AIAA-1999-1249-909

7/29/2019 AIAA-1999-1249-909

1/10

A99924642 AIAA-99-1249

THE ELEVATED TEMPERATURE COMPRESSIVE RESPONSE OFNOTCHED COMPOSITE LAMINATESJunghyun Ahn and Anthony M. WaastComposite Structures LaboratoryDepartment of Aerospace Engineering, University of Michigan, Ann Arbor, MI 48109-2118

February 1,1999AbstractA compression test at elevated temperature wasperformed to investigate the notch sensitivity andfailure mechanisms of fibrous laminated compositeplates containing a stress raiser, in the form of acircular cutout, under equi-biaxial and uniaxialcompressive loading. Strains measured at variouslocations on the specimen surface and cutoutedge were used to monitor the initiation of failure.Unloading the specimen before global failuremade it possible to understand the failuremechanism associated with the final failure mode.

1. IntroductionFor structural applications, polymer basedfibrous composite materials possess severalfavorable attributes when compared to othertraditional materials such as metals. A highspecific stiffness is one attribute that favorsapplications in the aerospace industry. Because oftheir complex microstructure, these materialsexhibit a variety of failure mechanisms whensubjected to mechanical and thermal loads.Understanding these mechanisms and developingmicromechanics based modeling capability havebeen subject areas that have received aconsiderable amount of attention in the recentpast. In the present paper, we report the results ofan experimental investigation into the failuremechanisms in fibrous composite laminatedplates. Plates containing a centrally locatedcircular hole were subjected to planar compressiveloads, both biaxial and uniaxial, under room (25C) and elevated temperature (200 C). The

motivation stemmed from a problem commonlyoccurring in aerospace applications where flatplates and panels containing cutouts have to bedesigned to carry a certain amount of mechanicalload under various operational temperatures. Animportant question that arises in this process is theidentification of the dominant failure mechanism ata given temperature and loading condition, whichis activated during service conditions. Asecondary, but equally important question is theamount of stiffness and strength degradation interms of the failure initiation in the presence of anelevated temperature environment.

In section 2, we present details of theexperiment, the specimens and the testprocedure. Next, we summarize and tabulate theresults obtained from the different t ypes oflaminates, with important observations deducedfrom the tests to ascertain the laminate stiffnessproperties as well as a discussion of the effect ofthe environment on the observed failure mode.Finally, we remark on how the failurecharacteristics are related to the lay-up andenvironment.

7 Fxperimental Method3 1 Test SpecimensAs pointed out in several papers (SeeKhamseh 8, Waas, 1997), difficulties associatedwith stress concentrations are encountered in thedesign of a cruciform shaped planar specimen forin-plane biaxial loading, in so far as achievingfailure in the center of the specimen.

Ph.D. Candidate, Department of Aerospace Engineering, University of Michigan, Ann Arbor, MemberAIAA Copyright @ 1999 by Junghyun Ahn, Published by the American Institute of Aeronautics andAstronautics, Inc. with permission.Associate Professor of Aerospace Engineering, Associate Fellow, AIAA.

518

(c) 1999 American Institute of Aeronautics & Astronautics

7/29/2019 AIAA-1999-1249-909

2/10

In order to select the size of the specimen, wegenerated a model of the notched laminated plateand load introduction grips(steel) using theABACUS* FEA software package. The stressanalysis program was used to examine the effectof several specimen geometries on the stressconcentrations due to the applied loads. The loadswere applied at the boundaries of the loading armsby use of displacement constraints. Eight nodedparabolic plane stress elements were used tomodel the plate material and steel grips. Plyproperties were entered from which ABAQUScalculates the equivalent plane stress constantsusing classical lamination theory. It was concludedfrom the results that a cruciform configurationmatching the dimensions given in Figure 1, wasthe desirable specimen shape.

calculate the stress field corresponding to theuniaxial and equi-biaxial loading. The obtainedsolution was next verified with the strain data (far-field and near the hole edge)that was measured inthe room temperature test for each laminate type.Such a check yielded good agreement betweentest data and the 2D-elasticity solution.

The word desirable is used to refer to astress state for which an in-plane region in theinterior of the specimen (shown as the crosshatched area in Figure 2) is not influenced by theeffects of the far field edges (i.e., edge curvaturesidentified as Rc in Figure 1) of the plate due to theloading.Specimen configurations corresponding toseveral values of F& were studied using the FEAprior to arriving at a optimum value of R, basedon the condition that the specimen with a holedoes not contain effects associated with the stressconcentration at the hole interfering withnonuniformities in the stress field generated onaccount of the edge curvatures. For the compositeplate we studied this working area translated intoa 6.35 cm (2.5 in.) square region in the middle ofthe cruciform configuration.

The inplane dimensions of the test specimensare given in Figure 1. The thickness of thelaminates were 0.635 cm (0.25 in.) with a 0.013cm (0.005 in.) tolerance. In the center of the steelgrips, a 0.724 cm (0.285 in.) channel, 3.048 cm(1.200 in.) in depth, was machined along thelength of the grip. A specimen arm sat in thischannel, bonded to the walls of the channel withthe use of Devcon@ brand plastic steel puttyadhesive, treated with release agent in order toallow for ease of separation of the two materials atthe end of an experiment. More importantly, theDevcon@ putty acted as an interface between thespecimen edge and the grip surface on which thespecimen sat effectively smoothening out theinterface surface irregularities between the twomaterials and ensuring a smooth load transfer. Toensure proper alignment of the specimen insidethe channel (i.e., ensuring reasonably centeredseating of the specimen inside the channel), metalshim plates of various thickness were added toboth sides of the specimen to eliminate any gapinside the channel.

The test specimens were made of graphite /toughened epoxy material composed of HerculesIM7 (Intermediate Modulus) fiber and 977-3toughened epoxy matrix, designed for operation athigh temperatures. In Table 1, we show the laminaproperties, based on data provided by themanufacturer. The stacking sequence for thesesymmetric laminates were as follows: Cross-Ply[O/+90],2s and Angle-Ply [+60/-60],,,. In Table 2,we have listed the laminate material properties,calculated using generalized classical laminationtheory(CLT). The linear elastic plane stresssolution that corresponds to the problem at hand,as given in Lekhnitskii (1968) was used to

The biaxial loading frame has four loadactuators capable of exerting tensile / compressiveloads of 222 kN (50,000 Ibf) when mated to a 20.7MPa (3000 psi) hydraulic power supply . It shouldbe noted than each of the actuators can beprogrammed to operate in a load or displacementfeedback control mode (via the use of a load cellor displacement transducer) fully independent ofthe other three cylinders, thereby allowing for bothuniaxial as well as biaxial tests. Two load cellswere used in the system, one along each loadingaxis, each rated for a magnitude of 222 kN(50,000 Ibf) in tension and compression.

* We are grateful to Hibbit, Carlson and SorensonInc. for making ABAQUS available underacademic site license.

519

2.3 To perform a compression test at elevatedtemperature, a feedback controlled radiationheating element with an insulation tunnel (Figure3) was designed and used. Special strain gagesfor high temperature (mounted with hightemperature glue and lead wire insulated for


7/29/2019 AIAA-1999-1249-909

3/10

thermal protection) were used for the test. Toensure a steady temperature in the enclosed testchamber, four locations in the test chamber wereused to monitor temperature. In addition,thermocouple gages placed inside the specimencutout edge and on the specimen surface wasused to verify that the specimen temperature isfairly even (less than 2C change from location tolocation). The load cell and the loading f rame werecooled by circulating water through copper p ipewrapped around both parts. The maximumtemperature of the tests was limited to 200 C(392 F).24 Data AcquisitionSurface strain measurements at variouslocations on the specimen were recorded withstrain gages. Strain gages were mounted alongthe edges of the hole, as well as in thecorresponding far field region. A set of twoadditional strain gages were mounted inside thehole, along the wall thickness, in order to recordchanges in strain in the out of plane(through thethickness)direction, revealing failure initiation.Strain gage readings, along with load cell readingswere monitored via a custom data acquisitionsystem.2.5 Test ProcedureThe 48 ply composites were tested underuniaxial and 1: 1 in far-field displacement) biaxialloading at both room temperature and 200 C inorder to understand the failure mechanisms andhow they are affected by temperature.The catastrophic nature of failure incomposites and the additional limitation placedupon the visual investigation of failure initiationand progression suggested a displacementcontrolled mode for the tests (as opposed to theload control). This facilitates examination ofspecimens loaded to initial (local) failure butrecovered prior to global failure. In a displacementfeedback control mode.Each opposing pair of pistons wasprogrammed to move a certain distance in aspecified time interval. Preliminary resultsindicated that a rate of 0.381 cm (0.15 in.) in 25minutes for piston travel would assure a quasi-static test. The influence of hole size on the failuremode had been established and verified before(Khamseh and Waas, 1992, 1997). Therefore, inthe present study one hole size (0.5 in. Diameter)was chosen for all specimens.

Each test consists of loading a particular typeof laminate until global failure to estimate theultimate strength of the specimen. This informationis used subsequently to aid in unloadingspecimens that have failure initiated but notreached their ultimate load carrying capacity. Asudden nonlinear (with respect to a far-field loadcomponent) increase in strain inside the hole(strain gage placed through the thickness on thehole edge) indicates that there is a failure initiationat the hole edge of the specimen, even thoughthere may not be a sign of global fai lure of thespecimen.After unloading the specimens, a small regionaround the hole edge was cut and observed underan optical microscope in order to examine thefailure initiating mechanisms (kink banding, micro-buckling, fiber / matrix interface cracking, matrixcracking etc.). Both the in-plane and out-of-planeviews of the failed region were examined anddigital photomicrographs were next acquired.3. Results and Discussion -

3.1 Room TemDerature3.1.1 Uniaxial Test

Cross ply laminates show the typical kinkbanding failure mode (also identified before by anumber of researchers, Ref. l-1 2) at around a farfield stress of 44 KSI. Failure initiates in the zeroplies, at the hole edge, and propagates in adirection that is perpendicular to the direction ofapplied far-field load. Both inplane and out-of-plane kink banding occurs in the zero plies. InFigure 4, we show a typical view of the out-of-plane kink band. The surface strain at the holeedge in the direction of applied load at failureinitiation is 7,500 ALEand the corresponding farfield strain is approximately 3,000 f.zThe failure of the angle ply specimenssubjected to the same type of loading is quitedifferent. As indicated in Figure 5, the angle plylaminates when stressed in the Y-direction, fail bya mechanism of fiber matrix shearing. The failureis sudden so that the initiation and final failureoccur almost at the same time. Because thestrains near the hole edge are much larger thanelsewhere, the failure initiates at the hole edgeand traverses along the 60 fiber angle (the crackruns through the matrix between the fibers). Itappears that the -60 fibers running at an angle to

520


7/29/2019 AIAA-1999-1249-909

4/10

the crack path do not provide the requisite crackbridging toughness to arrest the crack, so thatonce initiated the specimen is split cleanly alonga line at 60.3.1.2 Equibiaxial Test

The main difference in response of thelaminates between the uniaxial and the equi-biaxial loading is not in the mechanism of failurebut the magnitude of far-field loading necessary toachieve failure initiation. This was transpired bythe measurements we performed. Thus, thisshows that the mechanism of failure is intimatelytied to the laminate lay-up and architecture, moreso than the loading, for two cases we haveexamined so far.For the case of cross ply laminates, sincethere are only 0 / 90 plies, failure initiation occursby kink banding in the 0 plies, starting from thehole edge at a position that is perpendicular to thedirection of loading. Ideally, both kink bands (oneither side of the hole) should happen at the sametime, but due to material non-homogeneity or aslight perturbation of the symmetry of loading, oneof the two kink bands form first, then followed bythe other. The kink band formation leads tointerlaminar cracking between the kinked ply andthe adjacent 90 ply (Figure 6)3.2. Elevated Temperature

At elevated temperature, a new parameteremerges to influence the mechanism of failure.This parameter is associated with the shear-stressvs. shear strain response of the pure matrixbehavior. Because of this, the laminate materialproperties are changed. Furthermore, due to themismatch in the coefficients of thermal expansion,the nature of the fiber / matrix bond is alsochanged. In addition, the matrix rich regionbetween plies which control the interlaminarstrength is also affected.As in the room temperature tests, failureinitiation is sensed by a sudden increase in thethrough thickness (out of plane mode) strains,which allows the unloading of a specimen for post-failure microscopic examination of regions nearthe hole edge.Cross ply laminates show extensive kinkformation just as in the room temperature case.* A summary of the measurements correspondingto the room temperature experiments are shown inTable 3.

Compared to the room temperature case, failureoccurs at a lower load (See Table 4). Again, thekink banding initiates at the hole edge andpropagates in a direction perpendicular to thedirection of loading. Noticeably, the out of planefailure is very extensive. This observation maypoint to the reduced interfaminar strength onaccount of the elevated temperature, for thereasons discussed above. Figure 7 shows theappearance of the kink bands when viewed in thethrough the thickness direction.In the case of the angle ply laminates, themaximum load drop is not as drastic as the crossply laminates, but the maximum strain at the holeis reduced to about half when compared to theroom temperature case. This shows that thisparticular mechanism of failure in angle plylaminates is not geometric, but rather it iscontrolled by a strength parameter. In thisparticular case, an examination of the failedspecimens reveal that this strength parameter isthe fiber / matrix interfacial strength.

4 Concludina RemarksWe have reported the experimental results of aseries of uniaxial and equibiaxial inplanecompressive loading experiments on notchedcomposite laminated plates under roomtemperature and elevated temperature conditions.Two types of 48 ply graphite/epoxy compositeswere studied in order to understand theircompressive failure mechanisms.The laminate stacking sequence wasselectively designed to bring out the differentfailure mechanisms that operate in angle plies andplies oriented in the loading direction. Inspecimens containing plies along the direction ofthe load (cross ply), uniaxial compression leads tofiber kink banding. This same scenario persists atboth room and elevated temperature.The angle ply specimens show failure along thefiber / matrix interface. The failure persists either

at +0 or -8 angle or both. Fibers across the line offailure were broken due to the large amount ofenergy released as the interfacial crackpropagates rapidly. At elevated temperature, thesame failure mechanisms persist, but at differentmagnitudes of far-field loading.

5 AcknowledgementsWe are grateful for financial support from theAFOSR under grant F49620-95-l-0326, and the

521


7/29/2019 AIAA-1999-1249-909

5/10

Department of Aerospace Engineering at theUniversity of Michigan.

1.2.

3.

4.

5.

6.

7.

8.

9.

6 ReferencesLekhnitskii S., Theory of Elasticity of anAnisotropic body, test, Godden-Day, 1968Waas, A. M., and Schuftheisz, C. Ft., 1995,Compressive failure in Composites, Part II,Progress in Aerospace Science, Vol.32, pp. 43- 78.Stames, J., Rhodes, M.D., and Williams, J.G.,1979, Effect of Impact Damage and Holes onthe Compressive Strength of a Graphite/EpoxyLaminate, Nondestructive Evaluation andFlaw Criticality for Composite Materials},edited by R. B. Pipes, ASTM STP 696, pp.145-171. )Grape, J. and Gupta, V., 1995, Failure inCarbon / Polyamide Laminates under BiaxialCompression, J. Composite Materials, vol. 29,No. 14, ppl850-1872.)Waas, A. and Bacock, C.D., Observation ofthe Initiation and Progression of Damage inCompressively loaded Composite PlatesContaining a Cutout, GALCIT SM Report 86-34(1 986). See also GALCIT SM Report 85-12,by the same authors, 1985.)Waas, A., 1988, Compression Failure ofFibrous Laminated Composites in thePresence of Stress Gradients: Experiment andAnalysis, Ph.D. thesis, California Institute ofTechnology.Waas. ABabcock, CD. and Knauss, W.G.1990, An Experimental Study of theCompression Failure of Fibrous LaminatedComposites in the Presence of StressGradients, Infernational Journal of Solids andStructures, Vol. 26, No. 9/l 0, pp. 1071-1098.Guynn, E. G., and Bradley, W. L., 1989, ADetailed Investigation of the Micromechani smsof Compressive Failure in Open HoleComposite Laminates, Journal of CompositeMaterials, ~01.23, May, pp.479-504.Soutis, C., and Fleck, N-A., 1990, StaticCompression Failure of Carbon FibreT800/924C Composite Plate with a SingleHole, Journal of Composite Materials, Vol.24, pp. 536-558.

11.

12.

13.

14.

A. Khamseh and A. Waas, FailureMechanisms of Composite Plates with aCircular Hole under remote Biaxial PlanarCompressive Loads, ASME J, Materials andTechnology, vol. 119, ~~56-64, 1997Soutis, C., Fleck, N. and Smith, F., FailurePrediction Technique for Compression LoadedCarbon Fiber-Epoxy Laminate with an OpenHole, J. Composite Materials, 25, pp1476-1498,199lSoutis, C., Curtis, P. and Fleck, N.,Compressive Failure of Notched Carbon FiberComposites, Pmt. Royal Sot. London A, 440,~~241-256, 1991.Ahn, Junghyun, and Waas, Anthony, M., AMicromechanic-based finite element model forcompressive failure of notched uniplycomposite laminates under remote biaxialload, ASME J. Eng. Materials & Technology,1999 (to Appear)

10. A. Khamseh and A. Waas, FailureMechanisms in Uniply Composite Plates underUniaxial Compression, ASME J. Materials andTmhnology, October, 1992.)

522


7/29/2019 AIAA-1999-1249-909

6/10

Material 6, VW E, (Msi) G,, (MS0 V 12 Thickness (in)IM7 I 977-3

epoxy23.5 1.21 0.72 0.3 0.0052

Table 1 Zero ply material properties of the 48 ply graphite / 977- 3 epoxy composites

Laminate Type Stacking Sequence k VW E,-JMsi) G,(Msi) VvQuasi-isotropic [+45/O/-45/90], 8.828 8.828 3.372 0.309Cross-Ply 10 1 goI,, 12.401 12.404 0.7119 0.029Angle-Ply [+60 / -go],, 1.492 7.383 4.698 0.309Table 2 Laminate material properties as determined from Classical Lamlnation Theory (CLT)

SpecimenTypeCross-PlyCross-Ply

Anole-Plv

Loading Type Failure initiation load Max. Hole Strain Failure Mode(Ksi) (I4Uniaxial 44 7,500 Kink at 0 edgeBiaxial 37 6,600 Kink in 0, 90 edge

Shear Failure (+60 / -60)Biaxial 35 4.500 Shear Failure (+60 / -60)Table 3 Room Temperature (25 c) Experimental Data

Loading Type Failure initiation load Max. Hole Strain Failure ModeUniaxial 19 2,700 Kink at 0 edgeBiaxial 18 2,080 Kink in 0, 90 edgeUniaxial 22 3,284 Shear Failure (+60 / -60)Biaxial 25 2,753 Shear Failure (+66 / -60)

Table 4 High Temperature (299 Co) Experimental Data

523


7/29/2019 AIAA-1999-1249-909

7/10

Figure 1 Specimen Geometry

lJ3.4 : L&k Top. Right, nnd Bottom hydraulle cylinder& rapccliwly.NOTE: Each cylinder CM be driven tilh . load or dlspk~mcnl kd-

Flgure 2 Schematics of worklng wlndow in the biaxial loading frame, depicting the loadingapparatus

524


7/29/2019 AIAA-1999-1249-909

8/10

?

- Lofwl cellHeating Unit

/

Figure 3 High temperature setup

525


7/29/2019 AIAA-1999-1249-909

9/10

Figure 4 Cross Ply Laminates Failure Mechanism (Uniaxiai)

Line

SIXEN Crack Line

J

Section A-BView of typical kink band in zero ply

Figure 5. Angie-Ply Laminates Failure Mechanism (Uniaxiai)

526


7/29/2019 AIAA-1999-1249-909

10/10

I y Kink Line

Figure 6. Cross-Ply Laminates Failure Mechanism (Biaxial)

Figure 7. Cross-Ply Laminates Kink Formation Through Thickness

527


AIAA-1999-1249-909

Documents

Transcript of AIAA-1999-1249-909