7

explosive because

C v (T)

and the shock temperature along the Hlugoniot II The standard Ilugoniot centered
0

curve were known from the previous work. on the initial cc/g) in the state (p-v) (p
0

= 1

atmosphere,

To = 29? K, and

vo = 0.884

plane was constructed with the universal Ilugoniot where a, and a2 are constants; and U, u, and sound speed in

for liquids and co

U = alco + a 2 u

denote shock velocity, condition. b = (bp/bT) v

,

particle velocity,

the initial Let h O3h

Q = i/vo(v/bT) p=l

and let the subscript

denote states on the standard Ilugoniot curve shown schematically as in Figure 2. The (p-v-T) equation of state in the region v : Vo

spanned by the tfugoniot curve can be written as p ph(v) + bET - T I(v)], v vo (8)

where

)h(v)
and Th (v) is

Po + (aco)

(Vo

V )/[Vo

- a 2 (vo

vhI

(9)

calculated by integrating the differential equation The (p-v-T) equation (bv/BT)

governing temperature along the Hiugoniot curve. of state was extended to the right of was constant and therefore equal to P = Po = 1 of state in when v vo. vo 0vo

by assuming that

along the atmospheric isobar (p-v-T) equation

With this assumption the v > vo can be written as - 1/a( v/vO -

the region

P - po

bE(T - TO)

1)1,

v Z vO

(10)

Additional work on the universal Ilugoniot curve for liquids is presented in this report in Appendix C.

B-8
,/}

Equations (8)

-and (10)

have continuous first Vo

derivatives where In performing used

they are patched together along -the

isoch6re 042. 2),

an integration along an isotherfn-, ,say 543 (Fig. until v = vo at 4, and then Eq. (8) is

Eq. '(I0) is

used along 43 for

v - vo .

Equations for the specific Gibbs .free energy G, entropy (8) S , the specific enthalpy -1t1, and C

the specific

were derived with Eqs. But only the equa-

and (10) G

and incorporated into the TIGER code.

tions for

will be given here because the equations for the other identities.

state variables can be calculated from them with thermodynamic For v > vo , the equation for -G is
-

G(T,v)
with

G(T) - b/? avo[V2

(v(T))D
T T

(11)

G(T)

Go - So(T - TO)

- T

To

(C /T)

T+

CPT

and v(T)
=

vo(l +

(T-

TO))

(13)

For

v vo

it

is

jG(T,v)
where

G(T,vo)

+ pv

p(Tv

)vo-

bT(v

vo)

+ bIT

(14)

G(T,vo)

and

p(T,vo) I
=

are obtained by setting

V!V

v = vo

in

Eqs. (8) and (11), and intgras o shck

1d
V0

and

IT=

v h
V0

denote the

resureand temperature along the Hugoniot curve.

developed at Stanford Research Institute for

The SPECIAL routine, Picatinny Arsenal detonation,

under Contract DAAA21-71.C-0-454 to treat nonideal

was used to calculate detonation parameters of cavitated X with different

nitromethane as a function of the unreacted fraction equations of state for the reaction products.

Most of the calculations

B-9

-,,-:.::z :i: .. S..

1.'

'-

,;..p

were performed with the virial equation of state because we are interested in detonation pressures in the region of 10 kilobars. The virial

equation of state used in the TIGER code contains the second virial coefficient B and the third virial coefficient C. The results of (C = 0)

the calculations performed without the third virial coefficient for a 10 percent increase in initial volume are given in Table 2,

and

the results of the calculations performed with the BKW and ideal gas equations of state, and the virial equation with problems in in C

# 0

are also. given. the calculations

Because of convergence

the SPECIAL routine,

for a 10 percent increase

initial volume were limited to values of and the case of a 20 percent increase in

in the range

0 < X' 9 0.625,

the initial

volume was considered.

The results of the calculations performed for a 20 percent increase in initial volume using the virial equation with The virial equation with C = 0 is C = 0 are given in values of

Table 3.

used for all

for convenience because the SPECIAL routine was formulated to calculate detonation parameters along the frozen C-J locus from the equilibrium,

= 0,

C-J state. in

The calculated detonation pressures and velocities 0 < ! 0.75 9 are plotted as a function of the graphs were extrapolated

for

the range

in

Figure 3.

For values of

X > 0.75

smoothly

to satisfy the boundary conditions that the pressure approaches one atmosphere and the velocity approaches the sound speed in liquid as the cavitated liquid was

approaches one.

The sound speed in cavitated initial

estimated to be 1.88 mm/lsec when the increase in

volume was 10

percent and to be 1.08 mm/p1sec when the increase was 20 percent. It follows from Figure 3 that the calculations are not completely values of LVI) velocity and pressure and lie tn the region where the results

definitive because experimental correspond to values of

X 0.8

B-10

-. 11._

of the calculations had to be extrapolated. Figure 3, however,

Further examination of

shows that the curves for a 10 and 20 percent increase

in initial volume lie very close together in the lO-kilobar region and leads to the conclusion that a surprisingly small amount of reaction is required to support LVD. Important features of the cavitation model

with incomplete reaction can therefore be stated as follows: (1) Only reaction of about 20 percent of the liquid is required to support LVD. (2) LVD parameters do not depend strongly on the initial degree of cavitation.

Conclusions It is clear that additional work is required to formulate a satisThis

factory method for calculating the propagation velocity of LVD.

work should be concerned mainly with the degree of incomplete reaction. An attempt should be made to establish a condition for the termination of the reaction zone so that there is a unique propagation velocity for the cavitation model with incomplete reaction. A consideration of bubble

dynamics with heat and mass transfer and chemical reaction should be undertaken to try and determine a criti" ,iity condition for termination of reaction and a means of calculating a value of reaction zone. values of model.

X at the end of the

LVD velocities for different explosives calculated with could then ho used to test the validity of the cavitation

The detonation products of a liquid explosive undergoing first the amount

JIVD and then LVD should be collected and compared to test if

of reaction predicted by the present work with the cavitation model for LVD is reasonable.

B-Il

Acknowledgments The author thanks D. C. Wooten, R. W. Woolfolk, helpful discussions, and is incorporating the C (T)
v

and R. Shaw for

greatly indebted to IV.Zwisler for

equation of state for liquids into the TIGER

code and calculating the detonation parameters with the SPECIAL routine.

B-12

REFERENCES

1. 2.

L. G. Bolkhovitinov, Dokl. Akad.

Nauk 130,

1044 (1960).

I. M. Voskoboinikov, A. V. Dubovick, and V. K. Bobolev, Dokl. Akad. Nauk, SSSR, 161, 1152 (1965). R. W. Watson, C. R. Summers, F. C. Gibson, and R. V. Van Dolah, Proceedings, Fourth Symposium (International) on Detonation ACR-126, Office of Naval Research, Department of the Navy, Washington, D.C., p. 117. A. B. Amster, D. M. McEachern, Jr., and Z. Pressman, Proceedings, Fourth Symposium (International) on Detonation ACR-126, Office of Naval Research, Department of the Navy, Washington, D.C., p. 126. R. W. Watson, Twelfth Symposium (International) on Combustion, The Combustion Institute, Pittsburgh, Pa., 1969, p. 723. R. W. Woolfolk and A. B. Amster, The Twelfth Symposium (International) on Combustion, The Combustion Institute, Pittsburgh, 1969, p. 731. J. Taylor, Detonation in Condensed Explosives, Chapter X, Oxford at the Clarendon Press 1952. R. Courant and K. 0. Friedrichs, Supersonic Flow and Shock Waves (Interscience, New York 1948). Ya B. Zeldovich and A. S. Kompaneets, Theory of Detonation, Academic Press, New York and London, 1960. M. Cowperthwaite and J. 11.Blackburn, NBS Special Publication 326,

3.

4.

5.

6.

7.

8.

9.

10.

Accurate Characterization of the High-Pressure Environment, 11. M. Cowperthwaite and R. Shaw, J. Chem. Phys. 53, 555 (1970).

p. 137.

B-13

00 e

a.) 00 C.4E-

uo NNN

CD tD (D

) N' oD

to Mo U)

cli (0t q

ttNS

0to m 0)

00 CtoC) N' C Wo

00 00 CS) ' -

00 0 C-

M 0)

E-4

Co

r-4 C) 0 ()

~N

ND 0
;

C' m' '4 _;

0)

CD
C;

-4
140 4

m'
-4

U)
4

00
*

I*

) 0

,4 r r4

C^)1-

tU)
-

1.

P;

>

'3 U) o C U) .0

:1

$u4

C0 D C' 9' C'D M'

-4

t'-

CD
1'

Lo) Co 4

0 C'Ur)

t_-;

0
o
H

CD

Ur)

c')
'
-4

CD 0)
1-4

C to C'-

o4

) M'

,4

cq )

0 C -4 1-4

-4' 04

E-4
-4 El Q >

.0

14

P-4

4.)-P10 04 '

C r-4 00 ,-4 00 v' .I M' v) H-

Cm' rH m' r-4 m' .-

D (4'-

'

Cd$4uU2

4 r-4 4

.- I 4

0)
-

0 0

0 0
H

0 0
-

0 0
4

0 0
4

0 H O

C4 4~' 4 . 0 o00U E-4 0 CM CC CD ) 0

'-4 C)
'

00C C) 0 0-4 C' '

0 C C'

U) '

C)

) '

C) ')

C) C)

C) C)

C) C)

. C

a) >z,4
C

4')
C C)C r=) Q)

CH

0r.

o '

C
-

C)

C) C)

0 *4
CU a) .1.4 .44 )

-0 *..

o)
c';

m
oi 44
'4'4

0 Cm

0
4.4 .4.44rcd q*

~~.~
CoO4

'J

;.

13-14

Table 2 CALCULATED DETONATION PARAMETERS FOR NITROMETHANE WITH THE INITIAL VOLUME INCREASED TEN PERCENT

Equation of State

Unreacted Fraction

Detonation Velocity (mm/tisec) 4.04 3.91 3.79 3.66 3.53 3.29 3.05 2.80 6.49 5.55 4.47 5.20 4.70 4.24 3.82 2.32 2.62 2.44 (atm.

C-J Pressure x 103) 54.9 53.9 51.4 48.6 45.6 39.5 33.7 27.5 118.7 81.9 50.8 79.5 65.1 53.7 44.5 24.8 33.6 28.4

C-J Volume (cc/g) 0.650

0.635

Adiabatic Index (X) 2.02 1.88 1.83 1.80 1.78 1.78 1.80 1.88 2.59 2.82 2.99 2.45 'X.44 2.39 2.32 1.22 1.07 1.13

(0)
0.0 1/16 1/8

0.629 0.625 0.623 0.622 0.625 0.635 0.702 0.717 0.729 0.691 0.689 0.685 0.679 0.533 0.503 0.515

Virial (C
= 0)

3/16 1/4 3/8 1/2 5/8 0.0

BKW

1/4 1/2 0.0

Virial (B and C)

1/8 1/4 3/8 0.0

Ideal Gas

1/4 1/2

B-15

4)X P- s
*,1 4

00

to.

-4

In-

P->

r-4

t
-o -

u000

tf)

C')

00

1-4

1'-

co)

4'

C< 4'

v4 C')

>

t0-4 0 M M 00 0) 0 0e 04

C')

E-4
'-4

41

0 u 00 )
i

I-

a)

CD) 'C')

C' -12 0

or)

f)

0 > E4 -4-

~
'NI

~
t-

f '

~~~~taC)

mo

-4J:
to -

0 C'3

'

C00 C')

Ck)

C')

v'

0.~
4.) 4J

r-i'

.4

I)

cs

B3-16

~j'

0.i

)IN

pv

LN

I-1

0
V

5
GA-7771-21

FIGURE B-2

SCHEMATIC HUGONIOT AND ISOTHERM FOR NITROMETHANE

B-18

55101

50

O
Pressure 0 Velocity .

- 1o

45 .

J
J

Initial Volume Increase 20% Initial Volume Increase 10%

o
40

Pressure Velocity

8
$

35

-7

E E EE
0.0

cc2 uJ25-- D

5 >
u-J

ww

Q)-

25

10

-- 2

0-

0. 125

0.375

0.625

0.875 GA-7771-20

X,FRACTIO'4I OF UNREACTED NITROMETHANE

FIGURE 8-3

DETONATION PARAMETERS VERSUS FRACTION OF UNREACTED NITROMETHANE

B-19

Appendix C A "UNIVERSAL" HUGONIOT FOR LIQUIDS

Id

A "UNIVERSAL" R. W. Woolfolk,

HUGONIOT FOR LIQUIDS and Robert Shaw

Michael Cowperthwaite,

Physical Sciences Division

Stanford Research Institute Menlo Park, California 94025

This work was supported by the U.S.

Office of Naval Research under

Contract No. N00014-70-C-0190 and by the U.S. Air Force Office of Scientific Research under Contract No. FP4620-69-C-0079.

C-iii

The universal Hugoniot curve1temperature in cause it sures. liquid explosives 4 '

5

used previously to calculate shock is not valid below 20 kilobars be-

was constructed from shock wave data obtained at higher presMoreover, it does not satisfy conditions defined in the initial The object of tl'e presstate and thus

state by the Rankine-Hugoniot jump.conditions.

ent work was to extend the Hugoniot curve to the initial

improve the thermodynamic description of liquids in the 1-bar to 20kilobar region. The objective was attained by modifying the form of the conditions," calibrating it and checking its with

Hugoniot curve to satisfy the initial static pressure data for water,
7 8

validity with dynamic

9 and carbon tetrachloride. 10 shock wave data for glycerine

The Hugoniot curve defines the locus of shocked states obtainable from a given initial condition. It is obtained experimentally 1 1 from a

determination of the states produced by constant gating at different velocities. is Experimental

velocity shocks propa-

determination of a Hugonoit and requires considerable There is there-

usually very expensive and time-consuming,

amounts of materials which are destroyed in the process.

fore a need for estimating Hugoniots from easily measured physical properties. Following earlier work by Gibson and coworkers,' and Voskoboinikov, Afanasenkov, Woolfolk and Amster
2

and Bogomolov 3 have shown that the Hugo-

noits of liquids could be represented by a single normalized plot of the form

u/c

a1 + (a 2 u/c)

(0)

where U is velocity,

the shock velocity,

u is

the particle velocity,

c is

the sound

a, and a 2 are constants,

and subscript zero denotes the initial

C-1

I,

state at a pressure p of 1 bar. fitted

2

Eqn.

(1)

with a,=

1.37 and a 2

1.62

the experimental data which were in the range u/c = 0.5 to 2.5, 0

corresponding to shock pressures in the range of 20 to 150 kbar. Tile main problem with Eqn. ary condition U = c 0 at u = 0. region from 1 bar to 20 kbar. cause of its (1) Eqn. is that it does not satisfy the boundthe

(1)

therefore cannot be used in

This low-pressure region is

of interest be-

importance in low-velocity detonations. ) exp[-a u/c0 with

Jacobs 6 has suggested that an additional term (1-a a 3 constant, be added to Eqn. (1)

so that the boundary condition may be met.

The form of the "universal

tfugoniot would then be

U/c = 1.37

(0.37) exp:-a3u/c ] + 1.62 u/c

00

(2)

which reduces to U/c

= 1 at u = 0.

In Fig.

1 the experimental data are , 5, and

shown along with three calculated curves that correspond to a.= 1. Figure 1 indicates that the value of a should be ! 5.

An expross;on

for a3 was derived from thermodynamic curve and the isentrope. data for water. The identity for the initial The constant

identities relating the Hugoniot a was evaluated using echo-sounding

slope or the Hlugoniot

in the (U-u)

plane

v 3

du p=1

4c0

\v.)3)
0

where v denotes specific

volume and s Jenotes specific entropy,

and the

thermodynamic identity for an isentrope

C-2

Q1
'I j =

(4

y/s

2c s..( _c V 2 Lv\vs LVk V/

(4)

were used to determine the constant a3 in Eqn. (2)

(2).

Differentiating Eqn.

with respect to u gives the slope of the Hugoniot as

dJ/du = 1.62 + 0.37 a 3 exp(.-a3 u/c )

(5)

and the relationship between the initial slope of the Hugoftiot and a3 is obtained as

0.37 a 3

((dU/du) p=

- 1.62)

(6)

by setting u = 0 in Eqn.

(5).

The calculation of (6

p/av 2 )

at p = 1 bar The identity

with sound velocity data is

then sufficient to evaluate a3.

V2

aP

(7)

was used in the evaluation of a 3 because it derivatives Coc/v) and (av/6p)

7

is convenient to estimate the Data from echo8

with experimental data. together with p-v-T data,

sounding tables for pure water,

were used to

construct a graph of the variation of sound speed with specific volume at pressures of I, 50, 25 0 C, and c
=

100,

and 150 bar (see Figure 2). we estimate

At p = 1 bar, T

1493 m/sec,

c/yv)p 1 from the graph as

LCAV"p=1

1.027 X 101; g/sec cm 2 )

C-3

F?

Similarly,

from the graph,

at constant c = 1493 m/sec,

the following val-

ues of p and v were estimated:

-Av

(bar) ___ !
1 50 100 150

/g)
1.00295 0.00325 0.99970 0.00290 0.99680 0.00275 0.99405

Extrapolating to p = I,

we obtain

()=
0

(Lv)
0

=-7~ x 10-1

cms/(dyne g)

Substitution of c -7

= 1.493 x 105 cm/sec, g),

v = 1.00295 cm3 /g,

(av/ap
0

x 10-11 cm5 /(dyne

(7) gives (6c/6v)
1011 sec 2 /(g
3

and (6c/6v)p= = 1.027 1

X 106 g(sec cm2) into

Evaluating (C2p/av
2

Eqn. 2.12 Eqn.

= 5.66 x 105 g/(sec cm2 ). cm" 7 ) with Eqn. (4)

x
(3)

and (dU/du)p= 1 = 2.40 with (5).

gives a

3 =

2.1 by substitution in Eqn.

The region of Fig. 1 close to the origin has been replotted as Fig. 3, using Eqn. (2) and values of a= , 10, 5, 2, 1, and 0.1. Also plotted in Fig. 3 are experimental THugoniot points for glycerin obtained by Erlich 9
10

and for carbon tetrachloride obtained by Lysne. available, we conclude that a value of a3
=

Until further data are consistent with results

2 is

of the calculations on water and with the experimental data on glycerin and carbon tetrachloride. In other words, Eqn. (2) becomes:

U/c

= 1.37 -

0.37 exp[-2u/c

+ 1.62 u/c0

(8)

C -4

It

is

concluded that the modified universal Hugoniot curve formulat-

ed in this note improves the thermodynamic descriptior. of liquids in the

1-bar to 20-kbar region.

Acknowledgments We are greatly indebted to A. Systems Command, B. Amster of the U.S. J. Naval Ordnance

who initiated this research and S.

Jacobs of the U.S.

ifNaval

Ordnance Laboratory who suggested the form of the euto.We al-2

so thank D. F. Barnes of the U.S. Geological Survey for providing echo-

sounding data, and D. results on glycerin.

C. Erlich for permission to quote his unpublished

[
A,-

SC
4

-5

1
..

T7

REFERENCES

1.

F. C. Gibson, R. W. Watson, J. E. Hay, C. R. Summers, J. Ribovich, and F. H. Scott, "Sensitivity of Propellant Systems," Quarterly
Report October 1 to December 1, eau of Mines, Naval Weapons, Pittsburgh, 1964, U.S. Pennsylvania. Dept. of Interior, BurPrepared for Bureau of

Order 19-65-8023WEPS.

2.

R. IV.Woolfolk and A. B. Amster, Stanford Research Institute Project 4051, Technical Progress Report 67-2 (Semiannual), "Sensitivity Fundamentals," October 15, 1967. I. M. Voskoboinikov, of Exp. A. N. Afanasenkov, and V. and Comb.) No. 4 (1967) 585. Shaw, J. Chem. 3657. 1969. Phys. M. Bogomolov, FGV

3.

(Phys. 4. 5. 6. 7.

M. Cowperthwaite and R. R. Shaw, S. J. J. Chem. Phys.

53,

(1970)

555.

54 (1971)

Jacobs,

private communication,

D. J. Matthews, "Tables of the Velocity of Sound in Pure Water and Sea Water," Hydrographic Department, Admiralty, London, England, 1939. N. E. Dorsey, New York, Ehrlich, Lysne, J. "Properties of Ordinary Water-Substance," 1940. unpublished work, Chem. Phys. SRI, 1971. 5242. Phys. 26, (1957) 815. Reinhold Pub.

8.

Co., 9. 10. 11. D. P. J. C. C.

55 (1971) J. Chem.

M. Walsh and M. I1. Rice,

C-6

.4F

,U/co

U/c - 1.37

1.62 u/c1.62 u/c

Line a U/co - 1.37 - 0.37 cxp I-5 u/c 1

" Line b U/co
-

1.37 - 0.37 exp I-ulcol 4 1.62 ulco

0.5

1.0
ulco

1.5

2.0

2.5
TA-4051-448

FIGURE C-1

NORMALIZED U-u PLOT

C-7

--

1.006/

1.005

A _
I(1

1.004

1.003

1.002

S1.001
,, 1000 0 > 4 0.999
g.

I1bar

0.98 p =50 bar

0.997

0.996 0.995 0.994

p=100 bar lo25C

0.993 r 1400

I 1450 SOUND SPEED

-

1500

1550 TA-8525-f9

m/sec

FIGURE C-2

SOUND SPEED IN PURE WATER AS A FUNCTION OF SPECIFIC VOLUME AT

PRESSURES OF 1, 50, 100, AND 150 BAR

C-8

2.2

2.0

"03 00o u/c . 3

033

10
0a3

a3 2

1.4

Catrbo

Tetrachloride,

01

0.2

0.3
0

0.4

0.5 TA-8525.60

FIGURE C-3

iI

CALCULATED NORMALIZED U-u PLOTS FOR VALUES OF a3 =0.1, 1, 2, 5, 10, AND -IN THE r,)UATION U/c. 1.37 - 0.37 exp [-au/(, + 1.62 u/co 0
SHOWING ERLICH-Pi EXPERIMENTAL GLYCERIN DATA, AND LYSNE'S

EXPERIMENTAL CARBON TETRACHLORIDE DATA

C-9

Append ix D COLLAPSE OF A VAPORI BUBBLE' LEADING TO HIOT-SPOT INITIATION

17

L. iiili ii=.i

--l -,. il l i i ill im !!I i II

l - "l'l

II

COLLAPSE OF A VAPOR BUBBLE LEADING TO HOT-SPOT INITIATION

Bly

David C. 1,Yooten Ultrasystems, Inc.

Newport Beach, California. 92660

March 1972

This work was Supported by tile U.S. Research under Contract No.

Air Porce Office of Scientific

FI'41620-69-C-0079.

-?his is a I-eprint of a paper inten(ie(i for publication in a scientific journal. Since changes may be made prior to publication, this preprint is distributed with tile understanding that it will not be copied or reproduced u.ithout express permissio,, of the author. D-iii

Introduction The hot spots caused by adiabatic heating during the collapse of 1 bubbles or cavities w,,ere proposed by Bowden and coworkers in 1947 as a mechanism for the initiation of explosion in highly energetic liquids. This initiation mechanism was proposed to explain the high sensitivity of liquid or gelatinous explosives to mechanical impact. extensive subsequent investigations, As a result of

the effect of hot spots on the shock The early experidue to compressive

sensitivity of liquid explosives is well understood. 2 ments by Yoffe stpported the view that initiation is heating of entrapped gas or vapor spaces.

'[lie detailed mechanism of hot-spot initiation by bubbles in some cases may be due to processes other than heating by adiabatic compression. 3 For example, studies by Campbell ot al. show that initiation may take place by shock interactions caused by both solid inhomogeneities and 4 bubbles. Mader, in a two-dimensional numerical computation of the reactive flow around cylindrical and rectangular cross section rods or cavities placed parallel to the shock front, has given added support to the conclusions reached in these experiments. fleating sufficient to
S

cause initiation by cylindri.,',ily symmetric shock interaction with a spherical bubble was also comnputecd by Evans et al. 5Hot-spot ignition caused by shock interaction or focusing at a bubbl increasingly effective with higher shock strengths. it is is of course

On the other hand,

also conceivable that the Munroc jet effect could be responsible

for bubble and cavity initiation at higher shock strengths. 'Tis 6 mechanism was proposed by Bowden 6,nd subsequently microjet ignition 7 was olserved experimentally by Watson and Gibson and undeir different 8
conditions by Blowden and McOnJie.8 At lower shock strengths, however, 9 Gibson et al. have shown that thle velocity or the jet is not sufficient to cause shock initiation as tile jet of the bubble. collides with tthe opposite side

)-I1

(P'

Thus,

hot-spot initiation by a bubble due to both shock focusing and

microjetting are generally of interest for shock strengths in excess of about 50 kbar. 'Tihe important role of bubble hot-spot ignition in (1I,)), the initiation has recently

and propagation of low-velocity detonation stimulated renewed pressures in interest in

however,

hot-spot formation at lower shock overh'le most plausible model of LVD

the 1 to 10 kbar range.

10 a cavitation model recently suggested by Watson et al. and inde11 pendently by Voskoboinikov et al. In the cavitation model, precursor is container wall waves cavitate the liquid ahead of a liquid shock; the cavitation bubbles are subsequently compressed by the liquid shock causing hot-spot-initiated reaction; and the resulting energy release in turn supports the liciaid shock. This model has been further verified

by Watson 1 2 who hias photographed the cavitation ahead of the reaction zone and by Watson at al.13 who more recently have reported tentative

measurements of precursor wall waves capable of producing cavitation ahead of the reaction zone. In addition, the existence of individual

burning sites rather than a continuous reaction front has been observed 12 by Watson in x-ray photographs of the LVD reaction wave, and Gibson 14 et al. have experimentally verified the dependence of bubble ignition on the strength of the initiating shock. 'lThe cavitation model also explains the relatively long dark region 12 observed between the shock front and the luminescent reaction zone. Thus, the cavitation model, though it involves an incompletely under-

stood interaction between the container and the liquid explosive, the formation of cavitation and subsequent hot-spot reaction, and as yet

unknown criteria for stability, appears to offer the most plausible mechanism for the propagation of LIT).

D-2

(N

it..

... .... . .. .

Background 'Il dynamics of vapor bubbles has been a subject of extensive

theoretical

and experimental investigation for a number of years. In 15-17 ,asequence of early papers, Plesset and Zwick calculated the rate of growth of vapor bubbles in slightly superheate:l external pressure. variations in Their calculations steam undcr constant

assume that the temperature a thin thermal boundary

the liquid are appreciable only in

layer near the bubble wall. asymptotically proportional tension is negligible,

The predicted bubble radius R(t) is to t when It is large enough so that surface

a result that was confirmed experimentally by

18 Dergarabedian. More recent work has extended the work of Plesset 15ikl-17 19-21 and Zwick to more general cases. 2 Most of the work on the dynamics of vapor bubbles has been aimed, related phenomena. however, at boiling and

Of more interest here is

the collapse of a bubble

leading to high interior temperatures causing local initiation of reaction which may ultimately ignite the bulk material loading to explosion.

'Tlhoretical

studies of hot-spot initiation that are related to

22 the adiabatic compression of a bubble have been carried out by Zinn,-2 2"3 2'1 E'nig, and Gill. These are computations of ignition delay time
which are applicable to a bubhle with a stationary radiu, and with no however,

mass transfer at the bubble wall.

In the present computations,

we are interested in

the temperature rise and the onset of reaction We include here the erfects and

or ignition in a collapsing vapor bubble.

of both heat and mass transfer (vaporization) at the bubble wall,

formulate the problem in Lagrangian coordinates that move with the It is ass ,ned

bulbble wall during Collapse. liquid are lonviscous and,

that both the vapor and they have the same

prior to collapse,

temperature T 0 0 uniform,

The interior of the vapor bubble is that is

assumed to be

an assumption

justified provided that the thermal

D- 3

diffusivity of the vapor is much greater than that of the liquid and that the speed of the bubble wall motion is much less than the sound
speed in the vapor. The first condition is certainly true for most

substances for, although the thermal conductivity of the liquid is about ten times greater than that of the vapor, of the vapor is still the thermal diffusivity

much greater than that of the liquid. 'Te second 25 condition has been discussed by Hickling in connection with the collapse of a bubble of inert gas in a liquid. Ilickling found that the

sound speed in the gas was indeed much higher than the velocity of the -I -3 bubble wall for initial bubble radii of It = 10 - 10 cm. and
0

overpressures of about 4 atm.

Under such conditions,

the pressure in

the vapor bubble remains approximately uniform because the effects of disturbances at the wall will have time to be propagated throughout the bubble interior. It is clear that during the early stages of

collapse the assumption of uniform bubble interior will certainly be valid in the present calculation. Further, to keep the problem tractable,

a spherical bubble will be assumed even though shock compression in general produces asymmetric collapse and, in addition, it is known that

spherical bubbles arc generally unstable during the latter stages of collapse.

D-4I

Formulation of the Problem A very general formulation of the equations governing the motion of a bubble in a liquid was given by 11sieh, 26 as well as simplified With the

equations for a spherical bubble with a uniform interior. above assumptions and the additional assumptions that: (1) (2) (3) (4) (5)

No body force is present, There is no translational motion of the bubble relative to thle liquid; Viscosity coefficients are zero throughout; The liquid is incompressible; The equation for the temperature in the liquid T 2 (r, t)can be written as

S+ v 2

s--

(1)

whero

is

the radial coordinate from the bubble center and

is

tile thermal diffusivity of the liquid, which will be assumed constant. Since the liquid is vapor o treated as incompressible, and tile density of tile from the

is generally much less than that of the liquid,

continuity requirement the liquid velocity may be directly related to the bubble radius It(t) as follows:

2. V
=

(2)

That is,

the contribu tion of mass transfer (Vaporization) at tile bubble I is neglected. Note that thle assumnptinn of uniform bubble v 1' Tv v t = 0 are

surface to

interior requires that the quantities inside the bubble functions of time only. a bubble of radius It
0

The initial condition will be that at

is at rest, with aic vapor inside at initial

I)-5

density and pressure of uniform initial temperature

P, Pa,

respectively.

The assumption of

vo

'T 0

vo

for the system requires that

W (0)
v

= 'r(r,0)

'r'
0

(3)

Further,

the boundary condition at the bubble wall requires

that

r (t)

r'r(Ot)

(4)

andi - A and (0,tt) I =

-

1 4t 2

c d (it

43 11i 4 33

3 vRC

Ov)v IF dT v+ dt

+ v

+ (l T"it3)

QZ cxp

[-

P/~v

(5)

I when 1, is

0V the latent heat per unit mass, and ,is C is tile specific heat v the thermal conductivity

at constant volume of tihe vapor,

of the liquid. as are

T'ile quantities

X., L, and

are assumed constant,

Q, Z, and H which denote, respectively, tihe specific heat of

and the activation energy. 'Ilie last

reaction, the frequency factor, term in Eq. (5) is

the heat release in the bubble due to gas phase

generally negligible until 'I'v reaches the order of

reaction which is El R.

Since the (uniform)

temperature in

the vapor bubble is determined it follows that

by the temperature of the liquid at tile bubble wall, the (uniform) pressure in the bubble Pv is

given by the vapor pressure We will therefore the bubble

of the liquid at the wall or bubble temperature. make the additional

assumption that the vapor pressure is 'Tus,

wall and that the bubble contains pure vapor.

lhe possibility that

D-6

T4

absorbed gas in the bubble is

the liquid may contribute to the interior pressure of and the results will be applicable to a pure

excluded,

liquid or to cases in which the partial pressure of absorbed gases is negligible in comparison with the vapor pressure. certainly valid in The equilibrium vapor

pressure requirement is

the early stages of collapse the bubble becomes comparable

until the rate of change of the pressure in to the kinetic rate of vaporization. analytically,

To enable us to treat the problem and hence

we will relate the vapor pressure of the liquid, the bubble,

the pressure in

to the temperature at the bubble surface by

use of the Clasuis-Clapeyron equation

0 (1

(Pexp

T )T

(6)

where

is

the ideal gas constant and temperature T0 . 0 Further,

is

the vapor pressure at

the initial

the thermodynamic quantities in

the vapor phase will be related

by the ideal gas equation of state

'P

=S~

'T

(7)

The additional equation required

to complete the formulation of the

problem is the differential equation for the motion of the bubble wall 26
i" + =-(

where

, assumed constant,

is

the initial

pressure in

the bulk liquid

and corresponds to the pressure at large distance from the bubble during

collapse.

The initial conditions For Eq.

(8) are taken as

R(O) i

it

(9)

ko7

S1D-7

r.

Equations

(l),

(2),

(6),

(7),

and (8)

now form a set for the T v (t), and o t)

dependent variables

VA(r,t),

Te(r,t),

P v(t),

subject to the conditions of (3),

(4),

(5),

and (9).

Introduce the dimensionless variables

=

l
0

(r 3

R3(t))

:.,3R

t= R
=

t/T
C

R(t)"/R

'T SV

T
o0

(10)

'V 0- ~
, and the parameters

I'
0

77

L/RT
0

(P

2(/n

ST

/a

TrR =

,, /QZ

where

is

the characteristic

time for bubble collapse

is T
r

the is (5).

characteristic

time for heat transfer to the bubble wall,

and Eq.

the dimensionless frequency factor for the reaction term in Equation (1) can now be written as

D-8

aTh

--

.[(+. 1

f(12)

and the conditions (3),

(4),

and (5)

become respectivelyj

v (0)
vA t

=
=0(o,t)

0 (xO)

0,

(13)

t -o

(14)

ande

26

{0

(n
it

+1

-- ) L- )
It

i+e

+7-i
v

+r

is

the initial

density in

the vIpor phase.

Thc bubble equation (8) can be written as isteiiil densty i t" 2 vao phse
S\d II

3 --

ci

(16)

D- 9

with the initial

conditions

R(0) and dR
dt

(17)

()

0(18)

D)-10K

Small-Time Solution For small time defined by ' form

R

<< T
c

the bubble radius has the

C(o)

+ t/2

.2

(19)

(I R

(20)

dt Since

0 (0) = 0, the small-time expansion of

v

9 (t)
V

is

1 I<

A similar statement may be about about In general, for small time, since

9 (x,0). T

for situations of << T 11 c interest here, the temperature variation in the liquid occurs in a thin layer near the surface of the bubble such that in Eq. neglect x/J3
A3

(12) we may
,

relative to unity;

that is

, x/I1 = x

< 1

an

approximation that is discussed in decail in References 15 to 17 and 19. Thus Eq. (12) may be written for small time as

(0)

W
,=.(21)

where

T c/T

and where Eq. Similarly, Eq.

(1.9)

is

used and only first-order ro the first order is

terms are retained.

(15),

ON

S(o)(22)

D-13

and the first-order parts of Eq.

(13)

and

(14)

are

e()xo)

(0 (X,0)

(23)

,(On)
U (0,t)
=

(0)

(o)

(24)

The reaction time is neglected since it until

is generally very small Now write Eq. (22)

becomes suobstantially greater than unity.

in the form

(:)
(0,0) - A (I t where

(t )
B t, (25)

A = 0/3 (77

- 77+

7 -

(26)

I3 =

0()+l)

(27)

and

(---1) (

Y-o

"
c

(28)

The problem consisting of the differential operation (21) with initial conditions (23) and boundary conditions (22) 27 and (24) belongs

to a class of problems solved by Carslaw and Jaeger transforms. 'Tie solution for small time is

using Laplace

(0),) (X,t)

3 .

2) el

\XABA

x
+

t:

AB
-r
n=0

xP

2*

erfe

1 T

+(et)

-2&
AU

ineric

x
2 (&ZL)(

(29)

i)-12

Wher

Where

i erfey

refers to the

ni

integral of the

erfcy

function

(see Reference 27). surface (x = 0) is

Consequently,

the temperature at the bubble

Vv

O,

(o)

"2

(t)

___

+_1_3___

(30)

"In comparison with the temperature in the bubble for an adiabatic

collapse (no heat or mass transfer):

v u

77+1 77
2

1
(31)

vi

77+

where

9Va

is

the temperature for the adiabatic collapse of the same

bubble. Generally

77

is in

the range of 10-20

so the initial

temperature rise in a collapsing vapor bubble with heat and mass transfer at the bubble wall may be almost an order of magnitude less than the corresponding rate of temperature rise for the adiabatic collapse of a similar bubble. The temperatuire profile for small time is a representative set of parameters. plotted in Figure 1 for

DI13

Rieferences

1.

F.

P. Bowclen, M. F.

R. Mulcahy,

i.

G. Vines,

and A. Yotfe,

"lie

S~Detonation of Lilquidl Explosives by Gentle Impact:

Gas Minute Spaces," 2. A. Yoffe, Proc. Roy. Soc. A.188,

The Effect of

291 (1947).

"Influence of Entrapped Gas on Initiation of Explosion in Proc. Roy. Soc. A. 198, 737 (1949). "Shock Initiation 498 (1961).

Liquids and Solids," 3. A. W. Campbell,

W. C. D)avis,

and J. R. Travis, Phys.

of Detonation in Liquid Explosives," 4.. C. I. Mader,

Fluids, 4,

"Initiation of Detonation by the Interaction of Shocks Phys. Fluids, Meizner, 8, 1811 (1965). "Interaction of Shock 651 (1962).

with Density Discontinuities," 5. M. W. Evans, F. If. Harlow,

and B. 1). Phys.

or Rarefaction with a Bubble," 6. F. P. Bowden,

Fluids, 5,

"The Initiation and Growth of Explosion in the 9th Symposium (International) p. 499. ".Jets from Imploding Bubbles," on Combustion,

Condensed Phase,"

Academic Press, N. Y. (1963), 7. R. W. Watson, Nature, 204, 8. F. P. Bowden,

and F. C. Gibson, 1296 (1964). and M. P. McOnie,

II
9.

'Cavities

and Micro Munro -Jets in :380 (1965).

Licjiaids:

Their Role in Explosion,"

Nature, 206,

F. C. Gibson, 11. W. Watson,

and F. IH. Scott,

J. E. hlay, C. It. Summers,

J. Ribovich,
Bureau of

"Sensitivity of Propellant Systems,"

Mines Quarterly Report, -January I to March 31, 10. It. W. Watson,

Bureau of Naval Weapons Order 19-66-8027-WEPS, 1966. F. C. Gibson, and II. IV. Van D)ol ah1

C. It. Summers,

D)etonations in Proceedings,

l(iquid Exploslves--TIhe Low Velocity Regime, on I)etonation, Office I).C., ACH-126;

Fourth Symposium (International) Dept. or the Navy,

of Naval Research, p. 117.

Washington,

D)- 14'

11.

I. N.

Voskoboinikov,

A.

V. Dubovik,

and V. K. Bobolev, Inst. Chem. 161 (5),

"LowAca,".

Velocity Detonation in Sci. 12. USSR, Trans.

Nitroglycerine," Akad. Nauk,

Phys.

from Dokl.

1152 (1965)'.

It. W. Watson,

"The Structure of Low-Velocity Detonation Waves." on Combustion, 723. flay, and I. W. Van l)olah, "l-ie 'I'hc Combus' ','n

'i'welfth Symposium; (International) Institute, 13. I. Pittsburgh, J. (1969), J. p. E.

W. Watson,

Ribovich,

Stability of Low-Velocity Detonation Waves," (International) 14. F. C. Gibson, I. on Detonation, WV.Watson, J.

Fifth Sympos_,-f (1970), J.

,'.

ONR Report DIR 163, H. Hay, C.

71.

it. Summers,

Uiibovich,

and F. Report,

II. Scott, U. S.

"Sensitivity of Propellant Systems," of Interior, Bureau of Mines,

Ouarterly

Dept.

Pittsburgh,

February 2, 15. M. S.

1966. and S. A. Zwick, "A Nonsteady Heat Diffusion J. Appl. Phys. 23, 95 (1952).

Plesset,

Problem with Spherical Symmetry," 16. M. S. Plesset, and S. A. Zwick, J. Appl.

"The Growth of Vapor Bubbles in Phys. "On Phys. 25, 493 (1954).

Superheated Liquids, 17. M. S. Plesset, and S.

A. Zwick, J. Math.

the D)ynamics of Small Vapor 33, 308 (1955).

Bubbles in 18.

Liquids,"

P. l)ergarabedian, Heated Liquids,"

"Observations on Bubble Growths in Various Super,.. Fluid Mech. 9, 39 (1960). "Spherical Bubble

19.

G. Birkhoff, Growth,

It. S.

Margulies, 1, 201

and W. A. Horning, (1958).

Phys.

Flu.ds,

20.

S.

A.

Zwick, Fluids, Gibson,

"Growth of Vapor Bubbles in a Rapidly Heated _3,685 (1960).

Liquid,

Phys. 21. 1). C.

"The Kinetic and Thermal Uxpansion of Vapor Bubbles," 71-FHi-13, presented at Fluids Engineering Conference,

ASMI, Paper No. Pittsburgh,

May 9-12,

1971. l )-15

22.

J. Zinn, "Initiation of Explosions by Hot Spots," 36, 1949 (1962).

J. Chem. Phys.,

23.

J. W. Enig, "Approximate Solutions in the Theory of Thermal Explosions for Semi-Ihfinite Explosives," 205 (1968). Proc. Roy. Soc. A. 305,

24.

S.

P. Gill, "Initiation and Decomposition of Nitromethane at l0-kbar Physics International Technical Rleport-70-2, March 1970.

Pressure," 25. It.Hickling, J. Acoust. 26.

"Effects of Thermal Conduction in Sonoluminescence," Am., 35, 9F7 (1963).

Soc.

1).Y. I.sieh, "Some Analytical Aspects of Bubble Dynamics," J. Basic Eng., 87, 991 (1965). and J. C. Jaeger, Conduction of Hfeat in Solids, London (1959).

27.

HI.S.

Carslaw,

Second Ed.,

Oxford University Press,

C\2

'7 *15. 001 y~1.400

h 100.0 T- 00100 x 10

0,1

0.1

=,

7- 0.06-

S0.04

0.01

I--

---------

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

GA-7771-19

FIGURE D-1

TEMPERATURE PROFILES IN'LIQUID-FOR SMALL-TIME SOLUTION (EQ. 29)

D-17

S1

-i?

Appendix B ME'ASUREmEn' OF PflESSIJRP IN LOW-V~ELOC!'rY D)ETONATION

F
i*
MEASUREMENT OF PRESSURE IN LOW-VELOCITY DETONATfON

By David C. Erlich Stanford Research Institute Menlo Park, California 94025

May 1972

work was supported byv the U.S. Air Force Office of Scientific SThis Research under Contract No. F4'1620-69-C-0079. This is a preprint of a paper intended for publication in a scientific journal. Since changes may be made prior to publication, this preprint is distributed with the understanding that it will not be copied or reproduced without express permission of the author.

E-iii

ABSTRACT A pair of piezoresistant ytterbium stress gages were used to record pressure-time histories in liquld monopropellants in a study of the shock initiation of low velocity detonation (LIM). The

propellant was contained in a cylindrical steel tube, and the initiating shock was produced by a tetryl pellet-attenuator donor system. Variation of the propellant, the wall thickness-to-

diameter ratio of the tubes, 30 to 50 kbars resulted in (a)

and the initial shock pressure from shock decay with no appreciable initiation of high velocity

reaction,. (b) initiation of LVI), and (c) detonation (H\YD).

Peak pressures recorded at 4 to 10 inches down the and in the

tube were in the 1- to 2-kbar range for shock decay, 4-

to 7-kbar range for L\I), but no pressures were recorded for IIVI).

The average LVD propagation velocity measured with the gages was about 2.0 mm/gsec for ethyl nitrate and about 3.0 mnm/psec for FE.FO.

E-v

Introduction This paper presents an experimental investigation of the shock

initiation of low velocity detonation *(LVD) in liquid monopropellants contained in cylindrical steel tubes. The object of the work was to

achieve a more quantitative description of LO),

Since LVD propagation 1,2 velocities had been measured in previous studies, and the circum3 ferential strain histories had also been examined, a better description of states attained in the liquid was required to achieve this objective. Piezoresistant in-material ytterbium stress gages were therefore used in the experiments to measure the pressu.re time histories of liquid monopropellant during the initiation and propagation of LVD. A pair

of gages were used in each experiment so that the pressure records could be used to determine propagation velocities.

IK-

,dJ

Experimental Procedure Figure 1 is performed in a schematic diagram of the LW) experiments that were Institute explosives vault facility.

the Stanford Research

A cylindrical seamless cold-rolled steel tube was used to contain the liqliid. To the bottom of the tube was attached a 3-inch diameter disk two 2-inch diameter by 1 inch thick pellets of tetryl, The tetryl

of tlomalite,

and one 1/4-inch thick by 1/2-inch diameter tetryl pellet.

detonation was initiated by an exploding bridgewire detonator in contact with the small pellet. 'lThe Homalite disk served as a stress

pulse attenuator and its thickness was chosen to produce the desired peak pressure in the liquid.

To determine this pressure, a standard

the

four-terminal pie, )-sistant manganin stress gage was placed in

flomalite 1/32 o' aih inch from the surface in several of the experiments to record Uhe peak stress that propagates into the liquid.
T'|wo Ve-shaped photoetched 4-terminal piezoresistant ytterbium stress gages,

from 0.001-inch thick foil, were mounted on a long strip

of 0.003-inch-weave fiberglass cloth whose width was the same as the

inside diameter of the tube.

Long thin copper leads were soldered onto

the gage terminals and a layer of Hlomalite was spread over the fiberglass to contain the gages and leads in a stiff package. 0.007-inch thick

Fiberglass fins to ensure that the gage package remained in andi the package was slid into the

the center of the'tube were attached :ube and glued

into place with the leaus emerging from the top.

This arrangement of the gage package was motivated by se\,ervI factors. seC.*ioii in First of all, since the package had a very small cross

the plane of the expected shock front (perpendicular to it wouldn't move much as the shock andi/or detonation Second, any motion of the

the tube axis),

wao. ;orced the liquid to move up the tube. Trade name, llomalite Corporation.

E- 2

gage package would not stretch and break the leads as quickly as if leads had been attached to the tube wall. And third,

the

since the coupling to the 1,11)

of the liquid and tube wall shock responses is phenomenon, of the wall, it was desirable to avoid

very critical

interference with the integrity

as would have occurred had the leads needed to emerge from

the walls. One disadvantage of this orientation of the gage was an increase in the response time of the gage. Since the thickness of the active element in the direction of shock propagation was about 0.020 inch,

the fastest rise that could be seen by the gage was about 250 nsec. For several experiments, one or two Micro-Measurements Type EA-06-

5006B-120* strain gages were glued to the outside of the tube wall adjacent to the location of the stress gages on the inside of the tube. They were oriented to measure hoop or circumferential strain and thus 4 to compare the strain histories in these and ca. lier experiments, and possibly to determine the time lag between I\) initiation at the

center of the tube and the circumferential expansion of the tube. Eight experiments were performed as part of the L\I) program. The

relevant experimental parameter for each of the shots ij; given in the upper half of Table 1. For the first 5 shots, ethyl nitrate was used

for the liquid monopropellant and wall thickness-to-diameter ratios and attenuator thicknesses were chosen that had previously been successful in it LVI) initiation in similar materials. was difficult
3

'4

When it

became clear that ethyl nitrate

to create a steady-state 1I,\) it

wave in

(see next section), ethylformal), After shot No. was analyzed

was decided to switch to FEFO (fluorodinitro-

a liquid known to be more sensitive to LVI) initiatLon. 6 failed to cause detonation in the FEFO, tile liquid

and was shown to contain a considerable

amount of impurities.

A new batch of FEFO was prepared for shot Nos.

*

7 and 8.

Micro-Measurements,

Romulus,

Michigan.

E- 3

',4,.,

All of the shots were instrumented except for shot No.

purpose was simply to determine by terminal observation if

2, whose
LVI) had

taken place.

For all of the instrumented shots, a pulsed constantconjunction with the ytterbium and on.oscillographs coefficient of to stress) in

current power supply was used in

manganin stress gages and the signals were recorded the manner described by Keough. ytterbium (the used in 6

The piezoresl',tance in gage

factor relating change

esistance

that data analysis was that determined

by Ginsberg. 7

-!:fI'

"Fxperimental Results
The :'esults of the IAN) lhalf of Table 1. experiments are tabulated in the bottom

Terminal observations of the remains of the steel if any, the liquid

tube were made to determine \lhat type of detonation, underwent.

Figmro 2 contains photographs of the tube remains for the as well as a photograph of

three different outcomes that were observed, the tube prior to detonation. nation took place,

For the case in which no liquid deto-

the tube Is undamaged except for the first inch or For the case in which the liquid

two adljacent to the explosive pellet. underwent IAN,

the section of tube below the liquid level broke into Some of

several pieces which were often severely bent or twisted.

these pieces did not break off from the und Imaged upper section of the
tube (the section above the liquid line) but bent back beyond the undamaged section in the shape of a partially peeled banana fruit removed) . No. 3, 4I, or 5, (with the

'llTis was particularly true of the ethyl nitrate shots where analysis of the gage records indicated a nonstealdy

state ID,

as will be discussed shortly.

For the case in which the

liquid tunuerwent 1lvi),

the tube broke into many long very thin fragments, the tube (luring

as would be expected due to the very high pressures in an IIVD. For shot No. 7 in which IDVI) occurred,

no stress records were

obtained due primarily to high propagation velocity of the IMVI) wave (between 6 and 7 mrn/ILsec)), which caused the wave to pass the ytterbium

gage location long before the triggering of the gage power supplies and oscilloscopes, which had been set to expect a slowet "'-Aocity IAT) wave.

The peak pressures recorded by the various stress gage.'- are also given in 'T'able I and tI'e average propagation velocity of the peah stress soon by the ytterbium gages for the shots in which both gages produced nalyzable records. A typical manganin gage record, that from shot No. 1, is shown in Figure 3. is:There is an initial peak of -. 59 kbar which is

S.. ..

-5

the stress that first by a drop to and is

reaches the gage plane in which is

the Hlomalite,

followed

_-16 kbar,

the stress induced

into the liquid,

lower clue to the impedance mismatch between the Ilomalite and the Since the attenuator thickness was neorly the same for shots

liquid. Nos. 1-5,

and Jince the explosive pellet configuration was identical it can be assumed that very nearly the same 1-5. Manganin ;t43 kbar

for all of the shots, stress was induced

into the ethyl nitrate for shots Nos.

gage records for the shots using FEFO yielded a pressure of for shots Nos. and

6 and 7 using approximately a 1-inch thick attenuator, 8, which used a l.-inch thick attenuator.

-; 33 kbar for shot No.

"The ytterbium gages generally worked very well in measuring the liquid pressure histories for the shots in which the liquid underwent I,\). A couple of the ytterbium records (those from shot No. 8)

exhibited a significant amount of high-frequency noise,

but the The

remainder of the records had excellent signal-to-noise ratios. records from the two ytterbium gages in shots Nos. in Figures d and 5, 3, respectively. ,4, and 5,

3 and 5 are shown

Observation or the various .ecords showed that tile pressure histories for each shot. This

from shots Nos.

however,

at the two locations are qualitati'vely dissimilar is most clearly seen in Figure 4,. TIe first gage,

located at a distance

of 6 inches froa

the detonating end of the tube,

recorded a rapid rise

(less than 1 ;Lsec)

to a peak pressure of about 2 kbar and then a slow Furthermore, the gage remained rise. continuous for the

decay to zero pressure. at least 20 nsec second gage,

following the initial

On the other hand, of tile tube,

located at 10 inches from the end

exhibited

a slower rise (more than 3 t:sec) inearly level plateau until later. I1,V),

to a stress of' over 5 kbar and then a

the gage began to stretch and break 3 ILsec is more like that to be expected former is in an

The Iatter gage record

while ,VI)

the lower stress and 1ongevity of Lhe in

indicative The records

of no

the immediate vicinity of the gage package.

H>-6

shot No.

5 exhibit simnilar tendencies, Sfrom

as shown in Figure 5,

except that here the gage,

has a higher peak pressure the gage located 7. kbar). inches

located 41 inches from the end of the tube,

(4.1 kbar) and shorter recording life than

from tile ends (with a peak pressurc of 2.7

Since previous experimental studies and computer simulation

indicate that IM) should start at distances diameters down thle tube, it

'f less than a few inside

is reasonable to assume that if a steadytile tube, it would begin before tile appears

state I1,) were to be initiated in

pressure pulse reaches the first gage location. Therefore it

that a steady-state lArf) wave was not produced ill tile region between the two ytterbium gages.
prcdluced,

More likely a pulsating or oscillatory lI,)

few isolated hot spots were initiated

was
at

or perhaps merely a

various points in

tile tulbo.

The higher pressure seen by one of tile 5 as compared velocity to shot No. ', tile

gages in both shot No.

3 and No.

higher peak pressure p)ropagation

(about 2.00 mm/lIsec as compared

to 1.55 mm/gsec),

and tile fact that the tube in shot No. 4 was less

damaged 0than thle tube in shots f3 and 5 indicate that significantly less of the liquid in shot No. ,I underwent INq) than in shots No. 3 or No. 5.

Many explanations are possible for the failure to initiate a steadystate IVi)) In ethyl nitrate. Insufficient: experimental data are

available

to predict precisely the conditions under which such a phenoumi-

enon will occur.

It is

therefore very possible that the experimental

parameters chosen, such as the wall thickness-to-tube diameter ratio anod attenuator thickness, were slightly different from those needed to acllieve steady-st-ce ,VIAD. Perhaps ethyl nitrate is sufficiently insensi-

tive in tle region of these parameters.

The two stress gage histories for shot No. 8 using FIWFO as the

l ldtui(I were muclh more (1 zali tatively al ike, IfolSe obscured their structure somewhat.

although high frequency FurLthrmore the higher

p
-7

measured

peak pressures

(-Ito 7 kbar) and faster average propagation by a lower stress, FEFO

velocity (3 ,nmm/sec),

despite the liquid being hit

indicated that a steady-state LVq) was more likely initiated in than in the ethyl nitrate.

'llTe final experimental result to be reported was that of the strain recorded for shot No. 4 (the strain gage records in shot No. 1

were too noisy to analyze).

'Ihe peak circumferential strain seen at Comparison

the distance of 10 inches from the end of the tube was 0.2%,0. with the computer simulation reported

in Appendix A indicated that a

0.2% strain corresponds to an internal liquid pressure of a few kilobars, which is just that recorded by the stress gage. Furthermore LI) would

produce a circumferential strain in shortly after the wave passed

the tube wall of more than 1% This is yet

that section of the tube. in 4.

another indication of the absence of LVI)

the region 10 inches from

the detonated en(I of the tube for shot No.

CE1"8

Conclusions 4 Although peak pressure measurements have been reported the results reported here are the first direct measurements of pressure history in a liquid undergoing LVI). initiated, In the shots in which successful LVD was

peak pressures of about 4 to 7 kbar were observed which is 1-4 consistent with other observations of LVD. The measurements were taken with a minimum of interference with the detonation wave by the use of very small gages. Therefore, the rise times and pressure

history measured by the gages are believed to be closely representative of the true stress histories in an LVI) wave. This technique should

prove useful for recording pressure history ahead of and within LID waves for comparison with theoretical prediction and to gain a better understanding of LIM) behavior in various liquids as a function of confinement, initiating shock strength, and other experimental parameters.

E-9

Acknowledgment The author wishes to thank Dr. it. W. WVoolfolk for preparing the

liquid samples and for many helpful discussions concerning the initiation of low-velocity detonation.

E-10

References 1. I. W. Woolfolk, and A. B. Amster, "Low Velocity Detonations: Some

Experimental Studies and 'IlTir Interpretation," (International) (1969) p. 731. 2.

'T'welfth Symposium

on Combustion, 'lle Combustion Institute, Pittsburgh

R. W. Watson, C. R. Summers, F. C. Gibson, and R. W. Van Do]ah, "Detonations in Liquid Explosives--The Low Velocity Regime," Pr.oceedings, Fourth Symposium (International) on Detonation,

ACR-126, Office of Naval Research, D.C., 3. p. 117.

Dept. of the Navy, Washington,

R. W. Watson, J. Ribovich,

J. E. flay, and It. W. Van Dolah,

"The

Stability of Low-Velocity Detonation Waves," (International) 4. 1). C. Wooten, on Detonation,

Fifth Symposium (1970) p. 71.

ONR Report DR 163, I. W. Woolfolk,

H. H. Bredfeldt,

and It. J. Kier,

"Investigations of Low-Velocity Detonation Phenomena In Liquid

Propellants, Fuels, and Explosives," Annual Report, AFOSIH Scientific

Report AFOSH 70-1005111, March 1970. 5. 6. It.W. Woolfolk, Private Communication. 1). 1). Keough, "Development of High Sensitivity Piezoresistive Shock Final Report, (1968) Stanford

Transducer for the Low Kilobar Range," Research Institute Project No. 7. M. I). Ginsberg, PGI 6979,

(unpublished).

"Calibration and Characterization of Ytterbium Final Report, Stanford Research Institute

Stress Transducer, Project No.

7511 (1971)

(unpublished).

~~~ I'
0 t,.V

')

a)

04^1 -1 I y t 44 44-

.4

Lk

'k

41.4

~4J0

C
'4 0

I
.4

0 u

~ w*

CC) 4w
Ir.4.

4.

00 414 A44 c O N ~4 4 414 '-4

4IE -lr

LU W

2r j 0

1.

41 %44

CO

CD
14

.4

M4 W)
.4
.

k
U

Ud N .

4.

C-e

1-4 4-.
'~

r ON' el k'
.4

U) S.' 4N 0 4
V;404

C3

0 0

0. WCj 41v 0 'N 0

u.

fn)
o

a)1
ZJ44

0 Q .0

1-'-

LEADS FROM STRESS GAGES

SEAMLESS COLD-ROLLED STEEL PIPE LIQUID FILLED THIS LEVEL

-TO

STRAIN GAGES (Oriented to Measure Hoop

Strain)

4-TERMINAL YTTERBIUM STRESS GAGES

HOMALITE ATTENUATOR

4-TERMINAL MANGANIN STRESS GAGE

PL-2 DETONATOR TETRYL PELLETS

GA 1111. 1411

FIGURE E-1

SCHEMATIC DIAGRAM OF LVD EXPERIMENTS

F-13

U)

UJOD
4 -r-

0,.

wu
WW
-Z,

-4t

-.

;i-

_.

____

~U. Wui

0cr. U <
D) D 00w CC
0.(

(n Z0

LU.

1I5u

Reproduce best~L

L)o ti,~becoy

Reproduced from besl availabl~efcp

2pwt

Gl' 7771

16

FIGURE E-3

OSCILLOGRAM FROM MANGANIN STRESS GAGE IN SHOT NO I

il

GP-7771-17A (al 6 Inches DOWN THE TUBE

i)-/771-178

(Ib 10 if-ches DOWN THE TUBE

(-w 7771 I1

FIGURE C[--

OSCILLOGRTAMS FROM SHOT tN1O3 FOR THE

YTrIFi3IUI SRSS GAGES AT DISTANCES S1I.IVN IN i, I AND (b)

Reproducediom f best available copy..__ N1.;'l

4.1 kbar

6 psec

(a) 4-1/2 inches DOWN THE TUBE

2.7 kbar

10

usec

(b) 6 inches DOWN THE TUBE

GP 7771-18

FIGURE E-5

OSCILLOGRAMS FROM SHOT NO. 5 FOR THE YTTERBIUM STRESS GAGES AT DISTANCES SHOWN IN (a) AND (b)

1.-17

P-1 7