Impact Testing of Composites For Aircraf
Impact Testing of Composites For Aircraf
R=20010047831 2020-07-10T16:07:18+00:00Z
April 2001
Trade names or manufacturers' names are used in this report for
identification only. This usage does not constitute an official
endorsement, either expressed or implied, by the National
Aeronautics and Space Administration.
Available from
NASA/TM--2001-210887 1
traditional materials and designs. However, as engine obtained from full-scale testing and analysis since it
design moves toward higher bypass ratios, there will be depends on the particular engine design and operating
an increasing need to consider composite materials for conditions. The design of the fan containment system in
the larger diameter fan cases. a particular engine depends on issues such as weight,
cost, prior experience, and limitations on space
Explicit finite element analysis is beginning to be used available for case deflection. Two design approaches
earlier in the design cycle for metal fan cases to shorten are currently used for blade containment. In the
development time, reduce testing cost, and reduce the "hardwall" design, the wall thickness near the blade
risk of failure for the final design. This approach is also impact site is made sufficiently thick to resist
used in the automobile industry where explicit finite penetration, and ribs are used to control global stiffness.
element analysis is used in conjunction with In this design, local damage at the impact site must be
instrumented crash tests to improve vehicle minimized, and propagation of the damage must be
crashworthiness. Recent advances in the use of explicit limited so that global stiffness and strength of the
codes such as LsDyna3D and ABAQUS/Explicit allow structure are maintained. In the "softwall" design, the
numerical testing of an entire car before even a single metal case is penetrated by the blade and captured by an
car body is manufactured. A similar approach is needed outer wrap of fabric. In this design, stiffness is provided
to overcome the large development cost and technical by either an isogrid or honeycomb structure on the case.
risk associated with using a new composite fan case Although the blade is allowed to penetrate the case in
design. Use of finite element analysis to assist design is the softwall design, it is critical that the perforated area
most effective when a design methodology already be kept as small as possible and that the damaged area
exists and a large database (laboratory tests, does not grow to affect global stiffness and strength.
certification tests, and field service) is available for
validation. This is a particularly important issue for The high specific stiffness and low density of
composites because of the large number of possibilities composites are advantages for both hardwall and
for fiber orientation, failure mode, materials, and softwall designs, but poor impact resistance is a
manufacturing method. Unfortunately, there is no prior potential problem. Low velocity impact of composites
certification or service experience for a composite case, can cause extensive internal damage (microcracking,
and only a limited amount of laboratory data is delamination) with little visible surface damage. 3 High
available from government and commercial technology velocity tests on composites using fiat, hemispherical,
development programs. and conical projectiles have shown shear failure in plies
nearest the impact surface to be the initial failure
The two main goals of this work are: (1) to develop and mode. 4 Subsequent damage modes depend on the
demonstrate design concepts for a composite fan case materials, fiber architecture, and ply lay-ups, but often
and (2) to generate test data and analysis methods for include delamination and fiber tensile failure in the
future use as part of an overall design methodology. interior or near the back surface plies.
The first step is to evaluate the capabilities and
limitations of current materials, test methods, and In order to identify expected damage modes under
explicit finite element analysis methods. In this paper, simulated blade-out conditions, a 15" diameter, 0.420"
ballistic impact test results are reported for composite thick 2D laminated glass/epoxy half ring was impacted
panels. A larger full-ring structure is proposed as a with a wedge-shaped titanium projectile at a velocity of
possible subcomponent design. Explicit finite element 850 ftYs. The 0.731b (330g) wedge was 5" long by 3.75"
analysis is used to evaluate impact dynamics for the wide with the thickness increasing from 1/8" at the thin
proposed subcomponent test. edge to 3/8" at the thick edge to simulate the mass
distribution of a fan blade fragment. To simulate the
COMPOSITE IMPACT TESTS
blade dynamics, the desired orientation of the wedge
was 45 ° from the horizontal with the thin edge of the
Ballistic impact tests have been used on targets ranging
wedge making contact first. The thin edge curls after
in size from small fiat panels to full-scale contact as the wedge rotates and the thick edge makes a
subcomponents. Full-scale tests can simulate many of second contact. Figure 1 shows the wedge and a side
the conditions in a real blade-out incident. Small panel
view of the 15" half-ring after impact. Local fiber shear
tests attempt to simulate at least a few critical
fracture occurs along the lines where the thin and the
parameters. The impact velocity is usually chosen as
thick edges of the wedge make contact. Extensive
the center of mass velocity of the blade fragment. The
delamination occurs throughout the composite, and
projectile shape and orientation relative to the target is
tearing occurs in plies near the back surface. Damage
determined from the blade dynamics and the type of also appears to be concentrated at some points around
damage observed in the case. This information must be
NASA/TM--2001-210887 2
theedges wherethepanelwasheldinplacebythe location of impact can also cause variability in the
metalfixtures. results. The location and orientation of the blade has a
large effect on the damage induce by impact. Damage
Thedamage evident in Figure1indicates thata induced by the first contact of the thin edge of the
compositehardwall containment system will require wedge is more severe for pitch angles less than 45 °.
useofasurface layerthatis capableofeliminating Rotation of the wedge also causes more severe damage
localshearfailure.Inaddition, somemethod for because the wedge makes contact at its comers rather
controlling
delamination will beneeded. Lightweight than along its edges. All of these factors indicate that a
materials
andstructures thatarecapable ofabsorbing a simpler, less expensive test would be preferable for
largeamount of energy have been developed for armor preliminary screening of composite concepts. In this
applications. Although most of these materials are not paper, a simpler test was used to evaluate the
suitable for use as the main structural material in a performance of a ceramic layer on a glass/epoxy
containment case, a recent report 5 has examined the composite and the performance of 3D woven materials.
feasibility of applying some of these armor concepts to Some justification based on observed failure modes is
commercial aircraft in the form of protective barriers discussed below. Analysis of impact dynamics for a
that could enhance aircraft survivability in the event of full-ring structure, discussed in a later section, provides
an uncontained engine failure. One such concept is to further support for the use of the simpler test method.
place a ceramic plate on the impact surface of the
composite. The ceramic serves to fragment the Small (7"X7") flat panels were impacted with flat-
projectile and spread the impact load over a larger area ended cylindrical titanium projectiles (2.54 cm long,
of the backing plate. Projectile fragmentation occurs at 1.27 cm diameter). Panel and projectile shapes are
very high impact velocities as a result of shock waves shown in Figure 2. This test is less realistic than tests
within the projectile. This process is unlikely to occur with the wedge-shaped projectile in the sense that it
with large titanium blade fragments impacting a case at cannot simulate blade dynamics. However, preliminary
velocities characteristic of a blade-out event. However, tests on laminated composite panels with various fiber
a ceramic layer may still be useful for spreading the and matrix materials demonstrated that the impact
impact load and blunting sharp blade fragments. damage had features similar to those shown in Figure 1.
Delamination can be controlled by reinforcement in the Fiber shear fracture occurred on the front surface,
transverse direction by processes such as stitching, delamination occurred in the interior, and fiber tearing
pinning, or 3D weaving. Evaluation of these concepts occurred on the back surface. The ballistic limit was
using the test configuration shown in Figure 1 would be determined by testing approximately 6-8 panels over a
expensive and time consuming. The high cost is a result range of impact velocities. Ideally, the ballistic limit is
of fabrication costs for the panels, fabrication costs for the velocity below which the projectile does not
the projectile, and the cost of performing the test. perforate the panel and above which the projectile
Although the proposed concepts would not necessarily completely perforates the panel. The ballistic limit can
be too expensive for commercial production, then be defined as the average of the highest velocity
fabrication of a small number of test panels can be very that did not perforate and the lowest velocity that
costly. For example, set-up costs for weaving represent completely perforated the panels. In practice, there may
a small part of the total cost for a large production run be a small velocity region near the ballistic limit in
but a large part of the cost for fabrication of a small which non-perforation and perforation overlap. In
number of panels. In addition, scheduling such a small addition, failure may be defined by a criterion other
run often cannot be done in a timely manner. The than complete perforation. For example, a plate may be
wedge used for the test shown in Figure 1 requires considered to fail the test if a crack propagates through
machining of a block of titanium. A simpler projectile the panel, even though the projectile does not perforate
that could be cut from stock material would be less the panel. The above definition for ballistic limit can
expensive and more readily available. In addition, the still be used, as long as the failure criterion is defined.
impact test shown in Figure 1 is somewhat complex.
Since the projectile is not cylindrical or spherical, it Impact test data for five different types of composite
must be mounted in a cylindrical container (sabot) in panels are shown in Figure 3. For these composite
order to fit into the barrel of the gas gun. The 45 ° pitch panels there was a velocity region near the ballistic
angle is achieved by bonding the titanium wedge to a limit within which the projectile was caught in the
45 ° prism-shaped wood platform within the sabot. panel. This was interpreted as perforation (or failure)
Imprints of the wedge in the test panel and high-speed for the purpose of calculating the ballistic limit. The
video of the projectile during the test have shown that ballistic limits calculated from the data in Figure 3 are
the orientation of the wedge often deviates in pitch and shown in Figure 4 along with the ballistic limits that
roll from the desired orientation. Deviations in the have been measured for 304 stainless steel and for
NASA/TM--2001-210887 3
lnconel 718.Thestainless steelalloy was chosen for reduction. For material cost reduction, E-862 epoxy
comparison because it has been used in commercial fan resin was used in place of PR520, and AS4 fiber was
cases. The ballistic limits of Inconel 718 in the annealed used in place of IM7. For weaving cost reduction, tows
and heat treated forms are shown to indicate the range of the AS4 fiber were combined during weaving. This
of properties possible for some alloys by heat treatment. resulted in a courser weave for the AS4/PR520 and
The thickness (or areal weight) range of the metal alloy AS4/E-862 composites compared to the IM7/PR520
panels used to obtain the data in Figure 4 was chosen so and IM7/E-862 composites. Failure on the front surface
that the ballistic limits would fall within the range of of the 3D woven composites was similar to that of the
velocities which are characteristic of blade fragments 2D glass/epoxy laminated composites with local shear
during a blade-out. It is useful to have data over a range failure on the front surface. However, the subsequent
of areal weights because it is usually not feasible to damage pattern was different. For the 3D composites
fabricate panels of different materials with exactly there was no evidence of delamination, but there was a
equal areal weights for direct comparison. region of microcracking around the impact site,
particularly for the more brittle E-862 resin. The 3D
The first composite concept evaluated was the use of a composites had fiber breakage but no ply tearing on the
1.9mm (0.075") thick silicon carbide ceramic layer to back side. The IM7/PR520 composite had a ballistic
reduce the extent of shear failure at the impact surface. limit of 190rn/s (623tt/s), comparable to that of the
7"X7" glass/epoxy panels were cut from a larger heat-treated Inconel 718 at an equivalent areal weight.
16.5"X38.0" panel that was made by hand lay-up of 28 The ballistic limit of the IM7/E-862 composite was
plies of type 6781 $2 glass/E-761 epoxy prepreg with lower at 179m/s (587ti/s). Ballistic limits for
autoclave cure. Six of the 7"X7" panels were used to composites made using the courser weave and AS4
measure the ballistic limit of the composite (Figure 3). fibers were 156m/s (512ft/s) for AS4/PR520 and
In Figure 4, the ballistic limit of the $2 glass/E-761 159rn/s (522it/s) for AS4/E-862. This is much lower
composite is 19 lm/s (626if/s), which is well below that than the ballistic limits of the metal alloys in Figure 4.
of the metal alloys at an equivalent areal weight. The IM7 composites exhibited a slight permanent
Figures 2a and 2b show the front and back surfaces conical deformation in the impact region when tested
after impact for glass/epoxy panels with and without a near the ballistic limit. This type of permanent
ceramic front layer. The lower panel in Figure 2a has a deformation is typical for metal alloys but not for
nylon spall sheet bonded to the surface to catch ceramic composites. This effect occurred to a much smaller
fragments during the test. The ceramic fragments have extent in the AS4 composites and not at all in the 2D
been removed to view the surface of the composite. The laminated glass/epoxy composites.
panels in Figure 2 were impacted at 636ff/s (194m/s),
SUBCOMPONENT DEFINITION
just above the ballistic limit. The panel without the
ceramic layer exhibited the typical failure modes of
fiber shear failure on the front, delamination in the Although more small panel testing is needed to
middle plies, and tearing on the back. The ceramic layer optimize performance, there are three results from the
completely eliminated the fiber shear fracture on the small panel tests that indicate scale-up is justified. First,
front surface but did not reduce the extent of the ballistic limit of some composites approaches that
delamination as viewed from the back side of the of the metals, even though the materials and fiber
panels. The dashed circles in Figure 2b show the architecture have not yet been optimized. Second,
delamination areas visible from the back side of the transverse fiber reinforcement is effective in limiting
panels. (Delamination is clearly visible when viewing delamination. Third, the ceramic layer is effective in
the panels but does not show well in the photographs). eliminating fiber shear failure at the impact surface.
Although the ceramic layer is effective in eliminating The next level of testing should address the issues of
fiber shear failure at the impact surface, this benefit manufacturability and cost as well as demonstrating
comes at the expense of increasing the areal weight of technical feasibility at a larger scale. The subcomponent
the panel. Further tests are needed to determine the test in Figure 5 is proposed for this purpose. The 36"
optimum ceramic thickness and the effectiveness of the diameter ring will be manufactured by filament
ceramic on a total areal weight basis. winding. Filament winding is a low cost process for
fabricating cylindrical structures. This process also
The second concept evaluated was use ofa 3D woven allows for considerable flexibility in materials selection
architecture to reduce delamination. Composites were and fiber architecture. In addition, features such as ribs
fabricated using 3D carbon fiber preforms and resin and flanges can be included in the design. If the
transfer molding. The main system of interest was IM7 composite material properties are not strain rate
fiber with the toughened epoxy resin, PR520. However, dependent, scaling of displacement, contact forces, and
options were also considered for potential cost strains from a small test to a larger prototype is
NASA/TM--2001-210887 4
possible,
butscaling
ofdamage
isnot. 6 Therefore, velocity of the upper point on the projectile to increase
results of the small panel tests can be used as a guide to during this time interval. Between 0.7ms and 1.0ms the
the design of the subcomponent in Figure 5, but some projectile rebounds from the case. After 1.0ms the
preliminary testing and development of a numerical projectile releases from the case and continues to travel
model will be needed to implement a cost effective test at constant velocity. In the time interval from 0 to
program at this larger scale. 1.0ms, while the projectile is in contact with the case,
the motion of the projectile is very similar for the full-
FINITE ELEMENT ANALYSIS ring and the half-rings with both free and fixed
boundary conditions. The motion of the projectile
Explicit, transient, finite element analysis of the impacting the quarter-rings deviates from the half- and
proposed subcomponent in Figure 5 was performed full-ring configurations after 0.4ms when the initial
using the commercial code ABAQUS/Explicit. 7A impact of the upper point on the projectile with the case
preliminary analysis was performed to examine the is complete. The same conclusions result when
overall dynamics of the ring during and after impact. In displacements and velocities for points on the rings are
this analysis the impact velocity was 600ft/s and the analyzed. These results suggest that a half-ring
composite ring was taken to be 0.3" thick with the configuration could be used in place of a full ring to
following properties, which are representative of a evaluate local damage. Either free or fixed boundary
conditions could be used. Alternatively, two impact
glass/epoxy laminate with a 00/90 ° layup.
tests could be performed 180 ° apart on the same full-
Ell = E22= 2.5X106 lbf/in 2 ring. Either approach would reduce test costs. A
Gl2 = 1.0XI06 lbf/in 2 quarter-ring configuration could be used if the purpose
GI3= G23= 6.0X105 lbf/in 2 of the test is to evaluate local damage caused by a
x)12='021= 0.14 single impact with no subsequent complex motion of
the projectile. This supports the use of a simple
Linear elastic, rate independent properties were used, projectile and smaller panels for initial screening
and damage was not considered for this preliminary studies when local damage is the primary concern.
analysis. The projectile was a fiat circular cylinder with
diameter of 4.5", thickness of 0.75", and mass of 1.9lb. Strain energies stored in arcs of various sizes around
It was modeled as a rigid body with a density of the impact point are shown in Figure 9a for the full-ring
0.00040861bf_/in 4. The ring was inclined at an angle of configuration. The total energy (strain energy + kinetic
energy) is shown in Figure 9b. An additional curve
22 ° to the horizontal. The mesh in Figure 6 along with
showing the energy in the full-ring plus the energy in
the symmetry condition about the mid plane was used
the projectile is shown in each figure. Since the
for the full-ring analysis. Half-ring and quarter-ring
projectile is modeled as a rigid body, all of its energy is
structures with both free and fixed boundary conditions
in the form of kinetic energy. In Figure 9a the strain
were also analyzed. Four-node thin shell S4R elements
energy grows to a maximum value at about 0.45ms as
were used for the shell. Six-node and eight-node solid
C3D6 and C3D8R elements were used for the the projectile comes into contact with the ring. The
maximum strain energy is a little over half of the initial
projectile.
kinetic energy of the projectile. In Figure 9b the total
energy in the full-ring remains constant after about
Deformation of the full ring structure at various time
1.0ms because the projectile has rebounded from the
intervals is shown in Figure 7. At 0.1 lms the case
ring. The maximum strain energy density in the ring is
begins to bulge as the top edge of the projectile pushes
a useful parameter for preliminary design of the ring.
on the case. At 0.3 lms the center of the case bulges as
From Figure 9a it appears that the maximum strain
the center of mass of the projectile contacts and pushes
energy density would occur at about 0.40ms within a
on the case. At later times the projectile rebounds, and a
50 ° arc. This energy density could be used in a static
ftexural wave begins to travel around the case. The
composite analysis to determine how close the ring is to
displacement and velocity of the upper point on the
its ultimate strength. A static analysis could also be
projectile are shown in Figures 8a and 8b for the full-
used to suggest changes in materials, fiber architecture,
ring model along with the results for the half- and
quarter-models with both free and fixed boundary or other parameters that would improve performance.
conditions. In the time interval from 0 to 0.25ms, The static analysis could also be used interactively with
the filament winding design and machine control
contact between the upper point on the projectile and
the case causes the case to bulge and causes the software to identify structures that will perform well but
also be feasible to fabricate. Future efforts will focus on
projectile to slow down and rotate. Between .25ms and
0.7ms the center point of the projectile contacts the case refining the finite element methods to model damage
more precisely and include strain rate dependence.
and causes the case to bulge fin'ther. This causes the
NASA/TM--2001-210887
Effortswill alsocontinue
torefinethesubcomponent
design. REFERENCES
NASA/TM--2001-210887 6
Figure1.Glass/epoxy
15"diameter half-ring and wedge-shaped titanium projectile
NASA/TM--2001-210887 7
210 .....
÷
20O 0
0
0 6
E 190
O- ,4- m
o B
o 6
180
>
Q.
E 170
+ ÷
- Not Perforated
160 O Caught
+ Perforated $
o..
i
150 I I I L
IM7/PR520 IM7/E-862 AS4/PR520 AS4/E-862 Glass/Epoxy
Fiber/Resin
Figure 3. Composite impact test data from 7"X7" flat panel tests.
280
260
240
on
E
.n
..d 22O
L)
¢/)
140
6 8 10 12 14 16 18 20 22 24
NASA/TM--2001-210887 8
Titanium disk
(4.5" dia., 0.75" thick)
6 l'll_ .....
Composite ring
m
Figure 6. Finite element mesh for 36" diameter full-ring subcomponent.
NASA/TM--2001-210887 9
; ; ; ; ;
0.0 I I-- I I
I
Quarter-ring i I I
I
Free BC [ i i
Quarter-ring
Fixed BC Ii Fixed BC
Half-ring [
I I I
._= I
I
I
I
Full-ring
-5.0
I I ! I
O3 I I I I
._= I I I
I
III [ Half-ring
Free BC I I
I i I I
I I I I I
-10.0 1"------I"-----'1"----- =I'----'*'I
I I I I I I
I I I I I
I I I I I
I I I I I
I I I I I
I I I I I
I I I I I
4,000
FreeBC
Quarter-ring [--I---I---I--! I I
, = I I I
I I I I I
2,000 1---'1-- -"i--
Quarter-ring I I I
Fixed BC I I i
I I I
>
-2,000
-4,000
Figure 8. Finite element analysis results for a titanium disk projectile impacting full-ring, half-ring and
quarter-ring composite structures at 600fl/s. Free and fixed boundary conditions (BC) were used for
half- and quarter-rings. (a) Displacement and (b) velocity of the upper point on the projectile.
NASA/TM--2001-210887 10
f_ I I I I I 1
20,000 180 °
15,000
10,000 "J_ 90 °
f 20 ° _ 50° 1
(a) Strain energy in various arc sections centered on the point of impact
_----e-t.-e-_c c Io o 6 C---eL-o---tiC C, Iv
60.000 - Full-ring and -
/c-x Projectile
/
50,000 - .{ _
r
t
Full-ring
t
40,000
180 °
J
30,000
90 °
20,000
i
_ko_ ¸' ..... ......
500 .............. ....
10,000 20"3¢._, " ....... " " " -
0
0.0 1.0 2.0 3.0
Time, ms
(b) Total energy in various arc sections centered on the point of impact
Figure 9. Finite element analysis results for a titanium disk projectile impacting full-ring composite structures at
600ft/s. Energy stored in arc sections centered on the point of impact is indicated for various arc
angles. (a) Strain energy and (b) Total energy (strain energy + kinetic energy)
NASA/TM--2001-210887 11
REPORT DOCUMENTATION PAGE Form Approved
OMB No. 0704-0188
Public reportingburdenfor thiscollectionof informationis estimatedto average1 hour per response,includingthe timefor reviewinginstructions,searchingexistingdatasources,
gatheringandmaintainingthedata needed,andcompletingand reviewingthe collection of information.Sendcommentsregarding this burdenestimateor anyotheraspectof this
collectionof information,includingsuggestionsfor reducingthis burden,to WashingtonHeadquartersServices,Directoratefor InformationOperationsandReports,1215Jefferson
DavisHighway,Suite1204,Arlington,VA 22202_.302,andto the Officeof Managementand Budget,PaperworkReductionProject(0704-0188),Washington,DC 20503.
1. AGENCY USE ONLY (Leave blank) 2. REPORT DATE 3. REPORT TYPE AND DATES COVERED
April 2001 Technical Memorandum
4. TITLE AND SUuiiiLE 5. FUNDING NUMBERS
WU-708-24-13-00
6. AUTHOR(S)
Prepared for the 42nd Structures, Structural Dynamics, and Materials Conference and Exhibit cosponsored by the AIAA, ASME, ASCE,
AHS, and ACS, Seattle, Washington, April 16-19, 2001. Gary D. Roberts and Duane M. Revilock, NASA Glenn Research Center;
Wieslaw K. Binienda and Walter Z. Nie, University of Akron, 302 Buchtel Mall, Akron, Ohio 44325--0001; S. Ben Mackenzie and
Kevin B. Todd, Saint-Gobain Performance Plastics, Ravenna, Ohio (formerly Norton). Responsible person, Gary D. Roberts,
organization code 5150, 216--433-3244.
12a. DISTRIBUTION/AVAILABILITY STATEMENT 12b. DISTRIBUTION CODE
Unclassified - Unlimited
This publication is available from the NASA Center for AeroSpace Information, 301-621--0390.
13. ABSTRACT (Maximum 200 words)
Before composite materials can be considered for use in the fan case of a commercial jet engine, the performance of a
composite structure under blade-out loads needs to be demonstrated. The objective of this program is to develop an
efficient test and analysis method for evaluating potential composite case concepts. Ballistic impact tests were performed
on laminated glass/epoxy composites in order to identify potential failure modes and to provide data for analysis, Flat
7×7 in. panels were impacted with cylindrical titanium projectiles, and 15 in. diameter haft-rings were impacted with
wedge-shaped titanium projectiles. Composite failure involved local fiber fracture as well as tearing and delamination on
a larger scale. A 36 in. diameter full-ring subcomponent was proposed for larger scale testing. Explicit, transient, finite
element analyses were used to evaluate impact dynamics and subsequent global deformation for the proposed full-ring
subcomponent test. Analyses on half-ring and quarter ring configurations indicated that less expensive smaller scale tests
could be used to screen potential composite concepts when evaluation of local impact damage is the primary concern.
17. SECURITY CLASSIFICATION 18. SECURITY CLASSIFICATION 19. SECURITYCLASSIFICATION 20. LIMITATION OF ABSTRACT
OF REPORT OF THIS PAGE OF ABSTRACT
Unclassified Unclassified Unclassified