April 1998

NASA/TP-1998-207645

Evaluation of Composite HoneycombSandwich Panels Under CompressiveLoads at Elevated Temperatures

Sandra WalkerLangley Research Center, Hampton, Virginia

The NASA STI Program Office ... in Profile

Since its founding, NASA has been dedicatedto the advancement of aeronautics and spacescience. The NASA Scientific and TechnicalInformation (STI) Program Office plays a keypart in helping NASA maintain thisimportant role.

The NASA STI Program Office is operated byLangley Research Center, the lead center forNASAÕs scientific and technical information.The NASA STI Program Office providesaccess to the NASA STI Database, thelargest collection of aeronautical and spacescience STI in the world. The Program Officeis also NASAÕs institutional mechanism fordisseminating the results of its research anddevelopment activities. These results arepublished by NASA in the NASA STI ReportSeries, which includes the following reporttypes:

· TECHNICAL PUBLICATION. Reports ofcompleted research or a major significantphase of research that present the resultsof NASA programs and include extensivedata or theoretical analysis. Includescompilations of significant scientific andtechnical data and information deemedto be of continuing reference value. NASAcounter-part of peer reviewed formalprofessional papers, but having lessstringent limitations on manuscriptlength and extent of graphicpresentations.

· TECHNICAL MEMORANDUM.

Scientific and technical findings that arepreliminary or of specialized interest,e.g., quick release reports, workingpapers, and bibliographies that containminimal annotation. Does not containextensive analysis.

· CONTRACTOR REPORT. Scientific and

technical findings by NASA-sponsoredcontractors and grantees.

· CONFERENCE PUBLICATION.

Collected papers from scientific andtechnical conferences, symposia,seminars, or other meetings sponsored orco-sponsored by NASA.

· SPECIAL PUBLICATION. Scientific,

technical, or historical information fromNASA programs, projects, and missions,often concerned with subjects havingsubstantial public interest.

· TECHNICAL TRANSLATION. English-

language translations of foreign scientificand technical material pertinent toNASAÕs mission.

Specialized services that help round out theSTI Program OfficeÕs diverse offerings includecreating custom thesauri, building customizeddatabases, organizing and publishingresearch results ... even providing videos. For more information about the NASA STIProgram Office, see the following: · Access the NASA STI Program Home

Page at http://www.sti.nasa.gov · E-mail your question via the Internet to

help@sti.nasa.gov · Fax your question to the NASA Access

Help Desk at (301) 621-0134 · Phone the NASA Access Help Desk at

(301) 621-0390 · Write to:

NASA Access Help Desk NASA Center for AeroSpace Information 7121 Standard Drive Hanover, MD 21076-1320

National Aeronautics andSpace Administration

Langley Research CenterHampton, Virginia 23681-2199

April 1998

NASA/TP-1998-207645

Evaluation of Composite HoneycombSandwich Panels Under CompressiveLoads at Elevated Temperatures

Sandra WalkerLangley Research Center, Hampton, Virginia

Available from the following:

NASA Center for AeroSpace Information (CASI) National Technical Information Service (NTIS)7121 Standard Drive 5285 Port Royal RoadHanover, MD 21076-1320 Springfield, VA 22161-2171(301) 621-0390 (703) 487-4650

Abstract

Fourteen composite honeycomb sandwich panels were tested tofailure under compressive loading. The test specimens includedpanels with both eight and twenty-four ply graphite-bismaleimidecomposite facesheets and both titanium and graphite-polyimide corematerials. The panels were designed to have the load introducedthrough fasteners attached to pairs of steel angles on the ends of thepanels to simulate double shear splice joints. The unloaded edgeswere unconstrained. Test temperatures included room temperature,250°F, and 300°F. For the room and 250°F temperature tests, the24-ply specimen failure strains were close to the unnotched allowablestrain values and failure loads were well above the design loads.However, failure strains much lower than the unnotched allowablestrain values, and failure loads below the design loads were observedwith several of the 8-ply specimens. For each individual testtemperature, large variations in the failure strains and loads wereobserved for the 8-ply specimens. Dramatic decreases in the failurestrains and loads were observed for the 24-ply specimens as the testtemperature was increased from 250°F to 300°F. Due to initialgeometric imperfections, all specimens exhibited varying degrees ofbending prior to failure. All 8-ply specimens appeared to have failedin a facesheet strength failure mode for all test temperatures. The 24-ply specimens displayed appreciably greater amounts of bendingprior to failure than the 8-ply specimens, and panel bucklingoccurred prior to facesheet strength failure for the 24-ply room and250°F temperature tests. For the two 24-ply specimens tested at300°F, one panel failed due to facesheet strength and the other due toa disbond between the facesheet and core materials.

Introduction

The development of advanced lightweightstructural components is an importantrequirement for an economically feasible highspeed civil transport. Figure 1 displays aconceptual drawing of a high speed civiltransport. At supersonic speeds, both the wingand fuselage structure will be exposed toelevated temperature. For an aircraft designedto cruise at Mach 2.4, structural surfacetemperatures up to 350°F have been predicted(ref. 1). Since material strength and stiffnessproperties generally degrade at elevatedtemperatures, the high temperature operatingenvironment challenges structural componentsto maintain effective load carrying capabilities.

The quest for lightweight structural componentsable to carry high mechanical loads at elevatedtemperatures has led to the consideration ofcomposite honeycomb sandwich panels forsuch applications. However, the effect oftemperature on the load carrying capability ofcomposite honeycomb sandwich structure is notwell understood. Although thermal effects onthe individual sandwich components, thecomposite facesheets, adhesive, and honeycombcores, have been investigated, interactionsbetween these components in a sandwich panelmay exhibit a significantly differenttemperature dependence.

In the present study, fourteen compositehoneycomb sandwich panels were tested at

2

either; room temperature, 250°F, or 300°F.Each panel was tested to failure undercompressive loads at a uniform temperature.The panels were fabricated by MarionComposites under contract with The BoeingCommercial Airplane Group. The panels weretested in the Thermal Structures Laboratory atNASA Langley Research Center.

The purpose of the tests was to evaluate theeffect of temperature on the failure mode andload carrying capability of the compositehoneycomb panels under investigation. Thispaper presents test and analysis results fromfourteen panels that were tested in compressionto 1) evaluate the ability of the compositehoneycomb sandwich panels to carrycompressive mechanical loads at room andelevated temperatures, 2) determine the effect oftemperature on the compressive failure mode ofthese panels, and 3) evaluate the ability ofanalysis methods to predict the compressivefailure mode and load of the compositehoneycomb sandwich specimens.

Test Program

A test program was established to observe theeffect of temperature on the failure mode andfailure load of structural panels beingconsidered for a high speed civil transport. Thefocus of this test program was on compressiondominated areas of the aircraft where structuralresponse tends to be more complex. Compositehoneycomb sandwich panels are beingconsidered for compression wing panels duetheir structural efficiency. However, there areconcerns about the ability of these panels tomaintain their structural integrity at elevatedtemperatures. Interactions between thelaminated facesheets, adhesive bonds, andhoneycomb core add to the complexity of theproblem in predicting failure loads andtemperature effects. The fabrication procedure,a co-curing process in the present study, mayeffect the structural integrity of the panels. Inaddition, geometric imperfections of the panel,possible non-visible impact damage andfabrication defects (i.e., local areas of disbond

between the facesheets and core) exasperate theissue. Although an extensive test program isneeded to understand the behavior of acomposite honeycomb sandwich panel atelevated temperatures, the present study waslimited to fourteen panels.

To introduce the compressive load into thepanels, the test specimens were configured withbolted joint connections to simulate doubleshear splice joints. Each end of the compositehoneycomb sandwich panel was potted andattached to a pair of steel angles by two rows offasteners. The bolted joints provided load pathsfor the tests which are similar to those expectedin joints in a high speed civil transport, but alsoadded more complexity to the tests andincreased the difficulty in creating an accurateanalytical model.

Panel Design

It is standard practice to design lightweightaircraft sandwich structure for simultaneousfailure modes to increase panel structuralefficiency. For these particular sandwich panelsdesigned by Boeing, panel buckling andfacesheet strength failure were the two possiblefailure modes having nearly zero margins ofsafety. Facewrinkling, shear crimping, boltbending, and bolt shear were ruled out aspossible failure modes due to high margins ofsafety. Pad-up plies were added in the jointarea to increase the bearing capability andreduce the risk of bearing failures. Thehoneycomb core was potted in the joint areas toprevent core crushing from bolt clamp-up. Thepreliminary design analyses for the panels wereconducted with the panels at room temperature.The objective of the testing program is to definethe structural performance degradation due totemperature.

Materials selected for the design of thepanels were among those being considered foruse in a high speed civil transport. Thecomposite facesheet material is an toughenedgraphite-bismaleimide designated IM7/5260and manufactured by Cytec Engineering. Two

3

wing panel concepts were chosen for this study.One concept had minimum thickness 8-plyfacesheets representative of a lightly loadedstructure and the other concept had 24-plyfacesheets representative of a moderately loadedstructure. Both the 8-ply and 24-ply laminatedfacesheets had quasi-isotropic layups.

Cytec BMI X2550G dry film adhesive, with athickness of 0.01 inches, was used to bond thefacesheets to the core. Two core materials withthe same cell density and configuration, wereselected for direct comparison and evaluation.The core materials were a titanium honeycombfoil of the alloy Ti-3Al-2.5V and a non-metallichoneycomb fabricated by Hexcel Corporation,designated HFT-G-327, which is a bias weavegraphite fabric reinforced polyimidehoneycomb. Both core materials have a cellsize of 3/16 inches and a nominal density ofapproximately 6Êlbs/ft3.

Test Specimen Descriptions

Fourteen test panels were fabricated byMarion Composites using a proprietary co-curing process developed jointly with Boeingand Marion. A schematic diagram of a panelspecimen is shown in FigureÊ2. The panellength, l, and width, w, are depicted in the frontview, and the facesheet thickness, t, and the coredepth, c, are depicted in the side view shown inFigure 2. A testing matrix describing all testspecimens is provided in Table 1. All panelsmeasured approximately 12 inches wide by 35inches long with a one inch thick honeycombcore. Four of the panels had 8-ply facesheetswith titanium core, four had 8-ply facesheetswith graphite-polyimide core, and theremaining six panels had 24-ply facesheets withgraphite-polyimide core. All 8-ply specimenshad 2.75 inches of the core potted at each endand had steel angles attached with two rows offasteners using nine 3/16 inch diameter steelbolts per row. Four of the 24-ply specimenshad 3.25 inches of the core potted and two rowsof fasteners using eight 3/8 inch diameter steelbolts per row, and the remaining two had 3.25

inches of the core potted on the ends butwithout the bolted joint connections used tosimulate a double shear splice joint. In the jointarea of the panels, four additional plies wereadded to the outside of each 8-ply facesheetand eight additional plies were added to theoutside of each 24-ply facesheet.

Each specimen was instrumented with straingauges mounted back-to-back on the outersurfaces of the panel facesheets. The panelstested at elevated temperatures were alsoinstrumented with thermocouples. A typicalinstrumentation layout used for several of thespecimens is shown in Figure 3. The figuredisplays half of the panel length, where allinstrumentation is symmetric about the panelcenterline.

Test Apparatus and Procedure

A 500-kip hydraulically-actuated universaltest machine was used to load all the testspecimens to failure in compression. A photoof one of the failed specimens located in the testmachine is shown in Figure 4. Heated platenswere mounted through ceramic insulators to thesteel compression platens. Loads were appliedto the panel through the heated platens and thesteel angles on the test specimens as shown inFigure 4. The two unloaded edges wereunconstrained. Test data was recorded using aNEFF 470 data acquisition system coupled witha computer to store test data and display real-time data plots.

Prior to heating the panels, ten percent of thepredicted compressive strength failure load wasapplied to each panel. At that load, back-to-back axial strains near the four corners of eachspecimen were monitored. The lower platen,which is seated on a spherical ball, was adjustedso that the difference between the monitoredstrains were within one percent of the expectedfailure strain. Even though care was taken tomount the steel angles on the specimen so thatthe specimen ends would be flat and parallel, inseveral cases, adjustment of the platen was notsufficient, and metal shims were placed between

4

the steel angles and heated platen to achieve thedesired strain uniformity.

An insulated clamshell-type oven with fin-strip heaters and fans mounted on the insidesurfaces was used along with the heated platensto provide a uniform test temperature for theelevated temperature tests. Threethermocouples mounted on the inside surfaceof the oven and on the upper and lower platenswere used as input for the three zone thermalcontroller. Thermocouples mounted on the testspecimen were also monitored to ensure auniform test article temperature prior toapplication of mechanical loads. Heat input wasadjusted such that all thermocouple readingswere within 25 degrees of the desired testtemperature. For the elevated temperature tests,the specimens were heated while the testmachine was in load control and maintaining anapproximate load of 300 lb. Once the desiredtest temperature was achieved, the test specimenwas loaded at a constant displacement rate of0.02 in/min. Applied load, ram displacement,and back-to-back strains at the center of thepanel were monitored in real-time. The testswere continued until the panels fractured.

Analysis

The ability to predict the panel test resultswas also investigated as part of this study.Several analysis approaches, with varyingdegrees of fidelity and hence complexity, wereevaluated to determine their ability to reproducethe test results. Due to the lack of available datafor IM7/5260, the design allowable straindetermined at Boeing from coupon testing ofIM7/5250, a material with properties consideredidentical to IM7/5260 for the design effort, wasused in predicting failure modes anddetermining allowable design loads for each testtemperature. A table containing the allowablestrain data is presented in Table 2. The open-hole-compression average strain values ofIM7/5250 were used as the design allowablestrains of the IM7/5260 facesheets for thepanels of this study.

The most basic analysis method used herewas a one-dimensional strength of materialssolution for the compressive strength. Themethod assumes that the applied load is a pureaxial load uniformly distributed across thewidth of the panel and is the same for bothfacesheets (i.e., the panel is indeed flat). Afurther assumption is that the honeycomb coreprovides negligible stiffness in the axialdirection. The strength of materials approach tothe compressive strength reduces to the simpleequation, PÊ=ÊEAe, where P is the design load, E= 8.63 ´ 106 psi is the isotropic modulus ofelasticity of the facesheets determined fromlaminated plate theory, e is the design allowablestrain, and A is the effective cross-sectional areaof the panel facesheets.

As a second step in increasing fidelity, two-dimensional finite element models were created.The test specimen was modeled as a compositemulti-layer plate. A schematic drawing of thecomposite layers and their properties, used inthe finite element model, are shown in Figure 5.The facesheets were modeled as separate quasi-isotropic lamina, the honeycomb was modeledas a single orthotropic layer, and the steel angleplates at each end were modeled as if they werecontinuously bonded to the facesheets.Although the steel angles were actually boltedto the composite panel, the approximationsallowed for a very simplified plate model of thetest specimen. The resulting finite elementmodel consisted of 1113Ênodes as displayed inFigure 6. The panel was first modeled andevaluated as a flat panel. A linear static analysisand linear buckling analyses were conductedusing the general purpose finite element codeNASTRAN (ref. 2). Two buckling analyseswere performed to determine the range ofpossible buckling failure loads, one withassumed simply-supported end boundaryconditions and a second analysis with assumedclamped end boundary conditions.

The effect of geometric imperfections wasalso investigated. Two of the panels werescanned prior to testing to determine themagnitude of their imperfections in the out-of-

5

plane direction. The out-of-planeimperfections for Panel 1 were measured alongthe length of the panel at the two unsupportededges. The out-of-plane imperfection wassimilar to, and consequently modeled as, a halfcosine shape along the panel length. Theamplitude of the imperfection was set equal tothe average of the maximum out-of-planemeasurements along the two edges, 0.045inches. For Panel 2, the out-of-planemeasurements were mapped to the grid of thenodal locations shown in Figure 6 to generatethe initial imperfection for the finite elementmodel. Figure 7 is a contour plot of the surfaceout-of-plane measurements for Panel 2. Notethat this contour plot includes the pad up whichaccounts for a 0.036 inch contribution to theout-of-plane measurement after the ramp up tothe joint area (see Figure 2). This pad upcontribution was subtracted from themeasurements before they were mapped to thenodal grid. For the geometric imperfectionmodels, NASTRAN nonlinear analyses wereutilized to evaluate the structural responses.

Results and Discussion

Since it is not practical to present all of thetest data, only selected results are presented todescribe the typical response of the panelsunder compression loading at room andelevated temperatures. Failure data along withpertinent panel characteristics are presented inTable 3. For all fourteen panels tested, failureof the panels occurred at the maximum appliedload. Although the maximum strain wasrecorded at the failure load for all panels, thelocation of the maximum strain reading variedfrom panel to panel. It is important to note that,since a limited number of strain gauges wereused during the tests, the measured maximumstrains are not necessarily the maximum strainsincurred by the panels during testing.

All 8-ply specimens failed due to facesheetfractures which is referred to as the facesheetstrength failure mode. All panels referred to inTable 3 as having failed in the facesheetstrength mode, were very similar in appearance.

Figure 8 is a photograph of Panel 4 after failurewhich is typical of the appearance of all panelsthat failed in the facesheet strength mode.Increasing the test temperature from roomtemperature to 300°F did not appear to effectthe failure mode for the 8-ply specimens. Incomparing the failure data of the 8-ply titaniumcore panels with the 8-ply graphite-polyimidecore panels, the effect of the type of core on thestructural performance in compression isinconclusive.

For the 24-ply specimens, buckling of thepanels prior to failure was observed for both theroom and 250°F temperature tests, as indicatedin Table 3. However, ultimate failure of the 24-ply panels at the room and the 250°F testtemperatures was initiated by fracture of thefacesheets well above the design allowablestrain. Figure 9 is a photograph of Panel 11which buckled prior to failure and is typical ofthe appearance of all failed 24-ply panels thatexperienced buckling before failure. For the24-ply specimens, increasing the temperaturefrom 250°F to 300°F resulted in a differentfailure mode. At 300°F, one 24-ply panelexperienced a facesheet strength failure, and theother 24-ply panel experienced disbondingbetween the facesheet and core material. Figure10 is a photograph of Panel 14 which failed inthe disbond mode. The failure modes at 300°Fcan be attributed to a reduced strength of thefacesheets and adhesive due to increasingtemperature. Also, it is important to note thatPanel 14 had potted ends only and thus did nothave an associated clamp-up pressure that mightprevent a disbond failure mode from occurring.

The maximum measured failure strain datafor all the panels tested is displayed graphicallyin FigureÊ11, along with the laminate strainallowable data of Table 2. Ideally, all panelswould be expected to fail at strains near thelaminate unnotched allowable strain values for agiven test temperature. From Figure 11, onecan observe that the 24-ply panels achievedmuch greater measured strain levels then the 8-ply panels for the room and 250°F temperaturetests. The maximum measured strains for the

6

24-ply were very close to the unnotchedallowables. However, the 8-ply specimenmaximum measured strains fell well below theunnotched allowable and displayed more scatterthan the 24-ply panels. One 8-ply panel, Panel5, failed with a maximum measured strain belowthe design allowable. When going from the250°F test temperature results to the 300°F testtemperature results, a dramatic, much greaterthan expected, decrease in the maximum strainswere observed for the 24-ply specimens. At300°F, the maximum strain for the 24-plyspecimens fell well below the unnotchedallowable level.

Figure 12 displays the failure load data foreach panel tested as a function of temperature.Also plotted for comparison are the designloads calculated from a strength of materialsapproach. The design loads were calculatedusing the minimum specimen width which was11.375Êinches and 11.25 inches for the 8-plyand 24-ply specimens, respectively. A modulusof elasticity determined from lamination theoryand based on ply properties and facesheetthicknesses based on a 0.0056 inch plythickness for the IM7/5260 material were usedin the computations. Although the failure loadsfor the 24-ply panels were above their designloads, there was not as large a differencebetween the design loads and measured loads aswith the maximum design and measured strains.This is most likely attributable to panel bendingwhich was ignored in the strength of materialsapproach described previously. The issue ofbending is addressed later when theexperimental data is compared to analysisresults. As with the measured maximum strains,the failure loads for the 24-ply specimenshowed a significant drop with increasingtemperature, indicating a greater than expectedstrength reduction with increasing temperatureto 300°F. For the 8-ply specimens, the testfailure data showed greater scatter than obtainedfor the 24-ply specimens. The failure loadsbracketed the design loads and severalspecimens failed at loads below the design load.Unlike the 24-ply specimens, no significant

decrease in the failure load was observed at300°F for the 8-ply specimens.

Back-to-back longitudinal strains measuredat the center of the panel facesheets are shownwith applied load for Panels 1, 5 and 11 inFigures 13-15, respectively. The experimentaldata shown in Figure 13 for Panel 1 is typical ofmost of the panels that failed in the facesheetstrength mode or disbond mode. From Figure13 it is apparent that a significant amount ofbending was experienced prior to failure. Asshown by the figure, back-to-back longitudinalstrains diverge with increasing load. This panelbending can be attributed to geometricimperfections as discussed earlier. For Panel 5,Figure 14, bending was negligible, but the panelfailed at a much lower load than the other two8-ply panels tested at room temperature. Thedata shown in Figure 15 for Panel 11 is typicalof the response obtained for all the panels thatfailed in the buckling mode. The diverging ofthe back-to-back strain measurements to thepoint where there is little increase in the loadbeing carried by the panel as the facesheetstrains continued to drastically increase anddecrease on opposing facesheets is characteristicof panel buckling.

Analysis results are compared toexperimental data in Figures 16Ð17, for Panels1 and 2, respectively, which were the only twopanels that were evaluated for geometricimperfections prior to testing. Shown in Figure16 are the structural responses obtained forPanel 1. The strength of material approachassumes that the panel is perfectly flat to obtaina simplified linear solution. The linearbuckling loads were obtained from the finiteelement model of the flat panel. TheNASTRAN nonlinear structural analysis resultsinclude the effect of the geometricimperfections. The simply-supported boundarycondition solution and the clamped boundarycondition solution represent the two extremeconditions for edge support. Due to geometricimperfections which induce bending, theexperimental data, and nonlinear analysissolutions diverge from the linear strength of

7

materials solution as the load is increased. Incomparing the analysis solutions to theexperimental data, the nonlinear analysissolution using the clamped boundary conditionprovides the better estimate of the load beingcarried by the panel at the design allowablestrain value. The strength of materials approachover predicts the magnitude of the strain for agiven applied load since it neglects the effect ofthe geometric imperfection and consequentlybending of the panel. Although Panel 1, Figure16, failed at a strain level greater that the designallowable, the load carried by the panel atfailure was less than that predicted using thestrength of materials approach primarilybecause of panel bending. The NASTRANnonlinear analysis solution assuming theclamped end boundary condition provided themost accurate prediction of the experimentalresults. For Panel 2, Figure 17, the slope of theexperimental curve at low loads is slightlygreater than the slope obtained from thestrength of materials prediction, indicating ahigher than expected modulus of elasticity forthe 8-ply laminate facesheets. Also, the panelfailed at a load slightly below the design loadfor the panel.

In evaluating the test data, three critical issuesarise. First, the maximum measured strain forthe 8-ply specimens fell well below that of the24-ply specimens for the room and 250°Ftemperature

tests. The 24-ply panels failed close to theunnotched allowable strain levels, however, thereappeared to be a strength reduction related tothe thinner, minimum gauge, 8-ply specimens.Consequently, specimens taken from 8-ply and24-ply panel facesheets were prepared forcoupon compression testing to determine anyassociated strength reduction due to the co-curing process for the thinner laminates. Atotal of sixteen 8-ply and sixteen 24-plycoupon specimens were cut from bothfacesheets of four of the panels tested. Thecoupon specimens were all cut 0.5 inches wideand 2.625 inches in length and were

instrumented with back-to-back strain gauges.The core was carefully cut and ground awayfrom the facesheets. Examination of thespecimens revealed that the honeycomb corewas indeed imbedded in the laminated facesheetmaterial. Although much care was taken toapply a uniform axial compressive load to thespecimens, bending was appreciable in mosttests. The bending is believed to be due to acombined effect of geometric imperfections inthe specimens and the asymmetry of thelaminates due to the honeycomb core cuttinginto the inner surface and damaging theinnermost layers of the laminate. Only two 24-ply specimens did not exhibit any significantbending prior to failure and hence were able toprovide axial compressive strength data. Forboth specimens, an ultimate axial compressivestrain of 10500 min/in was measured. Theultimate compressive strain was above thedesign allowable but 12% below the unnotchedallowable of 12000 min/in. Bending was evenmore significant for the 8-ply specimens andunfortunately no comparable strength data wasobtainable. The larger amount of bendingobserved for the 8-ply specimens could beattributed not only to a larger geometricimperfection effect, but to a more severestructural degradation effect of the honeycombcore cutting into the thin laminate.

The second critical issue concerns thedramatic decrease in the maximum measuredstrains and failure load for the 24-plyspecimens in going from the 250°F tests to the300°F tests. The possibility of an unexpectedlylarge strength reduction in the adhesive or coreneeds investigation. Finally, the failure of Panel5, at such a low maximum measured strain andfailure load, even below the design allowable,raises the issue of a possible manufacturingdefect or a possible dramatic compressivestrength reduction due to co-curing.

Concluding Remarks

Fourteen composite honeycomb sandwichpanels were tested to failure under compressiveloading at various test temperatures. The test

8

specimens included panels with both 8-ply and24-ply graphite-bismaleimide facesheets andboth titanium and graphite-polyimide corematerials. For the majority of the panels, theload was introduced through fasteners loadedby steel angles representing double shear splicejoints at both loaded ends of the panels. Thestructural compressive performance of thedifferent panels were evaluated for roomtemperature, 250°F, and 300°F operatingtemperatures.

The 24-ply panels performed well in theroom temperature and 250°F operatingenvironments. Experimental results were fairlyrepeatable for the two panels tested at each testtemperature. The panels experienced failurenear the unnotched failure strains and far abovethe design allowable. The 24-ply panelsexperienced significant degradation of thepanel strength in going from 250°F to the300°F operating environment. Therefore, it isrecommended that further study be conductedbefore utilizing the present 24-ply panel designat temperatures above 250°F.

The thinner 8-ply specimens did notperform as well as the 24-ply specimens. Theyexperienced unexpectedly low failure strainsand failure loads, in addition to much scatter inthe failure data. Therefore, the present 8-plypanel design is not recommended as a reliablepanel design in strength critical areas. However,one factor contributing to the scatter in thefailure loads between each panel is the variationin the out-of-plane geometric imperfections andhence the degree of bending that each panelexhibits. Theoretically, the more bending apanel incurs, the lower the load for a givenstrain at the panel center. Coupon compressiontesting of specimens extracted from the panelfacesheets was performed with limited successbecause these coupons also exhibited largeamounts of bending. Bending of the couponspecimens was believed to be due to geometricimperfections and to the laminates beingunsymmetric as a result of the honeycomb corecutting into the facesheet laminates during co-

curing. The latter effect is believed to havecaused a strength reduction in the facesheets.

References

1. Williams, Louis J.: HSCT Research Gathers Speed.Aerospace America, April 1995, pp. 32Ð37.

2. MSC/NASTRAN User Guide, Version 68, The Macneal-Schwendler Corporation, March 1994.

9

Table 1. Description or Test Specimens

Facesheet Honeycomb core Panel dimensions

Panel Material Layup No.plies

Material Depth,in.

Densitylb/ft3

Length,in.

Width,in.

Testtemp, °F

1 IM7/5260 (45/-45/0/90)s8 Ti 1 6 35 11.875 RT

2 IM7/5260 (45/-45/0/90)s8 Ti 1 6 35 11.750 250

3 IM7/5260 (45/-45/0/90)s8 Ti 1 6 35 11.375 300

4 IM7/5260 (45/-45/0/90)s8 Ti 1 6 35 11.563 250

5 IM7/5260 (45/-45/0/90)s8 HFT-G 1 6 35 12.000 RT

6 IM7/5260 (45/-45/0/90)s8 HFT-G 1 6 35 12.000 250

7 IM7/5260 (45/-45/0/90)s8 HFT-G 1 6 35 12.000 300

8 IM7/5260 (45/-45/0/90)s8 HFT-G 1 6 35 12.000 RT

9 IM7/5260 (45/90/-45/0)3s24 HFT-G 1 6 36 12.063 RT

10 IM7/5260 (45/90/-45/0)3s24 HFT-G 1 6 36 12.000 300

11 IM7/5260 (45/90/-45/0)3s24 HFT-G 1 6 36 11.438 250

12 IM7/5260 (45/90/-45/0)3s24 HFT-G 1 6 36 12.000 250

a13 IM7/5260 (45/90/-45/0)3s24 HFT-G 1 6 36 12.000 RT

a14 IM7/5260 (45/90/-45/0)3s24 HFT-G 1 6 36 11.250 300

aPanel has potted ends with no joint attachment.

Table 2. Compressive Allowable Strain DataUsed for IM7/5260 Laminates

Temperature,°F

70

250

300

Unnotched,µin/in

Design,µin/in

12 000

Compressive allowable strains

10 440

9 900

6500

5725

5400

1 0

Table 3. Experimental Failure Data

250°F test temperature

Room temperature tests

Panel Plies Core Failure load,kips

Maximum strain,µin/in

Failure mode

1

5

8

9a13

8

8

8

24

24

Titanium

Composite

Composite

Composite

Composite

–58.3

–46.8

–65.2

–193.6

–198.5

–8 845

–5 907

–9 641

–12 324

–11 555

Facesheet strength

Facesheet strength

Facesheet strength

Buckling

Buckling

2

4

6

11

12

8

8

8

24

24

Titanium

Titanium

Composite

Composite

Composite

–48.5

–46.1

–51.9

–153.6

–164.3

–7 015

–7 132

–7 022

–10 346

–10 148

Facesheet strength

Facesheet strength

Facesheet strength

Buckling

Buckling

300°F test temperature

3

7

10a14

aPanel had potted ends with no joint attachment.

8

8

24

24

Titanium

Composite

Titanium

Composite

–47.6

–58.6

–142.6

–142.9

–7 146

–8 687

–7 462

–8 146

Facesheet strength

Facesheet strength

Facesheet strength

Disbond

1 1

Figure 1. Illustration of a high speed civil transport concept.

1 2

l

w

t

Honeycomb core

Compositefacesheet

Pad-up plies

Steel angle

Fastener

c

Front view Side view

Figure 2. Schematic diagram of test specimen. Note: drawing is not to scale.

1 3

Symmetry

w/4

w

w/2

6.5 in.

l/3

l/2

Axial strain gauge

Axial and transverse strain gauge

Thermocouple

Figure 3. Typical instrumentation diagram where all gauges are mounted back-to-back and mirrored in the length, l,direction.

1 4

Heatedplaten

Ceramicinsulation

Loadingplaten

One half ofclamshelloven

Angle iron forload introduction

Testpanel

Figure 4. Photograph of failed panel located in test stand.

1 5

z

Core: E = 0

ν = 0.3

Gxz = 82000 psi

Gyz = 59000 psi

Steel: E = 2.9 × 107 psi

ν = 0.28

α = 6 × 10–6

Facesheet: E = 8.63 × 106 psi

ν = 0.31

Gxz = 8.5 × 105 psi

Gyz = 8.5 × 105 psi

α = 1.49 × 10–6 in/in

x

Figure 5. Schematic diagram of layers represented in the composite plate finite element model and their materialproperties. Note: drawing is not to scale.

1 6

1113 nodes

1040 quadrilateral plate elements

x

yz

Figure 6. Finite element model of composite panel.

1 7

6

65

5

4

4

3

3

2

2

z, in.

0

.0322

.0483

.0161

.0805

.0967

.1289

6

5

4

3

2

1

1

Figure 7. Contour plot of Panel 2 out-of-plane measurements.

1 8

Facesheetstrength failure

Figure 8. Photograph of Panel 4, with 8-ply facesheets, after testing to failure.

1 9

Facesheetstrength failure

Figure 9. Photograph of Panel 11, with 24-ply facesheets, which buckled prior to failure.

2 0

Disbond area

Figure 10. Photograph of Panel 14, with 24-ply facesheets and potted ends only, after testing to failure.

2 1

–14000

–12000

–10000

–8000

–6000

–4000

–2000

030025020015010050

8-ply24-ply

Stra

in, µ

, in/

in

Unnotched allowable

Design allowable

Temperature, °F

Figure 11. Maximum measured failure strain data.

–200

–150

–100

–50

0 30025020015010050

8-ply 24-ply

24-ply design load

8-ply design load

Failu

re lo

ad, k

ips

Temperature, ° F

Figure 12. Panel failure load test data.

2 2

–70000

–60000

–50000

–40000

–30000

–20000

–10000

0 –10000–8000–6000–4000–2000

Loa

d, lb

Strain, µ, in/in

Figure 13. Back-to-back measured strains at center of panel facesheets for Panel 1 having 8-ply facesheets.

–70000

–60000

–50000

–40000

–30000

–20000

–10000

0 –10000–8000–6000–4000–2000

Loa

d, lb

Strain, µ, in/in

Figure 14. Back-to-back measured strains at center of panel facesheets for Panel 5 having 8-ply facesheets.

2 3

–200000

–150000

–100000

–50000

0 –10000–8000–6000–4000–2000

Loa

d, lb

Strain, µ, in/in

Figure 15. Back-to-back measured strains at center of panel facesheets for Panel 11 having 8-ply facesheets.

–100000

–80000

–60000

–40000

–20000

0

Loa

d, lb

–12000–10000–8000–6000–4000–2000

Design allowable strain

Nonlinear analysis: simply-supported endsNonlinear analysis: clamped endsExperimentStrength of materials

Linear buckling:simply-supported ends

Linear buckling:clamped ends

Failure

Strain at center of panel, µ, in/in

Figure 16. Room temperature NASTRAN nonlinear composite plate finite element analysis for Panel 1 (Geometricimperfection modeled from first cosine shape buckling mode).

2 4

–100000

–80000

–60000

–40000

–20000

0

Loa

d, lb

–12000–10000–8000–6000–4000–2000

Design allowable strain

Nonlinear analysis: simply-supported endsNonlinear analysis: clamped endsExperimentStrength of materials

Failure

Strain at center of panel, µ, in/in

Figure 17. NASTRAN nonlinear composite plate finite element analysis for Panel 2 at 300°F (Geometricimperfection modeled from actual out of plane measurements mapped to nodal points).

REPORT DOCUMENTATION PAGE Form ApprovedOMB No. 0704-0188

Public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources,gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of thiscollection of information, including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 1215 JeffersonDavis Highway, Suite 1204, Arlington, VA 22202-4302, and to the Office of Management and Budget, Paperwork Reduction Project (0704-0188), Washington, DC 20503.

1. AGENCY USE ONLY (Leave blank) 2. REPORT DATE

April 19983. REPORT TYPE AND DATES COVERED

Technical Publication4. TITLE AND SUBTITLE

Evaluation of Composite Honeycomb Sandwich Panels Under CompressiveLoads at Elevated Temperatures

5. FUNDING NUMBERS

537-06-34-20

6. AUTHOR(S)

Sandra P. Walker

7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES)

NASA Langley Research CenterHampton, VA 23681-2199

8. PERFORMING ORGANIZATIONREPORT NUMBER

L-17666

9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES)

National Aeronautics and Space AdministrationWashington, DC 20546-0001

10. SPONSORING/MONITORINGAGENCY REPORT NUMBER

NASA/TP-1998-207645

11. SUPPLEMENTARY NOTES

12a. DISTRIBUTION/AVAILABILITY STATEMENT

Unclassified-UnlimitedSubject Category 27 Distribution: StandardAvailability: NASA CASI (301) 621-0390

12b. DISTRIBUTION CODE

13. ABSTRACT (Maximum 200 words)

Fourteen composite honeycomb sandwich panels were tested to failure under compressive loading. The test spec-imens included panels with both 8 and 24-ply graphite-bismaleimide composite facesheets and both titanium andgraphite-polyimide core materials. The panels were designed to have the load introduced through fasteners attachedto pairs of steel angles on the ends of the panels to simulate double shear splice joints. The unloaded edges wereunconstrained. Test temperatures included room temperature, 250F, and 300F. For the room and 250F temp-erature tests, the 24-ply specimen failure strains were close to the unnotched allowable strain values and failureloads were well above the design loads. However, failure strains much lower than the unnotched allowable strainvalues, and failure loads below the design loads were observed with several of the 8-ply specimens. For eachindividual test temperature, large variations in the failure strains and loads were observed for the 8-ply specimens.Dramatic decreases in the failure strains and loads were observed for the 24-ply specimens as the test temperaturewas increased from 250F to 300F. All 8-ply specimens appeared to have failed in a facesheet strength failuremode for all test temperatures. The 24-ply specimens displayed appreciably greater amounts of bending prior tofailure than the 8-ply specimens, and panel buckling occurred prior to facesheet strength failure for the 24-plyroom and 250F temperature tests.

14. SUBJECT TERMS

compression testing, composite honeycomb sandwich panels, elevated temperatures15. NUMBER OF PAGES

2916. PRICE CODE

A0317. SECURITY CLASSIFICATION

OF REPORT

Unclassified

18. SECURITY CLASSIFICATIONOF THIS PAGE

Unclassified

19. SECURITY CLASSIFICATION OF ABSTRACT

Unclassified

20. LIMITATION OF ABSTRACT

NSN 7540-01-280-5500 Standard Form 298 (Rev. 2-89)Prescribed by ANSI Std. Z-39-18298-102