Sunteți pe pagina 1din 13

International Journal of Impact Engineering 75 (2015) 110e122

Contents lists available at ScienceDirect

International Journal of Impact Engineering


journal homepage: www.elsevier.com/locate/ijimpeng

A non-linear orthotropic hydrocode model for ultra-high molecular


weight polyethylene in impact simulations
€ssig a, *, Long Nguyen b, Michael May a, Werner Riedel a, Ulrich Heisserer c,
Torsten La
Harm van der Werff c, Stefan Hiermaier a
a
Fraunhofer Institute for High Speed Dynamics, Ernst-Mach-Institut, Eckerstraße 4, D-79104, Freiburg, Germany
b
School of Aerospace, Mechanical and Manufacturing Engineering, RMIT University, GPO Box 2476, Melbourne, Australia
c
DSM Dyneema, P.O. Box 1163, 6160 BD Geleen, The Netherlands

a r t i c l e i n f o a b s t r a c t

Article history: This paper presents detailed experimental characterization of quasi-static anisotropic directional
Received 6 November 2013 strength properties as well as the shock behavior of ultra-high molecular weight polyethylene
Received in revised form (UHMWPE) for the development of an advanced material model for this class of materials. Specifically,
3 July 2014
we consider Dyneema® HB26 e pressed from uni-directional (UD) tapes in a 0/90 stacking sequence. A
Accepted 7 July 2014
Available online 28 July 2014
material model based on a constitutive law with orthotropic, non-linear strength, shock response,
composite failure and softening criteria is presented. A set of material parameters is derived for appli-
cations in hydrocodes (here: ANSYS AUTODYN). High- and hypervelocity impact tests with different
Keywords:
UHMWPE
impact velocities are used for preliminary validation and discussion of the predictive capabilities in view
High velocity Impact of future application.
Constitutive model © 2014 Elsevier Ltd. All rights reserved.
Orthotropy
Material characterization

1. Introduction absorption the investigated material Dyneema® HB26 is a prom-


ising candidate for protection against a large field of impacting
In recent years there has been a continuously increasing de- threats such as bullets and fragments [7e9]. Typically, composites
mand for high-performance composite materials that are suitable under high velocity impact situations show several damage and
for ballistic armor applications, for personnel and vehicle protec- failure mechanisms such as interlaminar delamination, permanent
tion, respectively [1e7]. These very specific applications demand non-linear deformation and fiber breaking within the perforated
for an optimized light-weight armor solution. Thus, polymer- layers [8,10,11]. Furthermore, observations showed that the mem-
matrix composite armor systems became more and more benefi- brane properties significantly influence the ballistic performance as
cial in reducing weight [4e6]. Many modern armor systems include well as the bulging on the rear face. Additionally, Karthikeyan et al.
polypropylene fibers (Tegris®, Curv®), Aramid fibers (Twaron®, [12] found that the ballistic performance can be improved by
Kevlar®) and fibers of ultra-high molecular weight polyethylene decreasing the in-plane shear strength. In connection to the
UHMWPE (Dyneema®, Spectra®) [8]. Predictive hydrocode models, aforementioned use for ballistic armor applications there are
based on thorough material characterization, are necessary for several literature reports that provide investigations that show
effective and customized design of protection systems. This paper is fundamental improvements towards modeling composite mate-
focused on experimental analysis, derivation of material constants rials [13e15]. When considering the micromechanical properties of
and numerical validation of a non-linear orthotropic material the orthotropic material behavior, the linear elastic properties, the
model as implemented in ANSYS AUTODYN. Due to its high specific orthotropic yield surface with a non-linear hardening description
strength to weight ratio (in fiber direction) and its high energy [16], a non-linear shock equation of state [17], and a three-
dimensional failure criterion supplemented by a linear ortho-
tropic softening description [13,18], should be taken into account. It
* Corresponding author. Tel.: þ49 761 2714 469; fax: þ49 761 2714 1469.
is important to consider all relevant mechanisms that occur during
E-mail addresses: torsten.laessig@emi.fraunhofer.de, torsten.laessig@gmx.de ballistic impact, as the quality of the numerical prediction capa-
€ssig).
(T. La bility strongly depends on a physically accurate description of

http://dx.doi.org/10.1016/j.ijimpeng.2014.07.004
0734-743X/© 2014 Elsevier Ltd. All rights reserved.
€ssig et al. / International Journal of Impact Engineering 75 (2015) 110e122
T. La 111

contributing energy dissipation mechanisms. The relevant phe- stacked in alternate directions and hot-pressed, such that a desired
nomena are as follows: Initially, a transverse shock wave runs layer sequence of [0/90]n is obtained. The production scheme was
through the layers of the material. The damage mechanism domi- presented in detail by Russell et al. [25]. A microsection of the
nating within the top plies is through-thickness shearing caused by layered structure using dark field microscopy is given in Fig. 1.
the projectile cutting through the target. In the meantime, the Due to its cross-ply layup, Dyneema® HB26 is assumed to be
shock wave is reflected at the back face, causing through-thickness orthotropic. For the presented modeling purposes, the material
tensile stresses. These tensile stresses can cause separation of the properties of 0 - and 90 -directions are therefore taken to be equal.
plies (delamination). The final phase is dominated by tensile In this work the 0 - and 90 -direction of the laminate is asso-
membrane deformation, which manifests itself as the typical back ciated to the 11- and 22-direction, respectively, and therefore 33 is
face bulging and delamination [19,11]. A material model that takes the through-thickness direction. The 45 -direction, which will be
into account all of the required damage mechanisms described discussed below, is defined as the direction 12 for the in-plane
above for modeling ballistic impact on composite materials was shear test.
proposed and validated in Refs. [20e22]. Within the current paper
this material model is applied to simulating the response of
2.2. Tensile tests of 0/90 -specimen
Dyneema® HB26 subjected to high- and hypervelocity impact
loading. A substantial experimental testing program was carried
For determining the in-plane properties in fiber direction tensile
out to enable model calibration. This includes stiffness and strength
tests were carried out. For that purpose specimens were cut out
data for in-plane as well as through-thickness directions, mode I
from 2 mm thick plates, with the layer sequence ([0/90]2)5, via
fracture toughness [23], and the shock equation of state (EOS) [24].
water-jet cutting in 0 -direction. To prevent sliding of the inner
The resulting data set, in the following referred to as data set “TL3”,
layers in the clamping when using standard norm specimens such
is given in the Appendix of this paper.
as DIN EN ISO 527-4 [26], a form-fit clamping condition was used to
An overview of the experimental program is presented in
lock each layer simultaneously. An account of challenges to test
Table 1.
these composites is given in Russel [25] and Levi-Sasson [27].
For verification purposes, some of the aforementioned material
Hence, a specimen was developed by Russell et al. [25] and was
tests were simulated and the measured signals were compared to
adopted in this study as shown in Fig. 2.
the calculated ones. This step concerns the following character-
As shown in the drawing, the specimen was clamped form-fit
ization components: the orthotropic yield surface, characterizing
with M4-Bolts. The tensile tests were carried out using a servo-
the hardening effects and calibrating the equation of state against
hydraulic testing machine shown in Fig. 3.
inverse planar plate impact tests. Finally, a new orthotropic mate-
A force history curve and a displacement history curve were
rial data set is determined which can be used for further in-
obtained from the load cell and crosshead, respectively. Additional,
vestigations of the material behavior under highly dynamic
an optical strain analysis was carried out using a high-speed camera
loading. This will be illustrated by modeling some experimental
and the optical analysis software ARAMIS® [28]. To enable enough
impact tests in which an aluminum sphere (diameter 6 mm) was
contrast the specimens were marked on the surface with an
fired at Dyneema® HB26 with impact velocities from 2052 to
inhomogeneous black speckled pattern. The resulting true stress
6591 m/s. These numerical simulations shall highlight predictive
true strain curves are reported in Fig. 4 and Table 2 including the
capabilities and remaining deficiencies for future ballistic in-
arithmetic average and the coefficient of variation (COV).
vestigations on ultra-high molecular weight polyethylene com-
Here, grey solid lines show the results of tensile tests in 0/90 -
posite materials.
direction and the dashed black line the numerical validation
simulation using a one-cell FE-model providing material data set
2. Experimental investigations “TL3”.
The typical true stress e true strain curves of 2 mm thick
2.1. Laminate microstructure Dyneema® HB26 is given by Fig. 4. This chart represents the results
of five 0/90 -samples under quasi-static tensile loading. Note that
Dyneema® fiber is produced via gel-spinning followed by hot the stress-strain relation remains linear-elastic during the loading
drawing. The resulting fibers, consisting of highly oriented mole- procedure until fracture, clearly represented by the sudden drop of
cules, are coated with a polyurethane (PU) resin (matrix material) stress. In the five tests, an averaged maximum stress of 753 MPa
and form the UD ply precursor (for Dyneema HB26). These plies are was obtained. The Young's modulus E11 ¼ E22 ¼ 26.9 GPa was ob-
tained at values of strain between 0.05 and 0.25 percent, as spec-
ified in Ref. [26].The longitudinal strain was obtained using the side
Table 1 view to disregard obstruction by a failing top ply. Furthermore,
Material characterization program for UHMWPE Dyneema® HB26.

Type of test Repetitions Velocity Gauge Property


length

0/90 -tension 5 1 mm/min 50 mm E11, E22, s11fail,


s22fail
±45 -tension 5 1 mm/min 50 mm G12, t12fail
Through-thickness 5 1 mm/min 25 mm s33fail
tension
Ultrasonics [25] e e e E33
Through-thickness 5 1 mm/min e G13, G23, t13fail,
shear t23fail
Double Cantilever 3 1 mm/min e GIC
Beam test (DCB)
Inverse planar plate 5 229e868 m/s e shock and
impact test (PPI) release

11- and 22-directions unfold the plane of lamina. Fig. 1. Microsection analysis of Dyneema® HB26, dark field microscopy.
112 €ssig et al. / International Journal of Impact Engineering 75 (2015) 110e122
T. La

Fig. 2. Specimen for tensile tests in ±45 -(*) and 0/90 -direction (**).

noticeable delamination of the layers in through-thickness direc-


tion took place when failure of the specimen occurred. The
apparent Poisson's ratio was found to be negative. However, this is
not a material property. It is a structural effect caused by delami-
nation of the specimen. Non-obtained data for Poisson's ratios was
taken from literature [5].

2.3. In-plane shear tests on ±45 -specimens

Utilizing the same clamping condition, testing machine and


instrumentations with the alternative specimen geometry from
Fig. 2, quasi-static tensile tests were carried out for a ±45 -lami-
nate. The true shear-stress-shear-angle curve as an outcome of the
before mentioned local strain analysis is given by Fig. 5 and Table 3.
The principal trend of the curve shows the three typical mech-
anisms of a ±45 -laminate tensile test. At a shear angle from 0 to
0.25 rad, the matrix behavior is dominant for the nonlinear elas-
Fig. 3. Test setup for in-plane tensile tests. ticeplastic trend (Fig. 5, zone A). In zone B from 0.25 rad until about
0.6 rad, fiber reorientation and loading in fiber direction is
increasing. From about 0.6 rad until complete failure, the curve
increases exponentially (zone C), mainly because of increasing in-
fluence of tensile fiber strain. Complete failure occurs when fiber
reorientation reaches a critical angle. From that point on, the fibers
detach from each other in the same plane as well as through
thickness. The averaged failure shear stress was t12 ¼ 35.2 MPa. The
shear modulus was derived as tangent modulus at the origin with
G12 ¼ 42.3 MPa as the average within the experimental scatter
(Table 3).

Fig. 4. True stress- true strain curves for 0/90 -specimens (direction 11 or 22) under
quasi-static tensile loading compared to the verification simulation (dashed, see sec-
tion 3.1).

Table 2
Young's moduli and failure stress from 0/90 tension tests, with the average (avg)
and coefficient of variation (cov).

Property Test No

1 2 3 4 5 Avg. Cov

E11 & E22 [GPa] 25.2 27.2 24.4 22.5 35.3 26.9 0.18
Fig. 5. True shear stress-shear angle curves for ±45 -specimens (direction 12) under
s11fail & s22fail [MPa] 765 751 704 765 782 753 0.04
quasi-static tensile loading with verification simulation (see section 3.1).
€ssig et al. / International Journal of Impact Engineering 75 (2015) 110e122
T. La 113

direction was obtained by acoustic measurements. A maximum


tensile strength of s33 ¼ 1.07 MPa was measured (Table 4).

2.5. Out-of plane shear tests

In order to characterize the shear behavior in through-thickness


direction diverse test configurations have been examined in the
past, e.g. by Feraboli [30] and Kim [31]. These test configurations
were prone to bending however. On that basis a new test method
was developed at EMI [21] [32] which minimizes bending
compared to shear deformation. The test configuration is given in
Fig. 7.
The shear stresses in through-thickness directions t13/23 as well
as the corresponding shear angles g13/23 can be obtained by the
experimental force displacement curve along with Eqns. (2)e(4) for
the 13-shear- and 23-shear-plane:

Fig. 6. Engineering stress-strain curves for tension of cubes in through-thickness di-


w 
s
rection with verification simulation (dashed line). g13=23 ¼ arctan ; (2)
l

2.4. Through thickness properties Fl3 4Fl3


With wb ¼ ¼ and ws ¼ wt  wb ; (3)
3EI Ebt 3
The stiffness in through thickness direction was obtained as
follows: F
t13=23 ¼ : (4)
2bt
E 33 ¼ r0 $c2B (1)
Deduced from the geometric logic in Fig. 7 the shear angle g13/
In Eq (1), r0 is the initial density and cB is the acoustic bulk sound 23 can be formulated with Eqs (2) and (3). The shear stress-shear
speed. An initial density of 0.98 g/cm3 and an acoustic bulk sound angle relation is defined by Eqs (2) and (4). Here, wt, ws and wb
speed of 1922 m/s (see Section 2.7) were used. Eq (1) provides a denote total, bending and shear displacement, respectively. The
Young's Modulus in through thickness direction of E33 ¼ 3.62 GPa. shear angle can be derived by the pure shear displacement ws and
In order to determine the strength in through thickness direction the free length l. Furthermore, the pure shear displacement can be
quasi static tensile tests were carried out. The quasi static tensile determined by reducing the transverse displacement wt by the
tests were carried out using cubes with an edge length of 26 mm. bending displacement wb Eq (3). For the quasi-static shear test in
The front and rear sides were grinded planar and glued between through-thickness direction strips of 200  25  4.3 mm were
two loading cylinders using a two-component epoxy resin. The used. Due to the crosswise layup of the laminate the shear
resulting engineering stress-strain curves and the corresponding behavior in the 13- and 23-shear-plane (through-thickness) is
verification simulation of the modeling approach described in assumed to be equal.
Section 3.1 are given by Fig. 6 and Table 4. The quasi-static shear test results are given in Fig. 8 and Table 5.
The resulting engineering stress-strain curves (Fig. 6) show After limited initial setting in the clamping elastic deformation
significant scatter. This has frequently been observed for tensile between a shear angle of 0.1 and 0.15 rad (Fig. 8) can be observed. In
tests on composites through-thickness direction and is also this deformation phase the shear moduli G13 and G23, respectively,
mentioned in standards [29]. These tests are strongly dependent on were obtained as secant moduli. Shear deformation remains
the properties of the resin. For the investigated composites scat- dominant until 0.5 rad, which is then followed by increasing in-
tering is particularly pronounced as the resin volume fraction is fluence of tensile loading in fiber direction (Tables 3 and 4).
only 17 percent [25]. The combination of this fact and the varying In order to avoid collision between loading fork and fixed sup-
degree of fiber bridging between fracture planes results in signifi- port, the testing device was limited in displacement. Therefore, the
cant scatter in strength. This can be observed in the form of pla- maximum shear stresses that could be obtained was g13/
teaus for two specimens between the engineering strain of 0.005 23 ¼ 0.8 rad. As shown in Fig. 8, the stress at that point does not
and 0.018, (Fig. 6). For this reason, the stiffness in through thickness represent the failure shear stresses in through-thickness direction.

Fig. 7. Configuration for quasi-static shear tests in through-thickness direction.


114 €ssig et al. / International Journal of Impact Engineering 75 (2015) 110e122
T. La

Throughout the shear tests it was observed that no global failure of Table 3
the specimen was achieved. In the progressive leading phase Shear moduli and shear failure stress from ±45 tension tests.

(g13  0.4 rad) the specimen was pulled into the shear test device Property Test No
causing tensile stress in fiber direction but no more deformations in 1 2 3 4 5 Avg. Cov
through-thickness shear direction. For this reason, the yield shear
G12 [MPa] 38.5 42 48 46.5 36.5 42.3 0.12
stress t13yield was derived from the point of matrix failure at about
t12fail [MPa] 35.3 35.6 34.3 35.6 35.2 35.2 0.01
0.15 rad. The results of the shear tests in through-thickness direc-
tion are summarized in Table 5.
elastic potential, divided by the releasing fracture zone, yields the
Since Dyneema® HB26 was found to be orthotropic, the material
parameters in the two shear planes in through-thickness direction Table 4
were assumed to be equal. Tensile test results in through-thickness direction.

Property Test No
G13 ¼ G23 and t13yield ¼ t23yield :
1 2 3 4 5 Avg. COV
It is important to mention that no specimen failure could be s33fail [MPa] 1.08 1.13 1.08 0.94 1.12 1.07 0.07
detected and accordingly, no failure shear stress t13fail could be
derived from the current test. However, shearing can always be
represented as a combination of tension and compression under
45 rotation. As already observed in the ±45 tension tests it may be Table 5
part of the deformation phenomenology of this material which is Results of the shear tests in through-thickness direction.
particularly weak in shear and very strong in tension, that shearing
Property Test No
is rotated to tensile fiber loading locally upon excessive shear
1 2 3 4 5 Avg. COV
deformation.
G13 [MPa] 30.5 29.6 29.7 34.4 29.2 30.7 0.07
t13yield [MPa] 2.67 2.59 2.58 2.68 2.54 2.61 0.02
2.6. Determination of mode I fracture toughness
Mode I fracture toughness as given in Eq (6) where b is the beam
In order to derive the post failure response, double cantilever width.
beam (DCB) tests in through-thickness direction were performed as
proposed by Nossek [33] and ASTM D 5528 [34]. Because of the  
1 vPðP; aÞ P 2 ·a2
weak bending stiffness of Dyneema® HB26 in through-thickness GI ¼   ¼ (6)
b va P EI·b
direction the specimen's thickness was increased from 5 mm to
26 mm. The results of the three quasi-static DCB-tests are plotted in
Fig. 9. Presented are the experimental force-opening displacement Due to the uniaxially applied bending load and the fact that the
curves as well as the black dashed curves obtained from plotting dominating part of the bending stress is perpendicular to the fiber
the analytical formulation derived by Nossek [33]. Nossek used direction, the bending modulus was approximated by the Young's
elastic beam theory to derive Eq (7) from the elastic potential P Modulus in fiber direction, E.
defined by Eq (5).
Za Za 8 rffiffiffiffiffiffiffiffiffiffiffi
1 1 P 2 ·a3 >
> 3·EI·u a20 GIC ·b
P¼ sdε/ EI·k2 ¼ (5) >
> if u  ;
2 2 3EI >
< a30 3 EI
0 0 PðuÞ ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi (7)
>
>
> EI·ðGIC ·bÞ3
4
Here, the elastic potential is a functional expression of the >
>
: pffiffiffiffiffiffiffi otherwise:
bending stiffness EI, the curvature of the bending beams k, the 3·u
opening force P and the total crack length a. The integral of the
The first part of Eq (7) can be derived from Euler-Bernoulli beam
bending theory before crack initiation occurs. Here, I denotes the
area moment of inertia, GIC the fracture toughness, b the width of

Fig. 8. Engineering shear stress-shear angle measurements in through-thickness di- Fig. 9. Forceedisplacement curves of DCB-tests with different fit curves for the soft-
rection 13 or 23, shown in five overlaying tests, compared to the simulation (dashed). ening branch according to second part of Eq (7).
€ssig et al. / International Journal of Impact Engineering 75 (2015) 110e122
T. La 115

When the projectile impacts the target, pressure waves propa-


gate back and forth through all plates. Every time a pressure wave
arrives at the free surface of the witness plate, an acceleration of
this surface occurs [38], see Fig. 12. The experimental free surface
velocity history (vfs-t-curves) was measured using a VISAR system
(velocity interferometer system for any reflector). Extensive infor-
mation about the functionality of VISAR is given in Refs. [38e41].
Five velocity curves (solid lined) were used to derive the shock
properties as the Us-uP-relationship and to validate the equation of
state (dashed lines) in Fig. 12.
For obtaining material properties a Hugoniot curve with linear
shock vs. particle velocity relationship (US-uP-relationship) is
assumed as.

US ¼ cB þ SuP ; (8)
Fig. 10. EMI test facility for inverse planar plate impact tests.

with the acoustic bulk sound speed cB and the slope S [17]. The
the specimen, and u the opening displacement. The second part of Hugoniot-stress sh characterized by the first velocity plateau u2,
Eq (7) is valid for states when the crack propagates and P can be occurring between 1000 and 1800 ns in Fig. 12, can be expressed
derived by the equilibrium state GI ¼ GIC. A graphical evaluation through the momentum conservation as:
plotting the second part of Eq (7) against the averaged force-
sh ¼ r0 US uP ; where r0 is the initial density: (9)
opening displacement curves to obtain the fracture toughness GIC
is shown in Fig. 9. Mass conservation and the compressive strain εh at the
The Mode I fracture toughness was determined as Hugoniot-state are defined as
GIC ¼ GC11 ¼ 0.79 kJ/m2 (Table 6). Compared to the value given by
Grujicic et al. (544.7 J/m2) [5], the fracture toughness obtained by uP
rh ðUS  uP Þ ¼ r0 ðUS  u0 Þ and εh ¼ (10)
our experiments is 31% higher. It should be mentioned that, due to US
the lack of experimental data, the values for the fracture toughness
The specific energy can be calculated as
in other modes were taken from literature [5].
 
1 1 1 1
e  e0 ¼  s ¼ u2 : (11)
2 r0 r h 2 P
2.7. Inverse planar plate impact test
The momentum conservation Eq. (9) is combined with mass
Planar plate impact tests were carried out to obtain the shock conservation Eq. (10). As a result, the Hugoniot stress sh on the
response in through-thickness direction and particularly material- interface of sample and target plate can be calculated by Eq. (12).
specific US-uP-data at strain rates up to 106/s [24]. Five experiments The subscript w denotes the known material properties
were performed using an inverse test configuration described by (r0,w ¼ 7.8 g/cm3, cp,w ¼ 6000 m/s, Sw ¼ 1.332, uHEL_w ¼ 75 m/s, see
Riedel [35] [36], Grady [37] or Meyers [17]. During an inverse planar Refs. [36,38]) of the C45 steel target, which is often called “witness
plate impact (PPI) test a specimen of a material with unknown plate”.
8
properties, in this case Dyneema® HB26, is shot onto a target of a > 1
>
< r0;w cP;w u2 if u2 < uHEL;w ;
material which is already well characterized under shock loading. 2
The projectile and target configuration is shown in Fig. 11. The sh ¼
>1
> 1 
experimental program was carried out on a single stage gun. This : r cP;w uHEL w þ r0;w US umax  uHEL w otherwise:
2 0;w 2
can be operated using compressed air for vimp ¼ 200e500 m/s and
(12)
with powder charges for impact velocities of 500 m/s up to 1100 m/
s [38]. The test facility is shown in Fig. 10. Here, umax denotes the maximum velocity of the free surface,
The projectile consists of a steel backing plate bonded to the uHEL_w the free surface velocity at Hugoniot Elastic Limit (HEL) of C45
sample made of the investigated material. This layered projectile is steel witness plate. In contrast to the formulation used by Meyers
then glued onto a polymer sabot. The target is a stationary witness et al. [17] where the low velocities at HEL are not taken into account.
plate made of well characterized C45 steel. Eq. (12) is applicable for calculating the maximum Hugoniot-stresses

Fig. 11. Projectile and target configuration for inverse PPI tests on Dyneema® HB26.
116 €ssig et al. / International Journal of Impact Engineering 75 (2015) 110e122
T. La

shock properties are different, the shock equation of state (shock


EOS) should be anisotropic. Here, the shock EOS is modeled
isotropic due to the large dimensions of the target panel in 11- and
22-direction.

3. Material model and numerical application

The material model developed in Ref. [21,22,32], and imple-


mented in ANSYS-Autodyn [45] is used to model the deformation
phenomena of UHMWPE under impact. Hereafter, the derivation of
model parameters for shock response, orthotropic elastic proper-
ties, hardening effects as well as the failure and softening behavior
is shortly summarized. We close this section by demonstrating the
numerical applicability of the material model with hypervelocity
impact simulations up to 6591 m/s.

3.1. Equation of state

The material states reached by shock loading are described by an


equation of state (EOS) [24]. A polynomial form of the Mie-Grü-
neisen equation of state expresses the volumetric response under
Fig. 12. Comparison of inverse PPI with Dyneema® HB26 and simulated back surface compressive hydrostatic loading (Eq (15)). Since the material is
velocity histories for polynomial EOS data. orthotropic, the volumetric stress is also partly coupled to the
deviatoric strains as proposed by Anderson et al. [13] and adopted
with respect to the velocity at HEL of the witness plate material [38]. in Ref. [21].
Therefore, Eq. (12) is always beneficial for low impact velocities and
P ¼ A1 m þ A2 m2 þ A3 m3 þ ðB0 þ B1 mÞr0 e (15)
low pressures. The up range particle velocity of the initial shock in the
Dyneema® HB26 sample plate up,Dyn is derived using Eq (13).
r B þ B1 m
where m ¼  1 and G ¼ 0 (16)
r0 1þm
1
uP;Dyn ¼ vimp  u2 : (13)
2 Eq (15) describes a polynomial equation of state with the cur-
rent density r, the reference density r0 and internal energy e. The
The plate dimensions, initial velocities and resulting Hugoniot
effective bulk modulus A1 is calculated from the orthotropic stiff-
states are summarized in Table 7. Hugoniot stresses from 499 MPa
ness coefficients as described in Ref. [13,46]. A2 and A3 are free
to 2987 MPa were reached in the inverse impact experiments
parameters and are tuned to match the first shock induced plateau
(Table 7). Hence, five points in pressure-density-internal energy
u2 in the inverse PPI test results (Fig. 12) according to Ref. [21], [32].
space are obtained. The Hugoniot curve as an US-uP-relationship for
B0 ¼ B1 in Eq (15) are determined by the Mie-Grüneisen Gamma G
Dyneema® HB26 is displayed in Fig. 13 and is complemented with
coefficient, which controls compression and release states in the
data provided by Hazell et al. [42].
vicinity of the reference (Hugoniot) curve. The initial value is
Fig. 13 shows an increasing relationship between the transverse
approximated to G y 2S-1, as proposed by Dugdale and Mac-
shock velocity and the particle velocity, which can approximated by
Donald [47] for metals. S describes the linear slope of the US-uP
the linear fit Eq (14).
relationship (see Fig. 13 and Eq (14)). The release plateaus
following u2 before the arrival of the back reflection around
US ðuP Þ ¼ cB;Dyn þ SDyn $uP ¼ 1922 þ 2:432$uP (14) 400e500 ns in Fig. 12 can then be used to validate or refine the
Eq (14) contains the derived acoustic bulk sound speed
Grüneisen coefficient. The procedure and structure of the inverse
cB,Dyn ¼ 1922 m/s and the shockeparticle relationship coefficient
planar plate impact curves are further explained in Refs.
SDyn ¼ 2.432 for Dyneema® HB26. Additional US-uP-data was ob-
[32,36,46]. As seen in Fig. 12 it turns out, that the simple approx-
tained from former studies at EMI [43] and by Chapman [44]
imation gives a good match of the release plateau velocities and
(provided by Hazel et al. [42]). We note that this investigation
timings.
only considers the shock behavior in through-thickness direction
The comparison of the measured (solid) and simulated (dashed)
(33-direction) and not in in-plane direction. Indeed, the shock
back surface velocity curves in Fig. 12 shows very good overall
properties in in-plane directions are different because of different
replication of initial shock, release states, reflected shocks and final
stiffnesses and, therefore, various acoustic sound speeds. Con-
velocities. The described equation of state approach thus describes
cerning the influences towards the ballistic performance of
shock and release states between 370 MPa and 2.3 GPa to a good
Dyneema® HB26, although the in-plane and through-thickness
approximation.

Table 6
Evaluation of mode I fracture toughness GIC according to second part of Eq (7) and 3.2. Orthotropic strength and hardening
Fig. 9.

Property Test No The orthotropic linear elastic parameters for the macro-
1 2 3 Avg. COV
mechanical laminate behavior are specified with following
stressestrain relationship Eq (17) including nine independent
GIC [kJ/m2] 0.81 0.78 0.77 0.79 0.03
stiffness coefficients Cij.
€ssig et al. / International Journal of Impact Engineering 75 (2015) 110e122
T. La 117

Table 7
Configuration data and experimental results of inverse PPI for Dyneema® HB 26.

Test tb [mm] tdyn [mm] tw [mm] vimp [m/s] ufs1 [m/s] uP [m/s] sh [MPa] Us [m/s]

a 5 6.1 2 229 21.3 218 499 2335


b 5 6.1 2 274 27.0 260 631 2475
c 5 6.1 2 357 43.5 336 1017 3092
d 5 6.1 2 427 45.2 405 1057 2665
f 5 6.1 2 573 76 535 1777 3387
e 5 6.1 2 868 131.1 803 2987 3798

master curve can therefore be used to describe material hardening,


. .
ε ¼ C 1 $ s with which can principally be calibrated by any of the three in-plane or
2 3 shear stress-strain responses. Generally, the master curve is cali-
1 n21 n31
6 E 0 0 0 7 brated in the plane that experiences the most non-linear plastic
6 11 E22 E33 7
6 7 deformation. In this case the shear stress-strain response in the 12-
6 7 shear plane measured by the ±45 -tension tests in Fig. 5. Due to the
6 n12 1 n32 7
6 0 0 0 7
2 3 6 E11 E22 E33 7 2 3 formulation of the nine-parameter parabolic yield function, one of
ε11 6 7 s11
6 7 the parameters in Eq (18) can be set independently. In this case the
6 ε22 7 6 n13 n23 1 7 6 s22 7
6 7 6 0 0 0 7 6 7 plasticity coefficient a66 is set to 1. According to [45], the effective
6 7 6 E 7 6 7
6 ε33 7 6 11 E22 E33 7 6 s33 7 stress seff and the effective plastic strain εp are calculated from s12
6 7¼6 7$6 7
6 g23 7 6 7 6 t23 7 and ε12 from the ±45 -tension tests as follows.
6 7 6 1 7 6 7
6 7 6 0 0 0 0 0 7 6 7 rffiffiffiffiffiffi
4 g13 5 6 G23 7 4 t13 5
6 7 3 pffiffiffiffiffiffiffiffiffiffi
g12 6 7 t12 s≡ f ¼ 3a66 s12 where s12 ¼ 2t12 and (19)
6 1 7 2
6 0 0 7
6 0 0 0 7
6 G13 7
6 7 p
6 7 ε12 t12 g12
4 1 5 εp ¼ pffiffiffiffiffiffiffiffiffiffi where
p
ε12 ¼ ε12  with ε12 ¼
0 0 0 0 0 3a66 2G12 2
G12
(17) (20)

The experiments clearly show non-linear, permanent defor- The effective stress-effective plastic strain curve used in the
mation during the ±45 tensile tests in the 12-shear plane (Fig. 5). A material model is approximated by 10 stress-strain data points
quadratic yield function proposed by Chen et al. [16] with an tabulated in the Appendix. The plasticity coefficients a11 and a22 are
associate flow rule is used to describe this orthotropic hardening derived from the 0/90 -tensile test results in Fig. 4. The material
behavior. This is described in Eq (16). model is then validated against the experimental results using a
single element numerical model as shown by the dashed line
 in Figs. 4e6 and 8. Due to absence of experimental data the mixed
f sij ¼ a11 s211 þ a22 s222 þ a33 s233 þ 2a12 s11 s22 þ 2a23 s22 s33
coefficients a12, a13 and a23 were set to a value near zero to suppress
þ 2a13 s11 s33 þ 2a44 s223 þ 2a55 s231 þ 2a66 s212 ¼ k significant hardening effects under combined loading. Further-
(18) more, the plasticity coefficients a44 and a55 were set equal to a66. In
addition, several restrictions must be made to maintain the stability
The constant plasticity coefficients aij and the hardening of the Euler-Backward return algorithm [45]. This implies, that the
parameter k define the shape and size of the yield surface. The yield surface in the through-thickness stress space generates a fully
constant plasticity coefficients aij keep the shape of the surface but closed ellipsoid. For this purpose, Eq (21) is limited to DetE < 0, rank
replacing k by a hardening function modifies its size [45]. A general of e equals three and the rank of E must equal four. In addition, all
non-zero components of e must have the same sign.

2 3
a11 a12 a13 0 2 3
6 a21 a11 a12 a13
a22 a23 0 7
E¼6
4 a31
7 and e ¼ 4 a21 a22 a23 5
a32 a33 0 5
a31 a32 a33
0 0 0 k
(21)
The resulting data set for the plasticity coefficients represents a
highly elongated ellipsoid in the principal stress space and can be
found in the Appendix.

3.3. Composite failure criteria

The adopted material model uses an extended three-


dimensional Hill criterion (Eq (22)e(24)). It consists of three fail-
Fig. 13. Hugoniot data for Dyneema® HB26 derived from inverse plate impact tests
ure surfaces ( e2iif , i ¼ 1, 2, 3) with two additional quadratic terms to
with linear Us-up approximation complemented by data from former studies [43] and account for shear stresses in the in-plane and out-of-plane di-
data provided by Hazell et al. [42]. rections. Softening is initiated if a value of 1 is reached.
118 €ssig et al. / International Journal of Impact Engineering 75 (2015) 110e122
T. La

!2 !2 !2 between two steel frames. Projectile residual velocity was obtained


s11 s12 s31 by evaluating images captured by a digital high speed camera. The
e211f ¼ þ þ  1; for 11  plane;
s11;fail s12;fail s31;fail corresponding FE-model compared to the experiment for an impact
velocity of 3532 m/s is presented in Fig. 14.
!2 !2 !2 (22)
s22 s12 s23 Fig. 14 presents an exemplary sequence of eight pictures of the
e222f ¼ þ þ  1; for 22  plane; side view of the FE-model compared to the pictures taken by the
s22;fail s12;fail s23;fail
high speed camera during the ballistic experiment. The first (from
(23) the left) of the depicted pictures corresponds to the moment when
!2 !2 !2 the high speed cam was triggered. The position of the projectile was
s33 s23 s31 approximately 10 cm in front of the target panel. The subsequent
e233f ¼ þ þ  1; for 33  plane;
s33;fail s23;fail s31;fail pictures show how the penetration process causes an ejecting
(24) debris cloud on the back face. There, the front of the debris cloud
containing parts of the panel material and the projectile are indi-
A linear formulation describes exemplarily in 11-direction cated by dashed lines. Furthermore, the outermost diameters of the
the degradation of the corresponding stress component Eq visible bulging areas are marked with brackets (*). It can be noted
(25) and is controlled by the damage coefficient Dij. Eq (25) that those two distinctive phenomena during perforation of the
defining post-failure model in the 11-direction, including Dyneema® HB26 panel after hypervelocity impact, were captured
linear softening (Eq (26)). quite realistically, also for the other ballistic tests up to impact ve-
!2 !2 locities of 6591 m/s. As presented in Fig. 14, the full setup was
s11 s12 modeled. Fine mesh at the vicinity of the impact and coarse meshes
e211f ¼ þ away from the center were generated using eight-noded brick el-
s11;fail ð1  D11 Þ s12;fail ð1  D12 Þ
!2 (25) ements exclusively. At the vicinity of the impact an element edge
s31 length of 1.5 mm was kept in all directions. The sphere was meshed
þ 1 with 0.5 mm brick elements. The panel was constrained at the edge
s31;fail ð1  D31 Þ
by a zero-velocity boundary condition in impact direction (x-axis).
The external gap contact formulation with no friction was used
L$sij;fail $εcr between sphere and panel. To reduce hourglass deformations
with Dij ¼ i; j ¼ 1; 2; 3: (26)
2$GCij around the vicinity of the impact a damping factor 0.1 was chosen.
An erosion model was used with an instantaneous geometric
The softening is formulated as degradation of the failure stress
erosion strain of 150 percent for Dyneema® HB26 and 250 percent
(Eq (25)) based on the critical crack strain εcr, the characteristic cell
for the aluminum sphere in order to avoid excessive computation
dimension L and the fracture toughness GCij (Eq (26)). Mesh size
times. It is noted that the erosion criterion is triggered post-failure
influence is thereby excluded. At failure initiation Dij is initially zero
with respect to any strain direction and does not affect the me-
and increases to the value of 1, where the material loses all strength.
chanical behavior of the material. Fig. 15 plots the projectile impact
A detailed derivation can be found in Ref. [45]. The damage coef-
and residual velocity of the experimental and simulated results. A
ficient Dij is implemented time step wise and includes a coupling
Lambert-Jonas-form of the ballistic limit equation (Eq (30)) was fit
coefficient c* with a value between 0 and 1 in Eq 27e29.This allows
against the experimental results, using the definition from Ref. [48].
for damage coupling between the different modes. By setting c* ¼ 0,
The fit was carried out using GNUPLOT in accordance with the
it is possible to simulate shear failure for in-plane and through-
LevenbergeMarquardt-Algorithm [49,50].
thickness direction, while the material in the fiber direction
exemplarily still remains intact. The damage coefficients Dij are  1
p p p
exemplarily shown for the post failure behavior in 11-direction. vres ¼ a$ vimp  vbl ; vimp > vbl : (30)

* Here, vimp is the impact velocity and vbl denotes the ballistic
Dnþ1 n
11 ¼ D11 þ DD11 þ c ðD12 þ D31 Þ (27)
limit velocity. As suggested in Ref. [51] the parameter p is set equal
2 assuming constant energy absorption. Parameter a was used for
*
Dnþ1 n
31 ¼ D31 þ DD31 þ c ðD11 þ D12 Þ (28) calibration and was determined as 0.37 for experimental and 0.40
for numerical obtained values.
Dnþ1 n * Fig. 15 shows comparison of the numerical (X) and experimental
12 ¼ D12 þ DD12 þ c ðD31 þ D11 Þ (29)
(black squares) results in terms of the impact velocity - residual
For validation simulations a coupling coefficient of 0.5 was used velocity relation of a 6 mm Aluminum sphere and a Dyneema®
and showed satisfactory results at several impact situations. The HB26 panel with an area density of 15 kg/m2. Furthermore, the
complete input data obtained by experimental investigations, graph should clarify the sensitivity of the ballistic performance to
analytical calculations and calibration simulations is summarized the used hardening model and the shock-EOS, respectively. The
in data set “TL3” (Appendix). results of the numerical simulation, using the orthotropic elastic
version of the material model (without hardening formulation),
3.4. Numerical application and predictability of ballistic shows a significant underestimation of ballistic performance (grey
performance triangles, Fig. 15). In contrast, however, a slightly overestimation at
higher impact velocities was caused by switching off the poly-
In the following section, the orthotropic non-linear constitutive nomial shock-EOS (grey points, Fig. 15). The study shows, that it is
material model for the 0/90 cross-ply Dyneema® HB26 is validated an important issue to consider all deformation mechanisms that
by comparing with ballistic impact experiments. For a number of effect the energy dissipation during absorbing the kinetic energy of
ballistic experiments a two stage light gas gun at EMI was used to the projectile. The results from Fig. 15 are listed in Table 8.
propel a 6 mm diameter aluminum sphere projectile from 2052 m/s It was found that the ballistic limit was satisfactorily reproduced
to 6591 m/s. The Dyneema® HB26 target with an area dimension of between an impact velocity of 2500 m/s and 3000 m/s. The ballistic
200  200 mm and an areal density of 15 kg/m2 was clamped tests no. 1 to 3 showed that the projectile was stopped which was
€ssig et al. / International Journal of Impact Engineering 75 (2015) 110e122
T. La 119

Fig. 14. FE-model for simulating ballistic impact situations of a panel with AD ¼ 15 kg/m2 and a 6 mm diameter aluminum sphere projectile at an impact velocity of 3532 m/s
compared to the ballistic experiment.

correctly predicted by the FE-simulations. At higher impact veloc- state and failure and post-failure damage evolution. An extensive
ities the numerical model showed a slight trend in over-predicting experimental program was carried out to characterize the param-
the residual velocity after perforation. Starting with larger de- eters for the constitutive model. The experimental program in-
viations close to the ballistic limit, but reduces to 14 percent and less cludes tension in all principal directions, shear tests in-plane and in
at impact velocities of 3500 m/s and above. From this validation, the through-thickness direction, Mode I fracture toughness and inverse
model seems suitable for predicting ballistic response. Further test planar plate impact experiments. The material properties (data set
and validation simulations are however necessary to investigate the “TL3”) were implemented into an existing macroscopic composite
models predictive capabilities under a range of different conditions. model in ANSYS AUTODYN which enables explicit FE-simulations of
high velocity impact situations. Single cell tests and plate impact
4. Conclusions simulations showed that the basic static and dynamic deformation
phenomena could be reproduce by this approach.
A non-linear constitutive material model for Dyneema® HB26 is Validation simulations of seven impact situations for an impact
presented which describes the material's orthotropic elastic and velocity range of 2052 m/s e 6591 m/s showed a good applicability
plastic hardening behavior, shock response in a detailed equation of and realistic perforation predictions. The modeling approach seems

Fig. 15. Comparison of experimental and simulated results in terms of velocities of the debris fronts of the ejected material against the projectiles impact velocities.
120 €ssig et al. / International Journal of Impact Engineering 75 (2015) 110e122
T. La

Table 8 For a better applicability in ANSYS AUTODYN where the 11-


Ballistic data of the experiments and numerical simulations. direction indicates the impact or through thickness direction of
Property Test No the composite material a coordinate transformation must be
1 2 3 4 5 6 7
done.
According to the coordinate system used in ANSYS AUTODYN
v0 [m/s] in exp 2052 2438 2453 3100 3532 5370 6591
(11-direction ¼ through thickness direction) the data input
vres [m/s] in exp. 0 0 0 679 958 1730 2457
vres [m/s] in sim. 0 0 0 790 956 1894 2566 follows:
dvres [%], (exp. vs. sim.) 0 0 0 14.1 0.2 0.1 4.3
vres [m/s] in sim. 353 702 708 1423 1803 2897 3445
(w/o hardening model)
Table 10
vres [m/s] in sim. 0 0 0 633 956 1844 2229
Material data set “TL3” according to the coordinate system used in ANSYS AUTODYN.
(w/o shock-EOS)
Orthotropic linear Polynomial EOS Orthotropic failure and
very suitable to support ballistic impact analysis using UHMWPE elastic model (stiffnesses (coefficients) softening (tens. fail. stresses
and Poisson’s ratios) and fracture toughnesses)
composite materials for protection applications. Topics for future
investigations could be further validation in different impact situ- E11 [GPa] 3.62 A1 [GPa] 7.04 s11fail [MPa] 1.07
E22 & E33 [GPa] 26.9 A2 [GPa] 10 s22fail & s33fail [MPa] 753
ations, sensitivities towards the size effects, and the influence of the
n12 [-]* 0.013 A3 [GPa] 0 t12fail [MPa] 1.01E+20
production process. n31 [-]* 0 B0 [-] 3.864 t23fail [MPa] 35.2
n23 [-]* 0.5 B1 [-] 3.864 t31fail [MPa] 1.01E+20
G12 [MPa] 30.7 T1 [GPa] 7.04 GC11 [J/m2] 790
Acknowledgments
G31 [MPa] 30.7 T2 [GPa] 0 GC22 & GC33 [J/m2]* 30
G23 [MPa] 42.3 Tref [K] 293 GC31 [J/m2]* 1.46
The authors would like to acknowledge Dr. Frank Bagusat, for 2
Spec. Heat 1.85E+03 GC12 [J/m ]* 1.46
executing the inverse planar plate impact tests for obtaining the [J/kgK]
2
Thermal 0 GC23 [J/m ]* 1.46
material response at shock loading conditions, and Florian
Conductivity
Schneider for executing the expensive experimental program to Dam. Coupl. 0.5
gain material data at quasi-static conditions. They further wish to Coeff.
thank Dr. Shannon Ryan at DSTO Melbourne for constructive sci- Orthotropic hardening model (coefficients and effective s-ε-values)
entific discussions and his support of the topic.
Plasticity coefficients Effective stress-strain-values

a11 [-] 0.03 seff#1 [kPa] 1.76E+02 εeff#1 [-] 1.82E-04


Appendix a22 [-] 1.00E-05 seff#2 [kPa] 9.89E+02 εeff#2 [-] 1.20E-03
a33 [-] 1.00E-05 seff#3 [kPa] 1.74E+03 εeff#3 [-] 3.11E-03
a12 [-] 1.00E-06 seff#4 [kPa] 2.42E+03 εeff#4 [-] 6.92E-03
a13 [-] 1.00E-06 seff#5 [kPa] 3.10E+03 εeff#5 [-] 1.13E-02
Table 9 a23 [-] 1.00E-06 seff#6 [kPa] 5.97E+03 εeff#6 [-] 2.83E-02
Material data set “TL3” for Dyneema® HB26 (for ANSYS AUTODYN v14.5 and later). a44 [-] 1 seff#7 [kPa] 1.20E+04 εeff#7 [-] 5.78E-02
a55 [-] 1.75 seff#8 [kPa] 2.07E+04 εeff#8 [-] 1.06E-01
Orthotropic Polynomial EOS Orthotropic failure and
a66 [-] 1.75 seff#9 [kPa] 3.46E+04 εeff#9 [-] 1.061E-01
linear elastic (coefficients) softening (tens. fail. stresses
e seff#10 [kPa ] 2.02E+08 εeff#10 [-] 1
model (stiffnesses and fracture toughnesses)
Additional material Data
and Poisson’s ratios)

E11 & E22 26.9 A1 [GPa] 7.04 s11fail & s22fail [MPa] 753 Reference density [g/cm3] 9.80E-01
[GPa]
* taken from literature [5] and calculated using assumptions by Bower [52].
E33 [GPa] 3.62 A2 [GPa] 10 s33fail [MPa] 1.07
n12 [-]* 0 A3 [GPa] 0 t31fail [MPa] 1.01E+20
n13 [-]* 0.1 B0 [-] 3.864 t12fail [MPa] 35.2
n23 [-]* 0.5 B1 [-] 3.864 t23fail [MPa] 1.01E+20
G12 [MPa] 42.3 T1 [GPa] 7.04 GC11 & GC22 [J/m2]* 30 Material data for Al99%-sphere (projectiles material data used in
G31 [MPa] 30.7 T2 [GPa] 0 GC33 [J/m2] 790 validation simulation).
G23 [MPa] 30.7 Tref [K] 293 GC31 [J/m2]* 1.46
Spec. Heat 1.85E+03 GC12 [J/m2]* 1.46
[J/kgK] Table 11
Thermal 0 GC23 [J/m2]* 1.46 Material data set “TL3” according to the coordinate system used in ANSYS AUTODYN.
Conductivity
Dam. Coupl. Coeff. 0.5 Shock EOS (coefficients) Strength model
(Steinberg Guinan, coefficients)
Orthotropic hardening model (coefficients and effective s-ε-values)
Gruneisen coefficient [-] 1.97 Shear modulus [kPa] 2.71E7
Plasticity coefficients Effective stress-strain-values
C1 [m/s] 5.386E3 Yield stress [kPa] 4E4
a11 [-] 1.00E-05 seff#1 [kPa] 1.76E+02 εeff#1 [-] 1.82E-04 S1 [-] 1.339 Max yield stress [kPa] 4.8E4
a22 [-] 1.00E-05 seff#2 [kPa] 9.89E+02 εeff#2 [-] 1.20E-03 S2 [m/s] 0 Hard. Constant [-] 400
a33 [-] 0.03 seff#3 [kPa] 1.74E+03 εeff#3 [-] 3.11E-03 Rel. volume VE/V0 [-] 0 Hard. Exponent [-] 0.27
a12 [-] 1.00E-06 seff#4 [kPa] 2.42E+03 εeff#4 [-] 6.92E-03 Rel. volume VB/V0 [-] 0 Derivate dG/dP [-] 1.767
a13 [-] 1.00E-06 seff#5 [kPa] 3.10E+03 εeff#5 [-] 1.13E-02 C2 [m/s] 0 Derivate dG/dT [kPa/K] -1.67E4
a23 [-] 1.00E-06 seff#6 [kPa] 5.97E+03 εeff#6 [-] 2.83E-02 S2 [-] 0 Derivate dY/dP [-] 0.00268
a44 [-] 1.75 seff#7 [kPa] 1.20E+04 εeff#7 [-] 5.78E-02 Ref. temperature [K] 300 Melting temperature [K] 1220
a55 [-] 1.75 seff#8 [kPa] 2.07E+04 εeff#8 [-] 1.06E-01 Spec. Heat [J/kgK] 884 e e
a66 [-] 1 seff#9 [kPa] 3.46E+04 εeff#9 [-] 1.061E-01 Thermal Conductivity 0 e e
- seff#10 [kPa ] 2.02E+08 εeff#10 [-] 1
Additional material Data
Reference 9.80E-01
density [g/cm3]
The material “Al 1100-O” can be found in material library of
* taken from literature [5]. Autodyn 14.5.
€ssig et al. / International Journal of Impact Engineering 75 (2015) 110e122
T. La 121

References [30] Feraboli P, Kedward KT. Four-point bend interlaminar shear testing of uni-
and multi-directional carbon/epoxy composite systems. Compos Part a
2003;34:1265e71.
[1] Phoenix SL, Porwal PK. A new membrane model for the ballistic impact
[31] Kim WC, Dharan CKH. Analysis of five point bending for determination of the
response and V50 performance of multy-ply fibrous systems. Int J Solids Struct
interlaminar shear strength of unidirectional composite materials. Compos
2003;40:6723e65.
Struct 1995:241e51.
[2] Cheeseman BA, Bogetti TA. Ballistic impact into fabric and compliant com-
[32] Riedel W, Harwick W, White DM, Clegg RA. Advanced material damage
posite laminates. Compos Struct 2003;61:161e73.
models for numerical simulation codes. EMI report no. I 75/03. Freiburg:
[3] Grujicic M, Arakere G, He T, Bell WC, Cheeseman BA, Yen C-F, et al. A ballistic
Ernst-Mach-Institut; 2003. p. 39e42.
material model for cross-plied unidirectional ultra-high molecular-weight
[33] Nossek M. Multiskalenmodellierung von Impaktbelastungen auf Faserver-
polyethylene fiber-reinforced armo-grade composites. Mater Sci Eng A
bundlaminate: Methodenentwicklung, Parameteridentifakation und Anwen-
2008;498:231e41.
dung. PhD thesis. Munich, 198-204: Bundeswehr University; 2010.
[4] Grujicic M, Pandurangan B, Koudela KL, Cheeseman BA. A computational
p. 198e204.
analysis of the ballistic performance of light-weight hybrid composite armors.
[34] ASTM Test Method D 5528. Mode I interlaminar fracture toughness of uni-
Appl Surf Sci 2006;253:730e45.
directional continuous fiber reinforced composite materials. West Con-
[5] Grujicic M, Glomski PS, Arakere G, Bell W, Cheeseman B. Material
shohocken PA USA: ASTM International; 2007.
modeling and ballistic-resistance analysis of armor-grade composites
[35] Riedel W, Nahme H, Thoma K. Equation of state properties of modern com-
reinforced with high performance fibers. J Mater Eng Perform 2009;18:
posite materials: modeling shock, release and spallation. In: Furnish MD,
1169e82.
Gupta YM, Forbes JW, editors. Shock compression of condensed matter-2003.
[6] Jovicic J, Zavaliangos A, Ko F. Modeling of the ballistic behavior of gradient
American Institute of Physics; 2004. p. 701e4.
design composite armors. Compos Part a 2000;31:773e84.
[36] Riedel W, Wicklein M, Thoma K. Shock properties of conventional and high
[7] Vargas-Gonzalez LR, Walsh SM, Scott BR. Balancing and back-face deformation
strength concrete: experimental and mesomechanical analysis. Int J Impact
in helmets: the role of alternative resins, fibers, and fiber architecture in mass-
Eng 2008;35:155e71.
efficient head protection. In: 26th International symposium on ballistics,
[37] Grady DE. Impact compression properties of concrete. In: 6th Symposium on
Miami, FL; 2011.
interaction of nonnuclear munitions with structures-1993. Panama City,
[8] Greenhalgh ES, Bloodworth VM, Iannucci L, Pope D. Fractorgraphic observa-
Florida: Applied Research Associates Inc; 1993. p. 172e6.
tions on Dyneema® composites under ballistic impact. Compos Part a
[38] Rohr I, Nahme H, Thoma K. Material characterization and constitutive
2013;44:51e62.
modelling of ductile high strength steel for a wide range of strain rates. Int J
[9] Chocron S, King N, Bigger R, Walker JD, Heisserer U, van der Werff H. Impact
Impact Eng 2005:401e33.
and waves in Dyneema® HB80 strips and laminates. J Appl Mech May
[39] Barker LM, Hollenbach RE. Laser interferometer for measuring high velocities
2013;80.
of any reflecting surface. J Appl Phys 1972;43:4669e75.
[10] May M, Nossek M, Petrinic N, Hiermaier S, Thoma K. Adaptive multi-scale
[40] Barker LM, Schuler KW. Correction to the velocity-per-fringe relationship for
modeling of high velocity impact on composite panels. Compos Part A
the VISAR interferometer. J Appl Phys 1974:3692e3.
2014;58:56e64.
[41] Hemsing WF. Velocity sensing interferometer (VISAR) modification. Rev Sci
[11] Iremonger M.J. Polyethylene composites for Protection against high velocity
Instrum 1979:73e8.
small arms bullets, In: Proceedings of the 18th international symposium on
[42] Hazell PJ, Appleby-Thomas GJ, Trinquant X, Chapman DJ. In-fiber shock
ballistics-2009, edited by W.G. Reinecke, published by Technomic Publishing
propagation in Dyneema®. J Appl Phys 2011:0435. http://dx.doi.org/10.1063/
Company, 946e953, Pennsylvania.
1.3622294.
[12] Karthikeyan K, Russel BP, Fleck NA, Wadley H, Deshpande VS. The effect of
[43] Nahme H. Dynamic properties of dyneema plastic material. Germany: Frei-
shear strength on the ballistic response of laminated composite plates. Eur J
burg; 2000.
Mechanics/A Solids 2013;42:35e53.
[44] Chapman DJ, Braithwaite CH, Proud WG. The response of Dyneema to shock-
[13] Anderson CE, Cox PE, Johnson GR, Maudlin PJ. A constitutive formulation for
loading. In: Shock compression of condensed matter-2009 edited by M. Elert,
anisotropic materials suitable for wave propagation computer program-II.
W.T. Butler (American Institute of Physics, Melville, 2009) M.D. Furnish, W.W.
Comput Mech 1994;15:201e23.
Anderson, W.G. Proud, and W.T. Butler (American Institute of Physics, Mel-
[14] Anderson CE, O'Donoghue PE, Skerhut D. A mixture theory approach for
ville, 2009).
the shock response of composite materials. J Compos Mater 1990;24:
[45] ANSYS Inc. AUTODYN composite modelling revision 1.3. Canonsburg PA USA:
1159e78.
ANSYS Inc; 2010.
[15] Chen JK, Allahdadi A, Carney T. High-velocity impact of graphite/epoxy
[46] Ryan S, Wicklein M, Mouritz A, Riedel W, Scha €fer F, Thoma K. Theoretical
composite laminates. J Comput Sci Technol 1997;57:1268e379.
prediction of dynamic composite material properties for hypervelocity impact
[16] Chen JK, Allahdadi FA, Sun CT. A quadratic yield function for fiber-reinforced
simulations. Int J Impact Eng 2009;36:899e912.
composites. J Compos Mater 1997;31:788e811.
[47] Dugdale JS, MacDonald D. The thermal expansion of solids. Phys Rev 1953;89:
[17] Meyers MA. Dynamic behavior of materials. New York, NY: John Wiley &
832e4.
Sons; 1994. p. 124e78.
[48] Ben-Dor G, Dubinsky A, Elperin T. On the Lambert-Jonas approximation for
[18] Hou J, Petrinic N, Ruiz C. Prediction of impact damage in composite plates.
ballistic impact. Mech Res Commun 2002:137e9.
J Compos Sci Technol 2000;60:273e81.
[49] Levenberg K. A method for the solution of certain problems in least squares.
[19] Iremonger MJ, Went AC. Ballistic impact of fibre composite armours by
Quart Appl Math 1944;2:164e8.
fragment-simulating projectiles. Compos Part a 1996;27 A:575e81.
[50] Marquardt D. An algorithm for least-squares estimation of nonlinear param-
[20] Riedel W, Nahme H, White D, Clegg R. Hypervelocity impact damage in
eters. SIAM J Appl Math 1963;11:431e41.
composites part II e experimental investigations and simulations. Int J Impact
[51] Recht RF, Ipson TW. Ballistic perforation dynamics. ASME J Appl Mech 1963:
Eng 2006;33:670e80.
384e90.
[21] Clegg RA, White DM, Riedel W, Harwick W. Hypervolicity impact damage
[52] Bower AF. Solid mechanics; 2013 [Online]. Available: http://solidmechanics.
prediction in composites: part I e material model and characterization. Int J
org/Text/Chapter3_2/Chapter3_2.php.
Impact Eng 2006;33:190e200.
[22] Wicklein M, Ryan S, White DM, Clegg RA. Hypervelocity impact on CFRP:
testing, material modelling and numerical simulation. Int J Impact Eng Notations
2008;35:1861e9.
[23] Anderson TL. Fracture mechanics e fundamentals and applications. 3rd ed.
CRC Press Taylor & Francis Group; 2005. p. 344e57.
[24] Hiermaier SJ. Structures under crash and impact. Freiburg: Springer; 2008. The following notation is used throughout the paper unless stated otherwise locally in
p. 176e7. the text.
[25] Russell BP, Karthikeyan K, Deshpande VS, Fleck NA. The high strain rate
response of ultra high molecular-weight polyethylene: from fiber to laminate.
aij: plasticity coefficients defining shape of yield surface
Int J Impact Eng 2013;60:1e9.
Ai: material constants of the polynomial equation of state
[26] DIN EN ISO 527e4. Bestimmung der Zugeigenschaften; Teil 4: Prüfbe-
avg.: arithmetic average
dingungen für isotrop und anisotrop €rkte
faserversta Kunst-
Bi: material constants of the Mie-Grüneisen form of the EOS
stoffverbundwerkstoffe. Berlin: Deutsches Institut für Normung e.V; 1997.
C: material stiffness matrix
[27] Levi-Sasson A, Meshi I, Mustacchi S, Amarillo I, Benes D, Favorsky V, et al.
Cij: stiffness coefficients
Experimental determination of linear and nonlinear mechanical properties of
COV: coefficient of variation
laminated soft composite material system. Compos Part B: Eng 2014;57:
Dij: damage parameter
96e104.
DCB: double cantilever beam test
[28] GOM mbH, [Online]. Available: http://www.gom.com/de/messsysteme/
εP : effective plastic strain (isotropic)
systemuebersicht/aramis.html [accessed on 04.09.13].
Eij: Young's moduli
[29] DIN EN 2561. Kohlenstoffversta €rkte Kunststoffe, unidirektionale Laminate,
e2iif : failure functions in principle directions
Zugprüfung parrallel zur Faserrichtung. Berlin: Deutsches Institut für Nor-
εij: strains
mung e.V; 1995. .
ε : principal strain tensor
122 €ssig et al. / International Journal of Impact Engineering 75 (2015) 110e122
T. La

EOS: equation of state tdyn: thickness of Dyneema® HB26 specimen (PPI)


f(sij): function describing the anisotropic yield surface tw: thickness of the steel witness plate (PPI)
GIC/GCij: fracture energy Mode I (energy release rate) tij: shear stresses
Gij: shear moduli tijmax: maximum shear stresses
gij: shear angles Tref: reference temperature
r: material density UHMWPE: ultra-high molecular weight polyethylene
r0: initial material density uP: particle velocity
sij: stresses US: shock velocity
sh: Hugoniot stress nij: Poisson's ratios
sijfail: failure stresses vfs: free surface velocity of the witness plate (PPI)
s: effective plastic stress (isotropic) vimp: impact velocity
.
s : stress tensor vres: residual velocity
tb: thickness of the steel backing plate (PPI)

S-ar putea să vă placă și