Test Methods For Composites - William J. Hughes Technical Center

PaqeFIGURE35 . .36.Contours of Normalized Shear Stress ( w i t h respect t o appliedshear) for Varying Ratio of Doubler t o Specimen Stiffnessesfor a Rectangular Isotropic Panel [471 . . . . . . . . . . . . . . . .Variation of Stresses in the Test Specimen as a Function ofModulus-Thickness Ratio. R. and the Reason for thisVariation 1531 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Damage in a [0/45/90/ 4512,GraphiteIEpoxy ShearSpecimen [531 . . . . . . . . . . . . . . . . . . . . . . . . .The Crossbeam Specimen [541. . . . . . . . 85. . . . . . . . . 86. . . . . . . . . . . . . . . 87. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88Normalized rXry,Contours for a [ -451, Cruciform Specimen w i t h SharpCorners [561 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90Normalized a . Contours for a [ -451,Cruciform Specimen w i t h SharpCorners [561 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91Normalized a,, Contours for a [ -451, Cruciform Specimen w i t h SharpCorners [561 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95Slotted Shear Specimens [271Normalized Stress Contours for [OI Slotted Shear Specimens.Spacing Between Notches 0.6 x Depth of Specimen [271. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98Three-Point Shear FixtureFour-Point Shear Fixture. . . . . . . . 96. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99Shear Stress Distributions in Three-Point Loaded Beam 1651Shear Stress Variation Along Beam Axis in a Short BeamI l h 4[651 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100. . . . . . . . . . 101Plate T w i s t Test [691 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104Four-Point Ring T w i s t Test [701Split Ring Shear Test [691. . . . . . . . . . . . . . . . . . . . . . . . . . . 104. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106Slotted Biaxial Specimen [54. 7 1Ivii.107

PaaeBlock Shear Tests [721. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108Lap Shear Test. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109Button Torsion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109Slant Shear Test. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110viii

EXECUTIVE SUMMARYThis document, which constitutes Volume 3 of a three volume set, provides anevaluation of current test methods for shear properties of "advanced" compositesconstructed of high modulus, high strength fibers embedded in organic matrix materialssuch as epoxies. Mechanical testing for various structural properties is one of severalessential steps in the design of composite aircraft structures. Companion volumesaddressing: Tension Testing ( Volume 1 ) and; Compression Testing (Volume 2 ) ofcomposite materials, are also available. The intention is t o provide a comprehensive sourceof information by which the current test methods for these types of property tests can beevaluated and from which test methods which appear t o give good-quality test data can beselected.The document provides: ( 1 ) a comprehensive review of performance features,advantages and negative aspects of various test methods which have been introduced forobtaining shear properties of composite materials; (2)an extensive annotated bibliographycovering most documented test method development activity which has taken place sincethe introduction of advanced composites in the mid 1960's; ( 3 ) a ranking of the commonlyused test method for shear properties, and: ( 4 ) an assessment of problem areas thatcontinue t o exist in the available test methods.T w o types of shear test were evaluated in the survey, in-plane shear which relates t ogeneral structural behavior in an aircraft component, and interlaminar, or transverse, shearwhich relates t o behavior at joints and other design details. Results of the survey are asfollows:INPLANE SHEAR(i) Initial shear modulus can be determined using most of the test methods, providedappropriate correction factors are used t o determine shear stress in the test section basedon adequate stress analysis.(ii) A number of test methods are available which give satisfactory results for in-planeshear measurement of composites. The most popular because of its ease of use is A S T MD35 18 which involves tension loading of r 45' forms of the material t o be tested.

Although there are minor deficiencies recognized in t h stest because of a non-pure shearstress state, industry generally considers the test t o be adequate for structural qualificationof composites. More theoretically correct results are provided by several other tests w h i c hinclude rail shear tests ( A S T M Guide D42551, picture frame shear tests and the recentlyadopted A S T M D 5 3 7 9 losepescu test which involves shear loading of a notched specimen;these tests are somewhat more complicated and difficult t o conduct than t h e 4 5 "tension t e s t and for practical reasons m a y not be in as general use.(iii) The most nearly ideal test from a theoretical standpoint involves torsion o f thin walledfilament w o u n d tubes, currently under development as a n A S T M standard. The t e s t isdifficult t o conduct properly, however, and m a y not be representative of flat laminatesprocessed by approaches representative of that type of material. Other torsion tests w h i c hhave been studied include torsion of solid rectangular columns and circular rods. Theseperform reasonably satisfactorily and are convenient t o apply t o some forms ofcomposites.(iv) One type of test, A S T M D 3 8 4 6 , involving slots c u t on opposite sides of column loadedspecimens, is particularly undesirable for obtaining mechanical property data. Extremelyhigh stress concentrations induced by the slots w h i c h have a crack-like behavior render thetest results more-or-less meaningless from the standpoint of the stress condition causingfailure.(v) A number of other tests w h i c h are in various stages of adequacy are discussed in t h ebody of t h e report.(vi) Ply cracks and other forms of subcritical damage, w h i c h result in nonlinear stressstrain response of unidirectional composites, usually grow a t differing rates w h i c h dependon the fiber orientation or lay u p (0 , 9 0 " or cross ply 0190") as well as t h e t e s t method.For example, the off-axis tension test (tension i n a unidirectional laminate oriented a t a nangle t o the load direction) often yields very l o w ultimate stress and strain for c o m m o nbrittle epoxy systems. The use of cross ply lay u p in torsion tube, losipescu or rail sheartests results i n slow constrained damage g r o w t h (increasing number of ply cracks w i t hincreasing load) w h i c h is considered by many workers as the pattern expected i napplication laminates. Similar behavior is also expected i n 9 0 " rail shear and thick ( 32ply) k 4 5 " tension t e s t specimens. Thinner k 4 5 " tension specimens yield l o w ultimatestresses and strains i n c o m m o n graphite-epoxy composites. Additional studies are

suggested t o address these issues.(vii) Simple modifications t o some of the methods m a y yield better specimen performance.For example, redesigning the grip regions in solid torsion specimens appears worthwhile.On the other hand, significant modifications may be required in some other methods suchas picture frame or crossbeam.INTERLAMINAR SHEAR(i) losipescu tests w i t h specimens prepared by bonding several layers of a material appeart o be the only available method at this time for determining interlaminar stress strainresponse. Obviously specimen preparation needs some effort and the quality of the bondmay affect the results in some cases.(ii) short beam shear is one of the simplest test to conduct and it is often used, but testdata are usually not accepted as material shear properties, since failure can be influencedb y flexural and contact stresses. Further studies are suggested for three and four-pointloaded short beam as well as lap shear tests used for measuring shear strengths.(iii) the slotted or notched shear test is possibly most economical, and for this reason i t isoften used for quality control purposes, but the results are influenced by the severity ofstress concentration at the notches. Because of the economy and wide usage, additionalstudies are suggested for possible improvements of the test and careful comparison of testdata w i t h those from other tests.In summary, i t can be noted that a f e w standards are either available or under development for some in plane shear test methods. Further w o r k is suggested for improving thek 4 5 O tension test for in plane shear and three- or four-point loaded short beam shear test(or some other reliable method) for interlaminar shear. In the meantime, it is hoped that thediscussions and suggestions contained in this report will be useful in deciding on testmethods, specimens, procedures and interpretation of data.

OVERVIEWGENERAL REMARKSThis document which constitutes Volume 3of a three volume set, provides anevaluation of the state of the art of current test methods for obtaining shear properties of"advanced" composites constructed of high modulus, high strength fibers embedded i norganic matrix materials such as epoxies. Mechanical testing is an important step i n the"building block" approach t o design of composite aircraft structures, as illustrated in theFigure below. Companion volumes addressing: Tension Testing ( Volume 1 ) and;Compression Testing (Volume 2) of composite materials, are also available. The intentionis t o provide a source of information by which the current test methods for these typesQUALIFICATIONDES 1 GNALLOWABLES7#IELEMENTSMCMNI CAL PROPERTYTESTSIBU I LD l NG BLOCK DES l GN APPROACHFOR COMPOS l TE A l RCRAFT STRUCTURESFULL SCALEVERIFICATIONCERTIFICAT l ONMechanical Property Testing in Composite Aircraft Design

of property tests can be evaluated and from which test methods which appear t o givegood-quality test data can be selected.Mechanical property testing of advanced composites has been under development eversince the introduction of such materials nearly a generation ago. The first majorconference on test methods for advanced composites, for example, took place in 1 9 6 9and culminated in ASTM Special Technical Publication STP 4 6 0 which summarized resultsfrom a number of DoD programs that were ongoing at that time. The methods which werereported on that occasion formed the basis for a number of test methods which are stillin use.The methodology for obtaining mechanical properties of such materials contains anumber of inadequacies and is in need of continuing development. The purpose of thisdiscussion is t o review the issues which are significant drivers in efforts toward improvedtesting methodology, in order t o provide a framework for evaluating the state of the art.OBSERVATIONS ON MECHANICAL PROPERTY TESTING OF COMPOSITESMechanical property measurements in structural materials can be characterized in termsof three regions in the test specimen (illustrated in the following Figure for a generictension test): (1) a load introduction or gripping region, where large stress peaksassociated w i t h the load introduction method are compensated for b y a relatively largeloaded area; ( 2 ) a central ("gage") region of relatively small loaded area where failure ismeant t o be produced, and; (3) a transition region joining the gage and grip regions. ( Aclear c u t transition region, (3),is not present in many types of test specimen).The gripping region is characterized b y complex loading features, often involving verypeaky stress distributions associated w i t h hard contact points. Three dimensionality in theform of stress variations through the thickness is frequently present in the grip region. Int h e representative case shown on the following page, the load is introduced through thehard teeth of serrated surfaces of a wedge grip which results in through-the-thicknessshearing; in the transition region this translates into spreading of the load i n the lateraldirection via in-plane shearing. Softening layers which may include tabs, thin sandpapersheets or other approaches, may be present in the grip region. In beam-type specimensused for short beam shear and flexure testing, hard contact points represented b y small-

1. LOAD l NTRODUCT i ON (GR I P) REG I ON2. GAGE REGION3 . TRANSITION REGIONElements of Generic Test Specimenradius rods of a relatively rigid material such as steel m a y be present t h a t give rise t osevere stress peaks i n t h e load introduction region w h i c h are unrelated t o t h e desiredstress state.The ideal mechanical property t e s t specimen w o u l d provide a large effective loaded areain t h e grip region t o compensate for stress peaks caused b y t h e gripping arrangement,while allowing t h e stresses in t h e gage region t o approach a uniform condition of highstress w h i c h ensures t h a t failure takes place in t h a t region. Furthermore, sufficient volumeo f t e s t material should be involved in t h e gage region t o provide an adequate sampling oft h e variability w h i c h is characteristic of t h e material being tested. For various reasons,such an ideal f o r m of behavior is hardly ever achieved i n practical t e s t specimens forcomposite materials.Specific problems w h i c h hamper successful mechanical property measurements inorganic m a t r i x composites w i l l be summarized at this point.Measurement of mechanical properties in organic-matrix composites is difficult because

of a general lack of ductile response together w i t h large d fferencesin the mechanicalstrengths of such materials for stresses in various directions. The problem is relievedsomewhat for materials reinforced in more than one direction because t h e strengthdifferences are considerably less in such cases, but the requl dmentsof the technology arecurrently set b y those for unidirectionally reinforced materials.For the situation shown in the preceding Figure, for example, a metallic specimen willbe relatively insensitive t o the indentations caused b y the serrations of the loading grips,and no special difficulty will be caused by the details of the transition region, since localyielding will cause the stress at any cross section t o tend toward a uniform "P-over-A"value (i.e. nominal stress defined by load divided b y section area) applicable t o the sectionunder consideration; these "P-over A " stresses will be obliged t o have their maximumvalues i n the gage region b y the mechanics of the situation, specifically the f a c t that thesmallest section occurs there, so that satisfactory confinement of failure t o the gageregion will be obtained. Accordingly, there is little need for concern over the possibility ofnot obtaining representative failures in metallic test specimens.In the case of organic composites reinforced w i t h high strengthlhigh modulus fibers,on the other hand, achievement of representative failure is difficult. For example, it w a sfound early in the development of the technology of advanced composites t h a t for tensionand compression testing, width-wise tapering t o form a stress-focussing transition region(see the preceding Figure) usually leads t o splitting failures in the tapered region longbefore a valid failure can be obtained in the gage region. This tendency appears t o berelated t o excessively l o w shear strength of organic matrix composites in comparison w i t htheir tensile or compressive strength in the fiber direction. For the case of tension testing,the problem was dealt w i t h in early efforts b y the introduction of rectangular (i.e. uniformwidth) test coupons w i t h thickness-wise bonded-on doublers ( i.e. tabs) at the ends,through which the load w a s sheared in. This is generally accepted practice for tensiletesting, as well as a number of compression test specimen designs.On the other hand, t h e processes governing the behavior of the tabs lead t o high stresspeaks at the gage ends of t h e tabs, so that failures near or inside t h e tabs are quite likelyand are commonly observed. Even though a consensus developed for the use o f tabs, theyobviously do not achieve t h e type of behavior described previously as the ideal of a testspecimen design. Moreover, a number of practical difficulties are associated w i t h tabs.

Debonding of tabs is certainly n o t unusual, and is especially troublesome for t e s t situationsinvolving high temperature and humidity. In other words, t h e use of tabs as a supposedcure for t h e problem o f splitting in w i d t h tapered tension and compression specimens isn c t a completely adequate solution. Tilis kind of poor choice of alternatives characterizesm a n y situations in t h e testing of composites.A n

an assessment of mechanical property test methods for organic matrix composite materials. The present volume presents a review and evaluation of test methods for shear properties of fiber reinforced composite materials. Two companion documents, Volume I on Tension Test Methods and Volume I1 on Compression Test Methods, have also been prepared.