Energy Absortion By Kevlonorepoxy

Testing Methods for Energy Absorption of Kevlar/Epoxy Dean D. Dubey* and Anthony J. Vizzini† Gessow Center for Rotorcraft Department of Aerospace Engineering University of Maryland College Park, Maryland 20742

Abstract Twenty flat-plate specimens and six tubes were crushed under quasi-static conditions. The energy absorbency of each specimen was measured resulting in a value of the specific sustained crushing stress (SSCS). The plates and tubes were manufactured from 49/CYCOM 919 Kevlar/epoxy fabric and had the same layup providing a common laminate for comparison. The flat plates were scored to facilitate the fracture of the laminate along the constraints during crushing. Two different widths of flat plates were tested to determine the effect of the end conditions of the testing fixture on the observed energy absorbency. The results for the flat-plate specimens indicate that the energy absorbency per unit thickness is independent of the specimen test width while maintaining the crushing mechanism. Thus, the edges of the testing fixture provided no significant contribution to the energy absorbency. Furthermore, the results indicate that the flat-plate geometry provides comparable energy absorption values to those of the tube geometry.

Introduction Safe operation of aircraft and rotorcraft requires that occupants survive crash events at various attitudes and velocities. Crew areas must not become lethal. A liveable volume must be maintained, post-crash fires must not occur, and deceleration of the occupants must be limited to prevent serious injury or death. Typically, a systems approach is taken to decelerate the occupants. Initially, the landing gear stroke and then deform. Concurrently, the seat strokes. In extreme conditions, the subfloor of the helicopter is sacrificed as a result of being crushed. The Advanced Composite Airframe Program (ACAP) demonstrated that composite materials are an effective part of a crashworthy design [1]. High energy absorbency per unit mass is possible with composite materials if proper failure mechanisms are initiated and maintained during the crash event. Whereas metals absorb energy primarily through plastic deformation, composite materials absorb energy through a variety of failure mechanisms. For example, Kevlar reinforced composites absorb energy through a buckling failure mechanism similar to the accordion buckling modes of metal structures. Graphite- and glass-reinforced composites absorb energy through successive failures involving delamination, intraply cracking, and fiber fracture. Because energy absorbency of a composite structure is directly dependent on the failure mode that occurs and the failure mode is a function of the laminate stacking sequence, the loading history and environment, proper characterization ought to include off-axis crush tests [2]. * †

Graduate Research Assistant Associate Professor, member AHS

1

An initial geometry used by researchers to study the energy absorption capabilities of composite materials was the tube. This geometry is self stabilizing and allows testing of relatively thin-section laminates. The lack of edges along its length reduces the complexity of the boundary conditions and provides consistency throughout the cross section [2–4]. Using this specimen geometry, Farley observed various energy absorbent failure modes [3]. Moreover, he determined the influence of such factors as the friction of the crushing surface, strain rate, material, and stacking sequence. These tests, however, were limited to uniaxial crushing. Fleming and Vizzini used truncated cone specimens to investigate the effect of side loads and specimen taper on the energy absorbency of composite graphite structures [2]; Knack and Vizzini investigated these same issues for truncated Kevlar cones [5]. The tests on the truncated cones demonstrated the effects of side loads; however, the cone geometry resulted in a combination of failure modes which made further analysis difficult. Jackson, et al developed a fixture to crush flat-plate specimens [6]. The fixture developed stabilized the flat plates during crushing to avoid plate buckling. Flat-plate specimens are easier to manufacture and are less expensive than tubes. Fleming and Vizzini adapted this fixture type to allow the crushing of flat plates at specified angles of incidence [7]. The flat-plate specimens indicated a performance degradation due to off-axis loading conditions. Although the tube specimen with its finite D/t ratio may be used to indicate the potential energy absorption of a given laminate, it remains a structural test. The diameter to thickness ratio, D/t, of composite tube specimens affects the energy absorption capability of laminates. For thin tubes, corresponding to large D/t, specimen crushing can become unstable leading to specimen collapse [8]. Even for stable crushing situations, D/t may affect the measured energy absorbency of composite specimens. Typically, Kevlar/epoxy composite specimens have a bilinear relation between energy absorption and D/t, while graphite/epoxy specimens have a non-linear relation between energy absorption and D/t [9]. An increase in energy absorption was observed in general as D/t decreased, reaching as its limit the compressive yield strength of the material [10]. This increase in energy absorbing ability is caused by the reduction in interlaminar cracking, which reduces the characteristic damage length of the fibers. Thus, a minimum of bending occurs and more of the strength of the material is realized. These conclusions would indicate that the tube geometry should provide superior energy absorption capability to the flat-plate geometry which has an infinite D/t. Because resistance to bending is presumed to be the cause of the variance of energy absorbency to D/t, then a flat plate that is stabilized by means of a fixture may reduce this observed difference in measured energy absorbency. Researchers have shown that testing of flat-plates for energy absorption can be accomplished when they are stabilized in a fixture [6, 7, 11]. Moreover, the simplicity of the manufacture of the flat plate provides a low-cost alternative method to observe crushing phenomenon. Under off-axis loading, flat-plates experience the same load across the crush plane whereas a tube specimen experiences a varying load around its circumference. It is desirable to be able to correlate test results from flat-plate specimens with tube specimens. Although previous tests done with specimens manufactured from unidirectional graphite/epoxy showed that the plates readily cracked along the edge constraints [7,12], the effect of the constraint on the damage mechanisms occurring in Kevlar/epoxy specimens was undetermined. 2

To overcome these limitations, an experimental test program was conducted to first determine the effect of the fixture by testing specimens with two different widths and then to compare equivalent laminates tested as flat plates and tubes. The figure of merit for comparing energy absorbed for a specific structure is the incremental energy absorbed divided by the instantaneous mass consumed. Thus, the specific sustained crushing stress SSCS is defined as:

SSCS



P 1l P =  2Energy = = V olume A1l A

(1)

where P is the average crushing load, A is the cross sectional area, and  is the density of the material. The stacking sequence, (+45/–45/0/90)S , was used for the 49/CYCOM 919 Kevlar/epoxy laminates. This stacking sequence was chosen for similarity with previous studies.

Experimental Test Matrix A total of 26 (+45/–45/0/90)S specimens with various geometries were manufactured from 49/CYCOM 919 Kevlar/epoxy bidirectional preimpregnated fabric. Of these, 20 were flat-plate specimens and six were tubes. Of the flat plates, ten were machined to 165 mm 2 89 mm and ten were machined to 165 mm 2 64 mm. The tubes were 102 mm in height with an inner diameter of 114 mm. A steeple chamfer was machined into one end of each specimen to initiate damage. A comparison is to be made first between flat-plate specimens of different widths. By making this comparison, a SSCS per unit width and the energy associated with cracking along the edges of the specimen at the constraints can be determined. Next, equivalent laminates with two different geometries will be compared. Finally, the present testing method can be evaluated for broader application by including data of graphite/epoxy specimens that were tested in the same manner [12].

Manufacturing The flat-plate specimens were machined from larger 305 mm 2 356 mm laminates with a stacking sequence of (+45/–45/0/90)S . The laminates were made from 49/CYCOM 919 Kevlar/epoxy bidirectional cloth. The tube specimens were machined from two longer tubes with an inner diameter of 114 mm. Tubes were laid up and cured on aluminum mandrels. The manufacturer’s recommended cure cycle was followed for both the tubes and the laminates. It consisted of 1 h at 121  C with a total pressure of 621 kPa. The flat-plate laminates and the tubes were then machined into individual specimens. A milling machine with a water-cooled diamond-grit cutting blade was used. From the three plates, 20 flat-plate specimens were machined with one plate yielding six 165 mm 2 89 mm specimens, another plate yielding eight 165 mm 2 64 mm specimens, and the third yielding four 165 mm 2 89 mm specimens and two 165 mm 2 64 mm specimens. From the two tubes, six tube specimens were cut to a nominal length of 102 mm. The edge of each of the flat-plate specimens that was towards the original center of the laminate was chamfered. Also the tubes were chamfered along edges that were originally internal. These internal edges were chosen to initiate crushing in consistent regions of the specimens. 3

To chamfer the flat-plate specimens, they were held in a fixture on a milling machine and a two-flute titanium-nitride end mill was used to machine a steeple chamfer of approximately 34 as shown in Figure 1. At least two cutting passes were made per side with the tool to remove the material. Additional passes were made until the peak of the steeple chamfers was deemed acceptable. The end mill machined the Kevlar/epoxy without fraying the ends as commonly occurs. To chamfer the tube specimens, they were held with a circular indexing head so that their central axes were parallel to the milling machine table, as shown in Figure 2. Chamfering was accomplished by positioning the tool to chamfer a certain thickness and then rotating the composite tube around its central axis by rotating the cylindrical indexing head. Dimensional variations in the tube specimens resulted in a less-precise machining of the chamfer than was accomplished with the flat-plate specimens. The length, width, thickness, and mass of the flat-plate specimens and the length, diameter, thickness, and mass of the tube specimens were measured and averaged. These values were used to determine the densities of the specimens. Measurements are provided in Table 1. The low coefficients of variation (CVs) indicate the relative uniformity of the specimens.

Experimental Testing All specimens were tested on a 220 kip hydraulic testing machine. The flat-plate specimens were crushed using the fixture shown in Figures 3 and 4 [7]. This portion of the fixture in Figure 3 is an extension of the hydraulic grip and allows for the testing of specimens of different thicknesses with alteration of only the restraining plate. The edge constraints move with the specimen eliminating that source of friction. One drawback of this fixture is that the specimen must split along its length at the constraints provided by the guide rods as shown in the photograph of tested flat-plate specimens (Figure 5). The ruler in the photograph is 152 mm long. The center portion of the flat-plate specimen is crushed against the crush plane, illustrated in Figure 4, while the edges are gripped by the support rods to stabilize the specimen and pass by both sides of the crush plane. The crush plane is positioned between the bottom edge of the specimen and the rod tip support plate while resting on the bottom portion of the test fixture. The fixture was originally developed to crush flat specimens at an angle of incidence, , as shown in Figure 4. In the present test program, the load incidence angle is set to 0 . Two separate fixtures were used, one for each specimen width. The fixture used for the 64-mm-wide specimens was the same as that used in Reference 7. It has a gap of 38.1 mm between the edge constraints. The fixture used for the 89-mm-wide specimens has a gap of 63.5 mm between the edge constraints. The edge constraints are 12.7 mm diameter hardened steel. During the course of testing, the four rods of the fixture for the testing of 89-mm-wide flat-plate specimens were strain gaged to determine to what extent the axial load was being carried through the fixture. The gages were oriented longitudinally midway along the length of the rods opposite the point of contact with the specimen. The tubes were crushed between self-aligning platens. All of the specimens were crushed at a constant stroke rate of 0.0635 mm/s. Load, stroke, and strain data were recorded every 0.5 s using a computer data acquisition system. During initial testing of the flat-plate specimens, cracking occurred along the constraints of the test fixture and the sides of the crush plane. The Kevlar/epoxy flat-plate specimens were 4

resistant to tearing along the constraints and material bound between the support rods and along the crush plane resulting in Euler buckling of the specimens. Because of the friction caused by the binding, the measured load increased rapidly with stroke. To ease the tearing of the specimen along the constraint rods and to prevent the specimen material from binding within the test fixture, the specimens were altered. The flat-plate specimens were scored along the path of desired tearing adjacent to the location of the constraint rods. Scores were machined into the specimens with a jeweler’s blade of thickness 0.154 mm. Scores were machined into both sides of the flat-plate specimens and their approximate locations are shown in Figure 6. The distance between the score lines was set to the width of the crush plane. Eight of the 64-mm-wide specimens and all 10 of the 89-mm-wide specimens were tested with scoring. Specimens were scored to varying depths from 1/5 to 2/3 of the thickness of the plate. The machining of the score lines was difficult because of variations in the thickness of the specimen. The thickness varied by up to 0.25 mm as indicated by repeated positioning of the jeweler’s blade to the surface of the specimen. An average zero was determined. Scoring and testing of specimens was done sequentially. After each test was complete, the failure mode of the tested specimen was assessed and, using this knowledge, the scoring depth of the next specimen was adjusted as necessary.

Results and Discussion The primary data obtained from a crushing test is the load versus stroke curve. A schematic representation of such a curve for a composite material is shown in Figure 7 where three separate regions are highlighted. The first region is the initial linear-elastic response of the structure terminated by the initiation of damage. The peak load divided by the cross sectional area provides the initiation stress. In the second region, damage spreads across the entire specimen and transitions into the third region which is characterized by a sustained crushing load. The load-stroke curves are inspected to determine the point at which the crushing had transitioned into a sustained mode. The total energy absorbed from that point until the end of the test or a chosen end point is then used to calculate the SSCS. For comparison purposes, the instantaneous load is converted into a specific crushing stress (SCS) by dividing it by the crushing area and the material density. In the flat-plate specimens, a transition region could not be identified. Instead the flat-plate specimens immediately began stable crushing, although the load extremes were much greater than those seen in the tube specimens. For all flat-plate specimens, a stroke of 17.8 mm was assumed to be the start of the sustained crushing region for the calculation of the SSCS. During testing of the flat-plate specimens, the specimens would bow ahead of the point of crushing of the specimen on the crush plane. At a stroke slightly greater then 50.8 mm, this bowing curvature would approach the clamped condition of the gripped portion of the specimen. Thus, the sustained crushing region was taken to end at 50.8 mm of stroke. An exception was made for the first four flat-plate specimens. An endpoint of 38.1 mm was used to calculate the SSCS because the total stroke of each of these tests was only slightly greater then 38.1 mm. The SSCS was determined by numerically integrating the area beneath the load-stroke curve within the sustained crushing region and then dividing this energy by the mass of the crushed volume 5

as prescribed by Equation 1. Table 2 includes the initiation stress and the SSCS of all the specimens tested successfully. For each group of specimens the average is provided as well as the coefficient of variation. Three failure modes were exhibited by the flat-plate specimens, and they were a function of scoring depth. One failure mode consisted of stable crushing with a constant amplitude sinusoidal buckling evident. Score depths for those specimens were 0.36 mm for the 64-mmwide specimens and 0.53 mm for the 89-mm-wide specimens. After the load reached an initial peak the specimen would bow out-of-plane above the crush plane. The chamfered edge against the crush plane would begin to move out of plane in the opposite direction of the bow. This response would be accompanied by slight “popping” sounds. As a buckle developed, the material would begin to fold back over itself. This behavior would be accompanied by occasional loud cracking noises and the appearance of white linear bands across the width of the specimen on the crest of the buckle which indicated internal delaminations. Occasionally, outer plies would delaminate along the width of the crest of the buckle. Then, the process of a new buckle forming on the other side of the specimen would begin. Successive buckles would continue to form resulting in the deformation shown in Figure 5. This failure mode is consistent with descriptions of previous testing on Kevlar/epoxy 1 cone specimens [5]. The buckles of the 89-mm-wide flat-plate specimens were of larger amplitude and wavelength then the buckles of the 64-mm-wide flat-plate specimens. This behavior can be seen by comparing the frequency of the load-stroke curves for typical 64-mm-wide flat-plate and 89-mm-wide flat-plate specimens in Figures 8 and 9, respectively. Another failure mode consisted of the specimen buckling or bending exclusively to one side during all or most of the test. This failure behavior was indicative of too great of a score depth resulting in a loss of stability. This failure mode was only seen in the 89-mm-wide flat-plate specimens and may indicate that the wider 89-mm specimens are less stable. Figure 10 shows a photograph of a specimen which exhibited this type of failure mode. The final failure mode consisted of the specimen exhibiting sinusoidal buckling, but with binding of the specimen in the test fixture resulting in a rapidly climbing load. The specimen would deform along the constraints and the material within the constraints that had been stroked beyond the crush plane would bend towards the center of the plate. Figure 11 shows a photograph showing both a 64-mm-wide flat-plate and a 89-mm-wide flat-plate specimen which experienced this binding failure mode. This failure mode was also occasionally accompanied by buckling along the constraints; these were most predominantly visible on 64-mm-wide flat-plate specimens. Figure 12 shows a photograph illustrating the buckling of a 64-mm-wide flat-plate specimen. This failure mode was indicative of a shallow score depth. A typical load-stroke plot for a flatplate specimen that experienced this binding failure mode is shown in Figure 13. The SSCS depicted on the plot was calculated using the same sustained crushing region. Friction from the test fixture at the crushing plane increases this measured value substantially. Thus, for this type of failure mode, the SSCS value cannot be applied as a performance indication. Evaluation of the strain gage data from the rods on the 89-mm-wide flat-plate test fixture showed only minor strain in the top portion of the test fixture. The strain versus stroke plots for typical tests with the bending, sinusoidal buckling, and specimen binding failure modes are 6

shown respectively in Figures 14, 15, and 16. The strain readings are greatest for the specimen binding failure modes case and indicated that the specimen was pushing against the constraint rods. This force resulted in a bending moment on the constraint rods. Loads applied by the specimens to generate the maximum strains over the entire stroke for each failure mode are given in Table 3. The loads were calculated by modeling the constraint rods as clamped-clamped beams with a point loading acting normal to the constraint rods at the gage location. The loads indicated are an order of magnitude less than the sustained crushing loads (about 6000N). This indicates a low value of loading of the fixture during testing. The effect of the score depth was substantial. A survey of SCS versus stroke curves is shown in Figure 17. Different score depths from zero to 0.71 mm illustrate the variety of failure mechanisms from binding to sustained crushing to bending. The effect on the measured SSCS value corresponds to the failure mode as shown in Figure 18. The three failure modes were easily identified and the specimens that exhibited either bending or binding within the test fixture are excluded in further discussions. For the 64-mm-wide flat-plate specimens a total score depth of approximately 1/2 of the specimen thickness (0.53 mm on each side) resulted in a successful test. For the 89-mm-wide flat-plate specimens a total score depth of approximately 1/3 of the specimen thickness (0.36 mm on each side) resulted in a successful test. The variation of failure modes in the 89–mm-wide specimens at a score depth of approximately 0.36 mm is due to the difference between the nominal score depth and the as-machined depth which could not be measured over the entire specimen. Out of eight 64-mm-wide specimens, four were deemed to have failed in an acceptable manner. Of the ten 89-mm-wide specimens, four were also deemed acceptable. From the SSCS data presented in Table 2, the values for the 64-mm-wide and 89-mm-wide flat-plate specimens are nearly equal. This would indicate that whatever the contribution of the constraints is, it is small and can be neglected. This is only possible because of the dominant crushing mechanism is maintained within the widths tested. Note also, that this similar behavior is achieved only after effectively scoring the specimens to facilitate the fracture at the constraints. The crushing failure of the 114-mm-diameter tube specimens was consistent with that of the flat-plate specimens that failed in the preferred failure mode. Sinusoidal buckling failure modes were exhibited around the circumference. Figure 19 shows a photograph of a typical crushed tube specimen showing sinusoidal buckling. The width of the buckles was generally a small fraction of the total circumference and the buckles were generally not in phase. Thus, the SCS versus stroke plot shown in Figure 20 does not exhibit the same sinusoidal pattern evident in the successfully tested flat-plate specimens. Instead, during steady crushing the load fluctuation is small. Comparison of the SSCS for the flat-plate and tube specimens is made in Table 4. Also shown are SSCS values for previously tested graphite/epoxy specimens of the same geometry from Reference 12. The flat-plate and tube SSCS values for the Kevlar/epoxy specimens are within 2% of each other, while the graphite/epoxy specimens are within 10% of each other. Although the SSCS is reported to be a function of D/t for both graphite/epoxy and Kevlar/epoxy laminates [9], the effect in the Kevlar/epoxy specimens appears to be minimal. The corresponding D/t ratios are 52.7 and 47.4 for the Kevlar/epoxy and graphite/epoxy specimens. Another comparison can be made with previous work done by Knack [5]. Knack tested 1 7

tapered Kevlar/epoxy cones of stacking sequence [+45/-45/0]S manufactured from preimpregnated tape. Tested at a load inclination of 0 , the SSCS values averaged 40.5 kN-m/kg. This value is almost 20% greater then the value obtained for the tubes of the present study. The difference in the SSCS values between the two studies can be attributed to layup differences and thickness effects. One-third of the cone laminates were 0 plies, while one-fourth of the tube laminates were 0 plies. The greater percentage of the 0 plies in the cones helps account for the differences in SSCS between the two studies. Also, as noted in Reference 6, the SSCS values decreased 10–25% for graphite-Kevlar plate specimens when thicknesses were scaled by a factor of two. The Knack Kevlar/epoxy specimens consisted of six plies of preimpregnated tape, while the present specimens consisted of eight plies of preimpregnated cloth. The Knack specimens were substantially thinner than those of the present study. Therefore, if the scaling results for the graphite-Kevlar plates holds true for the Kevlar/epoxy tubes and cones of the present and Knack’s study, then the differences in SSCS values can be partially explained by the differences in thicknesses of the laminates.

Conclusions Flat-plate and tube specimens manufactured of Kevlar/epoxy preimpregnated fabric with equivalent layup and thickness were quasi-statically crushed. The energy absorbency of the specimens were measured and compared. In addition, the present data were compared with existing data. Based on the experimental observations and the comparisons among the data groups the following conclusions can be made. 1. Testing of flat-plate Kevlar/epoxy laminates with the present fixture is possible in spite of the material being resistant to tearing at the crushing plane. 2. Three distinct failure modes in the Kevlar/epoxy flat-plate specimens are observed: stable sinusoidal buckling, binding of specimens with test fixture because of shallow scoring, and bending of specimens to one side because of deep scoring. The failure modes are easily distinguishable from each other both visually and via the load versus stroke curves. Only sinusoidal buckling is deemed to be indicative of a stable crushing failure mechanism. 3. For the 64-mm-wide flat plate, a score depth of approximately 1/2 of the specimen thickness leads to a successful test. For the 89-mm-wide flat plate, a score depth of approximately 1/3 of the specimen thickness leads to a successful test for this layup. 4. The load applied to the fixture by the specimen during the testing is much less than the measured crush loads. 5. The specific sustained crushing stress for Kevlar/epoxy flat plates with a layup of (+45/–45/0/90)S is not affected by the width of the plate using the present fixture. 6. Flat-plate and tube geometry SSCS values are similar for the present data. Because the material represents a testing challenge, this result indicates the capability of the present fixture to determine the energy absorption capability of composite materials.

8

The ability to test Kevlar/epoxy composite laminates consisting of fabric laminae with a flat-plate specimen has been shown utilizing the present fixture. Binding between the specimen and the test fixture can be eliminated for this test fixture by scoring the flat-plate specimens along the constraint rods. Comparable SSCS values for the flat-plate and tube specimens were found. The flat-plate specimen is a capable, cost-effective alternative to the tube geometry for testing the energy absorbing capabilities of composite laminates.

Acknowledgments The material used in this study was graciously provided by Dr. Steven Peake of CYTEC. The authors wish to thank Mr. James S. Harris for his help in this effort.

References 1. Sen, J. and Dremann, C., “Design Development Tests for Composite crashworthy Helicopter Fuselage,” SAMPE Quarterly, Vol. 17, No. 1, Oct. 1985, pp. 29-39. 2. Fleming, D. and Vizzini, A., “Tapered Geometries for Improved Crashworthiness under Side Loads,” Journal of the American Helicopter Society, Vol. 17, No. 1, Jan. 1993, pp. 38-44. 3. Farley, G., “Energy Absorption of Composite Materials,” Journal of Composite Materials, Vol. 17, No. 5, May 1983, pp. 267-279. 4. Hull, D., “A Unified Approach to Progressive Crushing of Fibre-Reinforced Composite Tubes,” Composites Science and Technology, Vol. 40, 1991, pp. 377–421, . 5. Knack, J. and Vizzini, A., “Energy Absorption of Truncated Kevlar Epoxy Cones Under Side Loads,” Proceedings of the 35th AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics and Materials Conference, Hilton Head, South Carolina, Apr. 1994, pp. 2831–2837. 6. Jackson, K., Morton, J., Lavoie, J., and Boitnott, R., “Scaling of Energy Absorbing Composite Plates,” Journal of the American Helicopter Society, Vol. 39, No. 1, Jan. 1994 pp. 17–23. 7. Fleming, D. and Vizzini, A., “The Energy Absorption of Composite Plates under Off-Axis Loads,” Journal of Composite Materials, Vol. 30, No. 18, 1996, pp. 1977–1995. 8. Thornton, P. and Edwards, P., “Energy Absorption in Composite Tubes,” Journal of Composite Materials, Vol. 16, No. 11, Nov. 1982, pp. 521–545. 9. Farley, G., “Effect of Specimen Geometry on the Energy Absorption Capability of Composite Materials,” Journal of Composite Materials, Vol. 20, No. 7, July 1986, pp. 390-400. 10. Farley, G., “Energy Absorption of Composite Material and Structure,” Proceedings of the AHS 43rd Annual Forum, St. Louis Missouri, May 1987, pp. 613–627. 9

11. Lavoie, J. and Morton, J., “A Crush Test Fixture for Investigating Energy Absorption of Flat Composite Plates,” Experimental Techniques, Nov./Dec. 1994, pp. 23-26. 12. Dubey, D. and Vizzini, A., “Energy Absorption of Composite Tubes and Plates,” Journal of Composite Materials, Vol. 32, No. 2, 1998, pp. 158-176.

10

Grinder or Cutter held in Mill Composite Specimen

34° 900-1500 RPM

Aluminum Guide Plates Clamp Fixtured on Mill

17°

Figure 1 Chamfering of flat-plate specimens Dubey & Vizzini

Testing Methods

JAHS–1

Circular Indexing Head Fixtured on Mill Grinder or Cutter held in Mill

34°

114 mm

17°

Aluminum Mandrel Composite Specimen

Figure 2 Schematic section of chamfering of tube specimens Dubey & Vizzini

Testing Methods

JAHS–2

Subassembly 1

Holes for attachment of gripping fixture to testing machine grips

Composite specimen

Rod tip support plate

Figure 3 Gripping portion of crushing fixture Dubey & Vizzini

Testing Methods

JAHS–3

Crushing Plane

φ

Spacers

Figure 4 Incline feature of crushing fixture

Dubey & Vizzini

Testing Methods

JAHS–4

Figure 5 Photograph of crushed flat plate specimens

Dubey & Vizzini

Testing Methods

JAHS–5

70 mm

Scoring

165 mm

Grip Inserts

Support Rods

Steeple Chamfer 38.1 or 63.5 mm

Figure 6 Use of scoring to ease tearing of specimen along support rods Dubey & Vizzini

Testing Methods

JAHS–6

Damage Initiation

Load

Linear Elastic (1)

Crushing (3) Transition (2) 0 0 Stroke

Figure 7 Schematic representation of a crushing test

Dubey & Vizzini

Testing Methods

JAHS–7

100 SCS, kN-m/kg

SSCS = 35.7 kN-m/kg

50

0 0

25 50 Stroke, mm

Figure 8 Typical specific crushing stress (SCS) versus stroke for 64 mm flat plate (preferred failure mode)

Dubey & Vizzini

Testing Methods

JAHS–8

100 SCS, kN-m/kg

SSCS = 31.5 kN-m/kg

50

0 0

25 50 Stroke, mm

Figure 9 Typical specific crushing stress (SCS) versus stroke for 89 mm flat plate (preferred failure mode)

Dubey & Vizzini

Testing Methods

JAHS–9

Vise Specimen

Figure 10 Photograph of crushed 89–mm-wide flat-plate specimen (bending to one side failure mode)

Dubey & Vizzini

Testing Methods

JAHS–10

Figure 11 Photograph of crushed 64–mm-wide flat-plate and 89–mm-wde flat-plate specimens (binding failure mode)

Dubey & Vizzini

Testing Methods

JAHS–11

buckling

Figure 12 Photograph of crushed 64–mm-wide flat-plate specimen (binding failure mode) exhibiting buckling

Dubey & Vizzini

Testing Methods

JAHS–12

100 SCS, kN-m/kg

SSCS = 52.1 kN-m/kg

50

0 0

25 50 Stroke, mm

Figure 13 Typical specific crushing stress (SCS) versus stroke for 89–mm-wide flat-plate exhibiting a binding failure mode

Dubey & Vizzini

Testing Methods

JAHS–13

Strain, mstrain

400 300 200 100 0 -100 0

20

40 Stroke, mm

60

Figure 14 Typical fixture strain versus stroke for bending failure mode

Dubey & Vizzini

Testing Methods

JAHS–14

Strain, mstrain

800 600 400 200 0 0

20

40 Stroke, mm

60

Figure 15 Typical fixture strain versus stroke for sinusoidal buckling failure mode

Dubey & Vizzini

Testing Methods

JAHS–15

Strain, mstrain

2000 1500 1000 500 0 0

20

40 Stroke, mm

60

Figure 16 Typical fixture strain versus stroke for binding failure mode

Dubey & Vizzini

Testing Methods

JAHS–16

SCS, kN-m/kg

100

100

50

50

No Score 0 0

100

25

50

0 0

100

0.53 mm

50

0 0

0.36 mm 25

50

0.71 mm

50

25

50

0 0

25

50

Stroke, mm

Figure 17 The effect of score depth on the load stroke behavior (64-mm-wide specimens) Dubey & Vizzini

Testing Methods

JAHS–17

SSCS, kN-m/kg

100 64 mm 89 mm

80 60 40 20

Binding Valid Bending

0 0.0

0.2 0.4 0.6 Score Depth, mm

Figure 18 SSCS and failure mode versus score depth

Dubey & Vizzini

Testing Methods

JAHS–18

Figure 19 Photograph of crushed tube

Dubey & Vizzini

Testing Methods

JAHS–19

100 SCS, kN-m/kg

SSCS = 33.5 kN-m/kg

50

0 0

25 50 Stroke, mm

Figure 20 Typical specific crushing stress (SCS) versus stroke for 114 mm diameter tube

Dubey & Vizzini

Testing Methods

JAHS–20

Table 1 Kevlar/epoxy Average Measurements

64-mm-wide 89-mm-wide 114 mm tube

Dubey & Vizzini

Thickness, mm (CV)

Length, mm (CV)

2.11 (1.8%)

165.6 (0.2%)

2.16 (1.2%)

101.9 (0.6%)

Width, mm (CV) 63.9 (0.1%) 89.3 (0.0%)

Testing Methods

—

Inner Diameter, mm (CV)

Density, g/cc (CV)

—

1.41 (0.4%)

113.8 (0.1%)

1.43 (1.3%)

JAHS–21

Table 2 Initiation and Energy Absorption Geometry Flat Plate 64 mm Average (CV) Flat Plate 89 mm Average (CV) Tubes 114 mm Average (CV)

Dubey & Vizzini

Initiation Stress [MPa]

SSCS [kN-m/kg]

75.3 81.6 83.7 87.0

29.1 35.7 37.9 34.7

81.9 (6.0%)

34.3 (10.8%)

61.3 70.6 66.8 66.0

33.4 31.5 34.4 33.6

66.2 (5.7%)

33.2 (3.8%)

83.2 79.9 77.9 90.6 84.0 86.2

36.0 34.1 35.7 33.5 34.7 32.4

83.6 (5.4%)

34.4 (3.9%)

Testing Methods

JAHS–22

Table 3 Loads exerted by the specimens on the bending rods Failure Mode Bending Sinusoidal Buckling Binding

Dubey & Vizzini

Maximum Strain, strain 313 788 1850

Testing Methods

Load, N 97.7 246.3 578.6

JAHS–23

Table 4 Overall Comparison of SSCS

Present Kevlar/epoxy Previous graphite/epoxy Reference 12

Dubey & Vizzini

Flat-Plates [kN-m/kg] (CV) 33.8 (7.8%)

Tubes [kN-m/kg] (CV) 34.4 (3.9%)

77.6 (16.1%)

86.0 (2.1%)

Testing Methods

JAHS–24