NASA Technical Reports Server (NTRS) 20110016010: Nondestructive Evaluation (NDE) for Characterizing Oxidation Damage in Cracked Reinforced Carbon-Carbon

0.8 MB·English

by  NASA Technical Reports Server (NTRS)

Int.J.Appl.Ceram.Technol.,7[5]652–661(2010)
DOI:10.1111/j.1744-7402.2009.02372.x
CeramicProductDevelopmentandCommercialization
Nondestructive Evaluation (NDE) for Characterizing
Oxidation Damage in Cracked Reinforced Carbon–Carbon
Don J. Roth* and Nathan S. Jacobson
NASA Glenn Research Center, Cleveland, Ohio 44135
Richard W. Rauser
University of Toledo, Toledo, Ohio 43606
Russell A. Wincheski
NASA Langley Research Center, Hampton, Virginia 23681
James L. Walker
NASA Marshall Space Flight Center, Huntsville, Alabama 35812
Laura A. Cosgriff
Cleveland State University, Cleveland, Ohio 44115
Inthisstudy,coatedreinforcedcarbon–carbon(RCC)samplesofsimilarstructureandcompositionasthatfromtheNASA
spaceshuttleorbiter’sthermalprotectionsystemwerefabricatedwithslotsintheircoatingsimulatingcrazecracks.Thesespecimens
wereusedtostudyoxidationdamagedetectionandcharacterizationusingnondestructiveevaluation(NDE)methods.Thesespec-
imenswereheattreatedinairat11431Cand12001Ctocreatecavitiesinthecarbonsubstrateunderneaththecoatingasoxygen
*[email protected]
r2009TheAmericanCeramicSociety.NoClaimtooriginalUSGovernmentworks
www.ceramics.org/ACT NDEforCharacterizingOxidationDamage 653
reactedwiththecarbonandresultedinitsconsumption.Thecavitiesvariedindiameterfromapproximately1to3mm.Single-sided
NDEmethodswereusedbecausetheymightbepracticalforon-winginspection,whileX-raymicro-computedtomography(CT)was
usedtomeasurecavitysizesinordertovalidateoxidationmodelsunderdevelopmentforcarbon–carbonmaterials.AnRCCsample
havinganaturallycrackedcoatingandsubsequentoxidationdamagewasalsostudiedwithX-raymicro-CT.Thiseffortisafollow-on
studytoonethatcharacterizedNDEmethodsforassessingoxidationdamageinanRCCsamplewithdrilledholesinthecoating.
Introduction were created by drilling holes followed by oxidation
of the sample and subsequent nondestructive evalu-
Reinforced carbon–carbon (RCC) with a silicon ation (NDE) to characterize oxidation damage.2 In
carbide(SiC)coatingforoxidationresistanceisusedon that study, RCC samples were oxidized to create ap-
the NASA Space Shuttle Orbiter’s wing leading edge, proximately hemi spherical holes underneath the SiC
nosecap,andarrowheadattachmentpointtotheexter- coating and subsequently inspected using various NDE
naltankforthermalprotectionduringreentry.Thehigh methods. In the current study, small breaches were cre-
strengthandlightweightofRCCmakeitanidealaero- ated by machining slots of various widths to simulate
space material; however, oxidation is a major concern. cracks of various sizes in the coating, and NDE was
OxidationdamagetoRCCcanoccuriftheSiCcoating again subsequently used to characterize the oxidation
isitselfdamagedbutstillintactsuchthathotgaseshave damage.
accesstothecarbonbeneaththecoating.1Insuchcases, The NDE techniques used in this study included
itiscriticaltoevaluatetheextentoftheoxidationdam- state-of-the-art backscatter X-ray (BSX), ultrasonic-
age underneath the intact SiC coating. Even small guided waves, eddy current (EC), and thermographic
breaches in the RCC coating system have recently methods. All of these methods are single-sided
been identified as potentially serious. In a prior techniques thereby lending themselves to practical in-
study, small breaches in the coating of an RCC sample spections of components only accessible from
Fig.1. Schematicofsiliconcarbide(SiC)-protectedcarbon/carbonusedinthisstudy.ReprintedwithpermissionfromElsevier.
654 InternationalJournalofAppliedCeramicTechnology—Roth,etal. Vol.7,No.5,2010
Fig.2. Photographsofmachinedreinforcedcarbon–carbon(RCC)sampleswithmachinedslots.Nominalwidthsofslotsareshownabovethe
slotandspacingbetweenslotsandfromedgesisalsoshown.
oneside.SampleswerealsoinspectedwithX-raymicro- faces are painted with a sodium silicate glass, which
computed tomography (CT) to evaluate the true di- meltsandsealscracksonreentry.Allthesamplesusedin
mensions and morphology of the holes, as well as nat- thisstudyhadtheTEOStreatment.Inaddition,oneof
ural crack formations.3 The controlled oxidation the samples studied had the sodium silicate glass. The
damage provides standards for investigating the effec- samples were all approximately 5mm thick.
tiveness of various NDE techniques for detecting and The sample with SiC plus glass coating (RCC1)
sizing oxidation damage in this material. NASA Glenn was a flat plate with an approximately 1mm thick SiC
Research Center led this investigation that had some of plusglasscoatingandcoatedonallsides.Theplatehad
the top NDE specialists/facilities from NASA Glenn, an artificial craze crack of linear geometry made with a
NASA Langley, andNASAMarshallinspect these sam- diamond blade (Keen Kut Products, Hayward, CA) of
ples with the various NDE methods. 0.25mm thickness. This plate was used for ultrasonic
studies. The slot was cut to the SiC plus glass coating/
carbon–carbon interface on one side of the sample.
Experimental Procedure The samples with only the SiC coating had an ap-
proximately 1-mm-thick coating and were coated on all
RCC Material sides. Of these, one was a plate. The plate (RCC2) had
machinedslotsmadewithdiamondbladesof0.25,0.51,
Figure 1 is a schematic of RCC with a SiC con- 0.76, and 1.02mm thicknesses. These slots were cut to
version coating. Briefly, this material is made with a the SiC/carbon–carbon interface on both sides of the
two-dimensional layup of carbon–carbon fabric with plateample.TheseslottedspecimensareshowninFig.2.
repeatedapplicationsofaliquidcarbonprecursortofill OtherSiC-onlycoatedsamplesincludedseveralflat
voids.AnoxidationprotectionsystemisbasedonaSiC 1.91cm diameter and 1.52-cm-thick disks.3 Some of
conversion coating. Because of the difference in coeffi- these had slots machined in them; others were used in
cient of thermal expansion of the SiC coating and car- their as-fabricated form with the naturally occurring
bon/carbon substrate, the SiC coating shrinks more craze cracks acting as paths for oxidation. Polishing
than the underlying carbon/carbon on cooldown from B300mm of SiC off the surface showed the cellular
the coating application temperature. This leads to ver- crackpatternasshown inFig. 3atogether witha‘‘skel-
ticalcracksinthecoating,andthesecracksarepathways eton’’ trace of the cracks in Fig. 3b.
foroxygen toreachthecarbon/carbon substrate.Actual Controlled laboratory oxidation exposures were
RCC used on the Space Shuttle Orbiter is infiltrated performed as follows: (a) Plates and disks with ma-
withtetraethylorthosilicate(TEOS),whichdecomposes chinedslots:0.5hat12001Cinstaticlaboratoryair;(b)
on a mild heat treatment to silica. Then the RCC sur- Diskwithcrazecracksonly:2.5hinbottleflowingairat
www.ceramics.org/ACT NDEforCharacterizingOxidationDamage 655
Fig.3. (a)Reinforcedcarbon–carbondiskwithcrazecrackpatternonthesurface.(b)‘‘Skeleton’’traceofcrackpattern.Reprintedwith
permissionfromElsevier.
11431C. The carbon/carbon substrate below the ma- line condition), and after, oxidation. These methods
chinedslotoxidizedtoformanapproximatehemispher- includedBSX, EC,thermography,andguidedwaveul-
ical void (Fig. 4).This uniform attack pattern indicates trasonics. X-ray CT was used to size oxidation damage
diffusion control; and Jacobson et al.3 describes an ox- andcracking,anddevelopthree-dimensionalvolumetric
idation model for given slot and crack breaches in the visualizations. Dugan and colleagues4–8 provide basic
coating. The resulting size of the void would be ex- principles for the various methods, and experimental
pectedtoincreasewithincreasingslotwidth.Depthand parameters for the methods are given here. Specialized
diameterofvoidsinthisstudytendedtobeontheorder image processing was used as needed to highlight
of 1–3mm. This controlled oxidation damage provides indications from the NDE methods using software de-
standards for investigating the effectiveness of various veloped at NASA and available in the public domain.9
NDE techniques for detecting and sizing this type of Thesoftwareprocessinggenerallyconsistedofthefollow-
oxidation damage in this material. ing operations (in the following sequence): image crop,
contrast expansion, outlier (bad value) removal, and
NDE
waveletdenoise. If theprocessing was applied, it was ap-
Various single-sided state-of-the-art NDEmethods plied identically to the images of pre- and postoxidation
wereusedtocharacterizetheRCCsamplesbefore(base- NDE images.
Fig.4. Cross-sectionsofcavityresultingfromoxidationtreatmentfortheslotteddiskswithanoxidationtreatmentfor0.5hin12001Cstatic
laboratoryair.
656 InternationalJournalofAppliedCeramicTechnology—Roth,etal. Vol.7,No.5,2010
BSX: AscanningBSXsystemwasusedtoimagethe andprocessingwereallperformedusingsoftwareonthe
RCC samples.4 Several parameter settings were used to acquisitioncomputer.Experimental datawerecollected
determineoptimizedconditionsbutforthepurposesof usingthefollowingprocedure.Thespecimenwasplaced
comparingpre-andpostoxidationimages,thefollowing in front of the infrared (IR) camera at a distance that
settings were used in those cases. The settings were ap- allowed the sample to fill most of the active focal plane
erture of 2mm, voltage of 55kV, current of 11.6mA, and then focused. Flash lamps were set at a distance of
focal spot size of 1mm, finned collimator angle of 901 approximately 300mm. from the sample at an angle of
and exposure time per scan position of 0.2s. The scan 451. Along with the images captured after the flash, six
parametersincludedscan(X)andstep(Y)incrementsof preflash images were collected. Instantaneous and de-
0.5mm and scan velocity of 2.5mm/s. Scans were per- rivative images (from relative temperature versus time)
formed from both sides of the RCC2 sample. were obtained, and the operator normally selected the
best images foranalysis using asubjective process ofse-
EC: Scanning EC inspection of the RCC sample
lecting frames of maximum contrast. Thermography
was accomplished using a high-frequency EC surface
was performed from both sides of the sample. The
probe connected through a spring loaded z-axis gimbal
differences in surface temperature are displayed in var-
to an x/y scanning system. In the present work a probe
ious shades of gray in the image.
drivefrequencyof5MHzandscanspacingof0.025in.
wereused.Calibrationofthesystemisperformedusing Ultrasonic-Guided Waves: An ultrasonic-guided wave
an uncoated RCC test article of nominally matching measurement system7 was used to determine whether
conductivity tothe partunder test. The proberesponse total ultrasonic energy(M )ofthetimedomainguided
0
isrotatedsuchthatlift-offisinthehorizontaldirection. waveform was altered by the addition of the slot (arti-
Nonconductive plastic shims are then used to measure ficial crack) and oxidation of the RCC1 sample con-
the nominal SiC coating thickness (lift-off) and lift-off taining the single slot. M is calculated from the area
0
sensitivity. Oxidation damage under the SiC coating is underthecurveofthepowerspectraldensityS(f)ofthe
measured as a localized increase in lift-off due to the time domain waveform according to
increased spacing between the sensor and the conduct-
Z fhigh
ingcarbon–carbonsubstrateintheoxidizedareas.Scans M ¼ SðfÞdf ð1Þ
0
were performed from both sides of the sample. The flow
differences in conductivity/lift-off values are displayed
wheref andf arethelowerandupperfrequency(f )
in various shades of gray in the c-scan image. The sys- low high
boundsoftheintegrationrange,respectively.Totalenergy
tem used in this study minimized edge effects as com-
isaphysicallyunderstandableparameterthatwouldlikely
pared with the one used in Roth et al.2
bealteredduetoultrasonicscatteringbothbytheaddition
Thermography: A pulsed full-field thermographic oftheslotinto theultrasonicpath andafterfurtheralter-
NDE method utilizing flash lamps and a high-speed ationsoftheslotduetooxidationandglassfilling.Broad-
camerawasusedtoobtainimagesoftheRCCsample.6 band ultrasonic transducers were used with center
The system consists of two high-energy xenon flash frequenciesofapproximately1MHz(bothsenderandre-
lamps, each capable of producing a 1.8kJ flash with a ceiver were of the same frequency). Multiple mode exci-
5msduration.Theflashlampswereplacedatlocations tationislikelyduetotheuseofbroadbandtransducersand
that provide a relatively uniform distribution of heat theexistenceofmultipleplatewavemodesisconfirmedby
across the surface of the specimen. The transient ther- the complicated nature of the signal.7 Ultrasound was
malresponseofthespecimenafterflashingwascaptured coupledtothematerialviaelasticcouplingpads.Thedis-
using high-speed infrared camera. The camera used in tance between sending and receiving transducers was
the study is a 640 by 512 InSb focal plane array type 2.5cm. For the baseline (before slotting) condition of
witha14-bitdynamicrange.Thecameraoperatesinthe theRCC1sample(Fig.2),thetransducerswerepositioned
3–5mm wavelength range and is capable of capturing sothattheywouldstraddlethefuturepositionoftheslot,
thermaldataatratesof30Hzforthefull-arraysize.For and then after slotting, they were positioned identically
this study, a 320 by 256 portion of the full array was such that they straddled the slot.
utilized in order to increase the frame rate to approxi- Analog-to-digitalsamplingratefortheultrasonictesting
mately 60Hz. Flash initiation, data collection, storage, was 10MHz. A measurement was made (contact
www.ceramics.org/ACT NDEforCharacterizingOxidationDamage 657
Fig.5. BackscatterX-ray(BSX)results.
load53.6310.23kg[810.5lb]),thesender–receiverpair X-RayCT: X-rayCTwasusedtoprovideadditional
was lifted, moved to the next location, lowered to be in images of the oxidation damage and study oxidation in
contactwiththesample,andanothermeasurementmade. the naturally cracked RCC sample, without destructive
Twenty measurements were obtained (2 columns(cid:2)10 sectioning.8ThisSmartScanModel100(CITASystems,
rows) with measurements separated by 1mm, and mean Pueblo, CO) system utilizes a Feinfocus FXE–160
and standard deviation of (M ) was calculated. The iden- (COMET AG, Flamatt, Switzerland) microfocus X-
0
tical pattern of measurements was made before slotting, ray source to produce very high-resolution imaging of
after slotting, after oxidation, and after removing glass samples, approaching 0.025mm, in the CT mode of
sealant from the crack followed by a second oxidation. operation.Themajorhardwarecomponentsofthissys-
Additionally,thefinalscanwasdonefivetimestomeasure tem included theX-ray source,an areadetectorsystem,
reproducibilityofthetechnique.Inafutureinvestigation, a five-axis object positioning subassembly, and a lead-
theeffectoftheslotandoxidationonotherultrasonicpa- lined radiation cabinet. A dual-processor computer sys-
rameters that can be derived from broadband ultrasonic- tem controlled the data acquisition and image process-
guided wave signals will be considered.7–9 ing.Thesliceplanethicknesswas0.120mmperslicefor
Fig.6. Eddycurrent(EC)results.
658 InternationalJournalofAppliedCeramicTechnology—Roth,etal. Vol.7,No.5,2010
Fig.7. Thermographic(derivative)imageresults.Preoxidationimagesareframesat0.18sinthetemperatureversustimecooldownstreamof
frames.Postoxidationimagesareframesat0.164sinthetemperatureversustimecooldownstreamofframes.
thesourceand detector, which allowed positioning and
rotation to obtain the slice images. The differences in
X-ray density are displayed in various shades of gray.
Results and Discussion
Figures5–9showNDEresultsincludingimagesof
the samples bofore and after oxidation. Figure 5 shows
BSX results. Before oxidation, indications of the slots
were difficult to discriminate when the X-ray source
faced the RCC2 sample face with five slots (front) ex-
cept for the bottommost slot which was approximately
Fig.8. Ultrasonicguidedwave(meantotalenergyM )resultson 0.76mmwide(thewidestofthefiveslotsonthatface).
0
samplereinforcedcarbon–carbon1(Fig.2).Measurement When the X-ray source faced the sample face having
variabilityisabout2%basedonrepeatedtrials. with three slots (back), those three slots could be dis-
criminated fairly easily due to their large width (0.76
these samples. Putting together slices electronically and1.02mm)beforeoxidation.Postoxidation,alleight
gave a three-dimensional view of oxidation damage. slots were easily discriminated with the X-ray source
The sample was placed on a micropositioner between facing the front face indicating the extensive depth of
Fig.9. (a)Contrast-enhancedX-raycomputertomography(CT)sliceofportionofreinforcedcarbon–carbon2sample.Ringpatternisan
artifactoftheCTprocessing.(b)Solidthree-dimensionalvisualizationconstructedfromCTslices.
www.ceramics.org/ACT NDEforCharacterizingOxidationDamage 659
Fig.10. X-raycomputertomography(CT)ofsiliconcarbide(SiC)-coatedreinforcedcarbon–carbonoxidizedfor2.5hinairat11431C.(a)
LocationofCTslices.(b)TwoCTslices.ReprintedwithpermissionfromElsevier.
damage created by the oxidation. A postoxidation BSX frombothsidestoallowvisualizationofeachface’sslots.
image with X-ray source facing the back face is not Qualitatively, the larger slots appeared to give more
shown (nor necessary) because all slots were showed pronounced indications.
with X-ray source facing the front face postoxidation. Figure 7 shows thermography (derivative image)
Figure6showsECresults.Slotswereimpossibleto results.Slotsweremodestlydiscriminated beforeoxida-
discriminate before oxidation but were easily discrimi- tion and were easily discriminated after oxidation.
nated after oxidation. Eddy current required scanning The heat source and thermography camera were facing
660 InternationalJournalofAppliedCeramicTechnology—Roth,etal. Vol.7,No.5,2010
the slots in order for oxidation damage detection to the sample portions. X-ray CT allows the ability to
occur. sweep through various cross-sections and find the max-
Figure 8 shows ultrasonic-guided wave (mean total imumdiameternondestructivelyandthuscanresultina
energy M ) results for RCC1 with the single machined more accurate value for void diameter than destructive
0
slot.Theguidedwavetechniqueisextremelysensitiveto sectioning. Three-dimensional visualizations and ani-
surfaceconditionandtheopticalappearanceofthesur- mations composed of 11 consecutive CT slices were
face appeared the same both before (baseline) and after prepared on different sections of test samples, with one
oxidation.Controlled,identicalforcewaspresentonthe solid visualization shown in Fig. 9b. These help show
ultrasonic transducers for the ultrasonic measurements additionalmorphologicalfeaturesofthedamagedueto
beforeandafteroxidation.AscomparedwithmeanM the three-dimensional nature.
0
ofthetimedomainwaveforthebaselinecondition(be- The high resolution of the X-ray CT technique
fore machining slot), mean M had decreased after ma- makes it suitable for probing the oxidation damage be-
0
chining, after first oxidation, and after machining out lowcrazecracks,asthisdamagetendstobesmallerand
glass in the slot and performing a second oxidation. much more irregular than the damage below the ma-
However, the decrease was not monotonic for these chined slots. Figure 10shows twoX-ray CTslices fora
steps. The general decrease in M after slotting and naturally cracked RCC sample with the outlined area
0
subsequentoxidationtreatmentsisindicativeoftheslot indicatingtheoxidationdamagebelowthenaturalcrack
structure scattering the ultrasound such that less ultra- paths. The oxidation cavity shapes are highly irregular,
sonic energy reaches than receiving transducer as com- as expected from the varying coating thickness and ir-
pared with the baseline (no slot) condition. Note that regular nature of the cracks.3 Figure 11 shows a trans-
therepeatabilityofthemeasurementmethodwasgood. lucentthree-dimensionalvisualizationconstructedfrom
The five scans that were repeated after the final step 11 consecutive CT slices of this sample.
show variability of mean M to be about 2%, which is
0
significantly o8% decrease in M noted from baseline
0
to after slot machined condition. This variability is de-
pendent upon surface condition and measurement load Summary
remaining very similar.
Figure 9a shows an X-ray CT slice of a portion of In this study, RCC samples having slots in their
theRCC2sampleshowingthe0.25,0.51,and1.02mm coating machined to the depth of the SiC coating, un-
slots and the hemi spherical oxidation damage beneath derwentoxidationtreatmentstocreatevoid-likedamage
the SiC coating and slots. CT has been used to size the in the carbon–carbon substrate underneath the coating.
hemispherical regions in order to help validate and/or ThesizeofthevoidsshowedbyX-raymicro-CTwason
show deviations from the oxidation model discussed theorderof1–3mmfordepthanddiameter,depending
in.1–3 The size of the voids showed by CT was on the on the slot width. State-of-the-art single-sided NDE
orderof1–3mmfordepthanddiameter,dependingon methods, practical where access to only one side of the
theslotwidth.CTsizingwasgenerallywithin5–10%of structure is available such as for on-wing inspections,
actual values from destructive sectioning of several of were used to detect the damage. These methods in-
cluded BSX, EC, thermography, and ultrasonic guided
waves. All of the methods were successful at detecting
the oxidation damage, whereas only thermography un-
ambiguously showed the slots before oxidation. The
higher resolution of the CT technique made it suitable
to quantitatively assess oxidization damage both below
machined slots and craze cracks. Further this technique
showed the patterns of oxidation attack, which tended
to be smaller and much more irregular below the craze
cracks than under machined slots. Changes in ultra-
Fig.11. Translucentthree-dimensionalvisualizationconstructed sonic-guided wave total energy show this parameter is
fromX-raycomputertomographyslicesofsampleshowninFig.10. sensitive to structural changes in RCC.
www.ceramics.org/ACT NDEforCharacterizingOxidationDamage 661
References 5. B.A.WincheskiandJ.W.Simpson‘‘ApplicationofEddyCurrentTech-
niquesforOrbiterReinforcedCarbon–CarbonStructuralHealthMonitor-
ing,’’Contract#23-376-70-30-05,2005.
1. N.S.Jacobson,T.A.Leonhardt,D.M.Curry,andR.A.Rapp,‘‘Oxidative 6. S.M.Shepard,J.R.Lhota,B.A.Rubadeux,andT.Ahmed‘‘Onwardand
AttackofCarbon/CarbonSubstratesThroughCoatingPinholes,’’Carbon,37 inward:ExtendingthelimitsofthermographicNDE,’’ProceedingsoftheSPIE
411–419(1999). ThermosenseXXII,Vol.4020,2000,p.194.
2. D.J.Roth,N.S.Jacobson,J.N.Gray,L.M.Cosgriff,J.R.Bodis,R.A. 7. D. J. Roth, L. M. Cosgriff, M. J. Verilli, and R. T. Bhatt, ‘‘Microstruc-
Wincheski,R.W.Rauser,E.A.Burns,andM.S.McQuater,‘‘NDEfor tural and Discontinuity Characterization in Ceramic Composites Using
CharacterizingOxidationDamageinReinforcedCarbon–CarbonUsedon an Ultrasonic Guided Wave Scan System,’’ Mater Eval. 62 [9] 948–953
theNASASpaceShuttleThermalProtectionSystem,’’Ceram.Eng.Sci.Proc., (2004).
26[2]133–141(2006). 8. R.N.Yancey,G.Y.Baaklini,andS.J.Klima.‘‘NDEofAdvancedTurbine
3. N.S.Jacobson,D.J.Roth,R.W.Rauser,J.D.Cawley,andD.M.Curry, EngineComponentsandMaterialsbyComputedTomography,’’ASME,In-
‘‘Oxidation Through Coating Cracks of SiC-Protected Carbon/Carbon,’’ ternationalGasTurbineandAeroengineCongressandExposition,36th,Or-
Surf.Coat.Tech.,203372–383(2008). lando,FL,June3–6,1991.
4. E.Dugan,A.Jacobs,D.Shedlock,andD.Ekdahl‘‘DetectionofDefectsin 9. D. J. Roth, R. E. Martin, J. P. Seebo, L. B. Trinh, J. L. Walker,
FoamThermalInsulationUsingLateralMigrationBackscatterX-rayRadi- and W.P. Winfree ‘‘A Software Platform for Post-Processing Waveform-
ography,’’ProceedingsofSPIE49thAnnualMeeting,SymposiumonOptical Based NDE,’’ SAMPE Proceedings and Presentation, June 2007,
ScienceandTechnology,PenetratingRadiationSystemsandApplicationsVI,Vol. Baltimore,MD.
5541,Denver,August,2004.