IAC-09.A6.3.11

HONEYCOMB VS. FOAM: EVALUATING POTENTIAL UPGRADES TO ISSMODULE SHIELDING

S. RvanUSRA Lunar and Planetary Institute (LPI), 3600 Bay Area Blvd, Houston, TX, 77058, USA

shannon. j .ryan(a_)nasa. gov

E.L. ChristiansenNASA Johnson Space Center, 2101 NASA Pkwy, Houston, TX, 77058, USA

eric.l.christiansen(c,.nasa.gov

ABSTRACT

A series of 19 hypervelocity impact tests have been performed on ISS-representative structure walls to evaluate theeffect on micrometeoroid and orbital debris (MMOD) protective capability caused by replacing honeycombsandwich panel cores with metallic open-cell foam. In the experiments, secondary impacts on individual foamligaments were found to raise the thermal state of projectile and bumper fragments, inducing break-up and melt atlower impact velocities than the baseline honeycomb configuration. A ballistic limit equation is derived for thefoam-modified configuration, and in comparison with the honeycomb baseline a performance increase of 3-15% atnormal incidence was predicted. With increasin g impact obliquity, the enhancement in protective capabilityprovided by the modification is predicted to further increase. The reduction in penetration and failure risk posed byMMOD impacts is achieved by the foam-modified configuration without a significant decrease in mechanical orthermal performance, and with no additional weight. As such, it is considered a promising upgrade to MMODshielding on ISS modules which incorporate honeycomb sandwich panels and are yet to fly.

INTRODUCTION

The performance of a dual-wall protective spacecraftstructure against the impact of micrometeoroid andorbital debris (MMOD) particles is generallyconsidered to be degraded by the presence of ahoneycomb core. For impacts which penetrate theshield outer wall (bumper or front facesheet),projectile and bumper fragments disperse radially asthey propagate through the shield interior,distributing the load over an area of the shield rearwall significantly larger than that of the originalprojectile diameter. The presence of honeycomb cellwalls acts to restrict expansion, effectivelychanneling the fragments within a limited number ofhoneycomb cells for a more concentrated impactupon the rear facesheet. However, missionrequirements often prevent the inclusion of adedicated MMOD shielding structure, and as such,structural panels (i.e. honeycomb sandwich panels)also commonly serve as the protective system.

Metallic foams are a promising alternative tohoneycomb structures as they offer comparablestructural and thermal performance without thepresence of MMOD shielding-detrimental channelingcells. In this paper, modifications to a double-layer

honeycomb sandwich panel shielding configurationrepresentative of those used onboard the InternationalSpace Station (ISS) are evaluated. The modificationsentail the substitution of aluminum honeycomb foraluminum open-cell foams, while the total shieldweight in maintained.

BACKGROUND

Honeycomb sandwich panels

Given their conunon application in space vehicleprimary structures, the performance of honeycombwider impact of MMOD particles at hypervelocityhas been investigated in a multitude of studies. Jex etal. [1] and Sibeaud et al. [2] discussed that thepresence of a honeycomb core enhanced the shieldingperformance of a dual-wall structure at hypervelocity.They concluded that secondary impacts betweenejecta fragments and cell walls overcompensated forthe detrimental effect of channeling. A morecommonly held view is that the presence of ahoneycomb core is unfavorable to the shieldingperformance. Taylor et al. [3] quantified thedegradation in performance through inclusion of ascaling factor which acts to reduce the effective rearfacesheet thickness by 50% in definition of the panel

ballistic limit at hypen-elocities (i.e. molten and/orvaporized ejecta). Ryan et al. [4] defined adegradation in shielding performance due to thepresence of a honeycomb core equal to a --46%reduction in shielding capability at normal impact,reducing with increasing obliquity (e.g. for impact at60°, the degradation in performance drops to --18%).Sennett and Lathrop [5] also quantified the effect ofthe honeycomb core, stating that once the panelthickness increases above two times the honeycombcell size, no increase in shielding capability isachieved with an increase in shield thickness whenfragments were either molten or vaporized. For solidfragment ejecta, the effect was not nearly as severe.In Fig. 1, a comparison between the perfornance at 7km/s (normal impact) predicted for a dual-wall shieldwith and without a honeycomb core. For the Whippleshield configuration (i.e. no honeycomb core), thenew non optimum (NNO) equation [6] is used. Itshould be noted that the NNO equation (and hence,the Taylor and Sennett & Lathrop approaches) mayprovide non-conservative predictions for projectilediameter to shield spacing ratios (S/d t,) < 15.

UNNO

— — — Semmett & Lathrop----- Taylor— • — SRL

0 2 4 6 8 10

Spacing (No. of honeycomb cells)

Fig. 1: Predicted performance for a dual-wall shield withand without a honeycomb core at hypervelocity.

Open-cell foams

Preliminary investigations of the hypervelocityimpact performance of metal foam structures havedemonstrated their potential, particularly incomparison with traditional structural panels. In [7]alternative configurations for the ISS Columbusmodule shielding were evaluated, one of whichincluded an open-cell aluminum foam bLmiper. Thisconfiguration was found to provide increasedprotection over the reference Columbus stuffed

Whipple shield at high velocities (>6 km/s) andnormal incidence. For oblique angles, theperformance was comparable to the referenceconfiguration at high velocities. For low velocitytesting, the performance of the foam-bumperconfi guration was clearly worse than the referenceshield, due to the inability of the foam bumper toinduce projectile fragmentation. Although the foamconfi guration provided a similar level of protectionoverall to the reference stuffed Whipple shield, theauthors noted that the primary advantage of themodified configuration are related to the extendedarea of the pressure hull that can be protected (due toa concentration of mass in the outer later), and toother design aspects such as a reduction in non-ballistic mass (stiffeners, local reinforcements, etc.).

The shielding performance of sandwich panelstructures with open-cell aluminum foam cores wasevaluated in [8] against that of aluminum honeycombcore sandwich panels (Al HC SP). In Fig. 2 acomparison between damages induced by nominallyidentical impacts are shown. It should be noted thatthe facesheets of the HC SP were significantlythicker than those of the foam panel in order toprovide comparable areal densities. In the figure, thefoam core is shown to restrict fra gment radialexpansion to an equal or greater degree than the HC_However, while fragments are expected to bechanneled within the HC cells, the foam homogeneityshould ensure that resistance to fragment cloudexpansion is equal in all directions, therefore limitingthe degree of channeling. For these impact

conditions, the performance of the foam panel isshown to be clearly superior to that of the honeycombpanel.

TARGET DEFINITION

Double-laver honeycomb (DL-H)

The baseline tar get is constructed of two honeycombsandwich panels, with two outer layers of stainlesssteel mesh and a monolithic aluminum rear wall,shown in Fig. 3. Details of the target components areprovided in Table 1. The total areal density of theDL-H configuration is 1.57 g/=7^.

^. ^HrTFO94s0a

rs.6 mm A% bailn•Q. li wn(1CG

aMan., ..^aNeFO.,u ,v .ne

l^stFOM^Rt '-43?IIIt

w.

I a.6 m. A14e

%.6mm 1114p Fb O^•M i an:,ec t

Fig.

aa.o . , maeo on N, -,

_ACT TESTING AND RESULTS

A series of 19 hypervelocity impact tests wereperfornied on the double-layer targets using the twostage light gas guns at NASA JSC's White Sand'sTest Facility (WSTF). A summary of the testconditions and results are presented in Table 3. Forthese tests, failure was defined as the ejection ofmaterial within the simulated pressure hull (i.e.perforation or detached spall of the target rear wall).

Test Target Angle Diameter Velocity Result(deg) (can) (km/s)

1 8592 DL-F 0 0.877 6.76 Pass2 8593 DL-F 45 0.837 6.87 Pass3 8594 DL-F 60 1.114 66.9 Fail4 8595 DL F 0 0.717 3.29 Fail5 8599 DL-F 60 1.005 7.03 Fail6 8596 DL-F 0 0.637 3.67 Fail7 8597 DL F 45 0.662 3.68 Pass8 8598 DL-F 45 0.837 3.62 Pass9 9024 DL F 60 1.005 6.80 Pass10 9038 DL-F 60 1.115 6.69 Pass11 9064 DL-F 60 1.276 7.00 Fail12 7460 DL F 0 0.833 6.74 Pass13 7461 DL-F 45 0.873 6.89 Fail14 7458 DL-H 45 0.754 6.94 Pass15 7459 DL-H 45 0.650 6.88 Pass16 7504 DL-H 0 0.730 6.86 Pass17 7509 DL-H 0 0.754 6.93 Pass18 7510 DL-H 45 0.873 6.74 Fail19 7629 DL-H 0 0.833 6.91 Fail

Table 3. Impact test results.

In order to evaluate the effect of interchangingaluminum honeycomb for open-cell aluminum foam,a direct comparison can be made between impactdamages induced on both configurations atnominally-identical impact conditions. In Fig. 6damages induced in the DL-H and DL-F targets bythe impact of 0.833 cm diameter projectiles at6.83±0.09 kin/s with normal incidence are compared.Damage in the two mesh layers, and the entry hole onthe I" sandwich panel are similar for bothconfigurations. The diameter of rear facesheetmaterial peeled back from the I ` sandwich panel exithole is also similar: however the extension of coredamage is noticeably less in the foam. The throughhole in the 2nd panels is shown to be significantlylarger for the DL-H confi guration than the DL-Fshield (88x90 nun vs. 70x62 min), indicating that thedebris cloud is more finely concentrated by the foamsandwich panel bumper than the honeycombsandwich panel. The diameter of the through hole inthe 2nd panels is similar to that of the core damage inthe first sandwich panel for both configurations (--91vs. 84 1nm for DL-H; --58 vs. _ 66 nun for DL-F).This suggests that the facesheets on the 2 "d panel ofthe DL-H configuration have little effect on the

expansion of the debris cloud (i.e. they have minimalre-focusing effect).

The rear wall of the DL-H configuration isperforated, showing a large through crack (80 mm inlength; 5 inln wide) and multiple individual craters.Given the appearance of the through crack, it isexpected that failure of the rear wall occurredthrough penetration of individual solid fragmentswhich acted as crack initiation sites that werepropagated during the impulsive load of the fragmentcloud. The rear wall of the DL-F configuration issignificantly deformed, yet there is no perforation ordetachment of spalled material from the rear surface.The majority of deposits on the rear wall are frommolten aluminum, visible as the bright silver coatingin the target photograph. The rear wall shows somecratering from impact of individual solid fragments,which form small dimples on the rear side of thepanel. Under these impact conditions, theperformance of the DL-F shield is clearly superior tothat of the baseline DL-H shield.

Evaluation of shield performance

The effect of secondary projectile and bumperfragment impacts upon individual foam cellligaments is expected to lead to increasedfragmentation, melting and vaporization at lowervelocities than for conventional shieldingconfigurations (e.g. Whipple shield, honeycombsandwich panel). This mechanism is utilized in themulti-shock shield, which was shown in [9][10] toprovide damage features at 6.3 km/s representative ofthose seen at 10 km/s on single bumper shields. Anapproximation of effective impact velocities can bemade from projectile entropy (or internal ener gy). In[11] Swift calculates required impact velocities formelt and vaporization conditions based on theconcept of entropy trapping — in which the entropyinjected into projectile and target materials can becalculated from the Hugoniot and release isentrope.The increase in entropy acts to raise the materialinternal energy (or temperature); eventually reachingand exceedin g the material fusion energy (melting)and vaporization energy.

The rear walls of the DL-H target in Fig. 6 shows adegree of molten aluminum deposits, although thepredominant damage feature is cratering about thecentral damage zone. Alternatively, the DL-F targetshows significant molten aluminum over a largecentral area with only a small number of finitecraters. Clearly, therefore, secondary impacts on thefoam ligaments are effective in raising fragmententropy.

a*HrTF07a9

ow

4IT

SRI

F Y y^. ,.S

n^: h

.. ........i^

Fig. 6: Comparison of impact damages in the DL-H (left) and DL-F (right) targets impacted by 0.833 cm diameter Al2017-T4spheres at —6.9 lan/s with normal incidence (0'). From top to bottom: I" sandwich panel (rear view), 2"a panel (rear view);rear wall (front view).

5

BALLISTIC LEVIIT EQUATIONS

To evaluate the effect of the shielding modificationsover the complete range of expected in-orbit impactconditions, ballistic limit curves can be used.Calculated using empirical ballistic limit equations(BLEs), these curves demarcate between impactconditions leading to pass or fail, and are used inmodern risk assessment codes such as NASA'sBUMPER-II to determine mission risk tomicrometeoroid and orbital debris (MMOD).

The DL-H configuration is representative of theentranced zone 11 shield onboard the FGB module ofthe ISS (Zarya) [12]. For FGB shielding, a genericballistic limit equation was defined based on theNNO Whipple shield equation [6]. In order to adjustthe equation to suit the double-layer honeycombconfiguration, the bumper thickness was estimatedusing the areal density of the I" honeycombsandwich panel, and half the areal density of the 2"`1sandwich panel. The remaining 50% of the 2'dsandwich panel areal density was added to thethickness of the shield rear wall, and the equationconstants were empirically adjusted from test data.The enhanced zone 11 FGB ballistic limit equation isdefined (from [ 12]) as:

Hiah velocity: when V >_ 7/cos 0,

d,, = Cx (V COS 0)-2/3 pp-1/3

(1)

where V — Projectile velocity (km/s)VL — Low velocity regime upper limit (k111/s)d,— Critical projectile diameter (cm)CH — High velocity fit coefficient (-) = 4.6510 — Impact angle (deg)pP — Projectile density (g/cm3)

Intermediate: when 3/cos 0 > V > 7/cos 0,

d,, =Chipp-1/3(V COS 0— VL)+...

(2)cl pp-9/19 (COS 8)-18

/19

(Vx — V COS B)

where VH — High velocity regime lower limit (kln/s)Ch; — Inter.-high velocity fit coefficient (-) = 0.318C li — Utter.-low velocity fit coefficient (-) = 0.203

Low velocity: when V 5 3/cos 0,

d = C, (COS 0) -31/11 V-12119pp -9/19

(3)

where CL — Low velocity fit coefficient (-) = 1.629

The diameter of the steel wire used in the enhancedzone 11 shield was 0.280 nun, less than that of theDL-H configuration tested in this study (0.4064 nun).As such, the ballistic limit equation constants must be

adjusted in order to fit the test data reported in Table3. The low and high velocity coefficients, C L and CHrespectively, are y calculated based on the arealdensities of the individual shield components:

C = 3.11 t + (2AD._, + AD,, r + ADS z

O

)4w

2.8

CH = 3.52 + 3.OADwesh (5)

The intermediate fit coefficients are calculated as0.209 and 0.290 for C 11 and Chi respectively.

For the DL-F configuration, the areal densities of thespecific shield components are also included in theequation fit coefficients. For honeycomb sandwichpanels, the mass of the core is generally ignored indetermining effective shield thicknesses (i.e. treatedas non-ballistic mass). For foam core sandwichpanels, however, the foam is an active shieldingcomponent. The ballistic limit equation for the DL-Fconfiguration is defined as:

High velocity: when V >_ V H/cos 0,

de = C.x (V COS a)-a pp -113 (6)

where P — High velocity angle dependence constant (-) = 0.55

CH =3.0+2.4xADp^

Inter. velocity: when VL/cos 0 > V > VH/COS 0,

L) +d^(iV)—V

(VL).(V -VL) (7)de= dC(Vx L

Low velocity: when V:5 VL/cos 0,

de = CL (COs 9)—a

V -12119pp -9119 (8)

where a — LV angle dependence coefficient (-) = 1.75

C = 3(t., +(2AD..h + ADsPI + ADSP , )/2.8)

In Fig. 7 the ballistic limit curve of the modified DL-F shield is plotted a gainst the baseline DL-Hconfiguration. For normal impact, the modificationsresult in a small predicted improvement over therange of applicable impact velocities. At 3 km/s theDL-F target provides a 15% improvement in criticalprojectile diameter; while at 7 km/s a 3% increase ispredicted. The larger low velocity sizing constant(CL) leads to increasing performance gain withincreasing impact obliquity, although there is a lackof test data to support or disprove this extrapolation.

UN.o

0.8

0.6

12 15

12 15

Velocity (km/s)

0.4 '

0 3 6 9 12 15

Velocity (km/s)

1.4

DL-H

DL-F1.2

Velocity (kin/s)

Fig. 7: Ballistic limit curves of the DL-H and DL-F shields.From top to bottom: 0°, 45°, 60°.

DISCUSSION

Sandwich panels with open-cell metallic foam coresprovide comparable mechanical and thermalperformance to those with metallic honeycomb coresfor a minimal weight penalty. The homogenous foamstructure avoids the MMOD shielding-detrimentalchanneling cells of honeycomb panels, making thema promising alternative for spacecraft primarystructures which are also required to provide MMODprotection.

Destefanis et al. [13] reported on tests against a dualwall configuration with a bumper of open cellaluminum foam. In the tests a good deal of meltingwas observed at velocities as low as 2 km's, withcomplete melting reported at velocities as low as 4knvs. Similar enhanced fragmentation was reportedin [14] for nrm-sized projectiles at normal impact. Inthis study, clear evidence of melted deposits wasobserved on the target rear wall for test #4(HITF08595), performed at 3.29 krn/s. For lowvelocity impacts at oblique impact (e.g. test #6(HITF08596)) there was also clearly observabledeposits of melted aluminum upon the shield rearwall. Although the onset and degree of projectile andbumper melt is clearly increased by the open-cellfoam bumpers, in all impact tests performed there isevidence of solid fragment impacts upon the targetrear walls. For oblique impacts, these solid fragmentcraters are generally in-line with the projectilevelocity vector, indicating that they are most likelyprojectile remnants.

In [14], the velocity regime transition limits of aballistic limit equation for the foam bumper shieldingconfiguration were set at 2.7 and 6.5 km'srespectively, in recognition of the increasedfragmentation and melting provided by the structure(compared to a traditional Whipple shield). However,due to the evidence of individual solid fragmentimpacts upon the shield rear wall for impactvelocities up to 6.76 km/s in this study, and in theabsence of additional test data providing clearexperimental justification; the transition velocitiesdefined in [6] for aluminum Whipple shields and in[12] for the DL-H confi guration are maintained in theballistic limit equation derived for the DL-F shield.

Enhanced fragmentation and melting induced by thefoam microstructure was found in [13] to beineffective against projectiles in the cm-sized range atnormal incidence; and nun-sized projectiles atoblique angles. The authors concluded that secondaryimpacts were no longer able to induce fragmentationand melting of the entire projectile at these impact

1.2

1

0.8

0.6.o

0.4

0.2

1.2

b 0.8

UN.o

0.6

conditions. In this study, however, there was nonoticeable decrease in performance at obliquity, evenfor projectiles considerably larger than 1 cm indiameter (e.g. test 411 (HITF09064)). The doublelayer of mesh on top of the I" sandwich panel of theDL-F configuration is expected to break up theprojectile prior to impact on the sandwich panelfacesheet. Therefore, smaller projectile fragments arepropagated to impact within the sandwich panel foamcore and the size-limitations of secondaryfragmentation and melting discussed by Destefanis etal. are not valid.

CONCLUSIONS AND StWIMARY

In this paper, the effect on shielding performanceachieved by replacing metallic honeycomb cores formetallic open-cell foam cores in a double sandwichpanel MMOD shielding confi guration representativeof those used onboard the ISS was assessed. Abaseline double-layer honeycomb (DL-H), andmodified double-layer foam (DL-F) configurationwere subject to impact by projectiles athypervelocity, from which ballistic limit equationswere derived. These equations were based on theNNO IN'hipple shield [6] and general FGB [12]equations, and included fit coefficients based on arealdensities of individual shielding components. Atnormal incidence the foam-modified shield wasfound to provide a 15% improvement in criticalprojectile diameter at low velocity (i.e. 3 km/s) and a3% increase at high velocity (7 km/s). Withincreasing impact obliquity the foam shieldperformance enhancement increases at the low-shatter regime transition velocity, up to a 29%improvement in critical diameter at 60°. It should benoted that the double-layer honeycomb equationconstants are defined for consistency with theenhanced zone 11 shield described in [12], for whichthere is no low velocity test data.

The presence of honeycomb cells is considered to bedetrimental to the shielding performance of a dual-wall configuration due to thecell walls acting torestrict the expansion of projectile and bumper (orfront facesheet) fra gments — referred to aschanneling. However, the thickness of thehoneycomb sandwich panels in the double-layerconfiguration are less than twice the diameter of eventhe smallest projectile used in the testing. Thus,dispersion of the projectile and bumper fra gments isexpected to be uninterrupted prior to impact upon thesandwich panel rear facesheet. As such, theperformance enhancement gained by replacing thehoneycomb core with open-cell foams is not expected

to result as a simple absence of through-thicknesschanneling cells. Rather, secondary impacts ofprojectile and bumper fragments upon individualfoam cell ligaments induced repeated shocks,increasing fragment entropy and subsequentlyreducing Vfailure strengths. Evidence of increasedprojectile fragmentation and melting was shown forthe double-layer foam configuration (compared to thedouble-layer honeycomb configuration). Previousinvestigations on metallic open-cell foam bumpershave noted a decrease in performance for obliqueimpact, and normal impact of large cm-sizedprojectile due to an inability of the repeated shockingprocedure to fragment the entire projectile at theseconditions. However, the presence of the doublemesh outer layers breaks up the projectile prior toimpact upon the I't sandwich panel front facesheet,ensuring the propagation of smaller, moremanageable impactors within the foam core.

REFERENCES

[1] D.W. Jex, A.M. Miller, C.A. MacKay, "TheCharacteristics of Penetration for a Double-SheetStructure with Honeycomb", NASA MarshallSpace Flight Center, Huntsville, NASA TM X-53974,1970.

[2] J-M. Sibeaud, C. Thamie, C. Puillet,"Hypervelocity Impact on Honeycomb TargetStructures: Experiments and Modeling",International Journal of Impact Engineering; 35:1799-1807, 2008-

[3] E.A. Taylor, M.K. Herbert, B.A. Vaughan, J.A.McDonnell, "Hypervelocity Impact on CarbonFibre Reinforced Plastic / AluminumHoneycomb: Comparison with Whipple BumperShields", International Journal of ImpactEngineering; 23(1): 883-893, 1999.

[4] S. Ryan, F.K. Schaefer, R. Destefanis, M.Lambert, "A Ballistic Limit Equation forHypervelocity Impacts on CFRP/Al HC SatelliteStructures", Advances in Space Research; 41:1152-1166, 2008-

[5] R. Sennett, B. Lathrop, "Effects ofHypervelocity Impact on Honeycomb Structure",Journal of Spacecraft; 5(12): 1496-1497, 1968.

[6] E.L. Christiansen, "Design and PerformanceEquations for Advanced Meteoroid and DebrisShields", International Journal of ImpactEngineering; 14(1-4): 145-156 7 1993.

[7] R. Destefanis, "Enhanced Space Debris Shieldsfor Manned Spacecraft — Evaluation Report:Phase A and Phase B Tests", Alenia-Spazio,Turin, ESD-TN-A1-005, 2004.

[8] J. Yasensky, E.L. Christiansen, T. Prior."Hypervelocity Impact Evaluation of MetalFoam Core Sandwich Structures", GeoControlSystems, Houston; NASA TP-2008-214776,2008.

[9] B.G. Cour-Palais, J.L. Crews, "A Multi-ShockConcept for Spacecraft Shielding", InternationalJournal of Impact Engineering: 10: 135-146,1990-

[10]M. Alme, E. Christiansen, B. Cour-Palais,"Hydrocode Simulations of the Multi-ShockMeteoroid and Debris Shield", ShockCompression of Condensed Matter 1991, pp.975-978. 1991.

[II]H.F. Swift, "Hypervelocity Impact Mechanics",in: Impact Dynamics (Zukas, ed.), John Wiley &Sons, New York, 1982.

[ 12] G.A. Sanchez, E.L. Christiansen, " FGB EnergyBlock Meteoroid and Orbital (M/OD) DebrisShield Test Report', NASA Johnson SpaceCenter, Houston, JSC Document 27460, 1996.

[131R. Destefanis, F. Schaefer, M. Lmabert, M.Faraud, E. Schneider, "Enhanced Space DebrisShields for Manned Spacecraft', InternationalJournal of Impact Engineering; 29: 215-226,2003.

[14]R. Destefanis. F. Schaefer, M. Lambert. M.Faraud, "Selecting Enhanced Space DebrisShields for Manned Spacecraft', InternationalJournal of Impact Engineering, 33: 219-230,2006.