Journal of Composite Materials Dynamic impact behavior of ... · as the shear modulus of the core increased. The failure mechanism transitioned from shear cracks in the core to delamination

DRAFT

Dynamic impact behavior of syntacticfoam core sandwich composites

Journal of Composite MaterialsXX(X):1–11c©P. Breunig, et al. 2018

Reprints and permission:sagepub.co.uk/journalsPermissions.navDOI: 10.1177/ToBeAssignedwww.sagepub.com/

SAGE

P. Breunig1, V. Damodaran1, K. Shahapurkar2, S. Waddar3, M. Doddamani3, P. Jeyaraj3, P.Prabhakar1

AbstractSandwich composites and syntactic foams independently have been used in many engineering applications. However,there has been minimal effort towards taking advantage of the weight saving ability of syntactic foams in the coresof sandwich composites, especially with respect to the impact response of structures. To that end, the goal ofthis study is to investigate the mechanical response and damage mechanisms associated with syntactic foam coresandwich composites subjected to dynamic impact loading. In particular, this study investigates the influence of varyingcenosphere volume fraction in syntactic foam core sandwich composites subjected to varying dynamic impact loading,and further elucidates the extent and diversity of corresponding damage mechanisms. The syntactic foam cores are firstfabricated using epoxy resin as the matrix and cenospheres as the reinforcement with four cenosphere volume fractionsof 0% (pure epoxy), 20%, 40%, and 60%. The sandwich composite panels are then manufactured using the vacuumassisted resin transfer molding process with carbon fiber/vinyl ester facesheets. Dynamic impact tests are performedon the sandwich composite specimens at two energy levels of 80J and 160J, upon which the data is post-processedto gain a quantitative understanding of the impact response and damage mechanisms incurred by the specimens. Aqualitative understanding is obtained through micro-Computed Tomography scanning of the impacted specimens. Inaddition, a finite element model is developed to investigate the causes for different damage mechanisms observed inspecimens with different volume fractions.

KeywordsSyntactic Foam, Sandwich Composites, Dynamic Impact, VARTM, Cenospheres

Introduction

Composite materials typically allow for structural propertiesto be optimized such that the strength and weight constraintsare more easily met in the design of everyday structures. Thisis evident as composites continue to become more prevalentin aerospace, naval, and civil applications1. Sandwichcomposites typically consist of two stiff outer facesheets(away from the natural axis) usually made of fiber reinforcedpolymer that sandwich a lightweight core between them.This provides stiffness to the cross section and resistsmajority of the bending stresses. The lightweight coreconnects the two facesheets and assists with shear transferin the section.

A large body of research has been conducted by previousresearchers to better understand the behavior of sandwichcomposites using different materials and loading conditions.A few notable works reported on the low-velocity impactbehavior of foam core sandwich composites includes, but isnot limited to2–11. Schubel et al. investigated the low velocityimpact behavior of PVC foam cores with woven carbonfiber/epoxy facesheets and compared the performance toquasi-static tests for the same materials3. The results showedthat the damage observed in the low velocity impacttests was comparable to the damage in quasi-static testsat the same compressive strain level. Work by Hazizanand Cantwell12 reported the low-velocity impact responseof sandwich structures with foam cores and glass fiber

reinforced facesheets. Results showed that for a specificimpact energy, the maximum recorded force increasedas the shear modulus of the core increased. The failuremechanism transitioned from shear cracks in the core todelamination between the core and facesheet as the densityof the core increased. Elamin et. al13 evaluated the damageof sandwich structures under dynamic impact loading,which were exposed to arctic conditions with temperaturesranging from 23◦C to -70◦C. Polyvinyl chloride (PVC) foamcores were used with facesheets of 0◦/90◦ woven carbonfiber reinforced laminate with epoxy matrix. The articleconcluded that the peak impact force recorded decreased asthe in-situ test temperature decreased. Also, using micro-computed tomography, the authors noted that the specimensexperienced higher degree of damage at low temperatures ascompared to higher temperatures.

Syntactic foams are closed cell composite foams withhollow micro-spheres dispersed in a matrix resin. The closedcell structure provides excellent mechanical properties,

1University of Wisconsin-Madison2Sanjeevan Engineering and Technology Institute Kolhapur, India3National Institute of Technology Karnataka, India

Corresponding author:P. Prabhakar, University of Wisconsin-Madison, Engineering Hall Room2210, 1415 Engineering Drive Madison, WI 53706Email: [email protected]

Prepared using sagej.cls [Version: 2017/01/17 v1.20]

DRAFT

2 Journal of Composite Materials XX(X)

like high strength and low density, in addition to lowermoisture absorption as compared to open cell foams14.Hence, syntactic foam cores in sandwich composites ensurehigh rigidity and strength of the sandwich structures ascompared to other polymeric foam cores14–19. Few widelyknown applications of syntactic foams are in components forboat decks, ribs, hulls and floatation modules for offshorestructures20. In addition, they are also used in deep seaapplications like remote operated vehicles, submarines andunderwater pipelines. Few potential applications of syntacticfoam core sandwich composites could be in building facades,bridge decks, and other civil infrastructure.

In the current study, syntactic foams are used asthe core material in sandwich composites. Specifically,cenospheres (fly ash particles) are used as the hollowmicro-spheres in these syntactic foams. Past researchershave investigated the behavior of syntactic foams withengineering glass (Sodalime-borosilicate) microballons asthe filler material21–23. However, dealkalization of glass hasbeen shown to degrade such syntactic foams24. Fly ash beinga byproduct of coal plants primarily consists of alumina andsilica. Hence, use of cenospheres in syntactic foams can aidin minimizing waste from the environment, while creatingsyntactic foams with better properties as shown by previousresearchers25–30.

Extensive studies on the mechanical behavior of syntacticfoams have been performed by previous researchers explor-ing their suitability for a wide range of applications21,31–33.In the work by Gupta et al.34, it was shown that the com-pressive strength and modulus of syntactic foams increasedas the internal radius of cenospheres was reduced, whileholding all other parameters fixed. Different types of testshave been performed on syntactic foams, such as three-point bending tests in flexure35–37 and short beam sheartests38–40 to determine their response under such types ofloading. Previous work by the current authors41,42 investi-gated the behavior of cenosphere reinforced syntactic foamsin compression and flexure over a range of temperatures. InGarcia et al.41, it was observed that the flexural modulus ofcenosphere/epoxy syntactic foams increased and the flexuralstrength decreased with cenosphere volume fraction. Addi-tional analysis showed that the failure strains decreased asthe cenosphere volume fraction increased.

Although syntactic foams have not been explored as thecore material for sandwich composites under impact loading,they have been studied when subjected to quasi-staticloading. Work by Gupta and Woldesenbet43 investigated theflexural properties of sandwich composites with syntacticfoam cores. It was reported that the effect of micro-balloonwall thickness to diameter ratio had little effect on thestrength of the specimens under three-point bend tests asthe failure mechanism was tensile tearing of the facesheets.However, high shear stresses in short beams shear testsresulted in shear cracks within the syntactic foams.

The present study expands on the knowledge of sandwichcomposites and syntactic foams by using cenosphere/epoxysyntactic foams as the core of sandwichcomposites, andevaluate their dynamic impact response. The sandwichcomposites are tested under low-velocity impact loading toinvestigate their dynamic impact response, as well as identifykey failure mechanisms and elucidate the reasons for the

Figure 1. Manufactured cenosphere/epoxy syntactic foam core.Inset shows microCT image of internal distribution of cenospheresin epoxy matrix.

observed behaviors. The purpose of this study is to gain anunderstanding of the influence of cenosphere volume fractionin syntactic foams on the impact behavior of the sandwichcomposites subjected to different impact energies.

Methods and Materials

Constituent MaterialsSandwich composites in this experimental study consistedof three major constituent materials: a syntactic foam core,woven carbon fiber facesheets, and vinyl ester resin. Thesyntactic foam core was comprised of cenospheres andLapox L-12 epoxy resin with K-6 hardener. The cenospheresbeing hollow particles which are a byproduct of coalproduction. Each facesheet consisted of 4 layers of 3K plainweave carbon fiber procured from Fibre Glast DevelopmentsCorp., and was cut to fit the dimensions of the core. Eachfacesheet layer had identical orientation and stacking of[(0/90)w4/core/[(0/90)w4]. Here, the subscript w representswoven carbon fiber layers. Commercially available vinylester resin and Methyl Ethyl Ketone Peroxide (MEKP), bothprocured from Fibre Glast Developments Corp., were usedas the matrix in the facesheets.

ManufacturingFoam Core The syntactic foam cores were fabricated bymixing a weight fraction (equivalent to 20, 40 and 60 volume%) of cenospheres with Lapox L-12 epoxy resin and K-6hardener at room temperature. A homogeneous and uniformslurry was assured by gentle stirring. Further, 10% by weighthardener was added to the slurry, followed by degassing themixture for 4 minutes prior to pouring into aluminum molds.Curing of cast slabs was conducted at room temperature for24 hours and subsequently post cured for 3 hours. Differentcompositions of foam samples were fabricated by varyingthe cenosphere volume fraction. Specimens with 0% (pureepoxy), 20%, 40%, and 60% cenosphere volume fractionwere prepared for use in sandwich composites. Distribution

Prepared using sagej.cls

DRAFT

P. Breunig, et al. 3

of cenospheres in the syntactic foam core is shown viamicroCT scan in Figure 1.

The weight fractions of the constituent materials (Epoxyand cenospheres) for known volume fractions werecalculated using the formulas given in Equation 1.

Wf =

ρfρm

ρfρmVf + Vm

Vf ; Wm =1

ρfρmVf + Vm

Vm (1)

where,W represents weight fraction, V represents volumefraction, ρ represents density, the subscripts f and mrepresent filler and matrix, respectively. The density ofEpoxy and cenosphere particles are considered to be 1192kg/m3 and 920 kg/m3 based on measurements previouslyconducted by the authors.28,41,42 Based on the dimensionsof a mold, the volume of composite (Vc) to be preparedwas found. The weight of cenospheres and epoxy for knownvolume fractions were calculated using the values of weightfraction.

Sandwich Composite The sandwich composites weremanufactured with the vacuum assisted resin transfermolding (VARTM) process. The VARTM process wasconducted on a flat aluminum mold plate. Vacuum wascreated between the mold plate, vacuum sealant tape, andvacuum bag (Stretchlon 800 Bagging Film) using twovacuum pumps (60-80 MPa vacuum pressure). 100 parts ofvinyl ester resin was mixed with 1.25 parts of MEKP bymass as per the manufacturer’s instructions. The first pumpwas used to infuse this resin mixture through the specimenand was removed once the infusion process was complete (5minutes in duration). This pump corresponded to an innervacuum bag. The outer vacuum bag applied pressure duringthe resin curing process which lasted for 3 hours from thetime the vinyl ester was initially mixed. The specimens werecured for at least 24 hours after resin infusion, after whichthey were removed from the mold.

Several best practices were employed to ensure thefabrication of high quality sandwich composites. First, acombination of HDPE Infusion Flow Media, cotton breather,and 1586 PTFE Coated FG (all procured from Fibre GlastDevelopments Corp.) were used in addition to the carbonfiber to ensure a consistent distribution of vinyl ester resinand allow for easy removal of the specimens from themold upon curing. Second, the vinyl ester resin mixturewas degassed prior to the infusion to help prevent voidsforming in the facesheets. The degassing continued until airbubbles in the resin were no longer visually detectable inthe resin. Finally, resin dams were constructed to direct theflow of resin through the facesheets and not just around thespecimens. Due to the relatively low viscosity of vinyl esterresin and large thickness of the sandwich composite, thevinyl ester resin was susceptible to flow around the specimenforming many unwanted voids and poor bonding betweenthe core and facesheets. The dams consisted of extra vacuumtape applied to the sides of the specimen perpendicular tothe direction of resin flow. A schematic of the manufacturingprocess is shown in Figure 2.

The dimensions of the manufactured sandwich compositeswere 175mm long by 125mm wide, from which the sampleswere water jet cut to nominal values of 150 mm x 100mm rectangles as per the ASTM D776644 standard . The

Figure 2. Vacuum assisted resin transfer molding (VARTM)process during the sandwich fabrication stage. Section A-A depictsthe front face of the sandwich, which shows flow of resin in frontof the core. The dams assist in diverting the resin to flow throughthe carbon fiber facesheets. The red arrows depict the flow path ofthe resin.

Table 1. Summary of manufactured density of sandwichcomposite specimens.

Cenosphere Volume Fraction 0% 20% 40% 60%

Average (kg/m3) 1314 1220 1147 1095± 3 ± 12 ± 17 ± 17

Change (%) 0 -7.1 -12.7 -16.7

actual average dimensions of the tested specimens were152.5 mm x 101.5 mm x 28.0 mm with standard deviationsof 0.2 mm, 0.3 mm, and 0.3 mm, respectively. The densitiesof these samples were measured and are summarized inTable 1. These densities were measured after manufacturingthe sandwich composite, which allowed for comparison ofthe in-service state rather than determining those of justthe cores. The densities of the specimens decreased as thecenosphere volume fraction increased.

Dynamic Impact TestingThe sandwich composite specimens were tested underdynamic impact (ASTM D7766)44 at two different energylevels corresponding to 80J and 160J. These two energylevels were determined via a preliminary testing program ofthe manufactured sandwich composites. They were chosendue to distinct failure mechanisms observed at the twoenergy levels. In total, twenty-four sandwich composite


DRAFT


Figure 3. CEAST 9350 Impact Machine with anti-reboundmechanism and testing chamber (inset) with fixture to simulateclamped boundary conditions consistent with ASTM D7766.

specimens were tested with twelve at each energy level. Ofthe twelve specimens at each energy level, four differentvolume fractions of cenospheres in the core were usedyielding three specimens for each test case.

Drop-weight impact tests were performed using a CEAST9350 Accelerated Drop Tower Impact System fitted witha hemispherical striker at the University of Wisconsin -Madison. The clamped boundary conditions shown in Figure3 are consistent with ASTM D776644. Force, displacement,energy versus time responses were recorded by the dataacquisition system ‘CEAST DAS 8000 Junior’ for each testat a sampling rate of 500 kHz. A 5 m/s impact velocity, wellwithin the range of low velocity impact, was chosen for thetests to ensure uniform propagation behavior between thestriker and specimen at different energy levels. To achievethis constant impact velocity, additional mass was added tothe striker in order to attain impact energy of 160J. In all,6.5kg of total mass was used for the 80J tests, while 12.5kgwas used for the 160J tests. The impact testing machine’santi-rebound mechanism was activated to avoid multipleimpacts on the sample.

Finite Element Model DescriptionA finite element (FE) analysis was performed to understandthe micro-mechanical material response of the syntacticfoam at different cenosphere volume fractions. Theindentation of the steel striker onto the syntactic foam wasmodeled as a quasi-static loading rather than an impactanalysis to obtain a qualitative understanding of the straindistribution. Strain rate effects were ignored in the epoxyas the goal of this analysis was to obtain a qualitativeunderstanding of the strain distribution in the syntactic foamcore underneath the striker location. With the output fromthe FE model, strain contours were analyzed and comparedwith failure mechanisms observed in the micro-CT scans tohelp explain the possible causes for damage mechanisms atcertain volume fractions.

The finite element model was divided into two subregionsas shown in Figure 4. A finely meshed two-phase region withthe epoxy matrix and hollow spherical cenosphere inclusionsdirectly under the impact location, and a larger homogenizedmedia (shown in green in Figure 4) with coarser meshaway from the impact location. The purpose of the two

Figure 4. Domain of finite element model superposed on theexperimental test setup. The enlarged square is the extent of thetwo phase media region. The vertical centerline restricts horizontaltranslation and enforces symmetric boundary condition.

Table 2. Comparison between calculated homogenized foammodulus values using cenosphere modulus of 40 GPa andreported compressive modulus values for the syntactic foamcore.

Cenosphere Volume Fraction 0% 20% 40% 60%

Reported Modulus (GPa) 42 3.4 3.9 4.7 4.8Calculated Modulus (GPa) 3.4 3.9 4.4 4.9Reported Strength (MPa) 42 104.8 100.8 98.8 92.1

subregions was to simulate a larger syntactic foam domainwhile capturing the details of the behavior in the vicinity ofthe impact location and including the effects of boundaryconditions away from the impact location. A symmetricboundary condition was considered about the axis of thestriker, which was introduced as a restriction in the horizontaltranslation on the left face of the domain as shown in Figure4. Further, contact conditions between the syntactic foamcore and the steel striker at the top, and the support atthe bottom ensured more realistic boundary conditions. Toreduce the complexity of the FE model, carbon fiber facesheets were not modeled as the goal of this simulation was toinvestigate the damage patterns in the foam core.

Material properties for the homogenized core andthe epoxy matrix were obtained from recent work byShahapurkar et. al42 where the authors investigated thecompressive modulus and strength of cenosphere/epoxysyntactic foam cores. The average compressive elasticmodulus values from Shahapurkar et. al42 were used for thehomogenized region of the finite element model. Using arule of mixtures approach, the average compressive modulusof the cenospheres was calculated. In this calculation, thecenospheres were idealized as hollow spherical particles witha constant wall thickness of 5µm and a mean diameter of110µm42 to back-calculate the equivalent elastic modulus ofthe cenospheres which was found to be 40 GPa. The inputvalues for the modulus are summarized in Table 2.

To account for the crushing of syntactic foam cores,the cenosphere, epoxy matrix, and homogenized region


DRAFT


were modeled as elastic-perfectly plastic materials. Thematrix in the two-phase region was modeled to yield at104.8MPa, which is a mean value of the compressivestrength reported by the authors in Shahapurkar et al.42

for the 0% cenosphere volume fraction. Similarly, thehomogenized media was modeled to yield at the meanvalues reported for the respective volume fractions of 20%,40% and 60%. The cenospheres were modeled to crush ata stress value vastly different from that of the matrix tohelp differentiate the materials in the output strain field. Anassumed yield stress of 150 MPa was chosen. A verificationanalysis showed that both higher and lower assumed yieldstress values for cenospheres as compared to that of matrixstrength produced similar strain patterns, which help withqualitatively explaining the crack propagation.

Results and Discussion

Internal Damage through MicroCT ScanningMicro-Computed Tomography (microCT) scans of theimpacted specimens were obtained using a Zeiss Metrotom800 at the University of Wisconsin-Madison. The scansallowed for the analysis of the damage mechanisms inthe facesheets and crack propagation in the syntactic foamcores. MicroCT images are shown in Figure 5, whichare characteristic images corresponding to each cenospherevolume fraction and energy level. The rows of the tablecorrespond to cenosphere volume fraction percentage and thecolumns correspond to impact energy. A circular pattern isobserved in the center of some of the images, which was avisual anomaly from the microCT process and not a physicalcharacteristic of any specimen.

Figures 5a and 5b show two specimens with 0%cenosphere volume fraction impacted at 80J and 160J each.For both energy levels, a majority of the damage occurredlocally around the impact location with minimal damagein the core. For the 80J specimen, a small indentationwas observed in the facesheet which corresponded tomatrix cracking. For the 160J specimen, in addition tomatrix cracking, fiber tearing and delamination between thefacesheet and core was visible.

The 20% cenosphere volume fraction specimens exhibiteda high degree of shear cracking that propagated throughspecimen thickness as shown in Figures 5c and 5d. Thisis an undesirable failure mechanism as it typically leadsto loss of structural integrity. On the other hand, localizedcrushing would be a less undesirable damage mechanism asthe damage is contained in a localized region. At 80J, shearcracks emanated conically outward from the impact location.Delamination from excessive deformation was also observedbetween the bottom facesheet and core. An additionaldamage mechanism was visible directly underneath theimpact location which consisted of a slight discoloration ofthe core. This was attributed to localized compression of thecore caused by the collapse of individual cenospheres andcrushing of the surrounding matrix during impact. Similardamage mechanisms manifested at 160J, but to a greaterdegree.

The microCT images for the 40% cenosphere volumefraction specimens are shown in Figures 5e and 5f. Localizedcompression was observed under the impacted face at

both 80J and 160J impact energies, while shear crackingmanifested only in the 160J case. The damage mechanismsfor the 60% cenosphere volume fraction specimens werelocalized as compared to 20% and 40% cenosphere volumefraction specimens, and are shown in Figures 5g and 5h.Under both impact energies, the 60% specimens exhibitedlocalized compression in the foam core and fiber fracturein the top facesheet at the impact location. The localizeddamage in the 60% cenosphere volume fraction specimenswas higher than any of the specimens tested. No shearcracking in the core or non-localized delamination betweenthe core the facesheets was noticable. The damage trendshown in the microCT images between specimens ofdifferent volume fractions translated into distinct mechanicalresponses from the impact tests, as discussed next.

Mechanical Response from Impact TestsThe output data from each impact test was post-processedto determine their mechanical responses. Figure 6 showscharacteristic force-displacement responses for the tests.Since the carbon fiber facesheets are stiffer than the foamcore, the initial slope of the force-displacement plots areidentical up to the point of initial penetration through thetop facesheet. However, once the facesheets have beenpenetrated by the impact striker, the stiffness of the foam coreis dominant and the stiffness of the specimen is observedto decrease as the cenosphere volume fraction increases,regardless of the impact energy level. Based on the force-displacement responses, the initial stiffness ranged from 10-13 kN/mm for both the 80J and 160J tests. The peak contactforce recorded for 160J impact energy was higher than thosecorresponding to 80J for all volume fractions of cenospheres.Moreover, for both impact energies, the peak force reducedwith increasing cenosphere volume fraction. Sharp verticaldrops in the post peak regime of the force-displacementresponses were observed for specimens with cenospherevolume fractions of 20% and 40%. This corresponded tomore damage in the specimens, which resulted in largerimpact striker displacements and lower impact forces. Thestriker displacements and peak impact forces are summarizedin Figure 7. An increase in striker displacement and adecrease in peak impact force was observed for the 20%cenosphere volume fraction specimens tested at 160J. Inaddition, these specimens experienced the largest variationin test results as compared to the other volume fractions andimpact energy levels.

The summary plots shown in Figure 7 do not account forthe decrease in density due to increased cenosphere volumefraction in the core. To account for the changing densities,summary plots showing the specific striker displacementand peak impact force are shown in Figure 8. These plotsare very similar to those in Figure 7, however, the valueswere divided by normalized weight ratios which changefor specimens with different cenosphere volume fractions.The weight normalization, or in other words specific values,highlights the influence of weight reduction on the propertiesof the core. As a result, specific striker displacement andspecific peak impact force both increased as the cenospherevolume fraction increased.

Both the summary plots and microCT images have impliedthat the 20% specimens experienced the most damage and


DRAFT


(a) 0% volume fraction 80J (b) 0% volume fraction 160J

(c) 20% volume fraction 80J(d) 20% volume fraction 160J

(e) 40% volume fraction 80J (f) 40% volume fraction 160J

(g) 60% volume fraction 80J (h) 60% volume fraction 160J

Figure 5. Typical microCT cross sections for each cenosphere volume fraction and energy level tested.


DRAFT


Figure 6. Force-Displacement plots of impact specimens. (A) Specimens tested at 80J. (B) Specimens tested at 160J.

Figure 7. Summary plots of impact tests showing average values and standard deviation. (A) Striker displacement over the range ofcenosphere volume fractions. (B) Impact force over the range of cenosphere volume fractions.

variation in peak impact force among all of the testedspecimens. Further, the damage increased at the 160J impactenergy level. This is confirmed in Figure 9, which is asummary plot of the normalized absorbed energy for allthe tests conducted. The absorbed energy is the amount ofenergy absorbed by the specimen upon impact by the striker,which is determined graphically from the energy versus timegraph as depicted in Figure 9. To determine the normalizedabsorbed energy, the absorbed energy is divided by thecorresponding impact energy, which is the peak value in theenergy versus time plot.

It is evident from the normalized absorbed energy plotsin Figure 9 that 20% and 40% cenosphere volume fractionspecimens were relatively more damaged than 0% and60% specimens at 160J impact energy as compared to 80J.With that said, specimens with 0% and 60% cenospherevolume fractions experienced similar extent of damage atboth energy levels of 80J and 160J. Higher apparent damagemanifested by 20% and 40% cenosphere volume fractionspecimens can be related to the shear cracking damage

mechanism observed in these specimens as compared tothe 60% volume fraction specimens that exhibited localizedcompression and absorbed similar levels of normalizedenergy at the two energy levels tested.

Damage Mechanism Observations and Causes

The volume fraction of cenospheres in the syntactic foamcores influenced the damage mechanisms of the sandwichcomposites, especially under high impact energy of 160J.This was more evident upon comparing the mechanicalresponse results and microCT images. Large shear cracksand delamination in the microCT scans corresponded tosharp vertical drops in the average maximum impactforce recorded and correspondingly higher maximum strikerdisplacements. Shear cracking, a highly undesirable failuremechanism as it typically leads to global failure, wasobserved in the core along with delamination between thecore and the facesheets. This damage mechanism was mostcommonly observed in the 20% cenosphere volume fraction


DRAFT


Figure 8. Summary plots of impact tests showing average values and standard deviation divided by a normalized ratio of specimenweight. (A) Striker displacement over the range of cenosphere volume fractions. (B) Impact force over the range of cenosphere volumefractions.

Figure 9. (A) Energy vs. Time output for an arbitrary test specimen. (B) Average normalized absorbed energy for all cenosphere volumefractions and energy levels.

Figure 10. In-plane shear strain output of the two-phase media region shown in Figure 4 for the three different non-zero cenospherevolume fractions. (A) 20%. (B) 40%. (C) 60%.

specimens tested under 160J impact energy, but specimenswith 20% cenosphere volume fraction tested at 80J and

specimens with 40% cenosphere volume fraction tested at160J also experienced shear cracking in few specimens.


DRAFT


Localized compression (crushing) was the other commondamage mechanism associated with the core and occurreddirectly underneath the impact striker and manifested itselfin a discolored region in the microCT scan images.

From the damage mechanism observations, it can beconcluded that the behavior of the sandwich compositeis more favorable with higher cenosphere volume fractionin the syntactic foam core. That is, damage was morelocalized with increasing cenosphere volume fraction. Anexplanation for this observation stems from the behaviorof the cenospheres in the syntactic foam. In the microCTimages, it was observed that foam was crushed under thestriker regardless of the energy level and cenosphere volumefraction. In samples with low cenosphere volume fraction,stress redistribution would cause a crack to propagate in thematrix between two cenospheres that are situated a distancefrom each other. On the other hand, as the cenospherevolume fraction increases, the cenospheres are situated closeto each other. For the crack to propagate, it is hypothesizedthat it would either require crushing of more cenospheresor propagating cracks around additional cenospheres in atortuous path prior to manifesting as large cracks in the core.This supports the observation of large shear cracks at 20%and 40% volume fractions samples, whereas local crushingin the 60% volume fraction samples.

Corroboration of Damage Mechanisms throughComputational Modeling

The results of the finite element analysis show the correlationbetween the cenosphere volume fraction and the level ofdamage underneath the impact striker. Matrix shear straincontour plots are shown in Figure 10 for models with 20%,40% and 60% cenosphere volume fractions. The pure epoxymodel was not included because it did not exhibit stressconcentrations within the core. These plots correspond to astriker displacement of 1.6 mm for each case. The purposeof this finite element analysis was to compare the extent oflocalized strain underneath the striker, which is anticipatedto correspond to the level of damage.

From the 60% cenosphere volume fraction model, it isobserved that the areas of high strains are more dispersedand intermixed with areas with lower strain values. As aresult, a web of cracks are more likely to form in theseareas of closely packed cenospheres which helps promote thelocalized compression failure mechanism under the impactlocation. This localized web of cracks are expected todecrease as the cenosphere volume fraction decreases dueto the larger distances between adjacent cenospheres. As thedistance between cenospheres increases, failure mechanismwith few large cracks is expected, which can be seen asa continuous region of high strain as depicted in the 20%cenosphere volume fraction model in Figure 10. Therefore,having a higher volume fraction of cenospheres is favorableas the cenospheres help dissipate the strain energy by eithercrushing or driving the crack around them, thereby, splittinglarge shear cracks into multiple smaller cracks. These smallercracks are more localized and are constrained in the vicinityof the impact location, thereby containing the damage to alocalized region.

ConclusionAlthough there has been much effort to individuallycharacterize both syntactic foams and sandwich compositesunder dynamic impact loading, there has been relatively littlework to characterize and quantify the behavior of sandwichcomposites with syntactic foam cores under the same. Inthis study, sandwich composites with syntactic foam coresand varying cenosphere volume fractions were tested at twodifferent impact energy levels to gain an understanding oftheir mechanical responses as well as the damage level anddamage mechanisms. In addition, a finite element model wasdeveloped to investigate the causes of the observed failuremechanisms.

Syntactic foam cores were fabricated with averagecenosphere volume fractions of 0%, 20%, 40%, and 60% inepoxy. Then, sandwich composites were manufactured usingvacuum assisted resin transfer molding (VARTM) process.For the facesheets, dry woven carbon fabric was used asreinforcement and vinyl ester resin as the matrix material.Impact tests on sandwich composites were conducted attwo energy levels of 80J and 160J. The results of theimpact tests showed a higher extent of damage and undesireddamage mechanisms as the cenosphere volume fractiondecreased for non-zero volume fraction cases (that is, 20%,40%, and 60%). As observed from the microCT images,shear cracks within the syntactic foam core and face sheetdamage were visible in the 20% and 40% cenosphere volumefraction specimens. In contrast, only localized compressionunderneath the impact location was observed in the 60%volume fraction specimens. An explanation for the observedundesired damage mechanisms in the 20% cenospherevolume fraction specimens was elucidated by developinga finite element model. In the case with high cenospherevolume fraction, i.e. 60%, the strains redistributed aroundthe cenospheres which led to a dispersed web of cracksthat did not propagate the entire thickness of the syntacticfoam core. For the models with lower cenosphere volumefractions, the distance between adjacent cenospheres was toolarge to form a web of cracks, and instead larger shear crackswere expected to form. In summary, this study showed thatthe syntactic foam cores with 60% cenospheres by volumeare superior than other volume fractions investigated for thefollowing reasons:

• The sandwich composites with 60% syntactic foamsare less dense than the other specimens, being ≈18%lighter than the pure epoxy core samples.

• Even though the initial stiffness of the 60% specimenswas the most compliant of all specimens, it hadcomparable recorded average maximum impact forceas compared to the other specimens with non-zero(that is, 20% and 40%) cenosphere volume fractionstested.

• The 60% volume fraction specimens experiencedlocalized compression/crushing underneath the strikerimpact location at both energy levels. In contrast,other specimens with non-zero cenosphere volumefractions experienced at least some degree of shearcracking under high energy impact loading. Localizedcompression/crushing as compared to shear cracksdoes not drastically affect the structural integrity


DRAFT


of the core. Whereas, cascading effects such ascore/facesheet debonding is typically accompaniedwith large shear cracks in the core as seen in Figure5. Hence, 60% volume fraction specimens are deemedto perform better compared to other cases consideredin this study.

Although sandwich composites with high cenospherevolume fraction syntactic foams are shown to be desirablefor containing impact damage, more studies need to beconducted by varying other properties like, core thickness,distribution of cenosphere dimensions (diameter and wallthickness), etc. before they can reliably be used in structuralcomponents subjected to dynamic impact loading.

AcknowledgmentsThe authors would like to thank the U.S. Departmentof Defense (DoD) Office of Naval Research (ONR)Young Investigator Program (YIP) Grant [N00014-19-1-2206] through Sea-based Aviation: Structures and MaterialsProgram for their partial support towards conducting theresearch presented here. The authors would also like toacknowledge the partial support from the U.S. DoD DefenseIntelligence Agency (DIA) and ONR Solid MechanicsProgram through the Basic Research Grant [W911NF-15-1-0430]. In addition, the authors would like to thankMechatronics Lab at UW-Madison for use of the microCTmachine, and also the mechanical engineering departmentat the National Institute of Technology Karnataka (NITK),India for providing support and facilities for syntactic foammanufacturing. The authors declare there are no conflicts ofinterest.

References

1. Mouritz AP, Gellert E, Burchill P et al. Review ofadvanced composite structures for naval ships and submarines.Composite Structures 2001; 53(1): 21–24.

2. Daniel IM, Gdoutos EE, Abot JL et al. Deformation and Failureof Composite Sandwich Structures. Journal of ThermoplasticComposite Materials 2003; 16(2): 345–55.

3. Schubel PM, Luo JJ and Daniel IM. Low velocity impactbehavior of composite sandwich panels. Composites Part A:Applied Science and Manufacturing 2005; 36(10 SPEC. ISS.):1389–1396.

4. Zhang G, Wang B, Ma L et al. Energy absorption and lowvelocity impact response of polyurethane foam filled pyramidallattice core sandwich panels. Composite Structures 2014;108(1): 304–310.

5. Wang J, Waas AM and Wang H. Experimental and numericalstudy on the low-velocity impact behavior of foam-coresandwich panels. Composite Structures 2013; 96: 298–311.

6. Nemes JA and Simmonds KE. Low-Velocity Impact Responseof Foam-Core Sandwich Composites. Journal of CompositeMaterials 1992; 26(4): 500–519.

7. Mines RAW, Worrall CM and Gibson AG. Low velocityperforation behaviour of polymer composite sandwich panels.International Journal of Impact Engineering 1998; 21(10):855–879.

8. Erickson MD, Kallmeyer AR and Kellogg KG. Effect oftemperature on the low-velocity impact behavior of composite

sandwich panels. Journal of Sandwich Structures andMaterials 2005; 7(3): 245–264.

9. Salehi-Khojin A, Mahinfalah M, Bashirzadeh R et al.Temperature effects on Kevlar/hybrid and carbon fibercomposite sandwiches under impact loading. CompositeStructures 2007; 78(2): 197–206.

10. Yang P, Shams SS, Slay A et al. Evaluation of temperatureeffects on low velocity impact damage in composite sandwichpanels with polymeric foam cores. Composite Structures 2015;129: 213–223.

11. Tan KT, Watanabe N and Iwahori Y. X-ray radiographyand micro-computed tomography examination of damagecharacteristics in stitched composites subjected to impactloading. Composites Part B: Engineering 2011; 42(4): 874–884.

12. Akil Hazizan M and Cantwell WJ. The low velocity impactresponse of foam-based sandwich structures. Composites PartB: Engineering 2002; 33(3): 193–204.

13. Elamin M, Li B and Tan KT. Impact damage of compositesandwich structures in arctic condition. Composite Structures2018; 192(March): 422–433.

14. Gupta N and Woldesenbet E. Microballoon wall thicknesseffects on properties of syntactic foams. Journal of CellularPlastics 2004; 40(6): 461–480.

15. Gupta N, Woldesenbet E and Kishore. Compressive fracturefeatures of syntactic foams-microscopic examination. Journalof Materials Science 2002; 37(15): 3199–3209.

16. Gladysz GM and Chawla KK. Syntactic and composite foams:Proceedings of an engineering conferences international (eci)conference. Journal of Materials Science 2006; 41(13): 3959–3960.

17. Choqueuse D and Davies P. Ageing of composites inunderwater applications. ISBN 9781845693527, 2008. pp.467–498.

18. Ambika Devi K, John B, Reghunadhan Nair CP et al. Syntacticfoam composites of epoxy-allyl phenol-bismaleimide ternaryblendprocessing and properties. Journal of Applied PolymerScience 2007; 105(6): 3715–3722.

19. Banhart J. Manufacture, characterisation and application ofcellular metals and metal foams. Progress in Materials Science2001; 46(6): 559 – 632.

20. Gupta N, Zeltmann SE, Shunmugasamy VC et al. Applicationsof polymer matrix syntactic foams. JOM 2014; 66(2): 245–254.

21. Zhu B, Ma J, Wang J et al. Thermal, dielectric and compressiveproperties of hollow glass microsphere filled epoxy-matrixcomposites. Journal of Reinforced Plastics and Composites2012; 31(19): 1311–1326.

22. Yung K, Zhu B, Yue T et al. Preparation and propertiesof hollow glass microsphere-filled epoxy-matrix composites.Composites Science and Technology 2009; 69(2): 260 – 264.

23. Jayavardhan M, Kumar BB, Doddamani M et al. Developmentof glass microballoon/hdpe syntactic foams by compressionmolding. Composites Part B: Engineering 2017; 130: 119 –131.

24. Poveda RL, Dorogokupets G and Gupta N. Carbon nanofiberreinforced syntactic foams: Degradation mechanism for longterm moisture exposure and residual compressive properties.Polymer Degradation and Stability 2013; 98(10): 2041–2053.

25. Gupta N, Brar BS and Woldesenbet E. Effect of filleraddition on the compressive and impact properties of glass fibre


DRAFT


reinforced epoxy. Bulletin of Materials Science 2001; 24(2):219–223.

26. Satapathy BK, Das A and Patnaik A. Ductile-to-brittletransition in cenosphere-filled polypropylene composites.Journal of Materials Science 2011; 46(6): 1963–1974.

27. Qiao J and Wu G. Tensile properties of fly ash/polyureacomposites. Journal of Materials Science 2011; 46(11): 3935–3941.

28. Bharath Kumar BR, Doddamani M, Zeltmann SE et al.Processing of cenosphere/HDPE syntactic foams using anindustrial scale polymer injection molding machine. Materialsand Design 2016; 92: 414–423.

29. Kumar BR, Doddamani M, Zeltmann SE et al. Datacharacterizing tensile behavior of cenosphere/HDPE syntacticfoam. Data in Brief 2016; 6: 933–941.

30. Barkoula NM, Alcock B, Cabrera NO et al. Fatigue propertiesof highly oriented polypropylene tapes and all-polypropylenecomposites. Polymers and Polymer Composites 2008; 16(2):101–113. 1206.4529.

31. Gupta N, Ye R and Porfiri M. Comparison of tensile andcompressive characteristics of vinyl ester/glass microballoonsyntactic foams. Composites Part B: Engineering 2010; 41(3):236–245.

32. Gladysz GM, Perry B, Mceachen G et al. Three-phase syntacticfoams: Structure-property relationships. Journal of MaterialsScience 2006; 41(13): 4085–4092.

33. Shunmugasamy VC, Pinisetty D and Gupta N. Viscoelasticproperties of hollow glass particle filled vinyl ester matrixsyntactic foams: Effect of temperature and loading frequency.Journal of Materials Science 2013; 48(4): 1685–1701.

34. Gupta N, Woldesenbet E and Mensah P. Compressionproperties of syntactic foams: Effect of cenosphere radius ratioand specimen aspect ratio. Composites Part A: Applied Scienceand Manufacturing 2004; 35(1): 103–111.

35. Maharsia R, Gupta N and Jerro HD. Investigation of flexuralstrength properties of rubber and nanoclay reinforced hybridsyntactic foams. Materials Science and Engineering A 2006;417(1-2): 249–258.

36. Tagliavia G, Porfiri M and Gupta N. Analysis of flexuralproperties of hollow-particle filled composites. CompositesPart B: Engineering 2010; 41(1): 86–93.

37. Tagliavia G, Porfiri M and Gupta N. Influence ofmoisture absorption on flexural properties of syntactic foams.Composites Part B: Engineering 2012; 43(2): 115–123.

38. Kishore, Shankar R and Sankaran S. Short beam threepoint bend tests in syntactic foams. Part I: Microscopiccharacterization of the failure zones. Journal of AppliedPolymer Science 2005; 98(2): 673–679.

39. Kishore, Shankar R and Sankaran S. Short-beam three-pointbend tests in syntactic foams. Part II: Effect of microballoonscontent on shear strength. Journal of Applied Polymer Science2005; 98(2): 680–686.

40. Kishore, Shankar R and Sankaran S. Short-beam three-pointbend test study in syntactic foam. Part III: Effects of interfacemodification on strength and fractographic features. Journal ofApplied Polymer Science 2005; 98(2): 687–693.

41. Garcia CD, Shahapurkar K, Doddamani M et al. Effect ofarctic environment on flexural behavior of fly ash cenospherereinforced epoxy syntactic foams. Composites Part B:Engineering 2018; 151: 265–273.

42. Shahapurkar K, Garcia CD, Doddamani M et al. Compressivebehavior of cenosphere/epoxy syntactic foams in arcticconditions. Composites Part B: Engineering 2018; 135(June2017): 253–262.

43. Gupta N and Woldesenbet E. Characterization of flexuralproperties of syntactic foam core sandwich composites andeffect of density variation. Journal of Composite Materials2005; 39(24): 2197–2212.

44. ASTM. D7766-16 Standard Practice for Damage ResistanceTesting of Sandwich Constructions. Astm 2015; i: 1–9.


1206.4529

Journal of Composite Materials Dynamic impact behavior of ... · as the shear modulus of the core increased. The failure mechanism transitioned from shear cracks in the core to delamination

Documents