hydrocode in problems with limited experimental data, such ... · Render-safe impact dynamics of the M557 ... (EOD) community as it is ... axis of symmetry during standard render-safe

Render-safe impact dynamics of the M557

fuze

G.A. GazonasU.S. Army Research LaboratoryWeapons and Materials Research DirectorateAberdeen Proving Ground, MD 21005, USAe-mail: [email protected]

Abstract

This paper outlines the results of a combined experimental and computationalstudy which investigates the transient structural response of the M557 pointdetonating fuze subjected to low-speed (200 m/s) oblique impact by a hardenedsteel projectile. The problem is of interest to the explosive ordnance disposal(EOD) community as it is a method that is currently used to "render-safe"ordnance and other explosive devices found in the field [1]. An inert M557 fuzeis instrumented with a low mass (1.5 gram), 60 kilo-g uniaxial accelerometerand subjected to oblique impact by launching a 300 gram projectile from a 4-inairgun. The transient structural response of the fuze is compared to the predictedresponse using the Lagrangian finite element code DYNA3D. Peakaccelerations measured in two tests average 35 kilo-g whereas DYNA3Dpredicts a 40 to 50 kilo-g peak acceleration depending upon the amount ofprescribed damping. Both measured and computed acceleration histories areFourier transformed and the estimated spectral response at the base of the fuze isshown to be dependent upon the failure strength of the flash tube.

1 Introduction

The explosive ordnance disposal (EOD) community utilizes a variety ofdevices to disarm or "render-safe" ordnance and other explosivedevices found in the field. One class of render-safe methods can betermed "mechanical" in that it relies on a measured use of mechanicalforce to sever or disrupt the explosive initiation train within the fuze. Inthis study, the Navy sought to quantitatively assess the predictivecapabilities of a Lagrangian-based hydrocode in modeling the obliqueimpact and penetration of the M557 fuze by a hardened steel projectile.One of the motivations of the study is to assess the accuracy of the

Transactions on the Built Environment vol 32, © 1998 WIT Press, www.witpress.com, ISSN 1743-3509

342 Structures Under Shock and Impact

hydrocode in problems with limited experimental data, such as in EODwork, where data 'is both hazardous and costly to obtain.

A typical fuze geometry consists of a right-circular cone that threadsinto the ordnance at its base (Figure 1). The cutaway view alsoillustrates some of the important internal fuze components such as thefiring pin, detonator and booster. Although these components arecritical for proper operation of the fuze under normal impact conditions,replacing the "fine" internal structure of the fuze with a homogeneouscore material simplifies modeling the oblique impact problem.

The fuze selected for analysis is the M557 point detonating fuzecomprised of a thin-walled mild steel (1006) ogive that houses athreaded mild steel base and flash tube. The firing pin, safety spring,detonator, and other intricate nose cone components are modeled with ahomogeneous 6061-T6 aluminum nose cone. The solid 300 gcylindrical projectile consists of hardened Rc-57 steel, 1-in in diameterand 3-in long. The projectile is designed to strike the fuze normal to itsaxis of symmetry during standard render-safe procedures (Figure 2).

The explicit Lagrangian hydrocode DYNA3D (the 1994 version) [2],[3] is used to simulate the transient impact event since this code hasbeen successfully employed to model other highly transient physicalphenomena [4]. In order to quantitatively assess the predictivecapabilities of the hydrocode, the fuze is instrumented with a lightweight60 kilo-g uniaxial accelerometer and subjected to oblique impact bylaunching a 300 g projectile at constant velocity (200 m/s) from a 4-inair gun. Acceleration histories recorded by the accelerometer mounted onthe base of the fuze are compared to the accelerations predicted byDYNA3D. The accelerometer is modeled as a small rectangularlightweight mass attached to the base of the fuze.

Two oblique fuze impact tests reveal that peak accelerations at thebase of the fuze average 35 kilo-g while the predicted peak accelerationsrange from 40 to 50 kilo-g depending upon the amount of prescribeddamping. The measured acceleration histories are highly variable in thetime domain, so the signals are Fourier transformed to determine howwell the frequency content of the impact event compares with thefrequencies of the predicted response using DYNA3D. The measuredaccelerations contain a dominant energy peak near 30 kHz, whereas thecomputational results reveal the presence of an energy peak at 30 kHzand another at 120 kHz. It is shown that the predicted temporal andspectral response recorded at the base of the fuze strongly depends upthe failure strength of the flash tube.

2 DYNA3D computations

Computations are performed using the explicit Lagrangian hydrocode,DYNA3D, developed by the Lawrence Livermore National Laboratory(LLNL) in 1977. The code is in use by the aircraft industry (Boeing)and the automobile industry (Volvo, Saab, General Motors, andJapanese manufacturers) for crash and safety modeling and has seen


Structures Under Shock and Impact 343

BALL ROTOR

Figure 1: Cutaway view of a typical fuze illustrating internalcomponents.

• nose cone

- flash tube

• upper base assembly (UBA)

- accelerometer

_ _ -base

Figure 2: Cutaway view of the finite element model illustrating thegeometry and grid of elements of the fuze and projectile.

continuous development and use since its initial introduction. The finiteelement computational grid consists of 76,970 nodes and 53,516hexahedral finite elements (Figure 2). The ogive consists of 3hexahedral elements through the thickness. Computations proceedthrough three phases: pre-processor, main cycle and post-processorcorresponding to PATRAN, DYNA3D, and TAURUS routines,



respectively. The rate and temperature-dependent Johnson-Cook modelsimulates the constitutive behavior of the 1006 mild steel and 6061-T6materials. The accelerometer behavior is linear elastic. Fixing the outersurface coordinates of the fuze base simulates the experimental testcondition whereby the fuze base is threaded into a massive steel plate toprevent motion during impact. Non-reflecting boundary conditions arealso assigned to the nodes on this surface. The flash tube is threadedinto the upper base assembly (UBA) (Figure 2) which contains internalstructure not included in the finite element model. To simulate stresswave attenuation in the UBA, the global damping feature available inDYNA3D (Rayleigh damping extended to nonlinear analysis) is usedwith damping coefficients set at a = 0, p = 10~~8 [2]. The projectilestrikes the fuze at an initial velocity of 200 m/s. The impact position isidentical to that in the airgun tests described in the following section.Ten sliding interface definitions prevent interpenetration of the variouscolliding solids during impact. The simulation of penetration andfragmentation of the ogive employs the Slidesurfaces with AdaptiveNew Definitions "SAND" algorithm. Ogive failure commences whenthe equivalent plastic strain of a particular element in the ogive attains a

value of €** - 0.30 . Flash tube fracture is simulated using the "nodespotwelded to surface" slideline feature in DYNA3D. The slidelinereleases nodes along a "failure" surface when a predefined level of thenormal and shear failure strength of the flash tube is exceeded [2].

Computations are conducted on a 64-bit Cray C90 vector computer,and are terminated at 500 jus with a sampling rate of 2 jus/pt (500 kHz),corresponding to a Nyquist rate of 250 kHz. The computations cost anaverage of 0.1 CPU hr/|is. The computed accelerations of various fuzecomponents, such as the center of gravity (eg) of the accelerometer, aredirectly compared with measured accelerations. Some salient features ofrender-safe impact dynamics that depict the fuze deformation historyappear in Figure 3. The projectile velocity versus time plot (Figure 4)shows that projectile deceleration corresponds to impact into the ogiveand nose cone/flash tube. Constant velocity portions of the curverepresent periods of time when the projectile is essentially in "freeflight". The projectile pierces completely through the thin-walled ogiveat 45 jis and impacts the nose cone at 68 p,s. At 104 (is, the projectile isin full contact with the flash tube. At 140 p,s, the projectile severs theflash tube at its base and begins to severely deform the interior surfaceof the ogive. The projectile pierces completely through the ogive at 200(is, and at 500 jus the nose cone and flash tube rotate up and out of theflight path of the projectile.

3 Airgun experiments

Oblique fuze impact tests are conducted at an indoor airgun test range atAdelphi, MD. The test range includes a variety of airgun devicesdesigned to gradually accelerate projectiles to ballistic velocities for hi-gimpact experiments. The airgun used in these experiments consists ofeight 12-ft long sections, connected in tandem, to form a single gun tube



Figure 3: DYNA3D simulation results showing deformed fuze andogive fragmentation at 200 and 500 (is.

-6500

velocity (in/s)

1x10 1.5x10Time (s)

Figure 4 : Velocity history of the projectile center of gravity (eg).

96-ft long, with a 4-in inner diameter. The projectile is first loaded intothe breech (right-hand side of Figure 5), after which the entire gun tubeahead of the projectile is evacuated to ~1 torr. The projectile is releasedand gradually accelerated toward the muzzle of the gun by atmosphericpressure acting on its base. The exit velocity of the projectile isdetermined with either a streak camera or a light emitting diode systemmounted near the barrel exit (Figure 6). Since the projectile diameter inthese experiments is much less than the bore diameter of the airgun, acylindrical bakelite carrier, 4-in in diameter, transports the projectileduring flight down the gun tube. Several trial test shots establish theprecise weight (projectile + carrier) necessary to achieve an exit velocityof 200 m/s. The projectile flies through a 2-in hole in a steel plate atmuzzle exit which strips-off the bakelite carrier (Figure 6). Duringimpact, the projectile pierces through the ogive and severs the flash tube



Figure 5: View of 4-in airgun from breech end.

Figure 6: View of 4-in airgun from muzzle showing M557 fuze.



near its base. Each test uses a new bakelite carrier, as they shatter onimpact with the steel plate.

A 1.5 gram 60 kilo-g umaxial ENDEVCO Model 7270A-60Kpiezoresistive accelerometer is mounted on the base of the fuze justbelow the flash tube, and within the cavity which normally containsexplosive booster material (Figure 1). The vertical component of theacceleration vector at the base of the fuze is recorded. The 7270Aaccelerometer is a rugged undamped unit designed specifically for shockmeasurements. The mounted resonance frequency is 700 kHz, and anear-zero damping allows the accelerometer to respond accurately tofast rise time, short duration shock events. The accelerometer signal isconditioned using a Pacific Instruments signal conditioner (Model8655). Data are sampled at a rate of 0.5 p,s/pt (2 MHz), whichcorresponds to a Nyquist rate of 1 MHz, and stored on a Nicolet 2090oscilloscope for subsequent analysis.

4 Comparison of airgun test results with

DYNA3D predictions

Two airgun tests are conducted and the acceleration history for shot #1is shown in Figure 7a. The measured peak vertical accelerations are onthe order of 35 kilo-g, and decay rapidly with time. The variability inthe observed acceleration histories highlights one difficulty associatedwith data acquisition in structures subjected to shock and impact.Consequently, the requirement for model validation in the time domainis relaxed by comparing the spectral characteristics (frequency domain)of the measured and predicted signals. To this end, the power spectraldensity of the acceleration history is estimated by computing the so-called periodogram [5] which is based upon computation of the FastFourier Transform (FFT) [6]. The computational economy of the FFTmakes this approach one of the most popular methods for spectralestimation. Conventional FFT spectral estimation is based upon aFourier series model of the data: that is, the process is assumed to becomposed of a set of harmonically related sinusoids. For this problem,this is not a bad assumption given the fundamental nature of wavepropagation in media. FFT's of the measured accelerations are computedusing the Mathcad 7.0 software package [7], and from thesetransformations, periodograms are computed (Figure 7b). The FFT of

the vector v^ in Mathcad 7.0 is computed using,

with corresponding frequencies,

A =-&/,, (2)



that depend upon the sampling frequency f^, and the number ofsamples N. The periodogram plots are formed by squaring themagnitude of the vector in Equation 1 (see e.g. Oppenheim and Schafer[8]) and reveal the presence of a dominant energy peak at 30 kHz that isundoubtedly related to some physical phenomenon associated with theimpact event. The longitudinal wave speed 146,316 in/s in theaccelerometer is estimated from its modulus and bulk density. Hence,the fundamental frequency of an acoustic wave traveling verticallythrough the 0.11-in thick accelerometer is computed to be about 665kHz. Thus, the relatively low frequency peak at 30 kHz is produced bysome other mechanical disturbance, possibly due to the fracture failureof the flash tube. The discussion in a subsequent section examines thishypothesis. The same spectral estimation methodology is applied to thevertical accelerations computed for the eg of the modeled accelerometerin the DYNA3D analysis. The acceleration history for the DYNA3Danalysis is shown in Figure 8a. Comparison of the actual and simulated

3.5x10

1.75x10 _

-1.75x10 _

-3.5x10-5x10 0 5x10 0.001 0.0015 0.002

Time (s)

Figure 7a: Accelerometer response for shot #1.

Power

1.5x10

1.13x10 -

7.5x10

3.75x10

Periodogram

0 100 200 300 400 500

Figure 7b: Periodogram (power spectral density) for shot #1.



1x10*

-5xl(T

I I I I

Time (s)

Figure 8a: Simulated accelerometer response.

1.5x10' Periodogram

Power

5x10

I I I 1 h i I I I I

25 50 75 100 125 150 175 200 225 250

kHz

Figure 8b: Periodogram for simulated response.

periodograms in Figure 7b and 8b reveals that the measuredaccelerations contain a dominant energy peak at 30 kHz, whereas thecomputational results predict an energy peak at 30 kHz, and anadditional energy peak at 120 kHz. Additional airgun tests andsimulations are performed to gain a better understanding of the physicalphenomenon causing the observed energy peak at 30 kHz, and whyDYNA3D predicts an energy peak at 120 kHz.

4.1 Additional airgun tests

Two additional airgun experiments are performed whereby the M557fuze is replaced with a fuze "simulant" constructed from hot rolledsteel. In addition, the simulant is further simplified by removing theogive and nose cone, so that only the flash tube and base are subjected toimpact. The impact and boundary conditions are identical to those in theprevious airgun impact tests. The acceleration histories and



periodograms for these tests are very similar to those for the M557 fuze.Furthermore, the periodograms for the fuze simulant tests also revealdominant energy peaks at 30 kHz, and indicate that the ogive and nosecone structural components do not appreciably contribute to the spectralresponse of the M557 fuze. Although the projectile pierces through theogive and impacts the nose cone during render-safe operations, thesecollisions do not measurably influence the spectral response of the M557fuze.

4.2 Additional DYNA3D simulations

Another parameter which can influence the measured spectral responseis the failure strength of the flash tube. As described in Section 2, flashtube failure is modeled using the "node spotwelded to surface" slidelinefeature in DYNA3D. The flash tube failure strength is artificiallymodified by conducting two additional simulations whereby its strengthis first decreased and then increased, by an order of magnitude, from itsnominal value in prior simulations. The acceleration histories andperiodograms for the "weak", "nominal" and "strong" flash tubesimulations appear in Figure 9. The waveforms for the weak andnominal strength simulations are very similar as compared to the"strong" flash tube simulation which exhibits a high amplitudeacceleration response that does not decay appreciably with time.

Furthermore, the flash tube does not fail in the "strong" simulationas progressive distortion of the finite elements in the impact regionrapidly decreases the stable time step governed by the Courantcondition. Thus, the simulation is terminated at 180 jis. Periodogramsare computed for the three cases investigated and are plotted in semi-logform in Figure 10. The simulations involving the "weak" and"nominal" flash tube strengths are nearly identical in the estimate of thepower spectral density of the shock signal. However, the energyassociated with the impact into the "strong" flash tube is several ordersof magnitude greater, and is more uniform in strength across thefrequency spectrum; thus, increasing the flash tube failure strength hasthe effect of transmitting higher energy modes to the accelerometer. Theresults indicate that changing the failure strength of the flash tube cansubstantially modify the temporal and frequency response of the fuze.

4.2.1 Global damping

In order to simulate the effect of the UBA's intricate internal structure,the damping coefficient is P = 10~8 for the computations reported thusfar. As the precise effect of this internal structure on the resultingwaveform is unknown, the damping coefficient in the UBA is increasedto p = 10-6 in an effort to further attenuate the 120 kHz energy peak. Acomparison of the acceleration histories for the eg of the UBA for thecases p = 10~8 and (3 = 10~6 reveals that increasing the dampingcoefficient attenuates the resulting waveform (Figure 11). In addition,increasing the damping coefficient to P = 10'° attenuates the peakacceleration in the accelerometer to 41 kilo-g. Interestingly, the eg of



the UBA contains an energy peak at 30 kHz, but the 120 kHz energypeak is highly attenuated (Figure 12), relative to that predicted in theaccelerometer (Figure 8b). This observation is independent of the valueof P used in the analysis.

Another possible source of the anomalous energy at 120 kHz couldbe related to spurious frequencies induced by the finite element mesh inthe accelerometer. The 0.11-in thick accelerometer is initially modeledwith six hexahedral elements through its thickness resulting in the powerspectral density illustrated in Figure 8b. Halving the finite element meshdensity results in attenuation of the energy associated with the spuriousfrequency at 120 kHz, but the energy at 30 kHz is attenuated as well .

5 Discussion and conclusions

This paper outlines the results of a combined experimental andcomputational study which investigates the transient structural responseof the M557 point detonating fuze subjected to low-speed (200 m/s)oblique impact by a hardened steel projectile. Peak accelerationsmeasured in the airgun tests average 35 kilo-g whereas peakaccelerations predicted by DYNA3D range from 40 to 50 kilo-gdepending upon the value used for critical damping in the UBA.

A comparison of the estimated spectral response of the simulationsand experiments reveals that both contain spectral energy at 30 kHz.Additional tests on fuze simulants whereby the ogive and nose cone areremoved from the fuze also contain spectral energy at 30 kHz. Thisobservation indicates that the ogive and nose cone structural componentsdo not significantly contribute to the frequency content of the observed

0 5x10 ' 1x10nominalstrong Time (s)

- - - weak

Figure 9: Simulated accelerometer response to 200 ps for "weak" (top),"nominal" (middle), and "strong" (lower) flash tube failure strengths.



Power

1x10

1x10

Periodogram

250

weak

Figure 10: Periodograms for "weak", "nominal", and "strong" flashtube simulations.

Time (s)

Figure 11: Acceleration histories for damping coefficients P = 10~8, andP= 10-6 in the UBA.

waveforms. Furthermore, the estimated spectral response at the base ofthe fuze is shown to be largely dependent upon the failure strength of theflash tube; impact energy is transmitted to the accelerometer as long asthe flash tube remains in contact with the UBA. The DYNA3Dhydrocode also predicts the presence of a strong spectral peak in theaccelerometer at 120 kHz which is not observed in the impact tests.This spurious spectral peak appears to be an artifact of the finite elementanalysis as this spectral component is highly attenuated in the egresponse of the UBA. How then is energy at this frequency being excitedin the model accelerometer?



6x10

Power

PeriodogramI

-

1 1 1 1 1 1 1 1

(3=10-*

,,.,.-

L/\l J J ^L 1 1 1 10 25 50 75 100 125 150 175 200 225 250

kHz

Figure 12: Periodogcoefficients (3=10"

ams showing UBA spectral response with dampingand (3= 10-6.

A final investigation examines the spectral response of a vertical lineof nodes along the central axis of the UBA. These axial nodes are asubset of the total number of nodes which comprise the eg response ofthe UBA. When the nodes that comprise the "spotwelded" failuresurface are suddenly released during failure, a hi-g stress wave isintroduced into the mesh which attenuates as it travels downwardthrough the UBA. The stress wave excites the axial nodes in the UBAwhich contain a dominant spectral energy peak that is phase-shifted in aband around 120 kHz. If the acceleration histories of this subset of nodesare summed and spectrally analyzed the 120 kHz energy peak attenuatesand broadens. The summing procedure is known as signal stacking inthe geophysics literature [9] and has the effect of improving the signal-to-noise ratio. As an increasing number of UBA nodes are included inthe analysis the eg response is approached (Figure 12). Why the spectralpeaks of axial nodes shift phase in a band around 120 kHz is not known,but is currently an area of active study. In real media, however, phaseshifts are caused by wave reflections, caustics, and geometrical ormaterial dispersion effects. It is interesting to conjecture that the 120kHz frequency is related to the fundamental axial mode of vibration ofthe UBA. Since the height of the UBA is about 0.55 in and thelongitudinal wave speed is 129,816 in/s, this results in a fundamentalfrequency of about 118 kHz. The 120 kHz energy is further attenuatedby optimizing the Rayleigh damping coefficients, or postprocessing thefinite element transients, which inherently contain spurious frequenciesrelated to multiple wave reflection phenomena and element size effectsas discussed in [10].

6 Acknowledgments

This work was funded by the Naval EOD Technology Center, IndianHead, MD. The author thanks Ara Abrahamian, William Mclntosh, andEd Szymanski for conducting the airgun tests.



7 References

1. Gazonas, G.A., Segletes, S.B., Boyle, V.M., and Stegall, S.R.,Oblique impact modeling of fuzes, Intl. J. of Impact Engng., Vol.18, No. 4, pp. 435-457, 1996.

2. Hallquist, J.O. & Whirley, R.G., DFTV/UD [/ye Ma/zwa/ (NonlinearDynamic Analysis of Structures in Three Dimensions), UCID-19592, Rev. 5, Lawrence Livermore National Laboratory,Livermore, CA, May 1989.

3. Hallquist, J.O., Theoretical Manual for DYNA3D, UCID-19401,Lawrence Livermore National Laboratory, Livermore, CA, March,1983.

4. Gazonas, G.A., Lagrangian simulation of linear shaped demolitioncharge formation, Proc. of the 4*h Int. Conf. on Structures UnderShock and Impact IV, eds. N. Jones, C.A. Brebbia, and AJ. Watson,Computational Mechanics Publications, Southampton and Boston,pp. 361-372, 1996.

5. Schuster, A., On the investigation of hidden periodicities withapplication to a supposed 26 day period of meteorologicalphenomena, Terrestrial Magnetism, 3 , pp. 13-42, 1898.

6. Brillinger, D.R., Fourier analysis of stationary processes, Proc.7E&E, Vol. 62, pp. 1628-1634, 1974.

7. Mathcad 7.0 Software Program, Mathsoft, Inc., 101 Main Street,Cambridge, MA, 1997.

8. Oppenheim, A.V, and Schafer, R.W., Digital Signal Processing,Prentice Hall, Inc., Englewood Cliffs, NJ, 585 p., 1975.

9. Telford, W.M., Geldart, L.P., Sheriff, R.E., and Keys, D.A., AppliedGeophysics, Cambridge University Press, Cambridge, 860 p., 1976.

10. Holmes, N., and Belytschko, T., Postprocessing of finite elementtransient response calculations by digital filters, Computers and

c, Vol. 6, pp. 211-216, 1976.


Documents

hydrocode in problems with limited experimental data, such ... · Render-safe impact dynamics of the M557 ... (EOD) community as it is ... axis of symmetry during standard render-safe