CP620, Shock Compression of Condensed Matter - 2001edited by M. D. Furnish, N. N. Thadhani, and Y. Horie© 2002 American Institute of Physics 0-7354-0068-7/02/$ 19.00

TRANSIENT DETONATION PROCESSESIN A PLASTIC BONDED EXPLOSIVE

Keith A. Thomas1, Eric S. Martin1, James E. Kennedy1, Ismael A. Garcia1, andJoseph C. Foster Jr2

2Los Alamos National Laboratory, Los Alamos, NM 875452Air Force Research Laboratory, Eglin AFB, FL 32542

Abstract. Experiments involving the transfer of detonation from small booster charges of PBXN-5(95% HMX and 5% Viton A) into larger charges of various plastic-bonded explosives (PBXs) haveproduced some surprising results and have stimulated investigation into the factors governing observedresponses. To understand these results, we conducted a series of tests with different miniaturedetonator-booster configurations using laser velocimetry to quantify the pressure pulse that istransmitted from the PBXN-5 booster. Models were used to determine the ideal explosive behavior forcomparison with the measured results. The differences are interpreted as being due to transientbehavior and late-time energy release from the booster charge. We characterize these behaviors asevidence of microdetonics, where we define microdetonics as the study of less-than-CJ detonationperformance due to curvature and/or transient behavior. This provides useful insights into thefundamentals of the detonation process that can feed into advanced modeling approaches such asDetonation Shock Dynamics (DSD).1'2

INTRODUCTION

Research on the detonation of condensed phaseexplosives spans many decades in both length andtime scales. Early work was limited primarily bythe instrumentation response. Improvements in theinstrumentation have revealed many aspects of thedetonation process that are best studied during theshock-to-detonation transition phase underconditions which represent high geometricaldivergence. An asymptotic solution to the classicalZel'dovich, Von Neuman, Doering (ZND) theory ofdetonation, called Detonation Shock Dynamics(DSD), has emerged in recent years.1'2 Continuumvariables emerge out of DSD for the influence ofcurvature of the detonation front and the reaction-zone chemistry on the detonation wave propagation.It has been expanded to include transient effects of

accelerating detonation waves.3'4 Experimentalcalibrations of DSD to date have addressed only thecurvature effects, through measurements of thecurved front in steady-detonation experiments incylindrical columns only slightly larger than thefailure diameter5 and with high convergence.6Those efforts have focused on explosives that areused for the large-volume, main charge of anexplosive assembly. However, an essential elementof the detonation process is initiation, which isusually triggered by a detonator and booster systemthat occupies a small volume in comparison to themain charge.

Historically, very little attention has been paidto the performance of real initiation systems. It isgenerally assumed that the initiation-systemexplosives operate at steady-state C-J detonationconditions as they are used to transfer detonation

1031

into the main explosive of interest. However,depending on the experimental configuration, theminiature detonator-booster system's small sizeimplies a highly divergent, transient region ofdetonation physics, which we call microdetonics.Microdetonics refers to the combination ofbehaviors dominated by transient effects indetonation, such as detonation acceleration andspreading, or curvature effects associated withinitiation by small sources. This paper is an attemptto determine whether these dynamics manifestthemselves in miniature detonation systems withhighly localized, short-duration, shock pulseoutputs. Thus we will focus our efforts on thebehavior of the second element in the explosivetrain commonly referred to as the booster.

EXPERIMENTS

To determine the presence of transient effectsin the output of charges typically used in detonatorsor boosters, we consider a few differentexperimental configurations. For most of ourexperiments we use a typical detonator arrangementwith an electrically driven exploding foil initiator(EFI) detonating a pellet of pentaerythritoltetranitrate (PETN). The PETN output drivesdetonation into a booster pellet of PBXN-5, whichis a commonly used pressed explosive boostercomposed of 95% HMX with 5% Viton A binder.The PBXN-5 in all of our tests was pressed to adensity of 1.77 g/cm3. Since most real initiationsystems are sealed in cups, we observe the outputsthrough a barrier material. Although interaction ofthis barrier with the detonation front complicatesthe shock waveform, we feel that it represents areal-world constraint in designing or analyzingexplosive initiation systems.

To quantify the performance of these miniaturedetonator-booster systems, we used a VISARsystem to measure the particle velocity history of anexperiment driven into a coated lithium fluoride(LiF) window. The impedance of the Al barrier andthe LiF crystal are closely match to minimize anyreflections at this interface. The VISAR system forthe experiments had a dual-leg arrangementutilizing two interferometer cavities with differentfringe constants. This arrangement enablesunambiguous recognition of and correction for

missed fringes in the data trace, and thus providesmore accurate data.

In order to determine deviations from idealbehavior, we simulated the experimentalarrangements with the CTH hydrocode.7 The one-dimensional model we used applies a constantenergy release rate across the detonation fronttraveling at full detonation velocity to achieveproper C-J pressure. Thus, it represents an idealsteady-state explosive behavior and serves as abasis for comparison for the measured VISARrecords. Any observed deviations from thepredicted CTH models should indicate the presenceof transient or nonideal behavior.

Case 1: Aluminum Barrier

The first experimental setup consisted of anexploding foil initiator detonating a 4-mm-diameterx 2-mm-long PETN pellet in contact with a 5-mm-diameter PBXN-5 booster pellet. A 0.254-mm-thickaluminum barrier plate between the PBXN-5 andthe LiF crystal simulated the booster cup. VISARmeasurements were then made of the particlevelocity at the aluminum-LiF interface.

Figure 1 shows the measured particle velocitiesin the LiF for two lengths of PBXN-5 pellets, 6mmand 2.2 mm, as well as the CTH simulations forboth lengths. There is a large discrepancy betweenthe calculated and measured velocity histories forthe 2.2-mm-long pellet. The lower peak amplitude

— 6 mm N-5 / Al Calculated—2.2 mm N-5 / Al Calculated— 6 mm N-5 / Al Experimental—2.2 mm N-5 / Al Experimental

0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40

Time (us)

FIGURE 1. Particle velocity for the 6-mm and 2.2-mm longPBXN-5 pellets with an aluminum barrier. CTH calculationsrepresenting an ideal explosive behavior are shown forcomparison.

1032

in the measured waveform is approximately half ofthe C-J pressure. The shapes of the waveformsbehind the front are also considerably different. Themeasured waveform displays a relatively flatsegment just behind the front, and its subsequentdecline in velocity is slower than that in thecalculation. We interpret this to indicate thepresence of an extended reaction zone that extendsthrough the supersonic and the subsonic portions ofthe flow behind the front.

The measured velocity history for the 6-mmpellet is higher than that for the 2.2-mm pellet, butit is still lower than the calculated curve for 6 mm.Also, the slope of the wave immediately behind thefront more closely resembles the CTH model. Thisindicates a buildup of detonation has occurred inthe interval between the 2.2 and 6 mm lengths, butthe detonation front has still not completely attaineda full C-J detonation.

Case 2: Initiation Methods

In case 1 the initiation of the booster wasaccomplished by placing the output of a detonatorpellet in contact with the PBXN-5 booster. Todetermine what effect the form of initiation has onthe booster performance, we conducted tests with aconfiguration designed to initiate the booster via aflying plate that is believed to overdrive detonation.A PETN pellet, initiated using an EFI, propelled a0.127-mm-thick, 2-mm diameter stainless-steelflyer across a 0.91-mm gap. The booster pellet forthis case was a 2.2-mm PBXN-5 similar to thatused in case 1. The output of the PBXN-5 is indirect contact with the LiF crystal, i.e. there is nobarrier for this case.

Figure 2 shows the output from this experimentalong with the CTH estimate, and the experimentalresults are also shown again for the 2.2-mm-longpellet initiated by direct contact with a similarPETN detonator pellet. The particle velocity datafrom the flyer-initiated PBXN-5 pellet is morerepresentative of a C-J detonation. The first peakjumps to a velocity of 1810 m/s corresponding to ameasured pressure of 36 GPa, which agrees withthe accepted C-J value for PBXN-5.8 The steepvelocity drop behind the front resembles a Taylordecay following a more ideal detonation. Thedifferences between these two experimental

measurements have been seen in repeated tests andmust be the result of the difference in initiation inthe two cases.

—22 rrmN-5 Grated

—22 rrm N-5 Experimental

—22nrmN5/<AI Experimental

0.00 005 010 015 02D 0.25 030 035 040

Time (us)

FIGURE 2. Particle velocity for a 2.2-mm long N-5 pelletsat the LIF interface. A CTH calculation without thealuminum plate is shown for comparison.

DISCUSSION

The experimental results from the 6-mm and2.2-mm boosters shown in Figure 1 reveal verydifferent performances for the two different columnlengths of explosive. The fact that the peak of thelonger booster pellet more closely approaches theCTH estimate is not surprising since the longercolumn of explosive may be expected to produce amore "ideal" detonation. At first glance one mightconclude that the 2.2-mm pellet is still running upto detonation as is observed in a wedge test. Whilethe VISAR data is not particularly clean, it appearsthe energy is being released over a period of at least50 ns; implying a very extended reaction region thatis much longer than the 6-mm PBXN-5 pellet.

The presence of this extended reaction regionindicates that DSD theory may not be well suited todescribe these waves since it may violate one of thefundamental assumptions of the DSD approach, i.e.the reaction zone length is smaller than the radius ofcurvature.1'2 Perhaps these miniature systems will

1033

have to be modeled using direct numericalsimulation explicitly addressing the chemicalreactions and their associated energy release. Thismay provide insight to clarify whether this type ofwaveform represents a feeble detonation or a re-initiation of detonation following detonation failure.

The 6-mm PBXN-5 pellet has a more classicalwave shape with a negative slope immediatelybehind the initial peak indicative of a detonatingexplosive. However, the peak amplitude is wellbelow the C-J level implying that even at thislonger column length the PBXN-5 may beundergoing a buildup of detonation. This behavioris evidence for the presence of a transient (dDn/dt)term.3'4 Further experiments with PBXN-5 columnsof even longer lengths leading up to a full C-Jparticle velocity are needed to confirm thisobservation.

Comparing the 2.2-mm-long boosterperformance in Figures 2, we again seedramatically different behaviors. The peak pressureof the experimental arrangement for case 1 is 40%lower than that of case 2. Given the slopes behindthe shock fronts, this disparity cannot be attributedto differences in attenuation losses in the barrier.We attribute the disparity to the difference ininitiation methods. Late energy release from thecase 1 initiation scheme does not appear to bepresent in the data from case 2. This not onlyindicates that the initiation mechanism matters, butappears to be evidence of a transient state in thedetonation of case 1, i.e. a dDn/dt behavior asdescribed in the DSD theory.3'4

CONCLUSIONS

While the results shown here are preliminary,they indicate a few trends that are significantconsiderations for explosive train designers. Forshort explosive booster columns, the initiation

method affects the explosive output, as does thelength of the booster column. We have presentedevidence indicating that there is delayed energyrelease in short columns of PBXN-5 thatcorresponds to nonideal detonation. The dataimplies a transient detonation behavior buildingtoward C-J pressure with increased column length.While further experiments are needed in this area,we believe that these results indicate thatmicrodetonics, i.e. transient effects, occurs indetonator-booster systems.

ACKNOWLEDGEMENTS

The authors wish to thank Willard Hemsingand Michael Shinas for their assistance with theVISAR system and data reduction.

REFERENCES

J.B. Bdzil, W. Fickett, & D.S. Stewart, NinthSymposium (Intl.) on Detonation, pages 730-742,OCNR 113291-7 (1989).J.B. Bdzil & D.S. Stewart, Phys. Fluids A 1, 1261(1989).J. Yao & D.S. Stewart, J. Fluid Mech. 309, 225(1996).T.D. Aslam & D.S. Stewart, Combst. TheoryModeling 3, 77 (1999).L.G. Hill, J.B. Bdzil, & T.D. Aslam, EleventhSymposium (Int.) on Detonation, pages 1029-1037,Office of Naval Research, ONR 33300-5 (1998).L.M. Hull, Tenth Symposium (Intl.) on Detonation,pages 11-18, ONR 33395-12 (1993).J.M. McGlaun, S.L. Thompson, L.N. Kmetyk, &M.G. Elrick, Intl. J. Impact Eng. 10, 351, (1990).T.N. Hall & J.R. Holden, Navy ExplosivesHandbook, NSWC MP 88-116, Naval SurfaceWarfare Center, Dahlgren, VA.

1034