Availableonlineatwww.sciencedirect.com SCIENCE c00925 Part B ELSEVIER Composites: Part B 37(2006)47-53 In situ monitoring of damage in SiC/SiC composites using acousto-ultrasonics Andrew L. Gyekenyesia, *, Gregory N Morscher, Laura M. Cosgriff OAl/NASA Glenn Research Center, 21000 Brookpark Road MS 6-1, Cleveland, OH 44135. USA POAINASA Glenn Research Center, 21000 Brookpark Road MS 106-5, Cleveland, OH 44135,USA Cleveland State University/NASA Glenn research Center, 21000 Brookpark Road MS 6-1, Cleveland, OH 44135, USA Received 15 August 2004; revised 15 January 2005: accepted 15 May 2005 Available online 25 July 2005 Abstract Acousto-ultrasonics(AU) is a nondestructive evaluation(NDE) technique that utilizes two ultrasonic transducers to interrogate the condition of a test specimen. The sending transducer introduces an ultrasonic pulse at a point on the surface of the specimen while a receiving transducer detects the signal after it has passed through the material. The aim of the method is to correlate certain empirical parameters of the detected waveform to characteristics of the material between the two transducers. The waveform parameter of interest is the attenuation due to internal damping for which information is being garnered from the frequency domain. Here, the three parameters utilized to indirectly quantify the attenuation are the ultrasonic decay rate, the mean square value of the power spectrum, and the centroid of the power spectrum. The sensitivity for each of these AU parameters was assessed with respect to the damage state of two types of Sic/Sic ceramic matrix composites( CMC). The two composite systems both had Hi-Nicalon fibers with a carbon interface but had different matrix composition that led to considerable differences in damage accumulation. Load/unload/reload tensile tests were performed and in situ aU measurements were made over the entire stress range. After analyzing the aU parameters, the overall sensitivity of the AU technique to material change or damage was quantified and shown to correlate well with the observed damage mechanisms for both material systems. In addition, the AU response was shown to be dependent on the stress state of the composites. This stress dependent behavior was observed while unloading the pecimens from the maximum stress, thereby, maintaining a constant damage state. C 2005 Elsevier ltd. all rights reserved. Keywords: D. Non-destructive testing: D. Mechanical testing: D. Ultrasonics; B. Damage tolerance 1. Introduction (static or dynamic), failure occurs due to the accumulation of distributed damage. The damage is in the form of fiber- The acousto-ultrasonic technique was developed in the matrix debonding(interface failure); matrix cracking late 1970s as an nde tool for characterizing the mechanical can be intralaminar or interlaminar (i.e. delaminations); and properties of reinforced composite materials [1-3]. The fiber fracture or fiber micro-buckling [5]. Consequently, the materials of interest involved polymer matrix composites NDE of such materials involves assessing the combined (PMCs)and ceramic matrix composites (CMCs)[1, 4 effect of the materials damage condition rather than These composites, when fabricated using continuous fibers identifying and sizing a single flaw. AU is a technique in a two-dimensional form, can be loosely categorized as that empirically quantifies the distributed damage in either a non-woven tape system, or as a woven system. advanced materials When subjecting the composites to mechanical loads The fundamental aspect of the AU method entails introducing a mechanical excitation at one point on material surface and sensing the resulting disturbance at nor Tel +12164338155: fax: +12169777150. another spot on the material surface [1-3]. The method of andrew.l gyekenyesi@ grc. nasa.gov (A.L. excitation usually involves the utilization of piezoelectric transducers. A pulsing transducer is used to introduce a 1359-8368/S- see front matter 2005 Elsevier Ltd. All rights reserved broadband, ultrasonic pulse into the specimen. The posttest.2005.05.010 ultrasonic pulse is allowed to distribute itself diffusely
In situ monitoring of damage in SiC/SiC composites using acousto-ultrasonics Andrew L. Gyekenyesia,*, Gregory N. Morscherb , Laura M. Cosgriff c a OAI/NASA Glenn Research Center, 21000 Brookpark Road MS 6-1, Cleveland, OH 44135, USA b OAI/NASA Glenn Research Center, 21000 Brookpark Road MS 106-5, Cleveland, OH 44135, USA c Cleveland State University/NASA Glenn research Center, 21000 Brookpark Road MS 6-1, Cleveland, OH 44135, USA Received 15 August 2004; revised 15 January 2005; accepted 15 May 2005 Available online 25 July 2005 Abstract Acousto-ultrasonics (AU) is a nondestructive evaluation (NDE) technique that utilizes two ultrasonic transducers to interrogate the condition of a test specimen. The sending transducer introduces an ultrasonic pulse at a point on the surface of the specimen while a receiving transducer detects the signal after it has passed through the material. The aim of the method is to correlate certain empirical parameters of the detected waveform to characteristics of the material between the two transducers. The waveform parameter of interest is the attenuation due to internal damping for which information is being garnered from the frequency domain. Here, the three parameters utilized to indirectly quantify the attenuation are the ultrasonic decay rate, the mean square value of the power spectrum, and the centroid of the power spectrum. The sensitivity for each of these AU parameters was assessed with respect to the damage state of two types of SiC/SiC ceramic matrix composites (CMC). The two composite systems both had Hi-Nicalone fibers with a carbon interface but had different matrix compositions that led to considerable differences in damage accumulation. Load/unload/reload tensile tests were performed and in situ AU measurements were made over the entire stress range. After analyzing the AU parameters, the overall sensitivity of the AU technique to material change or damage was quantified and shown to correlate well with the observed damage mechanisms for both material systems. In addition, the AU response was shown to be dependent on the stress state of the composites. This stress dependent behavior was observed while unloading the specimens from the maximum stress, thereby, maintaining a constant damage state. q 2005 Elsevier Ltd. All rights reserved. Keywords: D. Non-destructive testing; D. Mechanical testing; D. Ultrasonics; B. Damage tolerance 1. Introduction The acousto-ultrasonic technique was developed in the late 1970s as an NDE tool for characterizing the mechanical properties of reinforced composite materials [1–3]. The materials of interest involved polymer matrix composites (PMCs) and ceramic matrix composites (CMCs) [1,4]. These composites, when fabricated using continuous fibers in a two-dimensional form, can be loosely categorized as either a non-woven tape system, or as a woven system. When subjecting the composites to mechanical loads (static or dynamic), failure occurs due to the accumulation of distributed damage. The damage is in the form of fiber– matrix debonding (interface failure); matrix cracking that can be intralaminar or interlaminar (i.e. delaminations); and fiber fracture or fiber micro-buckling [5]. Consequently, the NDE of such materials involves assessing the combined effect of the material’s damage condition rather than identifying and sizing a single flaw. AU is a technique that empirically quantifies the distributed damage in advanced materials. The fundamental aspect of the AU method entails introducing a mechanical excitation at one point on a material surface and sensing the resulting disturbance at another spot on the material surface [1–3]. The method of excitation usually involves the utilization of piezoelectric transducers. A pulsing transducer is used to introduce a broadband, ultrasonic pulse into the specimen. The ultrasonic pulse is allowed to distribute itself diffusely Composites: Part B 37 (2006) 47–53 www.elsevier.com/locate/compositesb 1359-8368/$ - see front matter q 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.compositesb.2005.05.010 * Corresponding author. Tel.: C1 216 433 8155; fax: C1 216 977 7150. E-mail address: andrew.l.gyekenyesi@grc.nasa.gov (A.L. Gyekenyesi)
A L Gyekemyesi et al /Composites: Part B 37(2006)47-53 into the specimen. a diffuse wave is a complicated when taking measurements during tensile loads, when superposition of wave modes and sample reverberations cracks may be open, versus taking measurements during a that resembles an acoustic emission signal. The decay of this zero or compressive stress, when cracks may be closed. resulting incoherent field is then examined as a function of frequency [4, 6]. Typically in AU, two separate transducers are applied at some fixed separation on the face of the specimen. An ultrasonic wave is introduced and allowed to 2. Acousto-ultrasonics: theory propagate along the length of the specimen. The ultrasonic waves observed in the composite specimens tested here There are basically two approaches for analyzing the were composed primarily of plate waves. This was due to received diffuse (i.e. multi-mode) AU signal. First to be the fact that the artificially produced waves had wavelengths discussed is the analysis method that is heavily utilized at on the order of 1-2 cm while the smallest dimension of the NASA GRC 6, 10]. Titled the diffuse field decay rate, the specimens (i.e. thickness) was approximately 2 mm [7]. method involves quantifying the internal damping of More specifically, the two dominant modes of ultrasonic vibrational energy in materials. Damping is measured wave travel within the specimens tested were the through determination of the volume averaged decay rate extensional wave and the flexural wave of the ultrasonic field as a function of frequency and time. A a desirable attribute of CMCs is the non-linear stress- greater decay rate indicates a greater amount of internal strain behavior when loaded in tension. Such 'graceful damping. Determining the decay rate is accomplished by failure[8] is contrary to the brittle and catastrophic failures dividing the recorded waveform into a number of time observed in monolithic ceramics. This attribute combined windows. after which fast fourier transforms are with high-temperature durability makes CMCs potential performed individually on each of the time windows to materials for applications such as combustor liners within obtain the power spectra. Note that the power spectrum future high-performance civil aircraft engines [9]. The non- the square of the amplitude function at a given frequency linear response is due to the accumulated damage The total energy of each time window is calculated using the mechanisms previously mentioned. The primary contribu- respective power spectrum (i.e. the area under the power tors to the non-linear response are the transverse matrix spectrum plot). Further refinement is achieved by obtaining cracks, although, additional damage mechanisms such as the energy content of various frequency windows within the interlaminar matrix cracks (i.e. interlaminar delaminations) power spectrums. The frequency windows (i.e. filtering) and fiber breaks may also contribute to this behavior. The allow for noise reduction by focusing on the frequency underlying feature that governs the energy absorbing ranges that contain the majority of the energy content. The process of damage accumulation in CMCs is the interface selection of the frequency ranges is based on experience and that lies between the fibers and the matrix. A properly may vary for material and sensor types. Lastly, an planned interface allows matrix cracks to propagate around exponential decay curve is fit to the plot of energy versus fibers, via the interface, rather than fracturing the fiber itself. time window for the decaying portion of the wave. The This results in matrix cracks that are bridged by unbroken exponent in the equation is defined as the diffuse field decay fibers In the vicinity of the matrix cracks, the bridging fibers rate. Mathematically the equation for total energy is experience higher strains than fibers in areas without matrix expressed as follows: cracks. This, in turn, causes the observed non-linearity. The extent of the non-linearity in a CMC containing matrix M(fow fhigh, t)=aflow, high )e ow/ugh ir cracks is predominantly due to the interfacial shear strength, i.e. lower interfacial shear strengths result in longer load- where t is time; fow and fhigh are the low and high frequenc transfer lengths between fiber and matrix, wider crack filter limits; M is the mean square value of the power penings at stress, and therefore, larger non-linear strains at spectrum as a function of frequency and time; a is the a given stress in comparison to a similar composite with a intercept as a function of frequency( this value is not utilized in this study); and B is the diffuse field decay rate as a In this study, the aU technique was utilized to unction frequency continuously monitor damage during load/unload/reload The second method of post processing involves working tensile tests. Two types of SiC/SiC(silicon carbide with the entire time domain signal and ag dIn cond ducting a fiber/silicon carbide matrix)composite systems were tested. FFT in order to obtain the power spectrum (4, 11, 12 One had a standard Sic matrix while the other had an Certain parameters concerning the power spectrum were enhanced formulation. The main objective was to show the shown to be sensitive to various types of damage in feasibility of AU as an in situ, NDE tool for monitoring the composite materials [4, 12]. Called the shape parameters, the damage for the given ceramic composites. Also of interest general equations are defined as follows was the stress dependence of AU. Because the AU parameters studied here are an indirect measure of =1.2.3 attenuation, it is important to understand the au behavior Mr-kfe =2,3,4..r>k(2)
into the specimen. A diffuse wave is a complicated superposition of wave modes and sample reverberations that resembles an acoustic emission signal. The decay of this resulting incoherent field is then examined as a function of frequency [4,6]. Typically in AU, two separate transducers are applied at some fixed separation on the face of the specimen. An ultrasonic wave is introduced and allowed to propagate along the length of the specimen. The ultrasonic waves observed in the composite specimens tested here were composed primarily of plate waves. This was due to the fact that the artificially produced waves had wavelengths on the order of 1–2 cm while the smallest dimension of the specimens (i.e. thickness) was approximately 2 mm [7]. More specifically, the two dominant modes of ultrasonic wave travel within the specimens tested were the extensional wave and the flexural wave. A desirable attribute of CMCs is the non-linear stress– strain behavior when loaded in tension. Such ‘graceful failure’ [8] is contrary to the brittle and catastrophic failures observed in monolithic ceramics. This attribute combined with high-temperature durability makes CMCs potential materials for applications such as combustor liners within future high-performance civil aircraft engines [9]. The nonlinear response is due to the accumulated damage mechanisms previously mentioned. The primary contributors to the non-linear response are the transverse matrix cracks, although, additional damage mechanisms such as interlaminar matrix cracks (i.e. interlaminar delaminations) and fiber breaks may also contribute to this behavior. The underlying feature that governs the energy absorbing process of damage accumulation in CMCs is the interface that lies between the fibers and the matrix. A properly planned interface allows matrix cracks to propagate around fibers, via the interface, rather than fracturing the fiber itself. This results in matrix cracks that are bridged by unbroken fibers. In the vicinity of the matrix cracks, the bridging fibers experience higher strains than fibers in areas without matrix cracks. This, in turn, causes the observed non-linearity. The extent of the non-linearity in a CMC containing matrix cracks is predominantly due to the interfacial shear strength, i.e. lower interfacial shear strengths result in longer loadtransfer lengths between fiber and matrix, wider crack openings at stress, and therefore, larger non-linear strains at a given stress in comparison to a similar composite with a higher interfacial shear stress. In this study, the AU technique was utilized to continuously monitor damage during load/unload/reload tensile tests. Two types of SiC/SiC (silicon carbide fiber/silicon carbide matrix) composite systems were tested. One had a standard SiC matrix while the other had an enhanced formulation. The main objective was to show the feasibility of AU as an in situ, NDE tool for monitoring the damage for the given ceramic composites. Also of interest was the stress dependence of AU. Because the AU parameters studied here are an indirect measure of attenuation, it is important to understand the AU behavior when taking measurements during tensile loads, when cracks may be open, versus taking measurements during a zero or compressive stress, when cracks may be closed. 2. Acousto-ultrasonics: theory There are basically two approaches for analyzing the received diffuse (i.e. multi-mode) AU signal. First to be discussed is the analysis method that is heavily utilized at NASA GRC [6,10]. Titled the diffuse field decay rate, the method involves quantifying the internal damping of vibrational energy in materials. Damping is measured through determination of the volume averaged decay rate of the ultrasonic field as a function of frequency and time. A greater decay rate indicates a greater amount of internal damping. Determining the decay rate is accomplished by dividing the recorded waveform into a number of time windows, after which Fast Fourier Transforms are performed individually on each of the time windows to obtain the power spectra. Note that the power spectrum is the square of the amplitude function at a given frequency. The total energy of each time window is calculated using the respective power spectrum (i.e. the area under the power spectrum plot). Further refinement is achieved by obtaining the energy content of various frequency windows within the power spectrums. The frequency windows (i.e. filtering) allow for noise reduction by focusing on the frequency ranges that contain the majority of the energy content. The selection of the frequency ranges is based on experience and may vary for material and sensor types. Lastly, an exponential decay curve is fit to the plot of energy versus time window for the decaying portion of the wave. The exponent in the equation is defined as the diffuse field decay rate. Mathematically the equation for total energy is expressed as follows: Mðflow; fhigh; tÞ Z aðflow; fhighÞe KbðflowfhighÞt (1) where t is time; flow and fhigh are the low and high frequency filter limits; M is the mean square value of the power spectrum as a function of frequency and time; a is the intercept as a function of frequency (this value is not utilized in this study); and b is the diffuse field decay rate as a function of frequency. The second method of post processing involves working with the entire time domain signal and again conducting a FFT in order to obtain the power spectrum [4,11,12]. Certain parameters concerning the power spectrum were shown to be sensitive to various types of damage in composite materials [4,12]. Called the shape parameters, the general equations are defined as follows: Srk Z Mr MrKkf k c k Z 1; 2; 3. r Z2; 3; 4. rOk (2) 48 A.L. Gyekenyesi et al. / Composites: Part B 37 (2006) 47–53
A L Gyekenyesi et al. Composites: Part B 37(2006)47-53 where Constituents of Tested Hi-Nicalon":Composit O) df (3) Composite VolumeMatrix Porosity Fraction Modulus S( is the power spectral density, and f is the frequency Pa Two parameters related to Eq (3)have been shown in the HN-C-STD 0.29 Standard past to correlate well with damage progression in CMCs CVI SIC subjected to mechanical loads [4]. The first is called the HN-C-ENH 0.31 Enhanced 31.5 CVI SIC nean square value of the power spectral density, Mo, and is defined as the area under the power spectrum. The second Hi-Nicalon Fiber: diameter=13 um: E=270 GPa parameter is the location of the centroid of the power spectrum, fe, found by the following expression experiment, the stress was increased to a stress level of 69 MPa and then unloaded to zero stress This was followed f e (4) by stress reversal points(i.e. the maximum stresses attained after which point the specimens were unloaded to zero stress) Note that Eqs.(2)-(4)provide scalar values that capture of 138, 207, 276, 345 MPa, and then finally to failure.The the general characteristics of the power spectrum function. increasing stress rate was 0.69 MPa/s, while the decreasing The diffuse field decay rate as well as Mo and fe are stress rate was 2. 1 MPa/s Load holds were conducted during calculated utilizing an in-house, AU software package the AU measurements. AU measurements were taken every developed at NASA GRC [13] 69 MPa during loading as well as unloading. Broadband, Lastly, although the above analyses are based on the 1 MHz piezoelectric transducers were utilized for sending premise that a diffuse ultrasonic field in an isolated sample and receiving the aU wave. a gel couplant was used for will decay only from internal absorption mechanisms, there contacting the specimen. The 12.7 mm diameter transducers may be a dependence of this empirical technique to local were attached to the specimen using spring-activated clips boundary conditions(e. g. specimen geometry, gripping force The transducers were placed on the same face of the transducer contact force, etc ). Ref. [14 summarizes the specimen at a separation of 60 mm as shown in Fig. 1. The results of a study comparing changes in the AU parameters computer analog-to-digital acquisition rate was set at due to material damage to 25 MHz and the time window was 49.92 us. The stress resulting from modifications to the boundary conditions. dependence of the AU response was studied while unloading Although material damage was the dominant factor concern- the specimens from the maximum stress, thereby, ing the aU results in the study, the boundary effects did have an influence and should be considered during measurements 3. Experimental procedure Loom temperature load/unload/reload tensile tests were performed on Hi-Nicalon(Nippon Carbon, Japan) reinforced chemical vapor infiltrated (CvI) Sic matrix transduce composites(Manufactured by Honeywell Composites). All the composites were composed of eight plies of balanced, 8-harness satin woven fiber cloth. The woven cloth was Clip-on stacked in a 0/90 fashion. An interphase layer, approxi- Extensometer mately 0.5 um in thickness, consisted of carbon that was applied by the chemical vapor infiltration( CVI)method. There were two CvI SiC matrices studied. one consisted of a standard CVI SiC (referred to as HN-C-STD)and the second consisted of an enhanced matrix(referred to as HN C-ENd) that included boron carbide(Bc) additions. The wo material systems tested in this study are listed in Table 1 along with relevant constituent and composite properties. The test procedure for each individual specimen follows: the specimen was mechanically loaded in a re Front view Side view load/unload/reload fashion with each reload the max Fig. 1. Schematic representation of tensile test specimen and sensor stress level was progressively increased. For a typical
where Mr Z ðfhigh flow SðfÞf r df (3) S(f) is the power spectral density, and f is the frequency. Two parameters related to Eq. (3) have been shown in the past to correlate well with damage progression in CMCs subjected to mechanical loads [4]. The first is called the mean square value of the power spectral density, M0, and is defined as the area under the power spectrum. The second parameter is the location of the centroid of the power spectrum, fc, found by the following expression: fc Z M1 M0 (4) Note that Eqs. (2)–(4) provide scalar values that capture the general characteristics of the power spectrum function. The diffuse field decay rate as well as M0 and fc are calculated utilizing an in-house, AU software package developed at NASA GRC [13]. Lastly, although the above analyses are based on the premise that a diffuse ultrasonic field in an isolated sample will decay only from internal absorption mechanisms, there may be a dependence of this empirical technique to local boundary conditions (e.g. specimen geometry, gripping force, transducer contact force, etc.). Ref. [14] summarizes the results of a study comparing changes in the AU parameters due to material damage to changes in AU parameters resulting from modifications to the boundary conditions. Although material damage was the dominant factor concerning the AU results in the study, the boundary effects did have an influence and should be considered during measurements. 3. Experimental procedure Room temperature load/unload/reload tensile tests were performed on Hi-Nicalone (Nippon Carbon, Japan) reinforced chemical vapor infiltrated (CVI) SiC matrix composites (Manufactured by Honeywell Composites). All the composites were composed of eight plies of balanced, 8-harness satin woven fiber cloth. The woven cloth was stacked in a 0/90 fashion. An interphase layer, approximately 0.5 mm in thickness, consisted of carbon that was applied by the chemical vapor infiltration (CVI) method. There were two CVI SiC matrices studied, one consisted of a standard CVI SiC (referred to as HN-C-STD) and the second consisted of an enhanced matrix (referred to as HNC-ENH) that included boron carbide (B4C) additions. The two material systems tested in this study are listed in Table 1 along with relevant constituent and composite properties. The test procedure for each individual specimen was as follows: the specimen was mechanically loaded in a repeated load/unload/reload fashion. With each reload, the maximum stress level was progressively increased. For a typical experiment, the stress was increased to a stress level of 69 MPa and then unloaded to zero stress. This was followed by stress reversal points (i.e. the maximum stresses attained after which point the specimens were unloaded to zero stress) of 138, 207, 276, 345 MPa, and then finally to failure. The increasing stress rate was 0.69 MPa/s, while the decreasing stress rate was 2.1 MPa/s. Load holds were conducted during the AU measurements. AU measurements were taken every 69 MPa during loading as well as unloading. Broadband, 1 MHz piezoelectric transducers were utilized for sending and receiving the AU wave. A gel couplant was used for contacting the specimen. The 12.7 mm diameter transducers were attached to the specimen using spring-activated clips. The transducers were placed on the same face of the specimen at a separation of 60 mm as shown in Fig. 1. The computer analog-to-digital acquisition rate was set at 25 MHz and the time window was 49.92 ms. The stress dependence of the AU response was studied while unloading the specimens from the maximum stress, thereby, Table 1 Constituents of Tested Hi-Nicalone,1 Composites Composite Volume Fraction of fiber Matrix Porosity % Elastic Modulus, GPa HN-C-STD 0.29 Standard CVI SiC 30.0 177 HN-C-ENH 0.31 Enhanced CVI SiC 31.5 180 a Hi-Nicalone Fiber: diameterZ13 mm: EfZ270 GPa A A A A Tabs Clip-on Extensometer 60 mm 25 mm transducer Front view Side view Fig. 1. Schematic representation of tensile test specimen and sensor alignment. A.L. Gyekenyesi et al. / Composites: Part B 37 (2006) 47–53 49
L Gyekemyesi et al /Composites: Part B 37(2006)47-53 maintaining a constant damage state (i.e. damage does not 1200 change during unload, although, crack closure may occur ). 1000 4. Results 4.. Mechanical behavior The stress-strain behavior observed for the mechanicall tested materials in this study were very similar to earlier 50 studies concerning comparable materials [15, 16]. The 0 stress-strain plots from load/unload/reload tensile tests are 00.10203040.50.60.7 shown in Figs. 2 and 3. As the extent of non-linearity Strain, increased, the width of the hysteresis loops also increased Another report [17], by the authors, is available and focuses Fig. 3. Hysteresis stress-strain curves and gage section AE energy cumulated during load/unload/reload tensile test of HN-C-EN composite on the in-depth analysis of modal acoustic emissions(AE) Specimen 3c ENH in respect to the damaged CMC specimens. Some of the information of that report is presented here in order to whether residual stresses are tensile, compressive or zero. If provide clarity. Figs 2 and 3 indicate the AE activity, in the the point of intersection is at a positive stress, then the form of accumulated AE energy. AE energy was found to matrix residual stresses are compressive; if the point of have a nearly direct relationship to the amount of transverse intersection is at a negative stress, then the matrix residual matrix cracking in these types of composites [15, 16]. stresses are tensile; and if the point of intersection is at zero Therefore, most matrix crack formation and propagation stress then there are no residual stresses. For these occurred between 0.05 and 0.35 percent strain for the composites, the intersection was essentially zero stress or composites. Matrix crack saturation, as evidenced by the a slightly negative stress. The goal of these calculations was drastic reduction in the rate of cumulative AE energy, to gain knowledge concerning the stress levels that define ccurred at 0.35%(200 MPa for the HN-C-STD crack closure stress and then try to characterize the composite and =250 MPa for the HN-C-ENH composite). relationship between ultrasonic attenuation and crack Also shown in Figs. 2 and 3 are lines through the hysteresis loops that were utilized for approximating the residual stresses within the given composite. These internal residual stresses, induced during the post processing cool 42. AU results down are the result of the thermal mismatch between the fibers and the matrix material. a method has been defined Figs. 46 show the results of the in situ au. for each chat can approximate the existence of these internal residual plot, the parameters were normalized by the preload values stresses[18]. The methodology of Ref [18] dictates that the (i.e. prior to the application of load or damage), therefore, location of the point of intersection of the average slopes allowing for the display of all the parameters on a single of the hysteresis curves above the closure stress defines lot. Each data point was based on a single calculation that 0. 4to 1.6 MHz filter Mo at zero stress 2500 3.51 6-partitions for decay fc at zero stress - Decay rate at zero stress 250 5b STD 2000 H Mo at Decay rate at maximum 15 stress 1000 00.10203040.50.60.7 Maximum hysteresis stress, MPa Fig. 2. Hysteresis stress-strain curves and gage section AE energy Fig 4 Normalized AU parameters as a function of maximum hysteresis cumulated during load/unload/reload tensile test of HN-C-STD composite. stress for the HN-C-STD specimen. AU measurements were taken at The lines through the hysteresis loops are for the graphical construct, after maximum load of current cycle(solid lines)and at zero stress after Steen [14].(Specimen 5b STD). unloading from current maximum(dashed lines
maintaining a constant damage state (i.e. damage does not change during unload, although, crack closure may occur). 4. Results 4.1. Mechanical behavior The stress–strain behavior observed for the mechanically tested materials in this study were very similar to earlier studies concerning comparable materials [15,16]. The stress–strain plots from load/unload/reload tensile tests are shown in Figs. 2 and 3. As the extent of non-linearity increased, the width of the hysteresis loops also increased. Another report [17], by the authors, is available and focuses on the in-depth analysis of modal acoustic emissions (AE) in respect to the damaged CMC specimens. Some of the information of that report is presented here in order to provide clarity. Figs. 2 and 3 indicate the AE activity, in the form of accumulated AE energy. AE energy was found to have a nearly direct relationship to the amount of transverse matrix cracking in these types of composites [15,16]. Therefore, most matrix crack formation and propagation occurred between 0.05 and 0.35 percent strain for the composites. Matrix crack saturation, as evidenced by the drastic reduction in the rate of cumulative AE energy, occurred at w0.35% (w200 MPa for the HN-C-STD composite and w250 MPa for the HN-C-ENH composite). Also shown in Figs. 2 and 3 are lines through the hysteresis loops that were utilized for approximating the residual stresses within the given composite. These internal residual stresses, induced during the post processing cool down, are the result of the thermal mismatch between the fibers and the matrix material. A method has been defined that can approximate the existence of these internal residual stresses [18]. The methodology of Ref. [18] dictates that the location of the point of intersection of the average slopes of the hysteresis curves above the closure stress defines whether residual stresses are tensile, compressive or zero. If the point of intersection is at a positive stress, then the matrix residual stresses are compressive; if the point of intersection is at a negative stress, then the matrix residual stresses are tensile; and if the point of intersection is at zero stress then there are no residual stresses. For these composites, the intersection was essentially zero stress or a slightly negative stress. The goal of these calculations was to gain knowledge concerning the stress levels that define crack closure stress and then try to characterize the relationship between ultrasonic attenuation and crack openings. 4.2. AU results Figs. 4–6 show the results of the in situ AU. For each plot, the parameters were normalized by the preload values (i.e. prior to the application of load or damage), therefore, allowing for the display of all the parameters on a single plot. Each data point was based on a single calculation that Fig. 2. Hysteresis stress–strain curves and gage section AE energy cumulated during load/unload/reload tensile test of HN-C-STD composite. The lines through the hysteresis loops are for the graphical construct, after Steen [14]. (Specimen 5b STD). Fig. 3. Hysteresis stress–strain curves and gage section AE energy cumulated during load/unload/reload tensile test of HN-C-ENH composite. (Specimen 3c ENH). –1.5 –1 –0.5 0 0.5 1 1.5 2 2.5 3 3.5 0 50 100 150 200 250 Maximum hysteresis stress, MPa Normalized AU parameters Mo at zero stress fc at zero stress Decay rate at zero stress Mo at maximum stress fc at maximum stress Decay rate at maximum stress 0.4to 1.6 MHz filter 6-partitions for decay rate 5b STD Fig. 4. Normalized AU parameters as a function of maximum hysteresis stress for the HN-C-STD specimen. AU measurements were taken at maximum load of current cycle (solid lines) and at zero stress after unloading from current maximum (dashed lines). 50 A.L. Gyekenyesi et al. / Composites: Part B 37 (2006) 47–53
A L Gyekenryesi et al /Composites: Part B 37 (2006)47-53 1.21 +In-situ Mo 0. 4 to 1.6 MHz filter fixed even though the wave arrival time changed, thereby 5b STD introducing large errors in the decay rate calculations. These issues are being addressed in the next version of the software. The parameters that appeared to track damage successfully were Mo(reductions of 75%o at stress ar at zero stress)and f (20%o at stress and no change at zero stress). Both values were reduced as a result of increased attenuation due to the accumulation of transverse cracks The most drastic reductions occurred prior to 138 MPa. This 0 correlated well with the ae data in Fig. 2 that showed the e少。es9s9eeA oo rate of AE energy moderating at approximately the same stress marking the reduced rate of transverse cracking Fig. 5. Normalized Mo as a function of in situ stress for the HN-C-STD and fe; a more drastic attenuation was seen for the aU data load/unload/reload cycles. Plot shows the stress dependence of the AU captured at load. This was due to the increased crack openings under tensile stress. Fig. 5 focuses on the Mo data as it was followed through the repeated load cycles involved a 4 wave average. For this study, the focus was on According to the aU results of Fig. 5, damage was initiated establishing an understanding of the aU trends. Future sometime between 69 and 103 MPa, after which Mo esearch will focus on creating statistical confidence continued to decrease, although with a saw tooth pattern through further testing involving a larger number of Again, this pattern was the result of the cracks opening and pecimen repetitions. Lastly, an Appendix a is provided closing. Measurements taken at zero stress indicated less at the end of the report in order to provide units as well as attenuation (i.e. increased ultrasonic energy as witnessed by typical values for the non-normalized au parameters Mo). In Fig. 5, the largest stress dependent change always Fig. 4 displays the au behavior for the HN-C-STD occurred between O and 69 MPa. Further explanations of the composite au behavior will be given after discussing the HN-C-ENH maximum hysteresis stress with the AU measurements results taken both at the maximum stress as well as at zero stress Fig 6 displays the results for the HN-C-ENH composite after unloading. For the HN-C-STD material, the diffuse The enhanced composite did not reveal a stress dependence field decay rate did not follow any real pattern and did not The diffuse field decay rate and Mo correlated well with appear to correlate to damage. The reason for this may be damage. The decay rate increased over 300% and Mo due to the fact that only the decaying portion of the received decreased by over 60%. It should be noted that no signal was supposed to be studied, and in turn, time bins modifications were made to the software, so the values for within the total window were predefined in the capture the decay rate, although appearing reasonable in Fig. 6, software to only focus on that portion of the curve Problems should be interpreted with caution due to the errors seen in arose when the ultrasonic velocity changed and caused Fig 4 regarding the HN-C-STD specimen A figure focusing different arrival times for the wave as a result of the on in situ au behavior during the stress cycling was not degrading material stiffness. The predefined time bins were necessary, since there was no apparent stress dependence for the HN-C-ENH composite Again, the most drastic changes Mo at zero stress in the AU parameters occurred at a maximum hysteresis fc at zero stress 0.4 to 1.6 MHz filter stress of 138 MPa. At this point, it was apparent that only HN-C-STD composites showed a stress dependence of th 3C ENH AU parameters. It was speculated that the stress dependent 351→◆ Decay rate at maximum stress attenuation behavior was dependent on the degree of crack opening as is discussed in Section 4.3. 寻2 43. Crack density and interfacial shear strength analysis The specimens were cut and polished in order to measure the transverse crack densities. The final. saturated. crack 0 50 100 150 200 250 300 350 400 densities for the HN-C-STD and the HN-C-ENH compo- Maximum hysteresis stress, MPa sites were found to be 2.1 and w4. 6 cracks/mm 6. Normalized AU parameters as a function of maximum hysteresis respectively. The significantly smaller crack density and for the HN-C-ENH specimen. AU meas ts were taken at the significantly wider hysteresis loop-width for the HN-C- aximum load of current cycle(solid lines) and at zero stress after STD composite inferred that the interfacial shear(sliding) unloading from current maximum(dashed lines) stress.t for composites was significantly lower than
involved a 4 wave average. For this study, the focus was on establishing an understanding of the AU trends. Future research will focus on creating statistical confidence through further testing involving a larger number of specimen repetitions. Lastly, an Appendix A is provided at the end of the report in order to provide units as well as typical values for the non-normalized AU parameters. Fig. 4 displays the AU behavior for the HN-C-STD composite. Shown are the AU values as a function of maximum hysteresis stress with the AU measurements taken both at the maximum stress as well as at zero stress after unloading. For the HN-C-STD material, the diffuse field decay rate did not follow any real pattern and did not appear to correlate to damage. The reason for this may be due to the fact that only the decaying portion of the received signal was supposed to be studied, and in turn, time bins within the total window were predefined in the capture software to only focus on that portion of the curve. Problems arose when the ultrasonic velocity changed and caused different arrival times for the wave as a result of the degrading material stiffness. The predefined time bins were fixed even though the wave arrival time changed, thereby introducing large errors in the decay rate calculations. These issues are being addressed in the next version of the software. The parameters that appeared to track damage successfully were M0 (reductions of 75% at stress and 50% at zero stress) and fc (20% at stress and no change at zero stress). Both values were reduced as a result of increased attenuation due to the accumulation of transverse cracks. The most drastic reductions occurred prior to 138 MPa. This correlated well with the AE data in Fig. 2 that showed the rate of AE energy moderating at approximately the same stress marking the reduced rate of transverse cracking. Another observation relates to the stress dependence of M0 and fc; a more drastic attenuation was seen for the AU data captured at load. This was due to the increased crack openings under tensile stress. Fig. 5 focuses on the M0 data as it was followed through the repeated load cycles. According to the AU results of Fig. 5, damage was initiated sometime between 69 and 103 MPa, after which M0 continued to decrease, although with a saw tooth pattern. Again, this pattern was the result of the cracks opening and closing. Measurements taken at zero stress indicated less attenuation (i.e. increased ultrasonic energy as witnessed by M0). In Fig. 5, the largest stress dependent change always occurred between 0 and 69 MPa. Further explanations of the AU behavior will be given after discussing the HN-C-ENH results. Fig. 6 displays the results for the HN-C-ENH composite. The enhanced composite did not reveal a stress dependence. The diffuse field decay rate and M0 correlated well with damage. The decay rate increased over 300% and M0 decreased by over 60%. It should be noted that no modifications were made to the software, so the values for the decay rate, although appearing reasonable in Fig. 6, should be interpreted with caution due to the errors seen in Fig. 4 regarding the HN-C-STD specimen. A figure focusing on in situ AU behavior during the stress cycling was not necessary, since there was no apparent stress dependence for the HN-C-ENH composite. Again, the most drastic changes in the AU parameters occurred at a maximum hysteresis stress of 138 MPa. At this point, it was apparent that only HN-C-STD composites showed a stress dependence of the AU parameters. It was speculated that the stress dependent attenuation behavior was dependent on the degree of crack opening as is discussed in Section 4.3. 4.3. Crack density and interfacial shear strength analysis The specimens were cut and polished in order to measure the transverse crack densities. The final, saturated, crack densities for the HN-C-STD and the HN-C-ENH composites were found to be w2.1 and w4.6 cracks/mm, respectively. The significantly smaller crack density and the significantly wider hysteresis loop-width for the HN-CSTD composite inferred that the interfacial shear (sliding) stress, t, for these composites was significantly lower than 0 0.2 0.4 0.6 0.8 1 1.2 0 346934 0 69 103 103 103 69 0 69 138 69 0 69 13820713869 0 69 138 207 In-situ stress, MPa Normalized Mo In-situ Mo 0.4 to 1.6 MHz filter 5b STD Fig. 5. Normalized M0 as a function of in situ stress for the HN-C-STD composite. AU measurements were taken at numerous points during the load/unload/reload cycles. Plot shows the stress dependence of the AU parameter. 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 50 100 150 200 250 300 350 400 Maximum hysteresis stress, MPa Normalized AU parameters Mo at zero stress fc at zero stress Decay rate at zero stress Mo at maximum stress fc at maximum stress Decay rate at maximum stress 0.4 to 1.6 MHz filter 6-partitions for decay rate 3c ENH Fig. 6. Normalized AU parameters as a function of maximum hysteresis stress for the HN-C-ENH specimen. AU measurements were taken at maximum load of current cycle (solid lines) and at zero stress after unloading from current maximum (dashed lines). A.L. Gyekenyesi et al. / Composites: Part B 37 (2006) 47–53 51
52 A L Gyekemyesi et al /Composites: Part B 37(2006)47-53 for the HN-C-ENH composite. The interfacial shear stress for significant increases in attenuation as a function of was estimated from hysteresis loop widths, dE max, from the increasing stress. The threshold for crack opening displace- relationship [ 19] ments that induced stress dependence was also assumed G\[b(1-a)(Rn) to be a function of the transducer's central frequency (5) AU stress dependence could probably be defined for the (4f-Em) HN-C-ENH composite if higher frequency transducers (shorter ultrasonic wavelengths) were utilized since the where a, is the peak hysteresis loop stress, f is the fiber shorter wavelengths would be more sensitive(i.e attenuate volume fraction, R, is the fiber radius, Pc is the matrix crack more)in regards to the smaller crack openings density, Em is the matrix modulus, and aI and b2 are constants based on the elastic properties of the constituents [20]. From Eq.(5), for the HN-C-STD composite T was 5 Conclusions estimated to be approximately 31 MPa whereas for the HN- C-ENH composite T was estimated to be approximately 350 MPa. These values may be overestimated due to the to the damage states of two types of SiC/Sic ceramic matrix stiffening of the hysteresis loop at low stresses and the composites(CMC) subjected to load/unload/reload tensile probable overlap of sliding lengths around cracks, however, tests. The two composite systems both had Hi-Nicalon the average pullout lengths of fibers on the fracture surfaces fibers with a carbon interface but had different matrix differed between the two composites by over an order compositions that led to differences in damage accumu- magnitude (>I mm for the HN-C-STD composite and lation. The majority of the damage for each of the composite 50 um for the HN-C-ENH composite)confirming at least systems was in the form of distributed transverse cracks. the qualitative difference in the interfacial shear stresses for The standard matrix composite, labeled HN-C-STD, had a the composite systems. This was important because the lower transverse crack density than the enhanced matrix amount of crack opening was strongly dependent on the interfacial sliding stress between the fiber and the matrix composite, HN-C-ENH (2.1 versus 4.6 cracks/mm, For the lower T material. HN-C-stD the stress respectively). After analyzing the AU parameters, Mo (defined as the mean square value of the power spectral dependent attenuation appeared at approximately 100 MPa density)was shown to be the most consistent at monitoring as was indicated by the growing separation between the the accumulated damage. The energy of the captured measurements taken at zero stress and at maximum stress ultrasonic wave, as quantified by Mo, was reduced by over (see Fig. 4). For the higher T material, HN-C-ENH, no 50% for each of the material systems when comparing the significant stress dependent attenuation was observed. undamaged and damaged states. In addition, AU measure- Based on AE activity and the measured crack densities ments related to the HN-C-STD composite showed a the failed specimens, the HN-C-ENH composites had twice dependence on the in situ tensile stress. Given a constant as many cracks as the HN-C-STD composites Therefore, it damage state, there were variations between AU measure- was not just the presence of matrix cracks that caused the ments taken at zero stress and au measurements taken at significant stress dependent attenuation. The degree of crack maximum stress. This stress dependence of the AU pening was assumed to have a major role in the stress parameters was not evident in the HN-C-ENH composites dependent attenuation of the acoustic signal. Assuming only Based on the analysis of the hysteresis loops, the crack frictional forces at the interface, the crack opening opening for HN-C-STD was an order of magnitude greater displacement, u, was estimated by Eq (6)[21] than the crack opening for HN-C-ENH at a given stress. It F-R was concluded that excessive crack opening displacements in the HN-C-STD caused the observed stress dependence of {47E1+a]} the AU parameter. As stated earlier, this study focused on defining the trends of the aU parameters as a function of note that Er is the fiber modulus. Based on the T values damage and stress for two types of composites. Although derived from the above analysis of the hysteresis loops, the statistical confidence needs to be developed by further crack opening for HN-C-STD was an order of magnitude testing, it is encouraging to know that similar results greater than the crack opening for HN-C-ENH at a given concerning stress dependent attenuation in these composites stress. Since, significant stress dependent damping occurred were observed in Ref [17]. In Ref [17]measurements were between 69 and 100 MPa for the HN-C-STD material and made utilizing modal acoustic emissions data. Future no real increase in stress dependent damping occurred for research involves further testing under various loading the HN-C-ENH material(see Fig. 4 and note growing conditions in order to build confidence in the use of AU separation between zero stress and at-load AU measure- as an in situ health monitoring tool for advanced ceramic ments), it was concluded both a certain level of matrix crack matrix composites. More specifically, an in-depth density and crack opening displacement (i.e. large u understanding of the frequency content of the ultrasonic resulting from a low interfacial shear stress, T) was required signal may allow for this NDE technique to characterize
for the HN-C-ENH composite. The interfacial shear stress was estimated from hysteresis loop widths, demax, from the relationship [19]: t Z s2 p 2d3max b2ð1Ka1fÞ 2ðRfrcÞ ð4f 2EmÞ (5) where sp is the peak hysteresis loop stress, f is the fiber volume fraction, Rf is the fiber radius, rc is the matrix crack density, Em is the matrix modulus, and a1 and b2 are constants based on the elastic properties of the constituents [20]. From Eq. (5), for the HN-C-STD composite t was estimated to be approximately 31 MPa whereas for the HNC-ENH composite t was estimated to be approximately 350 MPa. These values may be overestimated due to the stiffening of the hysteresis loop at low stresses and the probable overlap of sliding lengths around cracks; however, the average pullout lengths of fibers on the fracture surfaces differed between the two composites by over an order of magnitude (O1 mm for the HN-C-STD composite and w50 mm for the HN-C-ENH composite) confirming at least the qualitative difference in the interfacial shear stresses for the composite systems. This was important because the amount of crack opening was strongly dependent on the interfacial sliding stress between the fiber and the matrix. For the lower t material, HN-C-STD, the stress dependent attenuation appeared at approximately 100 MPa as was indicated by the growing separation between the measurements taken at zero stress and at maximum stress (see Fig. 4). For the higher t material, HN-C-ENH, no significant stress dependent attenuation was observed. Based on AE activity and the measured crack densities of the failed specimens, the HN-C-ENH composites had twice as many cracks as the HN-C-STD composites. Therefore, it was not just the presence of matrix cracks that caused the significant stress dependent attenuation. The degree of crack opening was assumed to have a major role in the stress dependent attenuation of the acoustic signal. Assuming only frictional forces at the interface, the crack opening displacement, u, was estimated by Eq. (6) [21]. u Z 2s2R 4tf 2Ef 1 C fEf ð1KfÞEm n o h i (6) note that Ef is the fiber modulus. Based on the t values derived from the above analysis of the hysteresis loops, the crack opening for HN-C-STD was an order of magnitude greater than the crack opening for HN-C-ENH at a given stress. Since, significant stress dependent damping occurred between 69 and 100 MPa for the HN-C-STD material and no real increase in stress dependent damping occurred for the HN-C-ENH material (see Fig. 4 and note growing separation between zero stress and at-load AU measurements), it was concluded both a certain level of matrix crack density and crack opening displacement (i.e. large u resulting from a low interfacial shear stress, t) was required for significant increases in attenuation as a function of increasing stress. The threshold for crack opening displacements that induced stress dependence was also assumed to be a function of the transducer’s central frequency. AU stress dependence could probably be defined for the HN-C-ENH composite if higher frequency transducers (shorter ultrasonic wavelengths) were utilized, since the shorter wavelengths would be more sensitive (i.e. attenuate more) in regards to the smaller crack openings. 5. Conclusions Trends for three AU parameters were assessed in respect to the damage states of two types of SiC/SiC ceramic matrix composites (CMC) subjected to load/unload/reload tensile tests. The two composite systems both had Hi-Nicalone fibers with a carbon interface but had different matrix compositions that led to differences in damage accumulation. The majority of the damage for each of the composite systems was in the form of distributed transverse cracks. The standard matrix composite, labeled HN-C-STD, had a lower transverse crack density than the enhanced matrix composite, HN-C-ENH (w2.1 versus w4.6 cracks/mm, respectively). After analyzing the AU parameters, M0 (defined as the mean square value of the power spectral density) was shown to be the most consistent at monitoring the accumulated damage. The energy of the captured ultrasonic wave, as quantified by M0, was reduced by over 50% for each of the material systems when comparing the undamaged and damaged states. In addition, AU measurements related to the HN-C-STD composite showed a dependence on the in situ tensile stress. Given a constant damage state, there were variations between AU measurements taken at zero stress and AU measurements taken at maximum stress. This stress dependence of the AU parameters was not evident in the HN-C-ENH composites. Based on the analysis of the hysteresis loops, the crack opening for HN-C-STD was an order of magnitude greater than the crack opening for HN-C-ENH at a given stress. It was concluded that excessive crack opening displacements in the HN-C-STD caused the observed stress dependence of the AU parameter. As stated earlier, this study focused on defining the trends of the AU parameters as a function of damage and stress for two types of composites. Although statistical confidence needs to be developed by further testing, it is encouraging to know that similar results concerning stress dependent attenuation in these composites were observed in Ref. [17]. In Ref. [17] measurements were made utilizing modal acoustic emissions data. Future research involves further testing under various loading conditions in order to build confidence in the use of AU as an in situ health monitoring tool for advanced ceramic matrix composites. More specifically, an in-depth understanding of the frequency content of the ultrasonic signal may allow for this NDE technique to characterize 52 A.L. Gyekenyesi et al. / Composites: Part B 37 (2006) 47–53
A L Gyekenryesi et al /Composites: Part B 37 (2006)47-53 a composite's behavior conceming crack opening displace- [9] Brewer D. HSR/EPM combustor materials development program. ments. Furthermore, the manual, point-by-point approach of AU is being enhanced by the development of a guided wave o1 Kautz HE. Determination of plate wave velocities and diffuse field scanning system at NASA Glenn Research Center [22] Lastly, AU parameters can be used by analytical models (111 Lott LA. Kunerth DC. nDE of fiber-matrix interface bonds and aimed at defining the current damage state(diagnostics)and materal damage in ceramic/ceramic ites In: Conference predicting remaining life(prognostics) nondestructive evaluation of modern ceramics, Columbus, OH [12] Telreja R, Govada A, Henneke EG. Quantitative assessment of Transonic Appendix A Appendix measurements. In: Telreja R, Govada A, Henneke EG, editors. In: Review of progress in quantitative nondestructive evaluation P. 3 Typical values and units for the individual AU [13] Gyekenyesi AL. Kautz HE, Cao W. Damage assessment of creep parameters as represented by the undamaged HN-C-STD tested and thermally aged Udimet 520 using acousto-ultrasonics omposite were as follows: Mo(defined as the mean NASA CR-2001-2109882001 value of the power spectral density)=889(Vrms)Hz; [14] Gyekenyesi AL, Harmon L. The effect of experimental conditions on centroid of the power spectrum, f=0.634 MHz; and the acousto-ultrasonic reproducibility. In: Gyekenyesi AL, Harmon L, ditors. In: NDE and health monitoring of aerospace materials and diffuse field decay rate=0.0165 w/s. civil infrastructures. San Diego, CA: SPIE: 2002. [15] Morscher GN Modal acoustic emission of damage accumulation in a SiC/SiC composite Compos Sci Technol 1999: 59: 419-26. [16] Morscher GN. Modal acoustic emission source determination References silicon carbide matrix composites. In: Thompson DO, Chimenti DE, editors. Review of pro [1] Vary A, Bowles KJ. An ultrasonic-acoustic technique for voL 19. Kluwer Academic/Plenum Publishers: 2000. destructive evaluation of fiber composite quality. Polym Eng [17] Morscher GN, Gyekenyesi AL. The velocity and attenuation of acoustic emission waves in SiC/SiC composites loaded in tension Compos Sci Technol 2002: 62: 1171-80 [2] Vary A. Acousto-ultrasonic characterization of fiber reinforced [18) Steen M, Valles JL. Unloading-reloading sequences and the analysis composites. Mater Eval 1982: 40 [3] Vary A. Acousto-ultrasonics. In: Sumn mechanical test results for continuous fiber ce destructive testing of fiber reinforced plastic composites, voL. 2. Jenkins MG et al. editor. Thermal and mechanical test methods and ondon: Elsevier; 1990. p. 1-54. behavior of continuous-fiber ceramic composites. West Con [4] Tiwari A Real time acousto-ultrasonic NDE technique for monitoring ocken, PA: ASTM STP 1309: 1997. P 49-65 damage in ceramic composites under dynmaic loads NASA CR [19] Domergue JM. Heredia FE, Evans AG. Hysteresis loops and the 1983741995 inelastic deformation of 0/90 ceramic matrix composites. J Am Cera 15] Gyekenyesi AL. Isothermal fatigue, damage accumulation, and life Soc1996:79:161-70 prediction of a woven PMC. NASA CR-1998-206593 1998 [20] Hutchnison Jw, Jensen HM. Models of fiber debonding and pullout in [ 6] Weaver RL. Diffuse field decay rates for material characterization. In brittle composites with friction. Mech Mater 1990: 9: 139-63. Achenbach JD, Rajapaskie Y, editors. Solid mechanics research for [211 Marshall DB, Cox BN, Evans AG. The mechanics of matrix crackin quantitative nondestructive evaluation. in brittle-matrix fiber composites. Acta Meta 1985: 33(11): 2013-21 7 an MR, Ziola SM. Plate waves produced by transverse matrix [22] Roth DJ, Vemilli M, Cosgriff LM, Martin RE, Bhatt RT. cracking. Ultrasonics 1991: 29: 245-51 Microstructural and discontinuity characterization in ceramic compo- [8] Evans AG, Zok FW. The physics and mechanics of fibre-reinforced sites using an ultrasonic guided wave scan system. Mater Eval 2004; brittle matrix composites. J Mater Sci 1994: 29: 3857-96 62(9)
a composite’s behavior concerning crack opening displacements. Furthermore, the manual, point-by-point approach of AU is being enhanced by the development of a guided wave scanning system at NASA Glenn Research Center [22]. Lastly, AU parameters can be used by analytical models aimed at defining the current damage state (diagnostics) and predicting remaining life (prognostics). Appendix A. Appendix Typical values and units for the individual AU parameters as represented by the undamaged HN-C-STD composite were as follows: M0 (defined as the mean square value of the power spectral density)Z889 (Vrms) 2 Hz; centroid of the power spectrum, fcZ0.634 MHz; and the diffuse field decay rateZ0.0165 m/s. References [1] Vary A, Bowles KJ. An ultrasonic-acoustic technique for nondestructive evaluation of fiber composite quality. Polym Eng Sci 1979;19. [2] Vary A. Acousto-ultrasonic characterization of fiber reinforced composites. Mater Eval 1982;40. [3] Vary A. Acousto-ultrasonics. In: Summerscales J, editor. Nondestructive testing of fiber reinforced plastic composites, vol. 2. London: Elsevier; 1990. p. 1–54. [4] Tiwari A. Real time acousto-ultrasonic NDE technique for monitoring damage in ceramic composites under dynmaic loads NASA CR 198374 1995. [5] Gyekenyesi AL. Isothermal fatigue, damage accumulation, and life prediction of a woven PMC. NASA CR-1998-206593 1998. [6] Weaver RL. Diffuse field decay rates for material characterization. In: Achenbach JD, Rajapaskie Y, editors. Solid mechanics research for quantitative nondestructive evaluation. [7] Gorman MR, Ziola SM. Plate waves produced by transverse matrix cracking. Ultrasonics 1991;29:245–51. [8] Evans AG, Zok FW. The physics and mechanics of fibre-reinforced brittle matrix composites. J Mater Sci 1994;29:3857–96. [9] Brewer D. HSR/EPM combustor materials development program. Mater Sci Eng 1999;261:284–91. [10] Kautz HE. Determination of plate wave velocities and diffuse field decay rates with broadband acousto-utrasonics In: Second international conference on acousto-ultrasonics, ASNT, Atlanta, GA 1993. [11] Lott LA, Kunerth DC. NDE of fiber-matrix interface bonds and material damage in ceramic/ceramic composites In: Conference on nondestructive evaluation of modern ceramics, Columbus, OH 1990. [12] Telreja R, Govada A, Henneke EG. Quantitative assessment of damage growth in graphite epoxy laminates by acousto-ultrasonic measurements. In: Telreja R, Govada A, Henneke EG, editors. In: Review of progress in quantitative nondestructive evaluation p. 3. [13] Gyekenyesi AL, Kautz HE, Cao W. Damage assessment of creep tested and thermally aged Udimet 520 using acousto-ultrasonics NASA CR-2001-210988 2001. [14] Gyekenyesi AL, Harmon L. The effect of experimental conditions on acousto-ultrasonic reproducibility. In: Gyekenyesi AL, Harmon L, editors. In: NDE and health monitoring of aerospace materials and civil infrastructures. San Diego, CA: SPIE; 2002. [15] Morscher GN. Modal acoustic emission of damage accumulation in a woven SiC/SiC composite. Compos Sci Technol 1999;59:419–26. [16] Morscher GN. Modal acoustic emission source determination in silicon carbide matrix composites. In: Thompson DO, Chimenti DE, editors. Review of progress in quantitative nondestructive evaluation, vol. 19. Kluwer Academic/Plenum Publishers; 2000. [17] Morscher GN, Gyekenyesi AL. The velocity and attenuation of acoustic emission waves in SiC/SiC composites loaded in tension. Compos Sci Technol 2002;62:1171–80. [18] Steen M, Valles JL. Unloading-reloading sequences and the analysis of mechanical test results for continuous fiber ceramic composites. In: Jenkins MG et al, editor. Thermal and mechanical test methods and behavior of continuous-fiber ceramic composites. West Conshohocken, PA: ASTM STP 1309; 1997. p. 49–65. [19] Domergue JM, Heredia FE, Evans AG. Hysteresis loops and the inelastic deformation of 0/90 ceramic matrix composites. J Am Ceram Soc 1996;79:161–70. [20] Hutchnison JW, Jensen HM. Models of fiber debonding and pullout in brittle composites with friction. Mech Mater 1990;9:139–63. [21] Marshall DB, Cox BN, Evans AG. The mechanics of matrix cracking in brittle-matrix fiber composites. Acta Meta 1985;33(11):2013–21. [22] Roth DJ, Verrilli MJ, Cosgriff LM, Martin RE, Bhatt RT. Microstructural and discontinuity characterization in ceramic composites using an ultrasonic guided wave scan system. Mater Eval 2004; 62(9). A.L. Gyekenyesi et al. / Composites: Part B 37 (2006) 47–53 53