Você está na página 1de 15

Composite Structures 67 (2005) 263277

www.elsevier.com/locate/compstruct

Performance of long ber reinforced thermoplastics subjected


to transverse intermediate velocity blunt object impact
S.D. Bartus, U.K. Vaidya *

Department of Materials Science and Engineering, The University of Alabama at Birmingham, Birmingham, AL 35294-4461, USA

Available online 27 August 2004

Abstract

An understanding of long ber reinforced thermoplastic (LFT) behavior while undergoing high energy, high strain rate impact is
essential in promoting its use in automotive and other markets. In this work, damage tolerance of polypropylene (PP)/E-glass LFTs
subjected to transverse blunt object impact (BOI) was investigated, treated from an experimental standpoint, in order to characterize
energy dissipation and damage modes. LFT subjected to BOI exhibited high impact energy dissipation, which increased linearly with
increasing areal density. The average impact energy dissipation at the critical velocity (ballistic limit) was 167 J and 121 J for a
4.61 g cm2 specimen impacted by at and conically shaped projectiles, respectively. The ber orientation also played a large role
in energy dissipation; failure appeared to occur along planes of preferential ber orientation. The critical energy for specimen per-
foration did not vary with the mass of the impactor, i.e. strain rate, in the range investigated. This overall work advances the state-
of-the-art in LFTs with an automotive focus.
 2004 Elsevier Ltd. All rights reserved.

Keywords: Automotive composites; Impact; Thermoplastic composites

1. Introduction United States market for such materials is in excess of


4.54 108 kg per annum, half of which is consumed by
1.1. Thermoplastic composites the automotive industry [1]. Long Fiber reinforced
Thermoplastic (LFT) composites have one of the high-
The use of thermoplastic composites has gained stea- est growth rates in the polymer material areas, sustain-
dy favor over traditional materials such as steel in struc- ing a projected 30% growth from 2000 to 2004 [2].
tural and semi-structural applications due to the Thermoplastic composites typically comprise a cost
prominent physical and mechanical behavior of these eective commodity matrix such as PP, polyethylene
materials such as specic strength, damping, corrosion (PE), nylon, etc.; reinforced with glass, carbon, or ara-
resistance, and impact properties. Moreover, closed mid bers. E-glass is the most common reinforcement
molded discontinuous LFT composites share the attrac- since the automotive market niche is driven more by
tive features of possessing greater strength, stiness and cost/performance ratio than weight/performance ratio
impact properties in contrast to short ber reinforced as demanded by the aerospace industry.
thermoplastics, in addition to high volume processabil- Thermoplastic composites used in these applications
ity, ability to ll complex geometries, intrinsic recyclabil- can be short ber reinforced thermoplastic (SFRT),
ity, and the capacity for part integration. The current glass mat thermoplastic (GMT), or LFT. Injection
molded SFRT composites with starting ber lengths less
* than 4 mm (before ber length degradation during
Corresponding author. Tel.: +1 205 934 9199; fax: +1 205 934
8485. processing) are currently the most prevalent of the afore-
E-mail address: uvaidya@uab.edu (U.K. Vaidya). mentioned composites. However, the full advantage of

0263-8223/$ - see front matter  2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compstruct.2004.07.023
264 S.D. Bartus, U.K. Vaidya / Composite Structures 67 (2005) 263277

the reinforcing ber is not realized due to the low ber testing [6]. Very little work has looked into the eect
aspect ratio. Injection molded LFTs also suer from of intermediate velocity blunt object impact (BOI) on
excessive ber length degradation in the plastication LFTs. Intermediate velocities are greater than low veloc-
and injection stages. In addition, diculties arise in ity drop tower impacts or pendulum type impacts
processing components with high ber content and (10 m s1), yet slower than high velocity ballistic type im-
starting ber lengths in excess of 13 mm due to the high pacts. The velocity range for this purpose simulates the
melt viscosity. eect of blunt objects such as rocks and debris traveling
The large ber aspect ratio of LFTs fully takes at highway speeds for automotive applications as well as
advantage of the ber strength in contrast to its short impact induced by debris from hurricanes and tornadoes
ber counterpart. Injection, injection-compression and for storm shelter and military housing applications. This
extrusion-compression molding techniques are em- work can also be extended to transverse loaded energy
ployed for LFT processing. Starting LFT ber lengths dissipation under high stain-rate for automobile crash
are typically greater than 13 mm and depend on the de- mitigation purposes.
sired properties, ber concentration, and processing
technique. LFTs oer several advantages in contrast to 1.2. Discontinuous long ber reinforcement
GMTs such as: the possibility to work without semi-n-
ished mats (e.g., in-line extrusion) making it less labor Traditional processing of LFT starts by rst hot melt
intensive, and lower compression forces due to a de- impregnating a tow of reinforcing bers with a thermo-
crease in melt viscosity, which results in capital cost sav- plastic matrix and subsequently chopping the continu-
ings in tooling and machinery. Moreover, LFTs oer ous tow into pellets of a set length. Hot melt
higher surface quality, less part rejection due to an in- impregnation is done by wirecoating, cross-head extru-
crease in the ability to ll complex features and inte- sion, or thermoplastic pultrusion techniques [7]. The
grated recyclability. Another advantage is greater LFT pellets are then fed into a single-screw plasticator
freedom in choosing ber and matrix materials. where they are fed down the barrel by the screw, heated
In the compression molding process, bers develop in above the melting point of the matrix, and extruded as a
plane orientations during ow, which can plague the charge (shot). The shot is then placed on a tool and
consolidated component. Preferential orientation during compression molded.
compression molding can reduce strength and stiness Extrusion/compression molding of LFT has rapidly
in critical areas and will induce warping through aniso- been gaining favor over traditional injection molding
tropic contraction upon cooling [3]. Although large ber (especially with in-line compounding) and GMT com-
aspect ratios and high ber loading decrease ber mobil- pression molding techniques due to superior mechanical
ity in the melt thereby decreasing the degree of preferen- properties at a comparable cost [8]. The ber aspect
tial orientation, a high degree of preferential orientation ratio, dened as the length to diameter ratio, dierenti-
may still occur depending on mold geometry, charge ates short ber from long ber reinforcement. The
location, and preorientation of the bers in the charge. aspect ratio of a long ber is typically an order of mag-
This is described in detail in previous work using nite nitude greater than that of a short ber [9]. While short
element methods to simulate the compression molding ber reinforced thermoplastics realize substantial gains
process with CADpress-Thermoplastic [4]. in mechanical properties over that of the neat material,
As the use of LFTs grows in automotive and other the full potential of the reinforcement is not obtained
industries, the need to determine impact response of because the ber is below a critical length. The critical
these materials increases to ensure the safety and stabil- ber length is given in Eq. (1):
ity of designed structures [5]. LFT was recently em-
rmax r
ployed for the underside belly pan of Daimler Lc 1
s
Chryslers PT Cruiser [2]. Sheet molding compound
(SMC), was rst used in this application. However, it where Lc is the critical ber length, r is the ber radius,
proved too brittle since the component needed to be rmax is the tensile stress acting on the ber, and s is the
exible and withstand impact from stones and other ob- interfacial shear strength [10].
jects [2]. Few authors have attempted to characterize the This equation is based on the several simplifying
impact performance of discontinuous, randomly ori- assumptions, rst of which is that the strain to failure
ented LFT thermoplastics. for the ber is less than that of the matrix. This is a rea-
The ber architectures inherent in LFTs make an sonable assumption in the case of thermoplastic matri-
accurate characterization of the failure mechanisms ces. A shortcoming of this equation is that it assumes
complex. Most eorts in understanding the impact per- the interfacial shear stress in constant over the ber
formance and failure mechanisms of LFTs have prima- length. It has been shown that bers produce higher
rily focused on Charpy and Izod impact testing, and to stresses at the ber tips resulting in a lower elongation
an even lesser degree, low velocity drop tower impact to failure [11].
S.D. Bartus, U.K. Vaidya / Composite Structures 67 (2005) 263277 265

Assuming that below the critical ber length, the Intermediate velocity is considered to fall between the
force required for debonding increases linearly with ber low and high velocity regimes. The main dierence being
length, the interfacial shear strength can be determined that, depending on the projectile mass, large deforma-
from the slope of the load required for pullout vs. ber tion may occur in the intermediate velocity range; par-
length. Above the critical ber length, sucient interfa- ticularly in the case of massive projectiles but may
cial shear stress exists for ber breakage to occur. Sub- dier from low velocity impacts with regard to loading
sequent micromechanical models for tensile strength, rate and momentum eects.
tensile modulus, and transverse notched impact strength
based on bers above and below the critical length can 1.4. Impactor mass and geometry
be found in Refs. [4,6,7].
The impactor size, shape, mass, material and angle of
1.3. Categorization of impact incidence all have a strong inuence on the response of
the specimen [12]. A blunt object is dened as a large
A considerable amount of work has gone into study- diameter, high contact area projectile that emulates the
ing transverse impact on composite structures [1219]. eects of debris to characterize energy absorption upon
The authors of the studies in [1219] have focused pri- impact. Jenq et al. [18] found that the momentum trans-
marily on continuous ber composites with thermoset fer to graphite/epoxy targets by at impactors is about
matrices. Research in thermoplastic matrices has mainly four times greater than that of sharp or conically tipped
been limited to poly ether ether ketone (PEEK), prima- impactors when the penetrator was red at high veloc-
rily relevant for aerospace applications. ity. Cantwell and Morton [19] studied the eect of pro-
Most of the literature on impact response of thermo- jectile mass on energy absorption in composites while
plastics pertains to either Charpy and Izod tests or low maintaining the impactor size and shape. It was found
or high velocity impact [46]. Charpy and Izod, pendu- that varying the projectile mass had a signicant eect
lum test congurations, are constrained in terms of spec- on the resulting damage. Cantwell and Morton [19] sug-
imen size, impact direction, and boundary conditions gest that lighter projectiles are more damaging to the
[5]. Moreover, Charpy and Izod testing, especially in overall load-bearing capacity of the composite because
the notched conguration, force the specimen to fail at the incident energy is dissipated over a very small area
a predetermined area rather than along the weakest immediately adjacent to the point of impact. This may
plane. This could skew results because of the highly or may not be the case with intermediate velocity impact
anisotropic nature of LFTs. The velocity range, impac- depending on the target material, boundary conditions,
tor geometry and, impactor mass should reect the type projectile shape and projectile mass.
of threats a component may encounter in service. In this
case, a more representative test would be in the interme- 1.5. Impact energy
diate velocity range with a blunt object impactor. This
would simulate the eect of stones and other debris a The energy absorbed in the system, Eabsorbed, can be
vehicle might encounter at highway speeds. In addition, described as having two components: the energy ab-
the importance of this test methodology lies not only in sorbed in creating damage in the specimen, Edamage,
the characterization of debris hits, but also energy dissi- and the energy absorbed by the system through vibra-
pation and failure mechanisms under high loading rate. tion, heat, elastic response of the specimen, elastic
The denition of the terms low, intermediate and behavior of the projectile/supports, ESL, Eq. (2):
high velocity often varies in literature [12]. Some authors
Eabsorbed Edamage ESL 2
contend that high velocity refers to conditions resulting
in complete perforation of the target. However, this phe- where the Edamage term includes the energies associated
nomenon can be readily observed in drop tower testing with the specimen indentation, matrix damage, ber
in which the maximum obtainable velocity is typically breakage, ber debonding, and ber pullout, and fric-
less than 10 m s1. For the remainder of this paper, the tion between the projectile and specimen during penetra-
term low velocity will be reserved for velocities less than tion/perforation. The energy absorbed in the system can
10 m s1 with large mass impactors and large target be equated to the energy put into the system, e.g., the ki-
deformations. The denition of high velocity impact will netic energy, KE, of the impactor shown in Eq. (3):
follow that dened by Abrate [12] as the ratio between
Eabsorbed KE 1=2mV 20  1=2mV 2f ESL 3
the impactor velocity and the transverse compressive
wave velocity is greater than the maximum strain to fail- where m is the projectile mass, and V0 and Vf are the ini-
ure in that direction. Damage induced by high velocity tial and residual projectile velocities. The energy spent in
impact is typically introduced by the rst few compres- the elastic response of the projectile and supports has
sion waves though the thickness when the global plate been shown to be small and will not be considered, sim-
motion has not been established [12]. plifying Eq. (3) to yield Eq. (4) [13].
266 S.D. Bartus, U.K. Vaidya / Composite Structures 67 (2005) 263277

Eabsorbed KE 1=2mV 20  1=2mV 2f 4 number W7650. The starting ber length was 25.4 mm.
The plasticator and 3.56 MN (400(US)-ton) compres-
The highest impact energy is absorbed at the ballistic
sion molder used were manufactured by C.A. Lawton
limit of the material in which the projectile imbeds itself
Co. The idle time was 5 s with a 30 s cooling time. This
in the specimen. However, embedding a projectile in
gave a cycle time of approximately 60 s. The system pres-
thin specimens is dicult due to an inadequate contact
sure in the plasticator was 1.25 MPa, the backpressure
area between the target and projectile for ctional forces
2.07 MPa, the screw speed 910 rpm, and the melt tem-
to act upon. Since the residual velocity of a non-pene-
perature at the knife was 232 C.
trating projectile is small, it is not included in the energy
The mold temperatures, checked every 40 cycles, re-
balance. Therefore, the Vf term is not part of the energy
mained constant at 54 C and 76 C for the top and bot-
balance unless the projectile perforates the specimen and
tom mold halves, respectively. The plaque temperature
has a detectible exit velocity. Otherwise, the initial veloc-
at the time of ejection varied from 87.5 C to 90.0 C
ity is assumed equal to the critical velocity, V0 = Vc. This
and from 45 C to 51 C 5 min after ejection for an ambi-
is similar to the V50 ballistic limit, which is dened as the
ent temperature of 3132 C. Twenty-four trial plaques
projectile having a 50% probability of completely pene-
formed before making a 160 plaque run to ensure the
trating the specimen at the critical velocity with a devi-
system was at steady state. The plaque ID is as follows:
ation in velocity less than 40 m s1 [12]. Given the low
(year, month, date, lot number, plaque number) e.g.,
critical velocity of blunt object impact, a more conserv-
(020919-1-#). The rst lot of plaques was manufactured
ative denition is appropriate. Since a standard test
on September 19, 2002. The mechanical properties are
method does not exist for blunt object impact, the criti-
given in Table 1 [20].
cal velocity is considered on the basis of a standard devi-
ation in velocity no greater than 10 m s1 in which 50%
2.2. Sample preparation
of the projectiles do not perforate the specimen.
Three equal size samples (470.4 0.7 cm2) were cut
from each of the LFT plaques produced for blunt object
2. Experimental impact testing and the tabs were removed. The samples
are shown schematically (with the tabs still in place) in
2.1. Celstran PP-GF40-03 processing and material Fig. 1. The average weight and Standard Deviation
properties (SD) for the three samples was 206.32 g (SD = 1.358 g),
219.80 g (SD = 1.822 g), and 225.27 g (SD = 1.056 g) for
Test plaques, with the geometry shown in Fig. 1 were the bottom, center and top samples, respectively. This
fabricated at Southern Research Institute, Birmingham, yielded an average areal density of 4.39 g cm2,
Alabama, using Celstran PP-GF40-03, 40 wt.% long 4.67 g cm2, and 4.79 g cm2 for the bottom, center,
glass ber reinforced, chemically coupled, heat stabilized and top samples. The areal density is of interest because
polypropylene manufactured by Ticona. The color num- impact energy absorption is often described in terms of
ber is AF 3001 (natural), stock number GL0175, and lot it. The variations in areal density are due to a thickness

Fig. 1. Schematic of the top and side views of the tab plaque (not shown to scale) and the representative sections that were cut from it and (b) Pro/E
isoperimetric drawing of the tab plaque.
S.D. Bartus, U.K. Vaidya / Composite Structures 67 (2005) 263277 267

Table 1 Table 2. In total, 147 specimens where tested. The goal


Material properties of Celstran PP-GF40-03 adopted [19] in the impact-testing phase was to establish whether a
Property Method Value Units relationship existed between the critical energy dissipa-
Density ISO 1183 1210 kg m3 tion and impact mass (e.g., strain-rate eects), the pro-
Tensile modulus (1 mm/min) ISO 527-2/1A 7900 MPa jectile geometry, ber orientation, and areal density.
Tensile stress at break ISO 527-2/1A 100 MPa The residual velocity due to rebound was measured
Tensile strain at break ISO 527-2/1A 2.0 %
Flexural modulus ISO 178 8000 MPa
using a Vision Research, Inc. digital high-speed camera
Flexural strength ISO 178 175 MPa (Model: Phantom V5.0), at 14,000 frames s1 with 40 ls
Charpy notched impact ISO 179/1eA 20 kJ m2 exposures, triggered manually. The maximum rebound
strength at 23 C velocity of a 100 g non-perforating sabot was calculated
to be approximately 4.4 m s1. Since the residual velocity
from a rebounding projectile was small, it was not in-
Table 2 cluded in the energy balance.
Blunt object impact test matrix
Projectile mass (g) Projectile geometry Areal density 2.3. Impact test apparatus
of specimen (g cm2)
25 Flat 4.39a A gas gun was used to propel a variety of projectiles,
25 Flat 4.67b Table 3, in a velocity range of approximately 42.8 m s1
25 Flat 4.79c to 140.4 m s1 with projectile masses varying between
50 Flat 4.39a
50 Flat 4.67b
23.7160.7 g with both conical and at congurations
50 Flat 4.79c to simulate the eects of blunt object impacts. The wide
100 Flat 4.39a range of projectile mass helped in determining velocity
100 Flat 4.67b eects, while the blunt projectile shape aided in charac-
100 Flat 4.79c terization of perforation mechanisms. The gas gun, Fig.
160 Flat 4.39a
160 Flat 4.67b
2, consists of a regulated high-pressure compressed uid
160 Flat 4.79c source (N or He), pressure transducer, pressure vessel,
25 Conical 4.39a ring valve, barrel, velocity sensors (Model: CED Mil-
25 Conical 4.67b lennium), capture chamber, specimen xture and projec-
25 Conical 4.79c tile arrest.
50 Conical 4.39a
50 Conical 4.67b
This range in projectile velocity and geometry will
50 Conical 4.79c simulate the eect of blunt objects such as rocks and
Average thicknesses:
debris traveling at highway speeds and greater. This
a
3.5 mm. work also provides insight into high strain-rate energy
b
3.8 mm. dissipation and failure mechanisms, such as a structural
c
4 mm. composite may undergo in the event on an automobile
accident.
A minimum of ve samples were tested for each
disparity caused by the lack of automatic leveling on the eighteen congurations, in an attempt to determine the
compression molder. The average thickness variation in critical impact velocity (or ballistic limit). Three test
the width and length directions was 20.3 0.8% and specimens of LFT PP/40 wt.% E-glass, shown in Fig.
5.5 0.9%, respectively, for the illustration shown in 1, were considered, representing ber orientation eects
Fig. 1. As determined later, the fracture path was found in addition to unintentional variations in areal density.
to be independent of the thickness. The results are discussed in terms of projectile mass,
A minimum of ve samples were examined for each projectile geometry, areal density, and ber orientation,
of the 18 congurations in the test matrix outlined in respectively.

Table 3
Blunt object projectile details
Projectile Weight (g) Shape Material Dimensions (mm)
Sabot 160 Flat Aluminum 37.71(/) 50
Sabot 100 Flat Aluminum 37.71(/) 50a
Sabot 50 Flat UHMWPE 37.71(/) 48
Sabot 50 Conical UHMWPE 37.71(/) 55, 60 shoulder
Sabot 25 Flat UHMWPE 37.71(/) 37a
Sabot 25 Conical UHMWPE 37.71(/) 41, 60 shouldera
a
Denotes sabots in which material was removed from the center to reduce mass.
268 S.D. Bartus, U.K. Vaidya / Composite Structures 67 (2005) 263277

Table 4
Statistical signicance of the critical velocity data for the top specimen
(4.79 g cm2)
Projectile Mean Standard 95% condence
velocity (m s1) deviation (m s1) interval (m s1)
25 g Flat 124.4 2.5 2.4
25 g Conical 103.5 2.4 2.1
50 g Flat 84.2 4.5 4.0
50 g Conical 76.4 3.9 3.4
100 g Flat 56.0 2.3 2.0
160 g Flat 46.9 2.1 1.6

Table 5
Statistical signicance of the critical velocity data for the center
specimen (4.67 g cm2)
Projectile Mean Standard 95% Condence
velocity (m s1) deviation (m s1) interval (m s1)
25 g Flat 127.2 9.8 8.6
25 g Conical 97.8 6.6 7.5
50 g Flat 83.3 6.3 4.7
50 g Conical 72.9 4.9 4.3
100 g Flat 53.8 1.1 0.9
160 g Flat 47.3 0.6 0.5

Table 6
Statistical signicance of the critical velocity data for the bottom
specimen (4.39 g cm2)
Projectile Mean Standard 95% Condence
velocity (m s1) deviation (m s1) interval (m s1)
25 g Flat 111.1 6.2 5.4
25 g Conical 97.9 4.9 4.8
50 g Flat 76.8 5.3 3.6
50 g Conical 68.2 4.1 3.6
100 g Flat 54.0 1.5 1.3
160 g Flat 44.7 1.1 1.0

the standard deviation increased with increasing velocity


however, the ratio between the standard deviation and
mean critical velocity did not exceed 5.3%, 7.7%, and
6.9% for the top, center, and bottom specimens,
respectively.
Factors that may have contributed to data deviation
Fig. 2. Gas gun test apparatus, (b) capture chamber, (c) entire
are impactor material properties and inconsistencies in
assembly.
the impact response of test specimen. Abrate noted that
the elastic material properties of the impactor could af-
3. Results and discussion fect the perforation energy; e.g., assuming no plastic
deformation of the impactor, higher modulus materials
3.1. Statistical analysis of the impact data result in higher contact stresses due to a decrease in elas-
tic deformation during the onset of damage [12]. Since
Statistical analysis was conducted on the velocity two dierent impactor materials were used in the mass
data shown in Tables 46. The largest standard devia- aect study, ultra high molecular weight polyethylene
tion in the velocity data at the critical velocity (ballistic (UHMWPE) and aluminum, a decrease in energy dissi-
limit) was 9.8 m s1, which corresponds to a standard pation could be expected in the case of the higher mod-
deviation in the energy data of 30.5 J for the center spec- ulus aluminum projectile. However, the standard
imen impacted with the 25 g at projectile. In general, deviation was large to make any conclusions as to
S.D. Bartus, U.K. Vaidya / Composite Structures 67 (2005) 263277 269

whether the impactor material type played a signicant point of contact on the target, Fig. 3, which resulted
role in energy dissipation. in a shear plug or punch-through failure. This is thought
Variations in data were most likely caused by irregu- to be a result of an increase in contact stresses at the
larities in the impact response of test specimens. point of impact from a decrease in contact area between
Although great care was taken in consistently impacting the projectile and target. All samples impacted with the
the same area of all the specimens, sample shifting is 100 g projectile were impacted under the exact test con-
impossible to control in the case of a simply supported guration and all data show a marked decrease in en-
sample. In addition, with the current sample holder, ergy dissipation. The gas gun was realigned before
there was no simple way to insure even clamping pres- further testing was carried out and the impression left
sure repeatedly, especially with variations in sample by subsequent impactors did not indicate projectile
thickness and process induced warping. Both factors tilting.
may have led to variations in the impact response of From the high-speed images, Figs. 3 and 4, the im-
the target. After impacting a specimen, the line of sup- pact phenomenon is noted to exhibit both low and high
port at the four support areas leaves an imprint. Varia- velocity characteristics. The rst impact event captured
tions in the imprint were noted, and care was taken to was below the critical velocity, resulting in no visible
avoid this however this may have inuenced energy damage. In the other two image series, where the impact
dissipation. occurred at close to the critical velocity for the 100 g at
projectile, incipient damage was observed before global
3.2. Observations from high-speed photography plate motion was established. This is a characteristic of
high velocity impact. Then, as momentum eects took
The primary explanation behind the deviation stems over, a large global displacement was observed (approx-
from observations of three series of high-speed images imately 25 mm), an attribute of low velocity impact, in
capturing the impact event of a 100 g, at tipped projec- which the specimen deected up to the back of the sam-
tile. Upon closer examination of the impact event, a ple support, because it was only simply supported on
slight tilting of the projectile was noticed. This is illus- two sides in order to observe the event.
trated in Figs. 3 and 4, which capture the projectile tilt- A marked increase in contact stresses is also thought
ing just prior to impact. Damage initiated at the rst to be responsible for a decrease in energy dissipation for

Fig. 3. High-speed digital images taken at 14,000 frames s1 of sample 020919-32C, illustrating the onset of damage, KE 150.2 J. The 100 g at tipped
projectile exemplied tilting.

Fig. 4. High-speed digital images taken at 14,000 frames s1 of sample 020919-15C, showing a 100 g at projectile after the initial impact and while
rebounding, KE 142.3 J.
270 S.D. Bartus, U.K. Vaidya / Composite Structures 67 (2005) 263277

the conical projectiles, discussed in Section 3.4. In cases gree, indicating that the specimens were not sensitive
where the projectile impacted normal to the specimen, to strain rate in the range investigated. Variation in en-
damage is thought to initiate at weak planes parallel ergy dissipation for the three dierent specimens tested
to areas of high ber orientation or in areas where ber are attributed to dierences in areal density and ber
clumping occurs, determined from previous work orientation. The eect of ber orientation is discussed
involving qualitative assessment of ber orientation via in greater detail in Section 3.7.
matrix burn-o and process modeling [4].
3.4. Eect of projectile geometry
3.3. Eect of projectile mass
In the preliminary testing using the at impactors, it
In the rst phase of testing, the eect of projectile was surmised that inadvertent projectile tilting aected
mass was examined by determining the critical impact the energy dissipation of the sample. It is thought that
velocity for three test specimens, top (areal den- this marked decrease in energy stems from an increase
sity = 4.789 g cm2), center (areal density = 4.674 g cm2) in contact stresses at the point of impact. As one would
and bottom (areal density = 4.386 g cm2). The goal in expect, a sharp or conically tipped projectile would also
this phase of testing was to establish if impact velocity result in a substantial decrease in perforation energy in
had an eect on energy dissipation and failure modes. contrast to a normally impacting at projectile. Projec-
Impact mass variation is an established technique in var- tile mass seemed to play even less of a role in the study
ying strain rate and often used in instrumented low of conically shaped projectiles, whose results exhibited a
velocity drop weight testing [21]. Impact velocity alone greater statistical signicance than the at shaped pro-
does not describe the impact properties of a material jectile study. If contact stress plays a dominant role in
but it can aect the energy dissipated in an impact. sample perforation, it would follow that less deviation
Tables 46 illustrate the variations in impact velocity would occur in conically shaped projectiles in which
with projectile mass as a function of the areal density projectile tilting would not adversely aect the contact
of each specimen. In all cases examined, the critical area.
velocity decreased exponentially as a function of projec- Both projectile geometries inicted planar cracking of
tile mass. the specimens upon impact. Typical damage modes for
The impact energy vs. projectile mass is shown in Fig. samples impacted with both geometries are shown in
5, where a linear regression was t to the mean of all the Fig. 6. Damage from at projectiles initiated along the
impact energy vs. projectile mass data. For the range periphery of the impact area and propagated radially
investigated, the impact energy and projectile mass did along two or three planes, away from incipient damage.
not inuence the specimen response to any notable de- Specimens impacted with conical projectiles however,

250

225

200 areal
density

175

150
Energy (J)

125 areal
density
100 bottom, flat impactor (average)
bottom, conical impactor (average)
75 center, flat impactor (average)
center, conical impactor (average)
50 top, flat impactor (average)
top, conical impactor (average)
25
0 20 40 60 80 100 120 140 160 180
Projectile Mass (g)

Fig. 5. Energy (J) vs. projectile mass (g) for the bottom, center and top specimens with a linear regression analysis, tted thought the mean of each
data set, indicative of the independent relationship between projectile mass (impact velocity) and the energy dissipation upon impact. The trend in
decreasing energy dissipation in the at and conical data sets is most likely due to a decrease in areal density and ber orientation eects.
S.D. Bartus, U.K. Vaidya / Composite Structures 67 (2005) 263277 271

Fig. 6. Typical fracture patterns for test plaques impacted with (a) at and (b) conical impactors. The dashed circles indicate the impactor location.
The large circles visible in all images are from the support used during image acquisition.

250
Energy (Flat) = 63.2*(Areal Density) - 124.6
2
225 R = 0.93
Energy (Conical) = 45.2*(Areal Density) - 86.5
2
200 R = 0.97

175 174.9 J 175.0 J


Flat impactors
Energy (J)

150 151.3 J

131.0 J
125 Conical impactors
112.0 J 122.8 J
100

75 bottom, flat impactor center, flat impactor


top, flat impactor bottom, conical impactor
50
center, concial impactor top, conical impactor
25
4.3 4.4 4.5 4.6 4.7 4.8 4.9
-2
Areal Density (g cm )

Fig. 7. Energy vs. areal density for at and conically tipped impactors showing a linearly increasing trend via linear regression t through the mean
of all the impact data. The results of 106 impacted samples are shown, which represent increasing areal density for the bottom, center and top
specimens, respectively, which illustrate the inuence of impactor geometry. The linear regression for the at and conically tipped projectiles is given
in the top right of the graph with the respective R2 values.
272 S.D. Bartus, U.K. Vaidya / Composite Structures 67 (2005) 263277

exhibited a higher degree of crack branching. Damage plane. The crack must then bifurcate in order to dissi-
from conical projectiles initiated at the point of initial pate excess energy.
contact, subjecting the specimen to high local contact Neglecting projectile mass eects and plotting projec-
stresses. It is possible that crack branching results from tile geometry as a function of energy dissipation and
the increased ber strain energy at the point of contact areal density, an increasingly linear trend develops.
for samples subjected to impact with conical projectiles. Fig. 7 illustrates this proclivity in which a linear regres-
When bers fracture, the release of strain energy is sion was t through the mean of the data for the at and
greater than the energy required to create a single crack conical projectile geometries resulting in R-squared val-

Fig. 8. SEM image approximately 650 to the fracture plane of sample 34B showing ber pullout and ber breakage. The rough ber surface indicates
strong interfacial bermatrix bonding. A mixture of crazing-tearing and brittle fracture is evident in the matrix.

Fig. 9. SEM image normal to the fracture plane of sample 39B showing variations in ber orientation through the thickness of the section, taken
normal to the fracture plane.
S.D. Bartus, U.K. Vaidya / Composite Structures 67 (2005) 263277 273

ues of 0.93 and 0.97, respectively. The mean numeric 3.5. Areal density eects
data is reported next to the respective data set, Fig. 7.
On average, a decrease in critical energy dissipation Fig. 7 summarizes the impact energy as a function of
for the bottom, center and top specimens was approxi- areal density, or weight area1 for the LFT specimens.
mately 26% (areal density = 4.39 g cm2), 30% (areal Variations in the energy dissipation of each of the sam-
density = 4.67 g cm2) and 25% (areal density = ples were rst attributed to the variations in areal den-
4.79 g cm2), respectively, when subjected to impact by sity, especially considering the expected linear increase
conically tipped projectiles. in energy dissipation. However, in the post mortem

Fig. 10. SEM image normal to the fracture plane of sample 34B showing ber pullout. Matrix failure appears to have a combination of crazing-
tearing (stress whitening) and brittle fracture (smooth appearance).

Fig. 11. SEM image normal to the fracture plane of sample 34B, taken normal to the fracture surface, showing the path of fracture following the
main ber orientation angle. The dominant failure modes are ber pull out (and shear out), ber breakage and matrix fracture.
274 S.D. Bartus, U.K. Vaidya / Composite Structures 67 (2005) 263277

examination of the impacted specimens, similarities be- were ber debonding/pullout, ber fracture, ber
tween the fracture patterns of the three dierent samples matrix pull away and matrix fracture, Figs. 813. It is
were qualitatively established, suggesting that a com- dicult to discern the dierence between ber fracture
mon failure mechanism might be at play. and ber pullout in discontinuous ber composites.
Upon close examination, part of the ber/matrix interfa-
3.6. Micrograph analysis cial surface can be seen where the ber fractured and
pulled out, Fig. 8. This type of failure also appears in
SEM analysis of the impacted LFT sample fracture Figs. 9, 11 and 13. Fiber pullout is evident in every
surface revealed that the predominant modes of failure micrograph.

Fig. 12. SEM image normal to the fracture plane of sample 34T showing a high degree of orientation and ductile pulling of brils. Fiber clumping
also appears to be present.

Fig. 13. SEM of sample 1C showing ber pullout (ber sliding) and ber matrix pull away.
S.D. Bartus, U.K. Vaidya / Composite Structures 67 (2005) 263277 275

3.6.1. Discussion This type of failure was seen in sections taken from
Fiber pullout occurs when the interfacial stresses at the specimens representing typical failure modes, shown
the ber matrix interface exceeds the interfacial strength, in Figs. 813 using SEM. The SEM images were taken
causing the ber to debond from the matrix. The impact normal to the fracture surface, Figs. 813. The ber ori-
load is then carried by the interfacial friction force im- entation was modeled in previous work [4] using CAD-
posed on the ber from the matrix and through matrix press-TP, a process modeling software that simulates
cracking. The magnitude of the interfacial friction is a mold lling and predicts ber orientation. A good corre-
function of the debonded ber surface roughness, the - lation was found between the fracture paths, which were
ber length, and the radial stresses on the ber imposed shown to follow planes parallel to areas with high ber
by thermal contraction. orientation predicted in the model. Matrix burn-os
The surface roughness of the ber is another factor were conducted on impacted samples, and they corre-
that is thought to inuence the ber matrix interface, lated with the CADpress-TP simulation [4]. Representa-
and thereby the failure. The surface roughness is mainly tive burn-os are shown in Fig. 14. The burn-o samples
dependant on chemical coupling and ber wetting at the
bermatrix interface. Fibers sized for the matrix exhibit
increased chemical coupling though an increase in ber
wetability. The debonded surface of a properly sized
ber is shown in Fig. 8, typically indicated by a rough
appearance when viewed at sucient magnication.
If the ber is above critical length, ber breakage
must occur for the ber to pullout. This failure mode,
referred to as ber fragmentation, occurs when the inter-
facial stresses exceed the ber strength. This is likely to
have occurred in the impacted LFT samples. However,
it is not easy to characterize the phenomenon without
low ber loading in a transparent matrix, clearly not
the case with the samples examined.
Three possible modes of failure appear in the matrix;
brittle fracture, crazing-tearing (stress whitening) and
separation of brils, Figs. 8, and 1013. A smooth frac-
ture surface indicates brittle fracture, Figs. 8, 10, 11 and
13. This was the most common failure mode seen in the
SEM micrograph analysis. The matrix fails in a brittle
manner because the ber reinforcement inhibits large
matrix deformations. Crazing-tearing, indicated by the
white areas in the micrographs shown in Figs. 8, 10,
11 and 13 is also common in thermoplastic matrices sub-
jected to strain. The ductile pulling of brils, Figs. 9 and
12 also indicates a brittle failure. No attempt was made
to quantify the matrix failure modes. The contribution
of energy dissipation from matrix is typically small in
contrast to the reinforcement.

3.7. Fiber orientation eects

Thomason and Vlug [6] surmised that a linear rela-


tionship exists between tensile and impact strength. As
the ultimate tensile strength increases, the impact
strength increases. Since tensile strength is a strong func-
tion of ber orientation, impact strength would be ex-
pected to follow similar trends. No known models of
impact strength as a function of discontinuous ber ori-
entation exist, and current models for uniaxially aligned
Fig. 14. Post impact matrix burn-os for representative samples (top,
bers have not been well-documented [6]. Intuitively, center, bottom) showing a high degree of ber orientation along the
fracture seems most likely to follow the weakest plane, fracture plane. The arrows indicate the direction of propagation and
e.g., the plane parallel to the main ber orientation. the dashed circle shows where the projectile impacted.
276 S.D. Bartus, U.K. Vaidya / Composite Structures 67 (2005) 263277

Impact area

Fracture path

Fibers blunting
crack propagation

Fig. 15. Illustration of a possible failure mechanism of the samples tested showing a transverse view of a test plaque with the fracture path following
a preferential plane of ber orientation.

qualitatively indicate that the highest ber orientation (average areal density = 4.61 g cm2) was 167 J and 121 J
occurred in the bottom sample (farthest from the charge for the at and conical projectiles, respectively. The pre-
placement), and the most random ber orientation was dominant energy dissipation mechanisms are ber frac-
in the top sample, where the charge was placed. Ran- ture, ber debonding, ber pullout, and matrix fracture.
dom ber orientation appears to result in the highest im- For panels impacted by the at tipped projectile with
pact strength in the case of BOI testing by creating a normal incidence (e.g., full contact of the projectile face
torturous crack path, while the Charpy and Izod impact with the sample), the limiting damage occurred by
testing; strength is greatest perpendicular to areas of punch-through (shear plug). The damage initiated
high ber orientation. around the periphery of the impactor as a result of high
The burn-os clearly show that failure occurred along transverse shear stresses. Assuming a uniform stress dis-
areas with a high degree of ber orientation, Fig. 14, and tribution around the periphery of the projectile, failure
the fracture path changed when the main ber orienta- occurred in areas with preferential ber orientation tan-
tion underwent a change, Fig. 14(b). This information gent to the impact area. Away from the impactor, the
is of great interest since areas with a high degree of ber dominant failure mode occurred by simultaneous tear-
orientation can be altered through choice of materials, ing (planar cracking) across planes of preferential ber
process parameters, and component geometry, in order orientation.
to improve impact resistance. In the 100 g at impactor study, the edge of the pro-
A simplied illustration of the failure progression is jectile made the rst contact resulting from a slight tilt-
shown in Fig. 15. The gure shows a schematic trans- ing upon or prior to contact. This consequently resulted
verse view of ber orientation and fracture path. The in an initial notch arising from increased contact stres-
impact occurs at the top of and the damage/crack fol- ses, followed by punch-through and subsequent tearing.
lows the plane with a higher degree of ber orientation. The increase in contact stresses decreased the critical en-
As the crack propagates, energy is dissipated though ergy required for perforation. This is also thought to be
ber debonding, ber pullout, ber breakage, and ma- the mechanism for a decrease in energy dissipation for
trix fracture. The crack continues to propagate until the conically tipped projectiles.
the energy for fracture is spent or the crack is blunted Energy dissipation (critical energy) for samples sub-
by bers oriented perpendicular to the fracture path. jected to conical projectile impact was approximately
27% less than samples impacted by at projectiles. How-
ever, samples impacted by conical projectiles exhibited a
4. Conclusions higher degree of damage, e.g., more fracture surface as a
result of increased crack branching. This may be a result
The eects of LFTs subjected to BOI were investigated of increased ber strain energy at the point of impact for
as a function of projectile mass and geometry. The aver- samples impacted with the conical projectiles. When the
age critical energy dissipated for PP/40 wt.% E-glass bers fracture, the release of strain energy is greater than
S.D. Bartus, U.K. Vaidya / Composite Structures 67 (2005) 263277 277

the energy required to create a single crack plane, result- [5] Lee SM, Cheon JS, Im YT. Experimental and numerical study of
ing in crack bifurcation in order to dissipate the excess the impact behavior of SMC plates. Compos Struct 1999;47:
55161.
energy. [6] Thomason JL, Vlug MA. Inuence of bre length and concen-
Impactor velocity did not result in any appreciable ef- tration on the properties of glass bre-reinforced polypropylene.
fects on the critical energy dissipation in the specimens. 4. Impact properties. Composites 1997;28A:27788.
This indicates that the material was not sensitive to the [7] Thomason JL, Vlug MA. Inuence of bre length and concen-
loading rate in the range investigated. Critical energy tration on the properties of glass bre-reinforced polypropylene.
1. Tensile and exural modulus. Composites 1996;27A:47784.
dissipation is linear with respect to areal density for each [8] Miller K, Ramani K. Process-induced residual stresses in
projectile geometry. In addition to areal density eects, compression molded UHMWPE. Polym Eng Sci 1999;39:1108.
ber orientation may also play a critical role in energy [9] Bush SF, Torres FG, Methven JM. Rheological characterization
dissipation. The energy dissipation of impacted LFTs of discrete long glass bre (LGF) reinforced thermoplastics.
appears to decrease as areal density decreases and as Composites 2000;31:142131.
[10] Chawla KK. Composite materials science and engineering. 2nd
ber orientation increases. ed., New York: Springer-Verlag; 1998.
The cracks had a tendency to advance along areas [11] Curtis PT, Bader MG, Bailey JE. The stiness and strength of
where an abrupt change in local ber orientation occurs. a polyamide thermoplastic reinforced with glass and carbon bers.
Abrupt changes in local ber orientation may also be a J Mater Sci 1978;13:37790.
mechanism for crack blunting. This indicates that com- [12] Abrate S. Impact on composite structures. Cambridge University
Press; 1998.
ponents exhibiting a high degree of anisotropy, or [13] Delfrosse D, Poursartip A. Energy-based approach to impact
abrupt changes in ber orientation, will be susceptible damage in CRFP laminates. Composites 1997;28A:64755.
to impact damage. A random orientation would likely [14] Dutta PK, Farrell D, Taylor S, Tadayon A, Hui D. CRREL
provide the maximum impact strength by creating a tor- Special Report 96-29, 1996.
turous path for crack propagation. More over, LFTs [15] Okafor AC, Otieno AW, Dutta A, Rao VS. Detection and
characterization of high-velocity impact damage in advanced
with a random ber orientation would also exhibit composite plates using multi-sensing techniques. Compos Struct
greater damage tolerance due to crack blunting. 2001;54:28997.
[16] Iremonger MJ, Went AC. Ballistic impact of bre composite
armours by fragment-simulating projectiles. Composites
References 1996;27A:57581;
Tarim N, Findik F, Uzun H. Ballistic impact performance of
[1] Hartness T, Husman G, Koenig J, Dyksterhouse J. The charac- composite structures. Compos Struct 2001.
terization of low cost ber-reinforced thermoplastic composites [17] Tarim N, Findik F, Uzun H. Ballistic impact performance of
produced by the drift process. Composites 2001;32:115560. composite structures. Compos Struct 2002;56:1320.
[2] Marsh G. Composites on the road to big time. Reinf Plast [18] Jenq ST, Wang SB, Sheu LT. A model for predicting the residual
2003;47:336. 3847. strength of GFRP laminates subject to ballistic impact. J Reinf
[3] Saito M, Kukula S, Kataoka Y. Practical use of the statistically Plast Compos 1992;11:112741.
modied laminate model for injection moldings. Part 1: method [19] Cantwell WJ, Morton J. The inuence of varying projectile mass
and verication. Polym Compos 1998;19:497505. on the impact response of CFRP. Compos Struct 1989;11:10114.
[4] Bartus SD. Long-ber-reinforced thermoplastic: process modeling [20] Available from: http://www.ticona-us.com/datasheet/DataSheet.
and resistance to blunt object impact. A thesis, The University of cfm?ID=827&STANDARD=ISO, August 9, 2002.
Alabama at Birmingham, Materials Science and Engineering [21] Hsiao HM, Daniel IM. Stain rate behavior of composite
Department, 2003. materials. Composites 1998;29B:52131.

Você também pode gostar