Development of reliable modeling methodologies for fan ... - CiteSeerX

ARTICLE IN PRESS International Journal of Impact Engineering xxx (2008) 1–11

Contents lists available at ScienceDirect

International Journal of Impact Engineering journal homepage: www.elsevier.com/locate/ijimpeng

Development of reliable modeling methodologies for fan blade out containment analysis – Part I: Experimental studiesq D. Naik a, S. Sankaran a, B. Mobasher a, S.D. Rajan a, *, J.M. Pereira b a b

Department of Civil Engineering, ECG 252, Arizona State University, Tempe, AZ 85287-5306, USA Structures and Dynamics Branch, NASA-GRC, 21000 Brookpark Road, MS 49-8, Cleveland, OH 44135, USA

a r t i c l e i n f o

a b s t r a c t

Article history: Received 9 July 2007 Received in revised form 20 February 2008 Accepted 7 March 2008 Available online xxx

High strength woven fabrics are ideal candidate materials for use in structural systems where high energy absorption is required. One of the more widely used applications for woven fabrics is in propulsion engine containment systems. In this first part of a two-part paper, details of the experiments to characterize the behavior of dry fabrics including KevlarÒ and ZylonÒ are presented. The experimental program to characterize the behavior of 1420 Denier KevlarÒ 49 17 17, 500 Denier ZylonÒ AS 35 35, and 1500 Denier ZylonÒ 17 17 are discussed. The primary objective is to use the experimental results in the development of a constitutive model that can be used in an explicit finite element analysis program. These include Tension Tests in both the warp and fill directions of the fabric, Trellising Shear Tests and Friction Tests between fabric layers. The results from these tests provide the basis for development of the constitutive model – relating stresses to strains, characterizing failure and interaction between fabric layers. In addition to these basic material tests, tests on systems built with fabric wraps were also conducted. Ballistic tests of containment wraps subjected to a high velocity projectile were carried out at NASA-Glenn Research Center. While these tests provide a comparison between the energy absorbing characteristics of the three fabrics, they also provide benchmark results to validate the developed finite element methodology discussed in the second part of this paper. Ó 2008 Elsevier Ltd. All rights reserved.

Keywords: Fabric material tests Kevlar Zylon Engine fan blade out containment Ballistic testing

1. Introduction High strength woven fabrics are ideal candidate materials for use in structural systems where high energy absorption is required. Their high strength per weight ratio and the ability to resist high speed impacts enable them to be very efficient compared to metals. One of the more widely used applications for woven fabrics is in propulsion engine containment systems. As a part of the Federal Aviation Administration’s (FAA) aircraft engine certification regulations, an engine must demonstrate the ability to contain a fan blade released at full operating speed. Currently the only woven fabric that is in wide use in engine containment systems is KevlarÒ49 developed by DuPont in the 1970s. It was the first organic fiber with sufficient tensile strength and modulus to be used in advanced composites. Originally developed as a replacement for steel in radial tires, Kevlar is now used in a wide range of applications. Like nylons, Kevlar filaments are made by extruding the precursor through a spinneret. The rod form of the para-aramid

q Research sponsored by Federal Aviation Administration under Grant No: 01-C-AW-ASU. * Corresponding author. Tel.: þ1 480 965 1712; fax: þ1 480 965 0557. E-mail address: [email protected] (S.D. Rajan).

molecules and the extrusion process make Kevlar yarns anisotropic – they are stronger and stiffer in the axial direction than in the transverse direction. The main longitudinal direction of a woven fabric is typically referred to as the warp direction and perpendicular to the warp direction is typically referred to as the fill direction. Hundreds of filaments are bundled together to form a yarn. The engine containment system is typically constructed by wrapping multiple layers of KevlarÒ49 around a stiffened metal structure. The fabric is then covered with a protective layer. Designing the containment system involves determining the type of fabric, the number of fabric layers and fabric width required. Currently the FAA’s design standards require that a full-scale test be completed to qualify an engine. Because of the extensive pre and post-test analysis, and the fact that equipment is unusable after testing, a typical fan blade out (FBO) test can cost several million dollars. With today’s advanced numerical techniques, modeling a propulsion engine and simulating an FBO event can be accomplished using an appropriate hydrocode. While there are proven constitutive models that can simulate the behavior of most of the materials, there are limited mechanistic based constitutive models for woven KevlarÒ49 (or any other) fabric especially for prediction of response subjected to high speed impact loads. The first part of this two-part paper addresses the determination of constitutive material properties of dry fabrics using

0734-743X/$ – see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.ijimpeng.2008.03.007

Please cite this article in press as: Naik D et al., Development of reliable modeling methodologies for fan blade out containment analysis – Part I: Experimental studies, International Journal of Impact Engineering (2008), doi:10.1016/j.ijimpeng.2008.03.007

ARTICLE IN PRESS 2

D. Naik et al. / International Journal of Impact Engineering xxx (2008) 1–11

laboratory testing as an extension of procedures developed earlier [1]. A four-part research report outlining the development of the current model is available from FAA [2]. In this paper laboratory tests were used in the constitutive model for both static and dynamic/explicit finite element analyses to calibrate a firstgeneration model. Tension Tests were conducted to develop the orthotropic stress–strain relationships, trellising ‘‘Picture Frame’’ Shear Tests were used in investigating the shear resistance properties of the fabrics, followed by friction tests to determine the coefficient of friction between fabric layers. In addition to these basic material tests, tests on systems built with fabric wraps were also conducted. Ballistic Tests of containment wraps subjected to a high velocity projectile were carried out at NASA-Glenn Research Center [3]. These tests have provided benchmark results to validate the developed finite element methodology discussed in the second part of this paper [4]. Our research goal is to correlate the constitutive models and responses of systems with several various parameters as variables. Response of several discrete layers lumped into an equivalent continuum element and the feasibility of the formulation as a design aid were also tested. An orthotropic response model was developed to predict the generalized loading conditions of a penetrator making a direct impact with the fabric along the principal directions. It is, however, not clear how the models based on material properties along principal directions respond to off-axis loading conditions. Cases that are subjected to variations in boundary conditions and directions of the load are also not well understood. The methodology is expected to be sufficiently general for use with other fabrics such as ZylonÒ. While it is understood that the durability concerns with Zylon material precludes its application in an engine environment, comparison of Zylon and Kevlar allows further validation of the material property models against one another.

There are several publications that discuss experimental efforts in characterizing the behavior of dry fabrics. Some involve testing of single yarns [5] where Kevlar KM-2 yarns are characterized as a transversely isotropic material in the small strain range. Chen et al. [6] discuss how they obtain shear properties using a trellis fixture. US Army Research Lab tested several square panels consisting of plain-woven 600-denier Kevlar KM2 to develop constitutive relationship in the form of second Piola–Kirchhoff stress vs. Green–St. Venant strain curves [7]. A recent publication [8] discusses experimental procedures to compare the ballistic performance of narrow and wide fabrics. These experimental efforts indicate that while rapid progress is being made in characterizing the mechanical behavior of fabrics, adequately describing the nonlinear relationship and quantification of failure predictions are still a challenge. The first part of this two-part paper is divided into three sections. First, a detailed experimental program in characterizing the behavior of 1420 Denier KevlarÒ 49 17 17, 500 Denier ZylonÒ AS 35 35, and 1500 Denier ZylonÒ 17 17 referred to as Kevlar, Zylon 500D and Zylon 1500D fabrics are presented. Second, ballistic tests that provide the data for calibration of the constitutive model are presented. Finally, the results from these studies are summarized. 2. Fabric experimental tests Several test methods addressing the uniaxial tensile, shear, and frictional aspects of woven fabrics are presented. 2.1. Tension tests Uniaxial tension testing was conducted in a 22 kips (98 kN) servo-hydraulic test frame operated under closed-loop conditions

Fig. 1. (a) Uniaxial tension test setup. (b) Button apparatus. (c) Side view of grips.


ARTICLE IN PRESS D. Naik et al. / International Journal of Impact Engineering xxx (2008) 1–11

Warp direction

a

Warp direction 3000

Kevlar Young's Modulus E11

a

Zylon AS-500 Young's Modulus E11

Stress, MPa

1600

Stress, MPa

3

1200

800

2000

1000

400

0.06

0.05

0.04

0.03

0.02

0.01

0

0

Lateral Strain, mm/mm

0.01

0.02

0.03

0.04

Axial Strain, mm/mm

0 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0

Transverse Strain, mm/mm

0.01

0.02

0.03

0.04

0.05

Axial Strain, mm/mm

Fill direction

Fill direction

3000

b

b Stress, MPa

Stress, MPa

1600

1200

800

Zylon AS-500 Fill Direction, E22

0 0.05

0.04

0.03

1000

400

Kevlar Young's Modulus E22 0.06

2000

0.02

0.01

Lateral Strain, mm/mm

0

0.01

0.02

0.03

0.04

Strain, mm/mm

Fig. 2. Stress–strain curves for Kevlar samples for E11, E22, and Poisson’s ratios n12 and n21. (a) Warp direction. (b) Fill direction.

using procedures similar to ASTM Standards [9]. A stroke control procedure with the loading rate of 0.1 in./min (0.254 cm/min) was used as shown in Fig. 1(a). Digital data acquisition was used to collect data at a rate of 2 samples/sec. The test was continued until complete failure of the specimen was achieved. The overall deformation of the specimen was measured by the stroke movement of the actuator in addition to an extensometer mounted onto the specimen. The load-deformation results were used to calculate the stress–strain response using the gage length and nominal cross sectional area. To ensure that slipping of the specimens from the grips did not influence the testing results, a set of gripping fixtures were developed as shown in Fig. 1(c). Flat steel plates 2.5 in. wide 2 in. long 0.25 in. thick (6.35 cm 5.08 cm 0.635 cm) with a curved groove on one side and a V shaped groove on the other side were used. A matching aluminum piece in the shape of the groove and the notch was used to clamp down the fabric in position. Two shoulder pins were used to hold the two sides of the grip, preventing slippage of the plates. The fabric was held between the V-notch and the aluminum piece. The two plates were held in the hydraulic grips thereby ensuring uniform pressure application set at 1000 psi (6.9 MPa). Gage length was determined based on the average distance of the grip face to grip face. A clip gage system was used to measure the transverse displacement and calculate the Poisson’s effect of the fabric. The primary part of the clip gage system consists of an extensometer mounted two rectangular wood buttons that are stitched onto the fabric as shown in

0 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0


0.01

0.02

0.03

0.04

0.05

Axial Strain, mm/mm

Fig. 3. Stress–strain curves for Zylon AS-500D samples for E11, E22, and Poisson’s ratios n12 and n21. (a) Warp direction. (b) Fill direction.

Fig. 1(b). Energy absorbed by the fabric was calculated based on the actuator displacement and also an extensometer attached to the specimen. The energy absorbed using both the displacement measurement techniques were within 2% of one another indicating that specimen slippage in the grips did not affect the test data significantly. Stress was calculated as the nominal load divided by the average yarn cross sectional area. The average yarn cross sectional area was calculated by measuring the length and mass of a number of warp and fill yarn replicates, and using the mass (bulk) density of the material. The total cross section (c/s) area of the specimen was defined as the cross sectional area per yarn multiplied by the number of yarns per inch of fabric times the total width of the fabric. Tension tests were first carried out on Kevlar samples in the warp and fill directions. In each case, both the axial and lateral deformations were measured under load. The ratio of axial strain to lateral strain was used as a measure of the Poisson’s effect. Fig. 2(a) shows the stress–strain response of five Kevlar 49 samples in the warp direction. In these fabrics, the initial portion of the stress– strain graph shows a large increase in strain up to 0.5% for a very small increase in load. Woven fabrics inherently have crimp (or waviness), and in this portion, the load essentially straightens the yarns by removing the crimp. As the load increases, the yarn alignment leads to an increase of the slope of the stress–strain response. The stress–strain response exhibits nonlinearity prior to the peak load, followed by a sudden drop in the stress beyond the ultimate strength that is characteristic of progressive yarn failure.


ARTICLE IN PRESS 4

D. Naik et al. / International Journal of Impact Engineering xxx (2008) 1–11 Table 1 Measured material properties for various fabrics

Warp direction

a Stress, MPa

Fabric

Zylon AS-1500 warp direction

3000

2000

1000

0

-0.05 -0.04 -0.03 -0.02 -0.01

0

0.01


0.02

0.03

0.04

0.05

Axial Strain, mm/mm

E (MPa)

Maximum Stress (MPa)

Ult. Strain (mm/ mm)

Toughness (MPa)

Kevlar

E11

Average Std. dev

92,862 2065

1618 62

0.02908 0.004786

24.24 4.61

Kevlar

E22

Average Std. dev

105,379 1967

1645 142

0.02164 0.00218

20.66 1.29

Zylon 500D

E11

Average Std. dev

133,136 3530

2938 115.8

0.0362 0.001725

45.38 3.29

Zylon 500D

E22

Average Std. dev

133,493 2459

2746 161.3

0.0264 0.0019

49.88 5.24

Zylon 1500D

E11

Average Std. dev

149,040 2576

3240 132.2

0.0336 0.001543

61.32 5.43

Zylon 1500D

E22

Average Std. dev

146,900 1803

3061 127.4

0.0296 0.0012

61.85 2.31

Fill direction

b

3000

Stress, MPa

Zylon AS-1500 Fill direction

2000

1000

-0.05 -0.04 -0.03 -0.02 -0.01

0

0

0.01


0.02

0.03

0.04

0.05

Axial Strain, mm/mm

Fig. 4. Stress–strain curves for Zylon AS-1500D samples for E11, E22, and Poisson’s ratios n12 and n21. (a) Warp direction. (b) Fill direction.

specimens of Kevlar 49 for the fill direction are utilized in computation of Young’s Modulus E22. The Young’s Modulus E11 and E22 are quite similar numerically; however, the amount of crimp in the warp direction is much more uniform than the crimp in the fill direction. The average stiffness values in the warp and fill directions are 13,490 ksi (93 GPa) and 15,230 ksi (105 GPa) respectively, while the average tensile strength values are 235 ksi (1618 MPa) and 239 ksi (1645 MPa), respectively. These results indicate that Kevlar has quite similar responses in the two principal materials directions. The stress–strain response for Zylon 500D and Zylon 1500D are presented in Figs. 3 and 4, respectively. Results indicate that the stiffness of Zylon 500D in the warp and fill directions is nearly identical at 19,291 ksi (133 GPa). The average stiffness of Zylon 1500D in the warp and fill directions are 21,612 ksi (149 GPa) and 21177 ksi (146 GPa), respectively. The close-knit density of Zylon 500D, characterized with 35 yarns per linear inch, results in the excessive crimp of the test results as shown in the transverse displacement. Note that as much as 1.3% axial and 4% transverse

W

Kevlar 49 ν21 Zylon AS 500 ν12

F W

Zylon AS 500 ν21

F

0.03 0.025 0.02

150000

3500

90000

3000 2500

W

F

2000

Poisson's effects

0.035

120000

Tensile strength, MPa

Kevlar 49 ν12 2

Tensile Stiffness, MPa

Ultimate Strain, mm/mm

Using the test data in the two orthogonal directions, one can calculate effective nominal Young’s Modulus E11, transverse modulus E22, and n12 from the linear portions of the response. As shown in Fig. 2(b), the results obtained from five replicate

Zylon AS 1500 ν12 Zylon AS 1500 ν21 1

0 W

F

W

F 0

1500 Kevlar 49

Zylon AS500

Zylon AS1500

Fig. 5. Tensile strength, tensile stiffness, and ultimate strain for Kevlar, Zylon 500D, and Zylon 1500D samples (first value E11, second value E22, W, warp; F, fill).

500

1000

1500

2000

2500

Applied Stress, MPa Fig. 6. Poisson’s ratio n12 and n21 for Kevlar, Zylon 500D, and Zylon 1500D within the ranges of applied stress.



2.2. Trellising (picture frame) shear tests A series of shear tests was conducted to calculate the shear response of the fabrics. Referred to as the Trellising Tests, specimens measuring 10 in. 10 in. (25.4 cm 25.4 cm) were tested. The square specimens had their corners cut (small squares were cut from the corners) and mounted on the trellis frame. A sample Zylon 500D fabric is shown in Fig. 8. A schematic of the shear (trellis) frame is shown in Fig. 8(a). Fig. 8(b) shows the steel frame developed at ASU. The frame consists of five basic parts – frictionless bearings, clamping plates, multiplier links, a long plate with a 3.5 in. (8.89 cm) slot and two connecting fixtures (for connecting to the top and bottom

Table 2 Poisson’s ratio for various fabrics within the applied stress range Fabric

Ratio

Stress range (MPa)

Average

Std. dev

Kevlar

n12

200–600 600–1000 1000–1400

1.844 0.705 0.618

0.09 0.069 0.079

Kevlar

n21

400–800 800–1200 1200–1500

0.611 0.222 0.087

0.267 0.201 0.22

Zylon 500D

n12

500–1200 1200–2000 2000–2500

0.676 0.152 0.052

0.147 0.027 0.014

Zylon 500D

n21

500–1200 1200–2000 2000–2500

1.293 0.762 0.817

0.124 0.13 0.443

Zylon 1500D

n12

500–1200 1200–2000 2000–2500

0.503 0.215 0.084

0.15 0.188 0.255

Zylon 1500D

n21

500–1200 1200–2000 2000–2500

0.525 0.147 0.093

0.122 0.098 0.079

actuators). Imposing a displacement at the bottom end of the trellis fixture causes the rotation of the hinges and the initial square shape changes into a rhombus. The geometry can be calibrated such that the imposed displacement is converted to the change in the angle of rhombus. Tests were conducted in a 22 kips (98 kN) servohydraulic test frame operated under closed-loop control. The test procedure was a displacement control test with the rate of displacement of actuator (stroke) set at 0.1 in./min (0.254 cm/min). Digital data acquisition was used to collect data at every 0.5 s. The test was continued until there was no perceptible increase in deformation of the sample. The overall deformation of the specimen was measured by the stroke movement of the actuator. Each edge of the rhombus consists of two plates that clamp the fabric in between. The fabric is held in the frame by gripping mechanism similar to the tension tests V-notch grips. The multiplier links increase testing rates roughly two and half times the cross head speed. Slight tension variations due to this procedure did not significantly affect the results. The center-to-center distance

4500 K-1

Z2-1

K-2

4000

Z2-2

K-3

Stress, MPa

strain is seen prior to the movement and alignment of the Zylon yarns when loaded in the warp direction. The stress–strain responses for Zylon 1500D are presented in Fig. 4(a) and (b). The magnitudes obtained for Young’s Modulus E11 and E22 are 21,612 ksi (149 GPa) and 21,322 ksi (147 GPa), respectively. These results show that Zylon 500D is fundamentally different in its warp and fill properties as compared to the other two systems. The initial nonlinear behavior is much more pronounced in the prepeak region of Zylon 500D (Fig. 3) as compared to the Kevlar, or the Zylon 1500D (Fig. 4). This indicates that the crimp is more a function of the architecture of the weave than the material type. The ultimate tensile failure of all the specimens is brittle in nature with little or no post peak response. The higher the weave density, the higher the degree of crimp and the two systems are different in the context of their elastic modulus. The modulus of the Zylon 500D is 19,001 ksi (131 GPa) that is as much as 16% lower than Zylon 1500D (22,047 ksi or 152 GPa). The average values of tensile strength, tensile stiffness, and the ultimate strain are compared in Fig. 5. Kevlar and Zylon 1500D have properties which are quite similar in both the directions. The highest degree of difference in directionality is observed for Zylon 500D. The measured material properties for the three fabrics are shown in Table 1. The energy absorbed by the fabric can be calculated from the total area under the stress– strain curve. There is a relatively small amount of energy dissipated in the initial straightening of the yarns corresponding to 3% and 7% of the total fracture energy required for the fracture of the samples. Fig. 6 represents the range of values for the apparent Poisson’s ratio as a function of applied stress level. The values are plotted at the average level of applied stress for which they were measured. It should be noted that the Poisson’s effect observed in these tests is highly nonlinear. Initially the lateral response is primarily due to yarn realignment (crimp and slack effects) and this measurement should not be interpreted as the mechanical response. There is a gradual decrease of the Poisson’s effect as the applied stress is increased. The Poisson’s ratio n12 starts at about 1.8 for Kevlar during the initial loading stages and then reduces to 0.7. Prior to the peak, the Poisson’s ratio approaches a much lower value of about 0.6. When tested in the fill direction, the Poisson’s ratio n21 starts at about 0.6 for Kevlar during the initial loading stages and then reduces to 0.22. Prior to the peak, the Poisson’s ratio approaches a very low value of about 0.087 indicating that the majority of the loading is taken by uniaxial tension, and no contraction of the yarns is observed at the final stages prior to the failure of the swatch. The highest Poisson’s effect is observed at the low stress ranges for Zylon at around 0.7 for 500D and 0.5 for 1500D. This level, however, decreases to about 0.2 in the intermediate stages and 0.09 at the highest stress levels. Parameters n12 were as much as twice the values of n21 for a majority of the samples tested. It should be noted that the reported values represent the macroscopic representation of the fabric properties. The results are summarized in Table 2. A comprehensive graph showing all the stress–strain curves for all the fabrics tested is shown in Fig. 7.

5

3500

K-4

3000

K-5 Z1-1

2500

Z1-3

2000

Z1-5

Z2-3 Z2-4 Z2-5

Z1-2 Z1-4

1500 1000 K - Kevlar Z1 - Zylon AS-500 Z2 - Zylon AS-1500

500 0

0

0.02

0.04

0.06

0.08

Strain, mm/mm Fig. 7. Stress–strain curves for all three fabrics.


ARTICLE IN PRESS 6


Fig. 8. Trellising picture frame test.

between the bearings on a single multiplier link is 2.75 in. (6.99 cm). Before the start of the test, the multiplier links are at 90 to each other. The long steel plate used behind the shear frame holds the shear frame in its original position (fabric is held at 45 direction to the loading direction). The slot in the plate facilitates the movement of the shear frame in the loading direction. The top cross head mount remains stationary while the bottom cross head mount moves at the rate of 0.1 in./min (0.254 cm/min). The shear angle may be defined as the angle through which the fabric has sheared. The multiplier link movement induces upward movement of the clamping plates in the machined slot. Initially, in the stationary position, the links are at 90 to each other. During the testing of a fabric, downward movement of the actuator results in increase in the distance between the vertical opposite bearings connecting the multiplier links. The top bearing is connected to the two bottom clamping plates and bottom bearing is connected to the lowermost multiplier links and the assembly that clamps into

the bottom hydraulic grips of the test system. Fig. 9 shows the multiplier links before and after deformation. Let L be the center-to-center distance between the bearings attached on the multiplier links along the links. Assuming that the links are much more rigid than the specimen, this distance remains constant throughout the period of the experiment. In the initial position, the diagonal distance between the bearings is O2L and during the course of the experiment the cross head is displaced by an amount d. Using the geometry of the loading, the change in the obtuse shear angle between the multiplier links, and the engineering shear strain is determined as follows. 1

g12 ¼ cos

! pffiffiffi 2 p 2 2Ld d 2 2L2

(1)

The total shear force of the test fabric in the clamping plate can be related to the load recorded by the machine. By defining a shear

L

Multiplier Links Initial Position

γ/2 θ

L

δ

Multiplier Links Position After Deformation

Fig. 9. Typical deformation of multiplier links and the free-body diagram for the picture frame.



7

Fig. 10. Kevlar test fabric during (a) initial stage of loading and (b) final stage of loading.

force F, actuator load P, length of the fabric in the clamping plate as S, and the thickness of the fabric as t, then

F ¼

q P sec 2 2

(2)

where q ¼ ðp=2Þ g. The nominal shear stress can be computed as follows.

s ¼

F tS

(3)

The test results can be used to plot actuator load versus actuator displacement and actuator load versus the fabric shear angle. The typical shear response shows an initial region with large increase in shear angle with minimal increase in the actuator load. During this initial phase, the yarns begin to rotate offering a small resistance to the applied shear loading. After this phase, the fabric load tends to increase rapidly. In this phase, it can be assumed that the yarns compress each other laterally at the crossover points. Finally, the ending part of load-shear angle curve shows there is a rapid increase in load with minimal increase in the displacement. During

100 Kevlar 49 Zylon AS-500 Zylon AS-1500

Shear Stress, ksi

80

60

40

20

0 0

0.4

0.8

Shear Strain, Radians Fig. 11. Shear stress versus shear strain comparison.

1.2

this ending stage, the trellis frame is subjected to the entire load and the experiment is stopped. Fig. 10(a) shows the Kevlar test in the initial stage of loading while Fig. 10(b) shows the sample in the final stage of loading. Fig. 10(b) indicates that the fabric buckles in directions aligned with the compression field during the final stages of loading. The shear stress–strain plots for the three fabrics are shown in Fig. 11. It should be noted that the initial portion of the plot contains the material behavior whereas, once the fabric buckles, the geometric nonlinear effect is captured in the plot. Fig. 12 shows the locking mechanisms of the Zylon fabric subjected to shear loading. Note that after the contact between the warp and fill yarns are made, the pinching of the two yarns contacting at a point results in contact forces to be transferred from one direction to the other direction. As we will see the second part of this paper, the shear stress–strain plot needs to be modified to be used correctly in the constitutive model.

2.3. Friction tests A series of friction tests to calculate the coefficients of static and dynamic friction were conducted on the three fabrics. These tests were conducted on a 55 kips (245 kN) horizontal test frame. The test setup is shown in Fig. 13. During the experiment, a layer of fabric was pulled using the horizontal actuator. This fabric layer was sandwiched between two layers of the same fabric. Normal loads were applied on the fabric such that the maximum normal load was 800 lb (3.56 N) for a contact area of 13.75 in.2 (88.7 cm2) for an effective pressure of 60 psi (0.41 MPa). These normal loads were applied through another actuator mounted vertically on an I-beam resting on two channel sections connected to the four columns. The second layer of fabric was allowed to move using zinc ball joint rod ends that were fixed to an I-beam. This beam was attached to two column sections. The coefficient of friction was computed by applying normal loads at loading rates of 2.0 in./min (5.1 cm/min) and 6.0 in./ min (15.2 cm/min), and was computed by plotting the maximum pull for each normal load against the respective normal loads. The coefficient of dynamic friction was computed by plotting the average pull for each normal load against the respective normal loads. A typical output from a friction test is shown in Fig. 14. Fig. 15 shows the force displacement response of nine Kevlar samples for a loading rate of 6.0 in./min (15.2 cm/min). Fig. 16 shows the comparison of the static and dynamic coefficient of friction obtained through the Kevlar friction tests for the two different loading rates. Note that the ranges are not significant and the response for a range of applied nominal stresses are linear. The slopes of these curves are shown in the figure and indicate static and dynamic coefficients of friction of 0.22–0.23 for the range of


ARTICLE IN PRESS 8


Fig. 12. SEM images of Zylon fabric during shear loading.

applied nominal normal stresses. Results of these tests are summarized in Table 3.

3. Ballistic impact tests Ballistic impact tests were conducted on ring-shaped specimens made up of layers of dry fabric. In these tests the impact energy was held constant, with the exception of a small number of tests, while the number of layers of fabric was varied. The fabric was wound around a circular fixture placed in front of a gas gun at a slight incline such that the projectile exited the gun barrel, passed over the leading edge of the ring and impacted the fabric from the inside. Impact tests were conducted on both KevlarÒ49 and ZylonÒ. The yarn and weave parameters of the materials used in this study are shown in Table 4. The projectile used in the impact tests was a flat rectangle-shaped piece of 304L stainless steel, 4.0 in. (10.2 cm) long, 2.0 in. (5.1 cm) high and 5/16 in. (0.8 cm) thick (Fig. 17) with a nominal mass of 320 g. The front edge of the projectile was machined with a full radius. It exited the gun barrel in such a way that the long dimension of the projectile was in the direction of travel, the height dimension was vertical and the thickness dimension was side to side.

The intended projectile velocity was approximately 900 ft/s (275 m/s) except in a few tests involving one or two layers of fabric, in which case the impact velocity was approximately 328 ft/s (100 m/s). Actual impact velocities were measured using a high speed digital video camera located above the target. Details of the gas gun and techniques used in accelerating the projectile are given in [3]. The fixture used to hold the fabric was a 1 in. (2.5 cm) thick metal ring with a 10 in. (25 cm) height and a diameter of 40 in. (102 cm). A 10 in. (25 cm) wide fabric strip was rolled around the ring under controlled tension to make up the desired number of layers. The ring had a 10 in. (25.4 cm) opening and was placed in front of the gun barrel at a 15 incline such that the projectile, after exiting the gun barrel passed over the front edge of the ring, passed through the opening in the fixture and impacted the fabric from the general direction of the center of the ring. The test fixture and specimen are illustrated in Figs. 18 and 19. Each fabric specimen consisted of a continuous 10 in. (25 cm) wide fabric strip wrapped around the test fixture under controlled tension to produce the

FrictionTests Static ( Bond) Friction Dynamic Friction

Force

22 Kip Actuator

F = 2*τ*W*L

Wood Plate

55 Kip Actuator

Linear roller Bearings Fig. 13. Schematic setup for friction test.

Slip Fig. 14. Typical friction test output.



Table 3 Comparison of friction coefficients for Kevlar, Zylon AS-500D and Zylon AS-1500D

400 58 psi

Force, lbf

300

Kevlar Friction Test Loading rate 0.1"/s Contact Area = 5.5 x 2.5 in2

200

100

14 psi 9 psi 6 psi 3 psi 0

2

4

6

Slip, in Fig. 15. Kevlar friction tests for loading rate of 6.0 in./min.

desired number of layers. The beginning and end of the continuous strip were held with an epoxy adhesive and were located 180 away from the impact location. The tension in the fabric was controlled to be 5.5 lb (24.5 N). High speed digital video cameras (Phantom 5, Vision Research, Inc., Wayne, NJ) were used to record the position and orientation of the projectile during the experiment. For each test the position of two or more points on the projectile was recorded as a function of time. The impact velocity and residual velocity (velocity after perforating the fabric) were determined by fitting a straight line to the position data, while in free flight before and after impact, and averaging the slopes of the resulting lines.

400

d=2"/min μs=0.233 μd=0.228

Maximum Pull Force, lbf

d=6"/min μs=0.221 μd=0.213 300 Kevlar Static and Dynamic Friction

Loading rate (in./min)

Coefficient of static friction

Coefficient of dynamic friction

Kevlar

2.0 6.0

0.233 0.221

0.228 0.213

Zylon 500D

2.0 6.0

0.188 0.194

0.183 0.171

Zylon 1500D

2.0 6.0

0.159 0.184

0.159 0.184

Twenty-nine impact tests were conducted, 14 on KevlarÒ 49, nine on 500D ZylonÒ material and six on 1500D Zylon. Table 5 summarizes the results of the test program. In all tests, except tests LG407, LG414 and LG428, the projectile perforated the fabric specimen. The failure was generally along the line defined by the leading edge of the projectile. In the initial plies, the failure was highly localized along this line, and to the naked eye resembled a cut in the fabric. As the projectile progressed through the layers the failure point remained generally along the same line, but there was significant fraying at the ends of the failed yarns. The fraying is indicative of individual filaments within the yarn failing at different locations. The same phenomena existed in cases where full perforation did not occur. In these cases it was clear that failure initiated at the corners of the projectile. There were several plies where there were holes at the corner locations while the material in between remained intact. Progressing from the outer layers to the inner layers, the holes grew in size until there was failure across the total leading edge region. The projectile impact and residual velocity data are shown in Table 5. As can be seen from the table, in all but three tests, the projectile perforated the fabric specimen. Fig. 20 shows the energy absorbed, normalized by the areal mass of the specimen, as a function of the number of fabric layers in each test. The arrows on selected symbols indicate that in these tests the projectile did not penetrate the specimen (all of the kinetic energy was absorbed) and more energy could have been absorbed. The areal mass of the specimen is defined as the areal mass per layer, in grams per square centimeter, times the number of layers. It is clear from Fig. 20 that for a given weight and under these impact conditions Zylon is able to absorb significantly more energy than Kevlar, and the heavier Zylon is more effective than the lighter version of the same material. The heavier weight Zylon and the Kevlar material were very similar in areal weight, yarn count, ply thickness and yarn denier. Fig. 20 illustrates that the normalized absorbed energy is relatively insensitive to the number of layers of material. For the Kevlar material the average normalized absorbed

Table 4 Fabric properties

200

Static, δ = 2.0 "/min

100

Static δ = 6.0 in/min Dynamic δ = 2.0 in/min Dynamic δ = 6.0 in/min

0

Fabric

38 psi

24 psi

0

9

0

200

400

600

800

Normal Load, lbf Fig. 16. Coefficient of friction for Kevlar for loading rate of 2.0 in./min (5.1 cm/min) and 6.0 in./min (15.2 cm/min).

Volume density Yarn Denier (measured) Yarn linear density Yarn count Yarn count Fabric ply thickness Fabric areal density Degree of Crimp Warp Yarns Degree of Crimp Fill Yarns

(g/cm3) (g/9km) (mg/cm) (yarns/in.) (yarns/cm) (mm) (g/cm2) (%) (%)

Zylon AS Poly-benzobisoxazol (PBO)

Kevlar-49 P-Aramid

500D

1500D

Standard

1.54 500 0.556 35 35 13.8 13.8 0.21 0.01575 3.1

1.54 1500 1.654 17 17 6.7 6.7 0.28 0.0223 2.2

1.44 1490 1.656 17 17 6.7 6.7 0.28 0.02275 1.1

0.6

0.9

0.8


ARTICLE IN PRESS 10


Fig. 17. 304L stainless steel projectile.

energy is 13.5 KJ-g/cm2. For the lighter weight Zylon this value is 22.9 KJ-g/cm2, and for the heavier weight Zylon the value is 38.9 KJg/cm2. From a practical point of view, this means that for the same weight of material, the thick Zylon can absorb almost three times as much energy than the Kevlar material under the conditions of this test.

4. Concluding remarks The first part of this two-part paper addresses the determination of material properties and failure characteristics of dry fabrics using laboratory testing. These include swatch Tension Tests both the warp and fill directions to develop the orthotropic stress–strain relationships, trellising Picture Frame shear tests to the shear resistance properties of the fabrics, followed by Friction Tests to determine the coefficient of friction between fabric layers. Kevlar fabric shows almost equal modulus of elasticity and tensile strength in both the warp and fill directions. However, the amount of crimp in the warp direction is much more uniform than in the fill direction. Similarly, for both Zylon 500D and Zylon 1500D fabrics, the modulus of elasticity are nearly the same in both warp and fill directions. A detailed analysis of the test results show that Zylon 500D is

a Fabric-wrapped steel ring

fundamentally different in its warp and fill properties as compared to the other two systems. The initial nonlinear behavior is much more pronounced in the pre-peak region of Zylon 500D as compared to the Kevlar, or the Zylon 1500D. This indicates that the crimp is more a function of the architecture of the weave than the material type. The Tension Tests were also used to obtain the Poisson’s ratio for the fabric swatch. However, we feel that the computed values are more due to geometric effects (change in the fabric architecture) than are fundamental material properties. The shear tests were used to obtain the shear response of the fabrics. The typical shear response shows an initial region with large increase in shear angle with minimal increase in the actuator load. During this initial phase, the yarns begin to rotate offering a small resistance to the applied shear loading. After this phase, the fabric load tends to increase rapidly. In this phase, it can be assumed that the yarns compress each other laterally at the crossover points. The fabric buckles in directions aligned with the compression field during the final stages of loading. All the three fabrics exhibited similar behavior with Kevlar 49 being slightly stiffer than the two Zylon fabrics. The friction tests were used to obtain the coefficient of static and dynamic friction between two Kevlar fabric layers. Additionally, Ballistic Tests of containment wraps subjected to a high velocity projectile were carried out. Test results show that for a given weight and under these impact conditions Zylon is able to absorb significantly more energy than Kevlar, and the heavier Zylon is more effective than the lighter version of the same material. From

Projectile

Gun Barrel

85°

Elevation

b

Cutout in steel ring exposing fabric Projectile

Gun Barrel

Fabric

Plan view

Fig. 18. Schematic of test setup showing (a) elevation and (b) plan view.

Fig. 19. Test fixture with Kevlar specimen in place.


ARTICLE IN PRESS D. Naik et al. / International Journal of Impact Engineering xxx (2008) 1–11 Table 5 Projectile impact and residual velocity

11

Test

Material

Number of layers

Projectile mass (g)

Impact velocity (m/s)

Exit velocity (m/s)

LG403 LG404 LG405 LG409 LG410 LG411 LG424 LG427 LG429 LG432 LG433 LG434 LG444 LG449 LG406 LG408 LG412 LG413 LG417 LG425 LG426 LG407 LG414 LG420 LG421 LG422 LG423 LG430 LG428

Kevlar Kevlar Kevlar Kevlar Kevlar Kevlar Kevlar Kevlar Kevlar Kevlar Kevlar Kevlar Kevlar Kevlar Zylon 500D Zylon 500D Zylon 500D Zylon 500D Zylon 500D Zylon 500D Zylon 500D Zylon 500D Zylon 500D Zylon 1500D Zylon 1500D Zylon 1500D Zylon 1500D Zylon 1500D Zylon 1500D

4 8 24 8 4 24 8 24 16 16 1 1 2 2 4 8 4 8 8 8 16 24 24 8 8 4 4 12 16

318.4 317.8 319.0 316.0 316.4 314.8 320.9 317.9 316.2 320.0 316.7 315.9 316.4 316.2 319.5 318.0 318.4 319.9 314.6 316.6 316.8 316.1 315.9 316.3 317.6 315.8 315.1 315.9 317.9

274 273 274 271 278 270 254 279 279 273 119 117 106 105 273 276 243 275 272 277 277 275 251 280 262 280 243 279 277

258 250 151 246 264 126 227 185 219 198 112 110 84.7 85.0 255 241 223 237 241 245 192 0 0 191 155 237 192 109 0

Energy Absorbed/Areal Weight (KJ-cm^2/g)

45 40 35

Heavy Zylon

30 25 Light Zylon

20 15 Kevlar

10 5 0 0

5

10

15

20

25

30

Number of Layers Fig. 20. Normalized energy absorbed as a function of number of fabric layers ([ denotes projectile was contained).

Acknowledgements The authors wish to thank Mr. William Emmerling, Mr. Donald Altobelli, Mr. Chip Queitzsch and Mr. Jim White of the FAA for their technical and financial support. Assistance of other research team members from Honeywell Engines and SRI Intl. is gratefully acknowledged. References

a practical point of view, this means that for the same weight of material, the thick Zylon can absorb almost three times as much energy than the Kevlar material. Finally, these tests provide most of the information for verification of the finite element methodology that is discussed in the second part of this paper. It is expected that more laboratory testing may be needed to improve the modeling capabilities. For example, to facilitate modeling capability for multiple layers of fabric one must better characterize the frictional sliding at various scales [10]. There are three different levels of frictional sliding that are operative – (a) filament to filament within a yarn, defined as inter-yarn sliding effects, (b) sliding of single warp or fill yarns within a fabric defined as the interfabric yarn friction effects, and (c) fabric layer vs. fabric layer, defined as interlayer friction effects. These three frictional mechanisms to a large degree control the response of the overall structural system. In the present work only fabric-to-fabric frictional sliding is discussed and tested. With high speed impacts, strain rate effects play an important role. Efforts are under way to characterize this behavior for various fabric materials.

[1] Sharda J, Deenadayalu C, Mobasher B, Rajan SD. Modeling of multi-layer composite fabrics for gas turbine engine containment systems. ASCE J Aerosp Eng 2006;19(1):38–45. [2] DOT/FAA/AR-04/40, P1–P4. Explicit FEA modeling of multi-layer composite fabric for gas turbine engines containment systems. Federal Aviation Administration; 2005. [3] Pereira JM, Revilock DM. Explicit finite element modeling of multilayer composite fabric for gas turbine engine containment systems. Part 2: ballistic impact testing. DOT/FAA/AR-04/40, P2; 2004. [4] Stahlecker Z, Mobasher B, Rajan SD, Pereira JM. Development of reliable modeling methodologies for engine fan blade out containment analysis. Part II: finite element analysis; Int J Impact Eng, in press. [5] Cheng M, Chen W, Weerasooriya T. Mechanical properties of Kevlar KM2 single fiber. J Mater Eng Technol 2005;127(2):197–203. [6] Chen J, Lussier DS, Cao J, Peng XQ. Materials characterization methods and material models for stamping of plain woven composites. Int J Forming Process 2004;4-3:4. [7] Raftenberg MN, Scheidler MJ, Moy P. Transverse compression response of a multi-ply Kevlar vest. Army Research Lab Publication; 2004. ARL-TR-3343. [8] Cork CR, Foster PW. The ballistic performance of narrow fabrics. Int J Impact Eng 2007;34:495–508. [9] ASTM D 3039-standard test method for tensile properties of fiber–resin composites. ASTM International; 2006. [10] Duana Y, Keefeb M, Bogettic TA, Cheeseman BA. Modeling friction effects on the ballistic impact behavior of a single-ply high-strength fabric. Int J Impact Eng 2005;31:996–1012.
