Вы находитесь на странице: 1из 12

INTERNATIONAL

JOURNAL OF
IMPACT
ENGINEERING
PERGAMON International Journal of Impact Engineering 26 (2001) 773-784
www.elsevier.com/locate/ijimpeng

SIMULATION OF HOLLOW SHAPED CHARGE JET IMPACTS ONTO


ALUMINIUM WHIPPLE BUMPERS AT 11 KM/S

E M M A A. T A Y L O R * ' **

'Astrium Ltd, Earth Observation and Science Directorate, Gunnels Wood Road, Stevenage, SG1 2AS, UK; *'Unit
for Space Sciences & Astrophysics, University of Kent at Canterbury, Canterbury, Kent CT2 7NR, UK

Abstract-- The computational technique of Smoothed Particle Hydrodynamics (as


implemented in the hydrocodes AUTODYN-2D and AUTODYN-3D) has been used to
simulate the impact of hollow shaped charge jet projectiles onto stuffed Whipple bumper
shielding. Due to limited availability of material models, the interim Nextel/Kevlar-Epoxy
bumper was modelled as an equivalent thickness of aluminium. Stuffed Whipple bumper
shields are used for meteoroid and debris impact protection of the European module of the
International Space Station (the Columbus APM). A total of 56 simulations were carried out to
investigate the impact processes occurring for shaped charge jet impact. Sensitivity studies
were carried out on the influence of projectile shape, pitch, yaw and strength at 11 km/s to
determine the range of debris cloud morphologies. The debris cloud structure was shown to be
highly dispersed, and no projectile remnant was observed at the centre of the cloud. The mass
of an aluminium sphere producing equivalent damage to a shaped charge jet projectile was in
the range 1.5 to 1.75 times greater than the mass of the shaped charge jet projectile. Upon
loading by the dispersed debris cloud, the interim bumper failed by spallation, producing
fragments moving at 2 km/s or less. The fragments distorted the rear wall (pressure wall) of the
shield but did not perforate it. The experimental data show rear wall deformation but to a
lesser degree. Perforation of the rear wall, observed for one test, was not reproduced by the
simulation. Nextel/Kevlar-epoxy material models are required to reproduce correctly the
interim bumper failure under debris cloud loading. © 2001 Elsevier Science Ltd. All rights
reserved.

Keywords: Aluminium, Columbus, Hypervelocity, Impact, Shaped charge, SPH, Whipple


bumper

INTRODUCTION

The Columbus A P M is the European pressurised module attached to the International Space
Station (ISS). The ISS pressurised compartments must be protected against perforation by
meteoroid and debris impacts. This is achieved by attaching a Whipple bumper type shielding
(multiple spaced bumper) to the ISS pressure wall. The impacting debris particle (or meteoroid)
is fragmented by the first bumper (aluminium) and the resulting debris cloud is disrupted by the
interim bumper (Nextel/Kevlar-epoxy/MLI). The dispersed and fragmented debris cloud impacts
the rear wall (pressure wall).
Validation o f the performance o f these shields under hypervelocity impact has primarily been
done by experimental testing, using hypervelocity impact facilities (particularly light gas guns
firing spherical projectiles). Over 100 shots have been carried out in the velocity range 3-7 km/s
using spherical projectiles [1]. Shaped charge jet testing o f Whipple-type shields at velocities
-11 km/s has also been carried out [2, 3, 4]. As current experimental techniques limit both the
velocity and projectile type (i.e. the projectile shape and material), computational simulation

0734-743X/01/$ - see front matter © 2001 Elsevier Science Ltd. All rights reserved.
PII: S 0 7 3 4 - 7 4 3 X ( 0 1 ) 0 0 1 2 9 - 4
774 E.A. Taylor / International Journal of Impact Engineering 26 (2001) 773-784

(using Smoothed Particle Hydrodynamics or hybrid methods) is required to validate the shield
performance. For complex projectile shapes, the mass of a spherical projectile causing
equivalent damage must also be established for direct comparison of results with spherical
projectile test data. This ratio is often called the ballistic limit mass ratio (BLMR). Previous
studies of shaped charge jet projectile impact have been carried out [5] and an estimation of the
ballistic limit mass ratio between the shaped charge jet projectile and a solid aluminium sphere
obtained.
Smoothed Particle Hydrodynamics (SPH) can be applied to simulation of non-standard
projectiles (e.g. shaped charge jet geometries) impacting Whipple-type bumper shields.
Limitations still remain regarding the representation of complex shield materials (e.g. Nextel and
Kevlar) within the hydrocode, although material models are under development [6]. In this
paper, the Columbus APM bumper shield is therefore represented as an all aluminium bumper,
with the interim bumper modelled an as equal areal density, equivalent thickness of aluminium.
Unlike the majority of tests carried out using spherical aluminium projectiles, the shaped charge
jet geometry and flight is more complex and may include more than one material phase [7, 8, 9,
10]. Hydrocode simulations of shaped charge jet impact should include a sensitivity study on
material density (or temperature), projectile shape and material state to determine their influence
on the simulation results.
This paper presents the results of a simulation program, based on the experimental data
presented in Ref. [4] with the following aims:

(i) to determine the diameter of a solid aluminium sphere producing the equivalent damage to
the hollow shaped charge jet, and
(ii) to evaluate the influence of material model, resolution and projectile shape and orientation
(pitch and yaw) on the simulation results, and
(iii) to reproduce the experimental test results from Ref. [4], using an equivalent thickness all
aluminium Whipple bumper geometry.

The simulations of hollow shaped charge jet projectile impacting an all-aluminium Whipple
bumper identified the influence of material model, resolution and projectile shape and orientation
(pitch and yaw) on the simulation results. The ballistic limit mass ratio (BLMR) was calculated
to be between 1.5 and 1.75 for the shaped charge jets simulated (for impact velocities of -11
km/s). The range of values represents the influence of the projectile shape and orientation on the
result. The dispersed debris cloud impact on the interim bumper produced by the simulations
highlighted potential problems with simulation of material failure with SPH. The
experimentally-observed rear wall failure mechanisms were also not well reproduced in the
simulation. Limitations in the available material models (in particular, representation of Nextel
and Kevlar-epoxy as equivalent thicknesses of aluminium) meant that there was less confidence
in the accuracy of the debris cloud loading on the backwall. Future simulations should use
material models for Nextel and Kevlar-epoxy, where available.

EXPERIMENTAL DATA ~:~

The experimental targets were identical in material composition and configuration to the
Cylinder Advanced Shield targets, as defined for use on the Columbus APM module. A
description of the shield components and a detailed review of hypervelocity impact testing using
spherical projectiles are given in Ref. [1]. Ref. [4] describes three tests that were carried out at 0,
45 and 60 degrees incidence at velocities in the range of 11 km/s (2097-14, 2097-15, 2097-16).
The two shots at 45 degrees incidence (2097-15, 2097-16) were above and below the ballistic
limit. The experimental test results are reported in Ref. [4]. Additional, preliminary
measurements were made of the target (in order to define the damage area on the Nextel/Kevlar-
E.A. Taylor / International Journal of lmpact Engineering 26 (2001) 773-784 775

epoxy bumper and thus the required simulation area) and to determine the separation between the
bumpers (required for the computational representation of the target). These are given in Ref.
[12].
Hypervelocity impact testing of the Columbus APM MOD shields was carried out using
shaped charge projectiles produced at the SwRI facilities in July 1998 [4, 11]. The projectiles
were observed in-flight before impact using two flash X-ray stations, each with orthogonal
views. Measurements of the projectile dimensions were made from the film from both views and
given in Ref. [4]. The data show that the outer diameter (OD) and inner diameter (ID) varied
along the length of the projectile, defining a 'conoid' section projectile (Fig. 1). The projectiles
tumbled in both pitch and yaw directions as they travelled along the line of flight. X-rays of the
debris cloud after impact with the first (aluminium) bumper show a dispersed debris cloud with
little clear structure, apart from darker regions, which suggest a higher spatial density of material,
probably the projectile remnant (example in Fig. 3).
It is not possible to model fully the distorted hollow conoid projectile shape, primarily due to
computer memory-driven resolution limitations. Therefore, the measurements of three mass-
equivalent projectiles with axial symmetry along the flight vector have been determined from the
data in Ref [4]. These projectiles have been called V1, V2 and V3. The equivalent-mass versions
1-3 projectile measurements for shots 2097-14, 2097-15 and 2097-16 are presented in Table 2.

• V1. Hollow cylinder with OD-ID constant along the projectile length and ODrear = ODfront
• V2. Hollow conoid with ODrear> ODfrontand IDrear> IDfron~. OD-ID varies along length.
• V3. Hollow conoid with ODrear > ODfront and IDrear > IDfront. OD-ID is constant along
length.

Fig. 1. Hollow shaped charge jet projectile


(shot 2097-15).

OD(rear)
I D fron,
Fig. 2. Measurements of simulated hollow Fig. 3. Debris cloud characteristics
shaped charge jet projectile. (shot 2097-14).
776 E.A. Taylor/International Journal of Impact Engineering 26 (2001) 773-784

Table 1. Experimental data and projectile measurements. (BL: ballistic limit).


Shot ID Expt.Data Proj. ODre~ ODfront IDre~r IDfront L Vol. Mass
(mm) (mm) (mm) (mm) (mm) (mm3) (g)
2097-14 V = 11.19 km/s V1 5.95 5.95 2.97 2.97 15.75 331.0 0.90
Below Pitch = 6.2 ° V2 6.80 5. l 1 3.54 2.40 15.75 331.9 0.90
BL Yaw = 14.9° V3 6.80 5.11 3.79 2. l 1 15.75 331.9 0.90
2097-15 V = 11.07 km/s V1 5.87 5.87 2.73 2.73 15.77 332.3 0.92
Below Pitch = 5.7° V2 6.58 5.16 3.00 2.48 15.77 335.5 0.92
BL Yaw = 0° V3 6.58 5.16 3.42 2.00 15.77 335.5 0.92
2097-16 V = 11.24 km/s V1 6.58 6.58 3.20 3.20 16.69 433.3 1.18
Above Pitch = 0° V2 6.84 6.34 3.24 3.18 16.69 434.4 1.18
BL Yaw = -3.7° V3 6.84 6.34 3.48 2.98 16.69 434.4 1.18

HYDROCODE SIMULATIONS

The simulation program was divided into Phase 1 and Phase 2. In the Phase 1 simulations,
the front bumper was represented by SPH particles and the interim bumper was represented by a
Lagrangian meshed bumper and acted as a 'witness plate', recording the extent and degree o f
damage caused by the different debris cloud distributions. Phase 2 replaced the Lagrangian
interim bumper with an SPH bumper, bringing the number o f SPH particles (SPH particle
diameter 0.5 mm) used to 400,000. The rear wall was represented by a Lagrangian meshed
bumper. The Columbus APM shield was defined by an all aluminium shield o f equivalent areal
density (2.5 m m bumper, 4.0 m m midbumper and an 4.8 m m rear wall). The material models
used are defined in Table 2. The projectile was modelled hydrodynamically, to reflect the
inferred material state o f the projectile (as in Ref. [13]). The conoid projectile geometries are
given in Table 1.

In Phase 1, three dimensional oblique incidence simulations were carded out to determine:

• the influence o f projectile shape (Table 1) and SPH particle size on the debris cloud
• the influence o f projectile yaw (Y) and pitch (P) on the debris cloud (P = + 15°; Y = + 15 °)
• the diameter o f a sphere causing the same damage as the simulated hollow conoid projectile

In Phase 2, three dimensional oblique incidence simulations were carried out to determine:

• the ballistic limit o f the equivalent aluminium bumper target


• the influence o f material model (projectile and target) on the rear wall failure modes

Table 2. Material models used in the simulation programme. Model data defined in Ref. [ 12]

EOS Strength model Failure model


Shaped charge jet A11100 None or A11100 Johnson None or All 100 failure
projectile Cook stress
Spherical projectile All 100 None or All 100 Johnson None or A11100 failure
Cook stress
Interim bumper A12024 A12024 Johnson Cook Failure stress for A16061
(Lagrange)
Interim bumper (SPH) A1 2024 A1 2024 Johnson-Cook 1 GPa or 2.5 GPa failure
stress
Back wall (A12219) A1 2024 A12024 Johnson Cook Failure stress o f 2.5 GPa
E.A. Taylor/International Journal of Impact Engineering 26 (2001) 773-784 777

RESULTS

Phase I simulations. Projectile sensitivity study and BLMR determination.

Figure 4 shows the influence of projectile shape on the debris cloud morphology 8 Its after
impact (just before impact on the Lagrangian witness plate), for version 1, version 2 and version
3 projectiles. For all three simulations, there is no remnant of hollow cylinder visible at the front
of the debris cloud, although a disc, representing the highest density material is present. The
hollow cylinder projectile remnant has been "peeled apart", to form a broadly dispersed debris
cloud region. The width of the debris cloud does not vary between projectile types although
quantitative inspection does show differences. Viewed head on, the debris cloud caused by a
version 1 projectile is broader in cross section than the other projectile geometries. In a side
view, the bulge above the central disk is more pronounced for debris clouds caused by version 2
and version 3 projectiles. The width of the cloud is 55 mm and the maximum width increases
from 38 to 43 mm from version 1 to version 3 type projectiles.

4'¸¸'¸¸ . ~ ~ . . . . . . . " ' . . . . . . . . . . . . . . . . . ' ~ ; . . . . . . . . . . . . . ". . . . . " " ........ • . . . . . . . . 4" . . . . . . . . . ~'~•

Fig. 4. Debris cloud morphology as a function of projectile shape for impacts at 45 °.


Left to right: version 1, version 2 and version 3.

The velocity distribution for all the three clouds is very similar, with the projectile remnant
(represented by a higher spatial density of material) travelling at velocities close to the initial
impact velocity. When the projectile first impacts the bumper, bumper material flows up through
the hollow cylinder. This up-flow disrupts the projectile and causes it to disperse into the 'peeled
apart' segments seen at 8its. The conoid form appears more stable than the hollow cylinder.
This may be due to the fact that the flow of material diverges as it passes upwards through the
hollow conoid.
This "peeled apart" morphology is independent of the SPH particle size. It was also observed
in the 2D simulations (not shown here), which have a resolution of 10 particles across the conoid
width. The 3D high resolution (6 particles across the conoid width) and low resolution (3
particles across the conoid width) simulations were compared and shown to have the same
velocity and dimensions. Note that a minimum of 5 particles are required for SPH to remain
consistent/conservative. The debris cloud can also be qualitatively compared with a flash X-ray
images taken during the SwRI tests (Fig. 3). It shows a highly dispersed debris cloud with no
distinct structure, except along the front part of the cloud.
A sensitivity study was carried out prior to the main simulations to assess (i) the influence of
strength on the debris cloud, as caused by a hollow conoid projectile (ii) the difference in debris
cloud morphology between clouds caused by hollow conoid projectiles and spherical projectiles
778 E.A. Taylor / International Journal of Impact Engineering 26 (2001) 773-784

(simulated with and without a strength model). Both the debris cloud structure and the damage
recorded on the witness plate were analysed.
The simulation results showed that the hollow conoid debris cloud sizes were independent of
strength model, although more bumper material is seen in the cloud when the strength model is
used. The central cloud high density region is more clearly defined for simulations with a
strength model, as expected. The difference between the hollow conoid projectile clouds and the
spherical projectile clouds is clearest in the side view, and shows the different distribution of the
central element (Fig. 5). The velocity of the central element is within 10% of the initial impact
velocity for both projectile types. Note that the debris cloud caused by the spherical projectile
extends further in the -x direction by 10-15 ram. The influence of strength model on the second
bumper damage for hollow conoid projectiles results in surface damage that is more tightly
focussed for the non-hydrodynamic projectile; this is consistent with the debris cloud simulation
observations. The total area is in the range 40-46 x 61-69 mm.

. . . . . . . . . . . . . . . . . . . . . ~"2£7'.~ *
. . . . . . . . . . . . . .

Fig. 5. The influence of projectile characteristics on debris cloud shape at 8 p.s. All V3
projectiles, except where noted. Row 1. Zero strength projectile. Row 2. Strength model used.
Row 3. Spherical projectile.

When pitch and yaw were added to the oblique incidence simulation, the projectile cross
section along the line of flight increases, causing greater damage to the front bumper (Fig. 6).
Viewed from the side, the debris cloud caused by a projectile with P = +15 ° has a dispersed front
projectile remnant, and a region consisting of bumper material. This latter region is caused by
the rear section of the long conoid projectile impacting the bumper after the front part has
perforated the region in front. It represents an increase in the volume of material in the debris
cloud, and thus increased damage causing potential. The volume of bumper material removed by
the projectile with P = -15 ° is much less and the front projectile remnant is less dispersed. The
influence of the projectile yaw can be clearly seen in Fig. 6. The debris cloud width along the y-
axis does not increase when pitch and yaw are added, but the material distribution varies, with
the projectile remnant (P = -15 °) showing less sign of asymmetry than for a projectile with P =
E.A. Taylor I International Journal of Impact Engineering 26 (2001) 773-784 779

+15 ° .
The effective cross-section of the projectile along the line of flight is altered when pitch and
yaw are included. The rotation to left and right of the yaw produces mirror symmetry effects.
However, the inclusion of positive or negative pitch on the projectile has a markedly different
effect on the perforation hole in the first bumper. For negative pitch, the perforation hole is
elliptical and not significantly different in size to the simulation without pitch and yaw (25 x 17
mm, compared with 22 x 17 mm). However, when positive pitch is included, the combination of
the projectile length (and thus erosion under impact) and rate of travel through the plane of the
bumper (impact velocity) results in the tail end of the projectile creating a secondary perforation
(total dimensions 28 x 17 mm) behind the first impact hole.

iF ~

Fig. 6. Influence of pitch and yaw on projectile. Row 1. P = Y = -15 °. Row 2. P = Y = +15 °.

IB IB v ~ m
r~lt f o ~

.....! ;::i..... 1,7:-'iii Izs;

~, ,~,, ~. . . . . . . , ....... ,5¸+.¸ ............

i ..... L

i 7.'~;

Fig. 7. Second bumper (Lagrange) damage pattems at 16 kts for a debris cloud resulting from a
version 3 projectile. Top row: front surface. Bottom row: rear surface. Column 1: P = Y = 0 °.
Column 2: P = -15 ° and Y = +15 °. Column 3: P = +15 ° and Y = -15 °.

The difference in the debris cloud structure reviewed above results in different damage, when
projected onto a Lagrange 'witness plate' (Fig. 7). The damage caused by the projectiles with P =
-15 ° and P = +15 ° bounded the simulation with P = Y = 0 °. The two regions of damage (along
the flight axis and behind it, offset by 50-60 mm) are no longer clearly visible, and the view of
the underside of the plate illustrates the differing damage patterns. Note that, for the simulation
with positive pitch, there is a region of extra damage 50-60 mm behind the line of flight vector.
780 E.A. Taylor/International Journal of lmpact Engineering 26 (2001) 773-784

This is probably caused by the bumper material removed during the rear segment of the projectile
impact on the first bumper.

.....t m m
. . . . . . 7; m

. . . . . ) ....
)

iN,,
m m '
m
m,,
Fig. 8. Second bumper (Lagrange) damage pattems at 16 IXS(front and rear plate views). Column
1. Hollow conoid projectile (with strength model). Column 2. Mass ratio (maSSspherdmaSSconoid)
equal to 1. Column 3. Mass ratio (massspherdmaSSco,oid)equal to 1.5.

Table 3. Lagrange backwall (rearview) damage area as a function of projectile size and type.
Projectile type Diameter of Damage area Diameter of Damage area
(Mass ratio projectile (mm) (x mm x y mm) projectile (mm) (x mm x y mm)
masssphere : 2097-15 2097-16
masSconoid)
Sphere (1:1) 4.4 none 4.7 14 x 20
Sphere (1.25:1) 4.7 10 x 18 5.1 16 x 31
Sphere (1.5:1) 5.0 10 x 20 5.4 19 x 28
Sphere (1.75:1) 5.3 20 x 34 5.7 20 x 44
Sphere (2:1) 5.5 18 x 40 5.9 28 x 40
Conoid (no str.) N/A 14 x 28 N/A 20 x 32
Conoid (str.) N/A 15 x 27 N/A 14 x 30

The BLMR was estimated by comparing the damage recorded on the computational witness
plates by the spherical projectiles and the hollow conoid projectiles. Figure 8 shows the damage
recorded on the computational witness plate for the spherical projectiles compared with the
hollow conoid projectiles modelled with and without a strength model. For increasing diameter
spherical projectiles, the area of damage on the front does not increase significantly but the
percentage area with material status "fully failed" within this area increases with increasing mass
ratio. This increased damage is reflected in the increase in failed material (shown by dark
regions on the rear side of the plate). The measurements of this damage area are given in Table
3. The results were then compared with the damage areas produced by simulations of the
hollow conoid projectiles to identify what range of spherical projectile diameter produced
equivalent damage areas. The areas of material failure, along the line of flight (lower region in
rear view of Lagrange plate) were measured and compared with the measurements on the rear
plate. The values are given in Table 3 and show that for both shots, the BLMR is between 1.5
and 1.75, the damage for these values encompassing the upper and lower limits of hollow conoid
projectile simulation (hydrodynamic and with strength as at room temperature). A previously
E.A. Taylor/International Journal of lmpact Engineering 26 (2001) 773-784 781

reported value of the BLMR at 11 km/s is 1.9 [5]. This value is slightly higher within the range
of values reported in this paper. Note that the BLMR is assumed to be velocity dependent,

Phase 2 simulations. Columbus A P M shield

In the Phase 2 simulations the midbumper was represented by an SPH layer. As for the
previous simulations, a sensitivity study was carried out to evaluate the effect of (i) simulating
the hollow conoid projectile with a strength model and (ii) varying the value of the failure stress
used in the simulation. For the baseline simulation, the projectile was modelled without a
strength model and tile failure stress for the midbumper was set to 2.5 GPa.
The simulation results at 20 ~ts showed the primary dispersed debris cloud causing
midbumper damage and perforation over nearly the whole simulation surface. There was little
variation in the perforation diameter as a function of the midbumper failure stress, although a
projectile modelled with strength produced a smaller perforation hole. This reflects the fact that
the central region of the debris cloud is less dispersed when strength is included. The hole
diameters are reported in Table 4.

Table 4. Hole diameter in the midbumper (x mm x y mm) as a function of time, projectile


strength and midbumper failure stress. All measurements in mm
ID t=20 ~ts t=30 ~ts t=45 ~ts t--70 Fs
Shot 2097-15 (baseline) 74 x 59 73 x 60 72 x 62 84 x 70
Shot 2097-15. Mid bumper failure stress 1 GPa 76 x 60
Shot 2097-15. Mid bumper failure stress 1 GPa + 61 x 37 70 x 50 -
projectile strength
Shot 2097-16 (baseline) 64 x 53 76 x 62 82 x 66 91 x 74
Shot 2097-16. Mid bumper failure stress 1 GPa 63 x 47
Shot 2097-16. Mid bumper failure stress 1 GPa + 64 x 53 79 x 61
projectile strength

For the two baseline simulations, the simulation was run until 70/as. The damage observed on
the midbumper at 70 ps was less than the experimental data observed (84 mm x 70 mm
compared with 130 mm x 100 mm; 91 x 74 mm compared with 200 mm x 150 mm). As the
brittle alumina fibres which form the Nextel weave fragment more easily than aluminium upon
passage of a shock wave this is to be expected. Note that the damage area produced
experimentally is the same size or larger than the simulated midbumper damage. Future
simulations aiming to reproduce this damage area will require a large number of SPH particles.
Figure 9 shows the velocity contours of the spallation region of the midbumper for the
baseline simulations (projectile modelled without strength and a midbumper failure stress of 2.5
GPa. Two regions are observed, corresponding to the spall caused by the debris cloud along the
line of flight and normal to the original impact point, where the velocities range from 1.9 to 2.1
km/s. The slight increase represents the influence of the strength model used to represent the
projectile. There is no difference between the velocity profiles of the spall debris clouds caused
by the two projectiles and the distribution of material is not strongly sensitive to material model
parameters. Fragments of solid midbumper material can also be seen to detach from the surface
in both the head on and side views. This may be an artefact of the SPH algorithms used.
As the midbumper debris cloud is travelling so slowly, the total computational time is
extended by a factor of 5 compared with propagation of the initial debris cloud. At 45 ~ts only
the portion of debris cloud travelling along the line of flight is impacting the bumper. The
bumper is deforming plastically. No hypervelocity impact craters or perforations are seen. At 70
~ts the bumper has deformed plastically by -4 mm in height for both simulations. Again, no
hypervelocity impact perforations have occurred. This is consistent with the velocity and
782 E.A. Taylor / International Journal of Impact Engineering 26 (2001) 773-784

material profile of the debris cloud discussed above. The experimental test results report small
bulges (~10-15 mm). These bulges are less than the computationally-derived deformations.

~toeav ~rr ~iv

i i::: i ....

Fig. 9. Velocity profile at 20 kts (left) and material density at 45 p.s and 70 Its (middle and right)
for simulation of shot 2097-16.

The highly dispersed nature of the initial debris cloud means that the midbumper fails by
spallation some time after the debris cloud impact. The absence of a high density projectile
remnant at the centre of the debris cloud means that rear wall damage will be caused by low
velocity spallation products. As Nextel and Kevlar-epoxy material models were not available,
the midbumper was simulated using an equivalent thickness of aluminium. The simulation
results show that the midbumper fails due to spall, with the peak velocity of the midbumper
debris cloud at ~ 2 km/s, with a slight increase in value when the projectile is modelled non-
hydrodynamically. Views of the midbumper debris cloud show that more material is present
when the initial projectile is simulated with a strength model and when the low bound value for
aluminium failure is used. This illustrates the importance of carrying out sensitivity studies. The
midbumper debris cloud loaded the rear wall, causing plastic deformation, but no perforation, for
both simulations.
Comparison of the results with experiment data showed that the perforation diameter of the
simulated aluminium bumper plate is less than the experimentally produced Nextel damage area.
This is as expected as the alumina fibres (from which Nextel cloth is woven) are brittle and fail
at lower tensile stresses than aluminium. For debris clouds with a solid and/or high density
central fragment, the equivalent thickness aluminium approximation may be more useful.

DISCUSSION

The simulation of a shaped charge jet projectile impact into a Whipple bumper requires a
number of assumptions and approximations. This section discusses the relative importance of
these assumptions, in the light of the results presented m previous sections.
Modelling of the hollow conoid shaped charge jet projectile within a hydrocode requires a
symmetrical shape (axial symmetry around the line of flight vector). This represents an
approximation to the observed asymmetrical shape. Varying the distribution of mass within the
volume produces changes in the distribution of the mass of the debris cloud, but not the overall
'footprint' of the cloud, when it impacts the second bumper. The shock equation of state is not
able to reproduce the changes of phase which occur at hypervelocity impact and thus represents
an approximation. The influence of the material models has been assessed. Experimental
observations of shaped charge jet temperature show that the jet is molten; although some
inconsistencies remain with hydrocode simulations. It is recommended that simulations of solid
and fully-molten projectiles are carried out to provide upper and lower limit data for comparison
with experimental results. Currently, due to computational limitations, the hollow conoid
projectile was modelled with only three SPH particles across the wall thickness. SPH requires at
least five particles to be consistent. In addition, at this low resolution, some debris cloud features
are lost.
E.A. Taylor/International Journal of Impact Engineering 26 (2001) 773-784 783

Utilisation of an equivalent thickness of aluminium to represent the second bumper


(midbumper) means that the failure of the Nextel is not correctly reproduced, nor is the
perforation of the Kevlar-epoxy plate. As MLI is not included in the simulation, its effect on the
emerging debris cloud is also not characterised. Correct reproduction of the simulation results is
probably not possible without use of appropriate material models to represent these materials.
In summary, the three key improvements required to allow simulation of shaped charge jet,
are (i) knowledge of the material state of the projectile and modelling of the hydrodynamics and
non-hydrodynamic projectile impact (ii) use of material models for Nextel and Kevlar-Epoxy,
and (iii) use of at least five SPH particles through the thickness of the shaped charge jet wall.
Use of (iii), in conjunction with the requirement to size the interim bumper such that the full
range of material damage can be modelled (e.g. Nextel damage area) will drive the simulation
size to the limits of current hardware performance. More efficient 3D SPH algorithms will be
required to explore fully the capabilities offered by the new Nextel and Kevlar-epoxy models.

CONCLUSIONS

The results from 56 simulations of shaped charge jet hypervelocity impact onto Columbus
APM Whipple bumper shields have been presented. The influence of projectile shape, pitch,
yaw, material model and SPH particle diameter on debris cloud morphology has been
characterised. Simulations of a shaped charge jet projectile (hollow conoid form) have shown
that after impact with the first bumper, a highly dispersed debris cloud is formed. The diameter
of an aluminium sphere producing the equivalent damage has been determined; the ballistic limit
mass ratio (BLMR) is between 1.5 and 1.75.
When this highly dispersed debris cloud impacts on the second bumper, the bumper failed
over a period of 10 ~ts, producing a large number of slow-moving fragments (velocities of 2 km/s
or less). These fragments cause the rear wall to distort plastically. In comparison, the
experimental results show deformation of the rear wall and perforation. Further simulations
should incorporate appropriate material models for Nextel and Kevlar-epoxy to simulate
correctly the failure of the interim bumper after impact by the highly dispersed debris cloud.

Acknowledgement--This simulationprogrammewas undertaken in the StructuresDivisionof ESTEC, ESA and the


author would like to acknowledgethe technical support given by the systemadministrators of ATOS (Per Flodstrom
and Dieter Suchar) and by Michel Lambert and Michel Klein (ESTEC, ESA). Century Dynamics Ltd (Colin
Hayhurst, Richard Clegg, John Ransom) provided technical support during the installation and running of
AUTODYN. Technicaldiscussions with Eric Christiansen (NASA)and Don Grosch(SwRI) are also acknowledged.

REFERENCES

[1] DestefanisR, Faraud M, Trucchi M. Columbus debris shielding experimentsand ballistic limit curves. Int. d.
Impact Engng. , 1999; 23:181-192.
[2] Bol J, Fucke W. Shaped charge techniquefor hypervelocityimpact tests at 11 km/s on space debris protection
systems. Proc. Second European Conf. on Space Debris, 1997; ESA SP-393:405-410.
[3] Bol J, Fucke W. Shaped changes for hypervelocity impact testing. Executive summary, ESTEC Contract
10.969/94/NL/PP(SC), BattelleReport No. R68.271.4, January 1997.
[4] GroschDJ. Inhibited shaped charge launchertesting of shield designs. Final Test report, SwRI Project No. 06-
2097, SouthWesternResearch Institute, September1998.
[5] Christiansen EL, Kerr JH. Projectile shape effects on shielding performance at 7 km/s and 11 km/s. Int. d.
Impact Engng., 1997; 20(1-5):165-172.
[6] HayhurstCJ, Hiermaier SJ, Clegg RA, Riedel W, Lambert M. Developmentof material models for Nextel and
Kevlar-epoxyfor high pressures and strain rates. Int. d. Impact Engng., 1999; 23: 365-376.
[7] ZernowL. The density deficit in stretchingshaped chargejets. Int. d. Impact Engng., 1997; 20(6-10): 849-860.
[8] Katayama M, Takeba A, Toda S, Kibe S. Numerical simulation of jet formation by shaped charge and its
penetration into bumpered target. Proc. Second European Conf. on Space Debris, 1997; ESA SP-393:411-
784 E.A. Taylor/International Journal of lmpact Engineering 26 (2001) 773-784

416.
[9] Walker JD, Grosch DJ, Mullin SA. Experimental impacts above 10 km/s. Int. J. Impact Engng., 1995; 17(1-6):
903-914.
[10] Jamet F, Thomer G. Flash radiography. Elsevier Scientific Publishing Company 1976.
[11] Prior TG. Summary of inhibited shaped charge launcher testing performed by Southwest Research Institute in
1998. JSC-28454, September 1998.
[12] Taylor EA. Numerical simulation of shaped charge impact on the Columbus APM shields, ESTEC working
paper, EWP-2029, June 1999.
[13] Hayhurst CJ, Livingstone IJ. Advanced numerical simulations for hypervelocity impacts. Final Report, ESTEC
Contract No. 12469/97/NL/GD, CDL Report No. R098:03, 12 June 1998.

Вам также может понравиться