《复合材料 Composites》课程教学资源（学习资料）第五章陶瓷基复合材料_j.1151-2916.2000.tb01457.x[1]

团购合买资源类别：文库，文档格式：PDF，文档页数：12，文件大小：1.38MB

J Am Cera Sac, 83 [7 1727-38(2000) urna Performance of Four Ceramic-Matrix Composite Divergent Flap Inserts Following Ground Testing on an F110 Turbofan Engine James M. stahl Air Force Research Laboratory, Metals, Ceramics, and NDE Division, Ceramics Development and Materials Behavior Branch, Wright-Patterson Air Force Base, Ohio 45433-7817 Four ceramic-matrix composite flap inserts were evaluated following ground testing on a General Electric F110 turbofan engine. Three of the composites accumulated -117 h of engine time. The fourth composite, a Nextel M 720 material with aluminosilicate matrix, accumulated 40 h. Large through- thickness cracks develop ed along the longitudinal edges of Nicalon /AL,O3 insert and the Nextel 720/aluminosilicate insert. The cracks developed because of high tensile stresses caused by the steep in-plane thermal gradients induced across he flap width during afterburner lights. The Nextel 720/ aluminosilicate insert also exhibited severe surface wear asso- ciated with the acoustic environment and contact with the adjacent divergent seals. Neither a Nicalon/silicon nitrocarbide insert nor a Nicalon/C insert exhibited significant signs of L. Introduction Exhaust Gases O R the past 5 to 10 years, ceramic-matrix composites(CMCs) e undergone testing in a number of military turbine engine applications. Some of the most extensive in-flight experience with CMCs has come from U.S. Navy efforts with Nicalon/C divergent flap and seal inserts for the afterburner(AB)of the General Electric F414 turbofan engine. Two such engines power the F/A-18E/F Super Hornet , In a recent Defense Advanced Navy program, four CMCs were considered for ich parallels the Seal flap flowpath elements on the ABs of General Electric F110 turbofan engines. These engines power the F16 fighter used by the U.S. Air Force and numerous other countries The AB for the F110 is comprised of a forward augmente section and a trailing variable exhaust nozzle The nozzle includes 12 divergent flaps and an equal number of divergent seals arranged an axisymmetric design which actuates in unison to change the ize of the exhaust opening. Depending on the size of the the width of the central region of the flap exposed at ar time to the hot flowpath gases can vary. When the nozzle open, the majority of the flap's surface will be exposed to hot exhaust gases as shown pictorially in Fig. 1(a). Conversely, when the nozzle is closed, as in Fig. 1(b), a comparatively narrow central strip will be When the AB is lit, a large amount of raw fuel is injected directly into the augmenter section with the combustion products expelled through the nozzle. The AB is necessary for certain flight and Exposition of the American Ceramic Society, Cocoa Beach, FL, January 15, 1997 1. Schematics of the nozzle aft looking forward, showing the overlapping of the flaps and seals in the(a) fully opened and(b) fully Systran Federal Corporation, Dayton, Ohio 45432-3068

Performance of Four Ceramic-Matrix Composite Divergent Flap Inserts Following Ground Testing on an F110 Turbofan Engine James M. Staehler* ,† and Larry P. Zawada* Air Force Research Laboratory, Metals, Ceramics, and NDE Division, Ceramics Development and Materials Behavior Branch, Wright-Patterson Air Force Base, Ohio 45433–7817 Four ceramic-matrix composite flap inserts were evaluated following ground testing on a General Electric F110 turbofan engine. Three of the composites accumulated ;117 h of engine time. The fourth composite, a NextelTM 720 material with aluminosilicate matrix, accumulated ;40 h. Large throughthickness cracks developed along the longitudinal edges of a NicalonTM/Al2O3 insert and the Nextel 720/aluminosilicate insert. The cracks developed because of high tensile stresses caused by the steep in-plane thermal gradients induced across the flap width during afterburner lights. The Nextel 720/ aluminosilicate insert also exhibited severe surface wear associated with the acoustic environment and contact with the adjacent divergent seals. Neither a Nicalon/silicon nitrocarbide insert nor a Nicalon/C insert exhibited significant signs of distress. I. Introduction OVER the past 5 to 10 years, ceramic-matrix composites (CMCs) have undergone testing in a number of military turbine engine applications.1–3 Some of the most extensive in-flight experience with CMCs has come from U.S. Navy efforts with NicalonTM/C divergent flap and seal inserts for the afterburner (AB) of the General Electric F414 turbofan engine. Two such engines power the F/A-18E/F Super Hornet.4,5 In a recent Defense Advanced Research Project Agency (DARPA) initiative which parallels the Navy program, four CMCs were considered for use as divergent flap flowpath elements on the ABs of General Electric F110 turbofan engines. These engines power the F16 fighter used by the U.S. Air Force and numerous other countries. The AB for the F110 is comprised of a forward augmenter section and a trailing variable exhaust nozzle. The nozzle includes 12 divergent flaps and an equal number of divergent seals arranged in an axisymmetric design which actuates in unison to change the size of the exhaust opening. Depending on the size of the opening, the width of the central region of the flap exposed at any given time to the hot flowpath gases can vary. When the nozzle is fully open, the majority of the flap’s surface will be exposed to hot exhaust gases as shown pictorially in Fig. 1(a). Conversely, when the nozzle is closed, as in Fig. 1(b), a comparatively narrow central strip will be exposed. When the AB is lit, a large amount of raw fuel is injected directly into the augmenter section with the combustion products expelled through the nozzle. The AB is necessary for certain flight R. J. Kerans—contributing editor Manuscript No. 189551. Received February 16, 1999; approved December 9, 1999. Portions of this paper were presented at the 21st Annual Cocoa Beach Conference and Exposition of the American Ceramic Society, Cocoa Beach, FL, January 15, 1997 (Engineering Ceramics Division). *Member, American Ceramic Society. † Systran Federal Corporation, Dayton, Ohio 45432–3068. Fig. 1. Schematics of the nozzle, aft looking forward, showing the overlapping of the flaps and seals in the (a) fully opened and (b) fully closed positions. J. Am. Ceram. Soc., 83 [7] 1727–38 (2000) 1727 journal

1728 Journal of the American Ceramic Sociery--Staehler and zawada maneuvers and to achieve supersonic flight. Because of the nature CMCs were considered in conjunction with a modified flap of the Ab, the materials of the nozzle must be able to withstand design wherein removable flowpath elements or inserts were used severe temperatures, rapid heat-ups, through-thickness and in- Each divergent flap consisted of a flat CMC insert which slid into plane thermal gradients, and acoustic loads. The relative motion of a"picture-frame"backbone structure. The new flap assembly was the flaps and seals, combined with the acoustic environment of the designed to be interchangeable with the current metal flap hard- engine, can lead to wear issues. The partial shielding by the seals ware. The intent was to improve the longevity of the flap as well so gives rise to in-plane thermal gradients across the flap inserts as provide for ease of replacement. The CMC flap inserts were which can in turn lead to high tensile stresses along the edges of tapered along their 530 mm length with the leading edge 118 the CMC flap inserts. Under extended AB lights, the maximum mm wide and the trailing edge about 140 mm. The thickness of temperatures of the exposed flap and seal surfaces can exceed 1000°C mm wide band along the longitudinal and leading edges where the The current flaps and seals on the F110 nozzle are fabricated of flowpath side was surface ground to a thickness of 1. 4-1.5 m Rene 41, a nickel-based superalloy. However, the combination of When installed, this step made the flowpath surface of high temperatures and thermal cycling on the nozzle leads to creep approximately flush with the top retaining rail of the picture cracking of the metal parts. At present, roughly 10% of the Rene General Electric on the Navy F414-GE-400 engle s that deformation which in turn causes severe warping and eventual backing structure. The basic design was similar to aps and seals must be removed or repaired after only about a third of their intended design life. As a consequence, they are a on an F110 engine by General Electric Aircraft Engines, Evendale high-maintenance, high-cost item for the Air Force. Figure 2 OH. During ground testing, the engine was run through several lows a Rene 41 flap and seal which were removed from an accelerated mission test(AMT)cycles. The purpose of the AMT ngine. Both warping and surface cracking are clearly evident. The to subject the(ground) engine to the cyclic demands of a fielded tensile response of Rene 41 at 23, 10000, and 1100C was engine but within a condensed period of time. The AMT for the measured by Finch and Zawada. The metal exhibits higher F110-powered F16 fighter is comprised of seven different cycles ultimate tensile strengths below 1000.C. However, the strength intended to simulate maneuvers such as air-to-ground gunnery and advantage that the metal holds over CMCs at low temperatures air combat. Each cycle is comprised of multiple throttle settings disappears above 1000C. The nozzle will see temperatures and specifically defines the duration and level of those throttle between 1000 and 1100 C. The metal shows significantly more settings. Many of these cycles include one or more AB lights with inelastic deformation at all temperatures compared with the CMCs durations ranging from a few seconds to a few minutes depending before failure on the simulation. The throttle settings for the cycles are derived The big advantage of the CMCs compared with Rene 41 from data recorded during F16 field and training exercises. Each of becomes apparent in creep rupture at 1000@ and 1100@C. Through the cycles to which the CMC flap inserts were subjected spanned this temperature range the CMCs are significantly more creep a period of roughly 35 min and included two AB lights. The resistant and can sustain a given static load longer before failure. desired design life for the CMc inserts was 2000 engine flight These are critical temperatures for the F110 flaps and seals. The hours. The "equivalent" AMT time based on the accumulation of clear-cut although two oxide/oxide composites stand out as better h with 5177 AB lights for a total time under AB conditions of 36 than the rest h In the field, 2000 engine flight hours might take 3 to 5 years to acquire. By comparison, a significant amount of engine time can be accumulated in a relatively short period through AMT ground Three of the four CMCs under consideration accumulated -408 AB lights or -117 h of engine time on two different engines These included Nicalon/C. Nicalon/AlO. and Nicalon/silicon nitrocarbide (SINC)composite inserts placed in positions DFI DF3, and DF4, respectively, on the nozzle. Figure I identifies the various flap and seal locations. The fourth CMC flap insert,a Nextel 720 with aluminosilicate(AS)matrix, was added latter and ccumulated -146 AB lights or 40 h of engine time in position DFS. At about the same time that the fourth CMC flap insert was added, four CMC seal inserts made of Nextel 610/AS were placed on the engine in locations DSl through DS4. The three nicalon fiber(Nippon Carbon Co., Tokyo, Japan) containing flap inserts accumulated roughly 10%11% of the intended design life whereas the insert with Nextel 720 fibers(3M Co., St. Paul, MN between 3% and 4% The purpose of this paper is to present the post-engine test (b) evaluations made of the CMC flap inserts. This includes the results from visual examinations. material characterization. and residual strength tension tests. While Nextel 610/AS seal flowpath ele- ments were engine tested, they were not subjected to the same post-test analysis as the flap inserts and, therefore, are not I. Materials The four CMCs considered for the F110 flap of matrixes and processing routes. Three of the Nicalon fibers which consist of submicromete in an ne 41 (a)flap and(b)seal currently amorphous mixture of silicon, carbon, and oxyg fourth F110 engine. lated damage includes surface used Nextel 720 fibers which are a mixture of a-alumina and crumpling, cracking, and the sever which is apparent in the seal mullite. The suppliers of the Nicalon-based CMCs were Dow

maneuvers and to achieve supersonic flight. Because of the nature of the AB, the materials of the nozzle must be able to withstand severe temperatures, rapid heat-ups, through-thickness and inplane thermal gradients, and acoustic loads. The relative motion of the flaps and seals, combined with the acoustic environment of the engine, can lead to wear issues. The partial shielding by the seals also gives rise to in-plane thermal gradients across the flap inserts which can in turn lead to high tensile stresses along the edges of the CMC flap inserts.6 Under extended AB lights, the maximum temperatures of the exposed flap and seal surfaces can exceed 1000°C. The current flaps and seals on the F110 nozzle are fabricated of Rene´ 41, a nickel-based superalloy. However, the combination of high temperatures and thermal cycling on the nozzle leads to creep deformation which in turn causes severe warping and eventual cracking of the metal parts. At present, roughly 10% of the Rene´ flaps and seals must be removed or repaired after only about a third of their intended design life. As a consequence, they are a high-maintenance, high-cost item for the Air Force. Figure 2 shows a Rene´ 41 flap and seal which were removed from an engine. Both warping and surface cracking are clearly evident. The tensile response of Rene´ 41 at 23°, 1000°, and 1100°C was measured by Finch and Zawada.7 The metal exhibits higher ultimate tensile strengths below 1000°C. However, the strength advantage that the metal holds over CMCs at low temperatures disappears above 1000°C. The nozzle will see temperatures between 1000° and 1100°C. The metal shows significantly more inelastic deformation at all temperatures compared with the CMCs before failure. The big advantage of the CMCs compared with Rene´ 41 becomes apparent in creep rupture at 1000° and 1100°C.8 Through this temperature range the CMCs are significantly more creep resistant and can sustain a given static load longer before failure. These are critical temperatures for the F110 flaps and seals. The fatigue performance of the CMCs and Rene´ 41 at 1000°C is less clear-cut although two oxide/oxide composites stand out as better than the rest. CMCs were considered in conjunction with a modified flap design wherein removable flowpath elements or inserts were used. Each divergent flap consisted of a flat CMC insert which slid into a “picture-frame” backbone structure. The new flap assembly was designed to be interchangeable with the current metal flap hardware. The intent was to improve the longevity of the flap as well as provide for ease of replacement. The CMC flap inserts were tapered along their 530 mm length with the leading edge ;118 mm wide and the trailing edge about 140 mm. The thickness of each was fairly uniform (nominally 2.00–2.25 mm) except for a 10 mm wide band along the longitudinal and leading edges where the flowpath side was surface ground to a thickness of 1.4–1.5 mm. When installed, this step made the flowpath surface of the insert approximately flush with the top retaining rail of the picture-frame backing structure. The basic design was similar to that used by General Electric on the Navy F414-GE-400 engine.4 The CMC flap inserts described in this paper were ground tested on an F110 engine by General Electric Aircraft Engines, Evendale, OH. During ground testing, the engine was run through several accelerated mission test (AMT) cycles. The purpose of the AMT is to subject the (ground) engine to the cyclic demands of a fielded engine but within a condensed period of time. The AMT for the F110-powered F16 fighter is comprised of seven different cycles intended to simulate maneuvers such as air-to-ground gunnery and air combat. Each cycle is comprised of multiple throttle settings and specifically defines the duration and level of those throttle settings. Many of these cycles include one or more AB lights with durations ranging from a few seconds to a few minutes depending on the simulation. The throttle settings for the cycles are derived from data recorded during F16 field and training exercises. Each of the cycles to which the CMC flap inserts were subjected spanned a period of roughly 35 min and included two AB lights. The desired design life for the CMC inserts was 2000 engine flight hours. The “equivalent” AMT time based on the accumulation of a comparable number of thermal cycles on the engine was ;1050 h with 5177 AB lights for a total time under AB conditions of ;36 h. In the field, 2000 engine flight hours might take 3 to 5 years to acquire. By comparison, a significant amount of engine time can be accumulated in a relatively short period through AMT ground testing. Three of the four CMCs under consideration accumulated ;408 AB lights or ;117 h of engine time on two different engines. These included Nicalon/C, Nicalon/Al2O3, and Nicalon/silicon nitrocarbide (SiNC) composite inserts placed in positions DF1, DF3, and DF4, respectively, on the nozzle. Figure 1 identifies the various flap and seal locations. The fourth CMC flap insert, a Nextel 720 with aluminosilicate (AS) matrix, was added latter and accumulated ;146 AB lights or ;40 h of engine time in position DF5. At about the same time that the fourth CMC flap insert was added, four CMC seal inserts made of Nextel 610/AS were placed on the engine in locations DS1 through DS4. The three Nicalon fiber (Nippon Carbon Co., Tokyo, Japan) containing flap inserts accumulated roughly 10%–11% of the intended design life whereas the insert with Nextel 720 fibers (3M Co., St. Paul, MN) between 3% and 4%. The purpose of this paper is to present the post-engine test evaluations made of the CMC flap inserts. This includes the results from visual examinations, material characterization, and residual strength tension tests. While Nextel 610/AS seal flowpath elements were engine tested, they were not subjected to the same post-test analysis as the flap inserts and, therefore, are not discussed here. II. Materials The four CMCs considered for the F110 flap included a range of matrixes and processing routes. Three of the composites used Nicalon fibers which consist of submicrometer b-SiC in an amorphous mixture of silicon, carbon, and oxygen. The fourth used Nextel 720 fibers which are a mixture of a-alumina and mullite. The suppliers of the Nicalon-based CMCs were Dow Fig. 2. Photographs of a Rene´ 41 divergent (a) flap and (b) seal currently used on the F110 engine. Typical service related damage includes surface crumpling, cracking, and the severe warping which is apparent in the seal. 1728 Journal of the American Ceramic Society—Staehler and Zawada Vol. 83, No. 7

Corning(Nicalon/SINC, trade name Sylramic S-200), Hitco material properties for comparison with the DC insert. This panel Technologies(Nicalon/C, trade name CeracarbMSC537EH), and was made after the DC flap insert had completed ground testing DuPont Lanxide(Nicalon/Al2O3). The Nextel 720 composite had and as such was not true witness material n aluminosilicate matrix and was fabricated by General Electric A second HT Fllo flap insert which never saw engine time was (Nextel 720/AS). Irrespective of the type of fiber, balanced used as"witness"material for this composite. A corner of the eight-harness-satin-weave(&HSW) cloths were used for the lay leading edge of this insert had been damaged during im ps In each CMC, the plies ran the full length and width of the installation on the nozzle and was never engine tested. The insert except where grinding removed surface material from along specimens used for witness purposes were cut well away from the the longitudinal and leading edges to accommodate the picture- damaged region and the test results were not affected by it. frame backing structure. Table I is a brief summary of each DuPont Lanxide made two F110 inserts of the DL compos omposite, to include compositions and processing routes. Some They performed a limited number of tension tests on witness of the details of processing and matrix fillers were omitted for specimens machined from the two panels used to fabricate the actual inserts. In one panel the room-temperature tensile strength The Nicalon 8HSW plies in the Dow Corning(DC)composite was about 154 MPa while in the second only about 97 MPa. Both were warp-aligned with the length of the flap insert. However, were significantly lower than the 200 MPa measured on material each successive ply was rotated about the warp direction relative to evaluated earlier. 4 The lower strengths were associated with the previous one such that in one case the warp-face of the cloth damage to the Nicalon fibers during fabrication which the manu- was up and in the next the fill-face was up In the DL composite facturer believed was a result of inexperience making parts to the the plies were not warp-aligned but instead each successive ply size and specifications of the flap insert. was rotated in-plane such that in one case the warp direction coincided with the length of the insert while in the next the fill direction did. This alternating warp and fill orientation cont Ill. Experimental Procedure throughout the lay-up. Despite the obvious difference in dure, the final appearances of the two lay-up schemes General Electric performed infrared (IR)ten indistinguishable ments of the CMc inserts in conjunction with the ground testing The Hitco Technologies (HT) composite, Ceracarb The IR camera was positioned in the test cell near the nozzle and SC537EH, has a fairly compliant, inhibited carbon matrix. The captured the Ir signals looking through the engine exhaust plume microstructure of this material was examined in detail by bourrat Only two of the CMC inserts the camera's field of view at et al via TEM. This composite does not employ an engineered any given time(positions DF4 and df5 in Fig. 1). This necessi- interphase per se but does develop to some extent a pyrolytic tated the repositioning of the CMC inserts during an engine down carbon fiber coating during processing. The mechanical properties time to obtain the temperature information from the remaining two of this and similar composites have been studied in detail else materials. During this repositioning the D insert was broken into where- The matrix itself contains a large number of two pieces. To keep the time on each of the Nicalon-containing rocessing-related shrinkage cracks, some of which are later filled inserts the same, all three were removed from the engine. Despite in with pyrolytic carbon via chemical vapor infiltration. While its limited engine exposure, the Nextel 720/AS insert was showing inhibitors help to suppress the oxidation of the carbon matrix, the wear and for this reason was removed as well. The IR measure- composite utilizes an overcoat for additional protection. This ments pleted using new DL and HT flap inserts in coating was a dual-layer system consisting of a CVD applied Sic positions DF4 and DF5. For practical reasons related to the environmental barrier coating( Chromalloy RT42, 100-150 um positioning of the IR camera in the test cell, only the rear thick) followed by a spray-applied seal coat( Hitco Technologies two-thirds of the CMC inserts were visible M185 glaze coating, 75-150 Hm thick). The individual plies for Initial examinations of the engine-tested Cmc inserts included the HT lay-up were alternately warp and fill oriented with the visual and low-magnification optical microscope inspections. In length of the flap insert similar to the DL composite addition, each insert was analyzed via ultrasonic C-scans. Because The General Electric(GE)composite does not use an engi- of the potential susceptibility to moisture of some of the eered fiber-matrix interphase and has a relatively strong fiber- each insert was enclosed in a sealed vacuum bag for th matrix interface. Toughness is achieved through crack deflection of the scan. The scans verified that no delaminations were within its highly porous, relatively weak matrix. This porosity in any of the inserts consists of numerous shrinkage cracks from processing and exten- Following the nondestructive evaluations(NDEs), each flap sive microporosity which is evenly dispersed throughout the insert was machined into multiple straight-sided tensile specimens matrix. The lay-up scheme used for the ge composite was similar oriented as shown in Figs. 3(a-c). The specimens were nominally to that of the dc material 10 mm wide by 90 mm in length. Specimen thickness was left as-is To supplement data collected on the DC flap insert from the parent insert Fiberglass tabs 1.6 mm thick X 10 mm X run on the engine, ning provided a 154 mm square panel 30 mm with tapered forward edges were affixed to the ends of each with the same fib composition, and processing route. specimen with epoxy for gripping during testing Specimen gauge were used to collect as-processed sections in all cases were nominally 30 mm in length. Figures 3(b) Table L. Summary of Composite Compositions and Processing Fiber-matrix interphase Matrix Processing technique Overcoat(yes/no Nicalon &HSW 0.5 um bn by CVI SiNC fillers Polymer infiltration No Midland. MI Nicalon/ cloth SING Hitco HT Nicalon/C Nicalon 8HS W None applied Inhibited carbon Pyrolysis followed Yes(see text) by CvI of carbon Gardena, CA DuPont Lanxide. DL Nicalon 8HSW-0.5 um BN, 3-5 um Alumina 240 grit Direct metal Newark DE Nicalon/ Sic filler oxidation General Electric. GE Nexto :I 720 None applied 7 wt%Al,,O3, 13 Pyrolysis Evendale. OH 720/AS 8HSW cloth

Corning (Nicalon/SiNC, trade name SylramicTM S-200), Hitco Technologies (Nicalon/C, trade name CeracarbTM SC537EH), and DuPont Lanxide (Nicalon/Al2O3). The Nextel 720 composite had an aluminosilicate matrix and was fabricated by General Electric (Nextel 720/AS). Irrespective of the type of fiber, balanced eight-harness-satin-weave (8HSW) cloths were used for the layups. In each CMC, the plies ran the full length and width of the insert except where grinding removed surface material from along the longitudinal and leading edges to accommodate the pictureframe backing structure. Table I is a brief summary of each composite, to include compositions and processing routes. Some of the details of processing and matrix fillers were omitted for proprietary reasons. The Nicalon 8HSW plies in the Dow Corning (DC) composite were warp-aligned with the length of the flap insert. However, each successive ply was rotated about the warp direction relative to the previous one such that in one case the warp-face of the cloth was up and in the next the fill-face was up. In the DL composite the plies were not warp-aligned but instead each successive ply was rotated in-plane such that in one case the warp direction coincided with the length of the insert while in the next the fill direction did. This alternating warp and fill orientation continued throughout the lay-up. Despite the obvious difference in procedure, the final appearances of the two lay-up schemes were indistinguishable. The Hitco Technologies (HT) composite, CeracarbTM SC537EH, has a fairly compliant, inhibited carbon matrix. The microstructure of this material was examined in detail by Bourrat et al.9 via TEM. This composite does not employ an engineered interphase per se but does develop to some extent a pyrolytic carbon fiber coating during processing. The mechanical properties of this and similar composites have been studied in detail elsewhere.9–13 The matrix itself contains a large number of processing-related shrinkage cracks, some of which are later filled in with pyrolytic carbon via chemical vapor infiltration. While inhibitors help to suppress the oxidation of the carbon matrix, the composite utilizes an overcoat for additional protection. This coating was a dual-layer system consisting of a CVD applied SiC environmental barrier coating (Chromalloy RT42, 100–150 mm thick) followed by a spray-applied seal coat (Hitco Technologies M185 glaze coating, 75–150 mm thick). The individual plies for the HT lay-up were alternately warp and fill oriented with the length of the flap insert similar to the DL composite. The General Electric (GE) composite does not use an engineered fiber–matrix interphase and has a relatively strong fiber– matrix interface. Toughness is achieved through crack deflection within its highly porous, relatively weak matrix. This porosity consists of numerous shrinkage cracks from processing and extensive microporosity which is evenly dispersed throughout the matrix. The lay-up scheme used for the GE composite was similar to that of the DC material. To supplement the tensile data collected on the DC flap insert run on the engine, Dow Corning provided a 154 mm square panel with the same fiber lay-up, composition, and processing route. Specimens from this panel were used to collect as-processed material properties for comparison with the DC insert. This panel was made after the DC flap insert had completed ground testing and as such was not true witness material. A second HT F110 flap insert which never saw engine time was used as “witness” material for this composite. A corner of the leading edge of this insert had been damaged during improper installation on the nozzle and was never engine tested. The specimens used for witness purposes were cut well away from the damaged region and the test results were not affected by it. DuPont Lanxide made two F110 inserts of the DL composite. They performed a limited number of tension tests on witness specimens machined from the two panels used to fabricate the actual inserts. In one panel the room-temperature tensile strength was about 154 MPa while in the second only about 97 MPa. Both were significantly lower than the 200 MPa measured on material evaluated earlier.14 The lower strengths were associated with damage to the Nicalon fibers during fabrication which the manufacturer believed was a result of inexperience making parts to the size and specifications of the flap insert. III. Experimental Procedure General Electric performed infrared (IR) temperature measurements of the CMC inserts in conjunction with the ground testing. The IR camera was positioned in the test cell near the nozzle and captured the IR signals looking through the engine exhaust plume. Only two of the CMC inserts were in the camera’s field of view at any given time (positions DF4 and DF5 in Fig. 1). This necessitated the repositioning of the CMC inserts during an engine down time to obtain the temperature information from the remaining two materials. During this repositioning the DL insert was broken into two pieces. To keep the time on each of the Nicalon-containing inserts the same, all three were removed from the engine. Despite its limited engine exposure, the Nextel 720/AS insert was showing wear and for this reason was removed as well. The IR measurements were completed using new DL and HT flap inserts in positions DF4 and DF5. For practical reasons related to the positioning of the IR camera in the test cell, only the rear two-thirds of the CMC inserts were visible. Initial examinations of the engine-tested CMC inserts included visual and low-magnification optical microscope inspections. In addition, each insert was analyzed via ultrasonic C-scans. Because of the potential susceptibility to moisture of some of the CMCs, each insert was enclosed in a sealed vacuum bag for the duration of the scan. The scans verified that no delaminations were present in any of the inserts. Following the nondestructive evaluations (NDEs), each flap insert was machined into multiple straight-sided tensile specimens oriented as shown in Figs. 3(a–c). The specimens were nominally 10 mm wide by 90 mm in length. Specimen thickness was left as-is from the parent insert. Fiberglass tabs 1.6 mm thick 3 10 mm 3 30 mm with tapered forward edges were affixed to the ends of each specimen with epoxy for gripping during testing. Specimen gauge sections in all cases were nominally 30 mm in length. Figures 3(b) Table I. Summary of Composite Compositions and Processing Supplier Composite Fiber Fiber–matrix interphase Matrix Processing technique Overcoat (yes/no) Dow Corning, Midland, MI DC Nicalon/ SiNC Nicalon 8HSW cloth ;0.5 mm BN by CVI SiNC 1 fillers Polymer infiltration and pyrolysis No Hitco Technologies, Gardena, CA HT Nicalon/C Nicalon 8HSW cloth None applied Inhibited carbon Pyrolysis followed by CVI of carbon Yes (see text) DuPont Lanxide, Newark, DE DL Nicalon/ Al2O3 Nicalon 8HSW cloth ;0.5 mm BN, 3–5 mm SiC, by CVI Alumina 1 240 grit SiC filler Direct metal oxidation No General Electric, Evendale, OH GE Nextel 720/AS Nextel 720 8HSW cloth None applied 87 wt% Al2O3, 13 wt% SiO2 Pyrolysis No July 2000 Performance of Four Ceramic-Matrix Composite Divergent Flap Inserts 1729

Journal of the American Ceramic Sociery--Staehler and zawada ol.83.No.7 Fig 3. Schematics showing the lay-out of straight-sided specimens cut from each of the inserts: (a)the DC and HT inserts, (b)the DL insert, and(c)the GE insert and(c) include the locations of edge macrocracks observed in the load, strain, and actuator displacement data were recorded digitally GE and DL composites as a result of engine testing. Two axially using a sampling rate of 20 Hz. Specimen thickness was normally oriented tensile specimens from both of these inserts did not the average of three digital caliper measurements taken along the survive the machining process due to known cracks. length within the gauge section, one at the center and two 8-10 Straight-sided tensile specimens were also cut from the 154 mmmm each side of center. Because of significant wear to the ge square as-processed panel provided by Dow Corning as well as the insert, many of those specimens did not have a uniform cross non-engine-tested HT witness insert. In the former case a total of section. In such cases, calipers were used to measure as accuratel seven warp-aligned and seven fill-aligned specimens were pre- possible the largest and smallest thickness dimensions at each of pared. Eighteen axially oriented (two groups of nine running the three locations along the length. The high and low values at across the width) and ten transversely oriented specimens were each location were averaged and the lowest average used prepared from the HT witness insert. conjunction with the uniformly machined specimen width to All tension tests were performed using an MTS servo-hydraulic calculate a"corrected"cross-sectional area. This corrected area ad frame. Specimens were face-loaded using MTS wedge grips was used in the derivation of all subsequent stresses. with serrated wedge inserts. A clip gauge (MTS Model 632. 26 Excess material from each flap insert following machining was B-30)with a 7.56 mm gauge length was used to measure strains. used for material characterization. This included density, porosity, All testing was conducted at room temperature in ambient air and fiber volume fraction measurements. Skeletal densities under stroke control using a displacement rate of 0.05 mm/s The measured with a helium pycnometer(Micromeritics AccuPyo Table Il. Material Physical Property Summary DC Nicalon/SiNC HT Nicalon/C DL Nicalon/Al O3 GE Nextel 720/AS Skeletal density(g/cm) 2.28 32①212

and (c) include the locations of edge macrocracks observed in the GE and DL composites as a result of engine testing. Two axially oriented tensile specimens from both of these inserts did not survive the machining process due to known cracks. Straight-sided tensile specimens were also cut from the 154 mm square as-processed panel provided by Dow Corning as well as the non-engine-tested HT witness insert. In the former case a total of seven warp-aligned and seven fill-aligned specimens were prepared. Eighteen axially oriented (two groups of nine running across the width) and ten transversely oriented specimens were prepared from the HT witness insert. All tension tests were performed using an MTS servo-hydraulic load frame. Specimens were face-loaded using MTS wedge grips with serrated wedge inserts. A clip gauge (MTS Model 632.26 B-30) with a 7.56 mm gauge length was used to measure strains. All testing was conducted at room temperature in ambient air under stroke control using a displacement rate of 0.05 mm/s. The load, strain, and actuator displacement data were recorded digitally using a sampling rate of 20 Hz. Specimen thickness was normally the average of three digital caliper measurements taken along the length within the gauge section, one at the center and two 8–10 mm each side of center. Because of significant wear to the GE insert, many of those specimens did not have a uniform cross section. In such cases, calipers were used to measure as accurately as possible the largest and smallest thickness dimensions at each of the three locations along the length. The high and low values at each location were averaged and the lowest average used in conjunction with the uniformly machined specimen width to calculate a “corrected” cross-sectional area. This corrected area was used in the derivation of all subsequent stresses. Excess material from each flap insert following machining was used for material characterization. This included density, porosity, and fiber volume fraction measurements. Skeletal densities were measured with a helium pycnometer (Micromeritics AccuPyc Table II. Material Physical Property Summary DC Nicalon/SiNC HT Nicalon/C DL Nicalon/Al2O3 GE Nextel 720/AS Skeletal density (g/cm3 ) 2.38 2.28 3.08 3.26 Vf (%) 38.3 38.2 (43.6)† 37.3 37.1 † Fiber volume fraction excluding volume associated with composite overcoat. Fig. 3. Schematics showing the lay-out of straight-sided specimens cut from each of the inserts: (a) the DC and HT inserts, (b) the DL insert, and (c) the GE insert. 1730 Journal of the American Ceramic Society—Staehler and Zawada Vol. 83, No. 7

ol.83.No.7 part to the heat-sink effect of the center longitudinal rib of the (a) metal picture frame which holds the flap insert. The location of Width Crack three of the picture frame's lateral ribs can also be identified in the IR image for the same reason End Plate abrasions In computer simulations performed by General Electric, flap surface temperatures were assumed to be symmetric about the center line. This was roughly the case with the dC profile in Fig 4. However, the IR image from the ge insert was not symmetric because of temperature irregularities in the nozzle, a common engine phenomena called"streaking. "Of the two flap positions used for the IR data collection, DF5 was more susceptible to streaking than DF4. The former position was occupied, in turn, by the ge and dl composites while the dc and ht inserts occupied the latter. Despite this issue, peak temperatures rece the two flap positions were reasonably comparabl rded between (b) Figures 5(a-d) show representative temperature profiles across Seal Rub wear Zones Most Severe Wear the midsection of each insert just before the end of an AB light. These profiles were a compilation of the ir data recorded from the exposed portions of the inserts and estimates of the surface O@题 temperatures near the edges hidden from the IR camera by the djacent seals. The IR images used were recorded about 25-30 s into an AB light. Based on General Electric simulations using for the ge insert, this should be about the time the worst thermal gradients arise in the flaps. Significantly higher peak temperatures, but with less severe thermal gradients, are achieved Aft-end errosion Edge Cracks 50 mm for longer times at full AB. The temperatures near the flap insert dges were estimated based on the anticipated worst-case thermal Fig. 6. Schematics of the macrocracks and wear patterns to the (a) DL gradients. The recorded peak temperatures were reasonably close and (b)ge inserts. computer simulations of the F110 nozzle following roughly the significant wear as a result of engine testing. This wear is included same amount of time under full AB in the sketch of Fig. 6(b). In flap position DF5, the ge insert was Figures 5(a-d)also include temperature profiles that occur djacent to a CMC seal element(Nextel 610/AS) located in DS4 across each inserts midsection during intermediate-rated power and a metal seal situated in DS5. Both sides of the ge insert (IRP). IRP is essentially full throttle but without lighting the AB experienced wear, although some of the deepest were on the side As such, temperatures are lower. However, over 95% of an F110s in contact with the metal seal. Over time the metal seals became operational time is spent at temperatures at or below those severely warped. This, combined with the acoustic environment of associated with an IRP throttle setting. Depending on the magni- the exhaust nozzle, created point impacts that accelerated the wear tudes of the temperatures encountered and the CMC systems In the most severely eroded regions, as much as 40%-50% of the involved, this information could be very important in judging the inserts thickness had been worn away. The wear on the side in long-term survivability of each composite. Peak temperatures contact with the CMC seal was generally less severe and more observed at irp varied from insert to insert but were between 600% uniform in material loss along the entire length of contact but still and750°C significant. In most areas the wear was no more than 10%0-15% but The HT and DC inserts, following visual inspections, exhibited in a few locations, nearly 35% of the thickness had been lost only minor signs of distress. In both cases this included surface The second DL insert used to complete the IR work took the smoothing as a result of contact with the adjacent seals. There were place of the ge insert in position DF5. While it did began to show a few spots along the leading and trailing edges of the hT insert surface rubbing patterns similar to those in the ge composite as where the protective overcoat was chipped. However, these were result of the warped metal seal in DS5, no ply loss occurred. This attributed to rough handling during insertion and removal of the was despite being on the engine for well over twice as long, insert from the picture frame. No indication of oxidation damage h. However, a portion of the 92 h of ground-test time included was evident at these locations. The DC insert did have two small several cycles which skipped the AB lights due to unrelated gouges into the flowpath surface. They may have arisen through augmenter problems. Only minor evidence of surface rubbing was contact with one of the adjacent metal seals which were on the evident on the side of the dl insert in contact with the CMC seal inal ground test engine. CMC seals were installed on either in DS4. Unlike the first DL insert, the second did not show any ide of the dc insert after it had already accumulated -77 h of sual evidence of macrocrack formations along its edges. This engine time was verified by X-ray radiography. The poor retained tensile Through-thickness macroscopic cracks were observed along the properties and presence of cracks in the first DL engine tested longitudinal edges of both the DL and ge inserts following engine insert suggested that it possessed the lower of the two as-processed testing. The sketches in Fig. 6 show the prevalence of these cracks tensile strengths reported by DuPont Lanxide in each part. As noted previously, the DL insert was broken into The ultimate tensile strength (UTS)results from each of the four two pieces. Two engine-induced cracks on opposing edges of the engine tested composites are summarized in Figs. 8 and 9 using a insert were believed to have linked up as a result of handling statistical format. In Figs. 8(a-d) the UTSs from the axial during repositioning on the nozzle to accommodate the IR mea- oriented specimens of each insert were collected into nine groups, surements. The numerous hairline cracks in the DL composite each corresponding to one of the nine rows of specimens in Figs ranged in length from 6 mm to over 40 mm. 3(a-c)running the length of the inserts. The same results appear in The two cracks observed in the ge composite were both about Figs. 9(a-d) but collected into five groups(four for the DL insert) 30 mm in length. Unlike the well-defined hairline cracks in the dl running across the width of each insert. In the latter case. the insert, the cracks in the ge material included a more tortuous path were numbered one through five(four for the DL insert) The micrographs in Figs. 7(a) and (b) contrast the cracks which from the trailing edge of each insert. Figures 8(a- formed in the two inserts. Both cracks in the ge part appeared the results from the transversely oriented specimens along the edge adjacent to the higher temperatures in Fig. 4, ig. 3 as well as those from as-processed or witness suggesting that more severe thermal gradients existed there as a imens from the ge and dl inserts fractured result. The ge insert was the only one of the four to experience during ning due to preexisting macrocracks were assumed to

part to the heat-sink effect of the center longitudinal rib of the metal picture frame which holds the flap insert. The location of three of the picture frame’s lateral ribs can also be identified in the IR image for the same reason. In computer simulations performed by General Electric, flap surface temperatures were assumed to be symmetric about the center line. This was roughly the case with the DC profile in Fig. 4. However, the IR image from the GE insert was not symmetric because of temperature irregularities in the nozzle, a common engine phenomena called “streaking.” Of the two flap positions used for the IR data collection, DF5 was more susceptible to streaking than DF4. The former position was occupied, in turn, by the GE and DL composites while the DC and HT inserts occupied the latter. Despite this issue, peak temperatures recorded between the two flap positions were reasonably comparable. Figures 5(a–d) show representative temperature profiles across the midsection of each insert just before the end of an AB light. These profiles were a compilation of the IR data recorded from the exposed portions of the inserts and estimates of the surface temperatures near the edges hidden from the IR camera by the adjacent seals. The IR images used were recorded about 25–30 s into an AB light. Based on General Electric simulations using properties for the GE insert, this should be about the time the worst thermal gradients arise in the flaps. Significantly higher peak temperatures, but with less severe thermal gradients, are achieved for longer times at full AB. The temperatures near the flap insert edges were estimated based on the anticipated worst-case thermal gradients. The recorded peak temperatures were reasonably close to the 800°–850°C anticipated by General Electric based on computer simulations of the F110 nozzle following roughly the same amount of time under full AB.15 Figures 5(a–d) also include temperature profiles that occur across each insert’s midsection during intermediate-rated power (IRP). IRP is essentially full throttle but without lighting the AB. As such, temperatures are lower. However, over 95% of an F110’s operational time is spent at temperatures at or below those associated with an IRP throttle setting. Depending on the magnitudes of the temperatures encountered and the CMC systems involved, this information could be very important in judging the long-term survivability of each composite. Peak temperatures observed at IRP varied from insert to insert but were between 600° and 750°C. The HT and DC inserts, following visual inspections, exhibited only minor signs of distress. In both cases this included surface smoothing as a result of contact with the adjacent seals. There were a few spots along the leading and trailing edges of the HT insert where the protective overcoat was chipped. However, these were attributed to rough handling during insertion and removal of the insert from the picture frame. No indication of oxidation damage was evident at these locations. The DC insert did have two small gouges into the flowpath surface. They may have arisen through contact with one of the adjacent metal seals which were on the original ground test engine. CMC seals were installed on either side of the DC insert after it had already accumulated ;77 h of engine time. Through-thickness macroscopic cracks were observed along the longitudinal edges of both the DL and GE inserts following engine testing. The sketches in Fig. 6 show the prevalence of these cracks in each part. As noted previously, the DL insert was broken into two pieces. Two engine-induced cracks on opposing edges of the insert were believed to have linked up as a result of handling during repositioning on the nozzle to accommodate the IR measurements. The numerous hairline cracks in the DL composite ranged in length from 6 mm to over 40 mm. The two cracks observed in the GE composite were both about 30 mm in length. Unlike the well-defined hairline cracks in the DL insert, the cracks in the GE material included a more tortuous path. The micrographs in Figs. 7(a) and (b) contrast the cracks which formed in the two inserts. Both cracks in the GE part appeared along the edge adjacent to the higher temperatures in Fig. 4, suggesting that more severe thermal gradients existed there as a result. The GE insert was the only one of the four to experience significant wear as a result of engine testing. This wear is included in the sketch of Fig. 6(b). In flap position DF5, the GE insert was adjacent to a CMC seal element (Nextel 610/AS) located in DS4 and a metal seal situated in DS5. Both sides of the GE insert experienced wear, although some of the deepest were on the side in contact with the metal seal. Over time the metal seals became severely warped. This, combined with the acoustic environment of the exhaust nozzle, created point impacts that accelerated the wear. In the most severely eroded regions, as much as 40%–50% of the insert’s thickness had been worn away. The wear on the side in contact with the CMC seal was generally less severe and more uniform in material loss along the entire length of contact but still significant. In most areas the wear was no more than 10%–15% but in a few locations, nearly 35% of the thickness had been lost. The second DL insert used to complete the IR work took the place of the GE insert in position DF5. While it did began to show surface rubbing patterns similar to those in the GE composite as a result of the warped metal seal in DS5, no ply loss occurred. This was despite being on the engine for well over twice as long, ;92 h. However, a portion of the 92 h of ground-test time included several cycles which skipped the AB lights due to unrelated augmenter problems. Only minor evidence of surface rubbing was evident on the side of the DL insert in contact with the CMC seal in DS4. Unlike the first DL insert, the second did not show any visual evidence of macrocrack formations along its edges. This was verified by X-ray radiography. The poor retained tensile properties and presence of cracks in the first DL engine tested insert suggested that it possessed the lower of the two as-processed tensile strengths reported by DuPont Lanxide. The ultimate tensile strength (UTS) results from each of the four engine tested composites are summarized in Figs. 8 and 9 using a statistical format. In Figs. 8(a–d) the UTSs from the axially oriented specimens of each insert were collected into nine groups, each corresponding to one of the nine rows of specimens in Figs. 3(a–c) running the length of the inserts. The same results appear in Figs. 9(a–d) but collected into five groups (four for the DL insert) running across the width of each insert. In the latter case, the groups were numbered one through five (four for the DL insert) starting from the trailing edge of each insert. Figures 8(a–d) also include the results from the transversely oriented specimens depicted in Fig. 3 as well as those from as-processed or witness material. Specimens from the GE and DL inserts which fractured during machining due to preexisting macrocracks were assumed to Fig. 6. Schematics of the macrocracks and wear patterns to the (a) DL and (b) GE inserts. 1732 Journal of the American Ceramic Society—Staehler and Zawada Vol. 83, No. 7

July 2000 Performance of Four Ceramic-Matrix Comp rows 1 and 9 of the GE insert. Recall that one row 9 specimen was testable because of an edge crack. The sample mean is influ- lowpath surface 9 enced more so by an outlier than is the median. Had median values been compared, significant differences would have been less likely. No significant difference was suggested between rows 2 and 8, where in the latter case one specimen had also been D untestable A compilation of the average tensile properties obtained from E each of the four divergent flap inserts following engine testing are included in Table IV. For brevity, the results of all longitudinally Crack oriented specimens were lumped together. The as-processed ten- sile properties for each composit ell. For the ht TImqUmmmmmm sed results obtained fro witness flap insert and extra as-processed panel, respectively, are oresented as a function of orientation axial and transverse in the former case, warp and fill in the latter. The axial direction of the DC insert was reported by the manufacturer to be the warp direction. However, while the transversely oriented (fill direction specimens of the insert were stronger than the axially oriente (warp direction) ones, the reverse was true in the as-processed panel. It seems doubtful the difference could be attributed to the Outboard surface engine test and no definitive explanation was available. There was no indication that the HT insert had been adversely affected by the EDGE The limited witness data provided by DuPont Lanxide for the DL insert demonstrated that it was not up to the standards of the earlier Nicalon/Al2O, material tested under the DARPa progran Crack In contrasting the results between the Dl witness specimens and those cut from the flap insert(see Fig. 8(c)and Table IV),one might conclude that the engine-tested material had been degraded further. Environmental effects. such as oxidation of the bn nterphase and the Nicalon fibers, can take place if matrix cracking occurs and temperatures are sufficiently high. This could cause a drop in UTS and strain to failure. An interesting test would have 20 mm been a comparison between witness material and those specimens from the center region of the insert(rows 4, 5, and 6 plus the 90s) Fig. 7. Distress to the DL and GE inserts following ground testing. Edge the region continuously exposed to high temperatures and the hot exhaust gases, conditions more conducive to oxidation. Unfortu- nately, the limited amount of witness data made it difficult to draw have zero tensile strength and were included in the plots as points any objective conclusions about further losses in strength. The along the abscissa. larger scatter in UTS for the DL specimens cut near the edges of The boxes in Figs. 8 and 9 are bounded top and bottom by the the insert were felt to be more a result of damage caused by upper and lower quartiles, respectively, of the data. The number of in-plane thermal induced stresses rather than elevated specimens within each sample population are shown in parenthe- temperatures and ion. Because the edges of the insert ses below each box. The lines running through the interior of the generally remain cool. there was less chance of environ- boxes are median values whereas the darkened symbols were the mental degradation. More discussion of the thermal gradients and sample means. The ranges in data are depicted by the bars resulting stresses will follow extending above and below each box. Not all of these features may The UTS of the Ge insert did not appear to be degraded relative be apparent in each case because of the small sampling popula- as-processed material. By itself, this could be misleading since tions. An open circle appearing above or below a box represents an the ge insert did experience wear and edge cracks as a result of outlier Outliers were defined as any data point which lies either engine testing. Because of its oxide/oxide makeup, environmental above the upper quartile or below the lower quartile by 1. 5 times effects were not a big concern. However, the loss of material does the interquartile distance. All outliers were included in the deriva- diminish the composite load-carrying capability tion of sample mean and median values Most of the differences in mean utss with location or orien The data used in Figs. 8 and 9 were also analyzed using analysis tation suggested by Figs. 8 and 9, the ANOVA results in Table Ill, of variance(ANOVA) techniques to allow objective comparisons and those of Table Iv were more than likely an artifact of of the results. As part of this analysis, orthogonal contrasts were processing rather than engine testing. Unfortunately, complete used in an effort to identify any location or orientation dependence as-processed flap inserts of each composite were not available to to the UTS results. Table Ill lists the orthogonal contrasts used for be cut up and used for a more complete analysis of tensile property each material along with the results of the ANOVA. In each case, variations with location and orientation a type I error level of 5%(o=0.05) was used for all ANOVA Fracture surfaces of residual strength tensile specimens from applications. Type I error is the probability of incorrectly rejecting each of the four composite inserts were examined by scanning a null hypothesis when the null hypothesis is true. The probability electron microscopy (SEM). For the dC, ht, and GE composites of type I error decreases with decreasing a. Details of the aNova the fracture features were similar to those observed with as- methods used here can be found in Chapter 3 of Ref. 16. While a processed material fractured in tension at room temperature. This number of significant differences were suggested for both the ht included fibrous fracture surfaces indicative of good composite and DC inserts, there did not appear to be any clear trends material behavior and consistent with the excellent retained tensile consistent with the workings of the nozzle and the engine envi- properties of these three CMCs ronment. As such, these differences were felt to be a ref Fracture surfaces of the Dl engine-tested material, on the other as-processed variability. The one significant and showed features which were consistent with the materials ould possibly be attributed to the engine test poor strength. A typical DL fracture surface is shown in Fig. 10. There

have zero tensile strength and were included in the plots as points along the abscissa. The boxes in Figs. 8 and 9 are bounded top and bottom by the upper and lower quartiles, respectively, of the data. The number of specimens within each sample population are shown in parentheses below each box. The lines running through the interior of the boxes are median values whereas the darkened symbols were the sample means. The ranges in data are depicted by the bars extending above and below each box. Not all of these features may be apparent in each case because of the small sampling populations. An open circle appearing above or below a box represents an outlier. Outliers were defined as any data point which lies either above the upper quartile or below the lower quartile by 1.5 times the interquartile distance. All outliers were included in the derivation of sample mean and median values. The data used in Figs. 8 and 9 were also analyzed using analysis of variance (ANOVA) techniques to allow objective comparisons of the results. As part of this analysis, orthogonal contrasts were used in an effort to identify any location or orientation dependence to the UTS results. Table III lists the orthogonal contrasts used for each material along with the results of the ANOVA. In each case, a type I error level of 5% (a 5 0.05) was used for all ANOVA applications. Type I error is the probability of incorrectly rejecting a null hypothesis when the null hypothesis is true. The probability of type I error decreases with decreasing a. Details of the ANOVA methods used here can be found in Chapter 3 of Ref. 16. While a number of significant differences were suggested for both the HT and DC inserts, there did not appear to be any clear trends consistent with the workings of the nozzle and the engine environment. As such, these differences were felt to be a reflection of as-processed variability. The one significant difference which could possibly be attributed to the engine test was that between rows 1 and 9 of the GE insert. Recall that one row 9 specimen was untestable because of an edge crack. The sample mean is influenced more so by an outlier than is the median. Had median values been compared, significant differences would have been less likely. No significant difference was suggested between rows 2 and 8, where in the latter case one specimen had also been untestable. A compilation of the average tensile properties obtained from each of the four divergent flap inserts following engine testing are included in Table IV. For brevity, the results of all longitudinally oriented specimens were lumped together. The as-processed tensile properties for each composite are included as well. For the HT and DC materials, the as-processed results obtained from the witness flap insert and extra as-processed panel, respectively, are presented as a function of orientation; axial and transverse in the former case, warp and fill in the latter. The axial direction of the DC insert was reported by the manufacturer to be the warp direction. However, while the transversely oriented (fill direction) specimens of the insert were stronger than the axially oriented (warp direction) ones, the reverse was true in the as-processed panel. It seems doubtful the difference could be attributed to the engine test and no definitive explanation was available. There was no indication that the HT insert had been adversely affected by the engine exposure. The limited witness data provided by DuPont Lanxide for the DL insert demonstrated that it was not up to the standards of the earlier Nicalon/Al2O3 material tested under the DARPA program. In contrasting the results between the DL witness specimens and those cut from the flap insert (see Fig. 8(c) and Table IV), one might conclude that the engine-tested material had been degraded further. Environmental effects, such as oxidation of the BN interphase and the Nicalon fibers, can take place if matrix cracking occurs and temperatures are sufficiently high. This could cause a drop in UTS and strain to failure. An interesting test would have been a comparison between witness material and those specimens from the center region of the insert (rows 4, 5, and 6 plus the 90’s). The gauge sections of each of these specimens lie more or less in the region continuously exposed to high temperatures and the hot exhaust gases, conditions more conducive to oxidation. Unfortunately, the limited amount of witness data made it difficult to draw any objective conclusions about further losses in strength. The larger scatter in UTS for the DL specimens cut near the edges of the insert were felt to be more a result of damage caused by in-plane thermal gradient induced stresses rather than elevated temperatures and oxidation. Because the edges of the insert generally remain relatively cool, there was less chance of environmental degradation. More discussion of the thermal gradients and resulting stresses will follow. The UTS of the GE insert did not appear to be degraded relative to as-processed material. By itself, this could be misleading since the GE insert did experience wear and edge cracks as a result of engine testing. Because of its oxide/oxide makeup, environmental effects were not a big concern. However, the loss of material does diminish the composite load-carrying capability. Most of the differences in mean UTSs with location or orientation suggested by Figs. 8 and 9, the ANOVA results in Table III, and those of Table IV were more than likely an artifact of processing rather than engine testing. Unfortunately, complete as-processed flap inserts of each composite were not available to be cut up and used for a more complete analysis of tensile property variations with location and orientation. Fracture surfaces of residual strength tensile specimens from each of the four composite inserts were examined by scanning electron microscopy (SEM). For the DC, HT, and GE composites the fracture features were similar to those observed with asprocessed material fractured in tension at room temperature. This included fibrous fracture surfaces indicative of good composite material behavior and consistent with the excellent retained tensile properties of these three CMCs. Fracture surfaces of the DL engine-tested material, on the other hand, showed features which were consistent with the material’s poor strength. A typical DL fracture surface is shown in Fig. 10. There Fig. 7. Distress to the DL and GE inserts following ground testing. Edge cracks in the (a) DL and (b) GE inserts. July 2000 Performance of Four Ceramic-Matrix Composite Divergent Flap Inserts 1733

Journal of the American Ceramic Sociery--Staehler and zawada 出。喜中*面+“ 200 一E 正量是 d GE flap, Nextel 720/AS 中 (4)(441(10)(2) Fig. 8. Ultimate tensile strength results from the residual strength tests presented as a function of longitudinal row location: (a)DC, (b)HT, (c)DL, and (d)GE. Also included for each composite were the results from the transversely oriented and as-processed specimens were some instances of fiber pullout, but in most cases the lengths experienced some adverse effects. In the former case this included were on the order of a few fiber diameters. More than half of the 0 edge cracks and in the latter a combination of edge cracks and Nicalon fibers, even many of those exhibiting pullout, were charac wear. While this first round of engine testing was considered a terized by completely smooth fracture surfaces. The smooth fiber success for both the DC and ht co, of the IsSues important to the osites, longer-term surviv- fracture surfaces suggest brittle behavior which was consistent with ability could still be a concern. Some the low strengths and linear stress-strain response of the tensile long-term survivability of CMC flap inserts on an exhaust nozzle specimens. include fatigue, wear, oxidation damage, and moisture exposure Severe but localized environmental attack, as shown in Fig. 11 The axisymmetric design of the F110 nozzle plays a critical role tensile specimens. Similar features were observed in"good"as- rubbing action and acoustic environment ofthenlong term.The was observed in several of the fiber fracture surfaces of the dl processed material fractured during room-temperature tension tests wear. The partial shielding of the flaps by the adjacent seals gives DuPont Lanxide reported similar features in the insert witness rise to in-plane thermal gradients and stresses which, as docu- specimens that they tested, indicating the damage was a result of mented by the dl and ge inserts, can lead to edge cracking. The processing. The localized nature of the damage was consistent with reduced thickness associated with the stepped cross section along the longitudinal edges will aggravate the situation further by mental attack of the fibers. Examination of damaged fibers via amplifying the gradient-induced stresses. Recall that these steps were machined along the sides and leading edge so that the insert electron dispersive X-ray techniques indicated the presence of alumI- would lie flush with the top retaining rail of the picture-frame num. The aluminum used in the direct metal oxidation process to form the matrix can attack the Nicalon fibers during processing if the Sic backing structure. The stepped thickness did not appear to give rise fiber coating were cracked. This appeared to have been the case. It to ply delamination in any of the four CMC inserts would be difficult to isolate fiber or interphase damage which could The thermal stresses generated in four CMCs, assuming an be specifically associated with the engine test. in-plane thermal gradient predicted by General Electric for a Nextel 610/AS flap insert, have been analyzed by John et al. Two of those CMCs were the hT and Dl composites. The maximum Discussion tensile stresses arise along the cooler edges of the insert as the center portion heats up and tries to expand during an aB The DC and HT flap inserts held up well following about 117 h naximum tensile stresses developed along the edges of the Ht and of engine ground testing. Neither the nondestructive inspections DL inserts were predicted to be 104 and 140 MPa, respectively nor the residual strength tests suggested any detrimental effects to For the HT material this was well below the room-temperature either CMC. On the other hand. both the dl and the ge inserts atigue limit of 185 MPa In the DL composite the thermal stresses

were some instances of fiber pullout, but in most cases the lengths were on the order of a few fiber diameters. More than half of the 0° Nicalon fibers, even many of those exhibiting pullout, were characterized by completely smooth fracture surfaces. The smooth fiber fracture surfaces suggest brittle behavior which was consistent with the low strengths and linear stress–strain response of the tensile specimens. Severe but localized environmental attack, as shown in Fig. 11, was observed in several of the fiber fracture surfaces of the DL tensile specimens. Similar features were observed in “good” asprocessed material fractured during room-temperature tension tests under the DARPA program but with a significantly lower frequency. DuPont Lanxide15 reported similar features in the insert witness specimens that they tested, indicating the damage was a result of processing. The localized nature of the damage was consistent with cracks in the matrix and SiC fiber coating which allowed environmental attack of the fibers. Examination of damaged fibers via electron dispersive X-ray techniques indicated the presence of aluminum. The aluminum used in the direct metal oxidation process to form the matrix can attack the Nicalon fibers during processing if the SiC fiber coating were cracked. This appeared to have been the case. It would be difficult to isolate fiber or interphase damage which could be specifically associated with the engine test. V. Discussion The DC and HT flap inserts held up well following about 117 h of engine ground testing. Neither the nondestructive inspections nor the residual strength tests suggested any detrimental effects to either CMC. On the other hand, both the DL and the GE inserts experienced some adverse effects. In the former case this included edge cracks and in the latter a combination of edge cracks and wear. While this first round of engine testing was considered a success for both the DC and HT composites, longer-term survivability could still be a concern. Some of the issues important to the long-term survivability of CMC flap inserts on an exhaust nozzle include fatigue, wear, oxidation damage, and moisture exposure. The axisymmetric design of the F110 nozzle plays a critical role in how well a CMC insert will survive over the long term. The rubbing action and acoustic environment of the nozzle can lead to wear. The partial shielding of the flaps by the adjacent seals gives rise to in-plane thermal gradients and stresses which, as documented by the DL and GE inserts, can lead to edge cracking. The reduced thickness associated with the stepped cross section along the longitudinal edges will aggravate the situation further by amplifying the gradient-induced stresses. Recall that these steps were machined along the sides and leading edge so that the insert would lie flush with the top retaining rail of the picture-frame backing structure. The stepped thickness did not appear to give rise to ply delamination in any of the four CMC inserts. The thermal stresses generated in four CMCs, assuming an in-plane thermal gradient predicted by General Electric for a Nextel 610/AS flap insert, have been analyzed by John et al.6 Two of those CMCs were the HT and DL composites. The maximum tensile stresses arise along the cooler edges of the insert as the center portion heats up and tries to expand during an AB light. The maximum tensile stresses developed along the edges of the HT and DL inserts were predicted to be 104 and 140 MPa, respectively. For the HT material this was well below the room-temperature fatigue limit of 185 MPa. In the DL composite the thermal stresses Fig. 8. Ultimate tensile strength results from the residual strength tests presented as a function of longitudinal row location: (a) DC, (b) HT, (c) DL, and (d) GE. Also included for each composite were the results from the transversely oriented and as-processed specimens. 1734 Journal of the American Ceramic Society—Staehler and Zawada Vol. 83, No. 7

were again below the material fatigue limit of 170 MPa, but by a smaller margin. The analysis assumed that the inserts had good as-processed properties. This was not the case for the DL insert actually engine tested. A third material considered by John et al. was a predecessor to the DC composite. Made by Kaiser Ceramic Composites, a joint venture between Dow Corning and what was then Kaiser Aerotech Engine Co. (San Leandro, CA), the composite used different fillers and processing temperatures compared with the DC insert. The mechanical properties of the Kaiser composite were presented elsewhere.17 The maximum tensile stresses in the Kaiser material due to the gradient were on the order of 115 MPa along the edges. These stresses were again below the material room-temperature fatigue limit of 160 MPa. It was assumed that the DC insert would have properties comparable to those of the Kaiser material and would experience similar thermal stresses. The fourth composite considered by John et al., 6 a Nextel 610/AS, was expected to develop 233 MPa tensile stresses along Fig. 9. Ultimate tensile strength results for specimens grouped in rows across the width of each insert: (a) DC, (b) HT, (c) DL, and (d) GE. Table III. Orthogonal Contrast Results from ANOVA Calculations Orthogonal contrasts† Reject null hypothesis (a 5 0.05)? DC HT DL GE Grouped as in Fig. 8 C1 5 d1mr1 1 d2mr9 Yes Yes No Yes C2 5 d3mr2 1 d4mr8 No Yes No No C3 5 d5mr3 1 d6mr7 No Yes No No C4 5 d7mr4 1 d8mr6 No No No No C5 5 d9m90s 1 d10mr4 1 d11mr5 1 d12mr6 Yes Yes No No Grouped as in Fig. 9 C19 5 b1mg1 1 b2mg5 No Yes No‡ No C29 5 b3mg2 1 b4mg4 Yes No No§ No C39 5 b5mg2 1 b6mg3 1 b7mg4 No No n/a No C49 5 b8mg1 1 b9mg2 1 b10mg3 1 b11mg4 1 b12mg5 Yes No No No † di ’s and bi ’s are coefficients; mij’s are UTS sample means. ‡ Substitute mg4 for mg5. § Substitute mg3 for mg4. Orthogonal contrast coefficients: d1 through d12, in order: DC and HT: 1, 21, 1, 21, 1, 21, 1, 21, 5, 21, 22, 21 DL: 1, 21, 1, 21, 1, 21, 1, 21, 2, 22, 21, 22 GE: 1, 21, 1, 21, 1, 21, 1, 21, 4, 21, 22, 21 b1 through b12, in order: DC and HT: 1, 21, 1, 21, 21, 2, 21, 3, 22, 22, 22, 3 DL: 1, 21, 1, 21, 0, 0, 0, 1, 21, 21, 1, 0 GE: 7, 29, 7, 29, 29, 16, 29, 10.9375, 27, 27, 27, 10.9375 July 2000 Performance of Four Ceramic-Matrix Composite Divergent Flap Inserts 1735

1736 Journal of the American Ceramic Sociery--Staehler and zawada Table IV. Summary of Tensile Results from Engine-Tested and As-Processed Specimens Conditions orentation UTS (MP Elastic modulus (Gh Prop. limit(MPa) Failure strain (% No, of specs tested DC Axial 3276(337)2 958(34) 127.6(8.5) 0.55(0.11) 45 Eng. test rans 93.9(1.5) 133.3(5 067(0.06 Warp 0.0(33.8) 01.6 120.1(7. 0.57(0.07 283.0(8.5) 117.0(5.6) Axial 178.1(12. 55.0(44) 44.6(58) 0.4600.05) Trans 191.8(66) 662(27) 57.0(106) Axial 1757(7.5 53.6(64) 45.5(7.9) 045(0.08 1854(52) 64.8(9.6) 46.9(11.4) Axial 79.6(146) 1434(32.9) 78(10.5) 0.1000.04 79.3(7.8) 2(17.6) 52.3(8.2) 0.1200.03 (witness) 103.5(9.2) .5(6.4) 190.7(5.03) 193.3(8.1) 67.0(1.0) 0.54(0 GE Eng test 173.4(11.4) :3 90.0(12.1) 804023942 Eng test Trans 172.5(54) 942(18.7) 02700.03) 172.0(14) 73.5(12.0) 0.33(0.01) Based on a 0.01% strain offset Numbers in parentheses are sample standard deviations. Due to loss of strain output for one specimen, elastic modulus, proportion Sk(CTE)compared with the Nextel 610 fibers which should in turn reduce the tensile stresses arising from the gradient. However, the reduced tensile stresses were still expected to be near or possibly above the material fatigue limit. Stresses arising from the inboard and outboard pressures on the nozzle, acoustic loading, and constraints from the picture-frame backing structure will add to those associated with the thermal gradient. Considering that the thermally induced stresses along the edges of the GE and dl inserts were already quite large, it was not surprising that these two components cracked. Because the cracks which developed in the ge insert occurred early on in the engine test, there did not appear to be any clear connection between the location of the cracks and the wear patterns which resulted through contact with the adjacent seals. The engine manufacturer considers 50 um the details of the stresses and stress distributions to be proprietary However, the peak tensile stresses anticipated from sources other Fig 10. SEM micrograph of a DL specimen fractured during room. than the thermal gradient were in the range of 35-70 MPa.The h of engine ground testing flight maneuver. The gradient-induced stresses comprise a major percentage of the total stresses encountered by each of the CMC were also expected These gradients are most severe in regions exposed directly to the hot exhaust gases. They should be less of a concen near the edges of the flap insert where the surface is shielded by the adjacent seals. No evidence of delamination was observed in any of the CMC inserts engine tested the in-plane thermal stresses could exceed each materials pip? The analytical calculations used by John et al.b indicated tha tional limit. If this occurs, matrix cracking is expected along the edges of each CMC insert during ground testing. Because the dc material exhibited a higher proportional limit than the Kaiser Nicalon/SINC, matrix cracks may or may not occur. Assuming they do, matrix cracks can allow oxygen and moisture to gain our access to the composite interior. For the dc and dl composites oxygen access could result in oxidation of the bn fiber coating Fig. 11. SEM micrograph of the localized damage observed in many of and/or the Nicalon fibers. Water vapor, which comprises as much the Nicalon fibers the DL insert. The damage originated during as 10%15% of the exhaust gases, can react with the bn fiber processing. coatings as well. 8, 9 Loss of the BN and/or oxidation of the fibers would eventually degrade the composite properties. Fortunately, because of the shielding by the adjacent seals, temperatures ale the edges may remain too low for significant oxidation-related the edges of the insert. stresses exceeded the material damage to occur. However, if buckling takes place because of the room-temperature tensile of 206 MPa and would quickly compressive stresses through the center of the insert, matrix lead to edge cracks. The stress analysis provided much of cracking may be more widespread. The cyclic nature of the the impetus for switching to Nextel 720 fibers in the ge compos- stresses caused by the intermittent AB lights and changing pres- have a lower coefficient of thermal expansion sures of the nozzle create a fatigue loading environment for the

the edges of the insert. These stresses exceeded the material room-temperature tensile strength of 206 MPa and would quickly lead to edge cracks. The thermal stress analysis provided much of the impetus for switching to Nextel 720 fibers in the GE composite. These fibers have a lower coefficient of thermal expansion (CTE) compared with the Nextel 610 fibers which should in turn reduce the tensile stresses arising from the gradient. However, the reduced tensile stresses were still expected to be near or possibly above the material fatigue limit. Stresses arising from the inboard and outboard pressures on the nozzle, acoustic loading, and constraints from the picture-frame backing structure will add to those associated with the thermal gradient. Considering that the thermally induced stresses along the edges of the GE and DL inserts were already quite large, it was not surprising that these two components cracked. Because the cracks which developed in the GE insert occurred early on in the engine test, there did not appear to be any clear connection between the location of the cracks and the wear patterns which resulted through contact with the adjacent seals. The engine manufacturer considers the details of the stresses and stress distributions to be proprietary. However, the peak tensile stresses anticipated from sources other than the thermal gradient were in the range of 35–70 MPa. The actual magnitude of these stresses depends on several factors including altitude, air speed, the size of the nozzle opening, and the flight maneuver. The gradient-induced stresses comprise a major percentage of the total stresses encountered by each of the CMC inserts. Through-thickness temperature gradients were also expected. These gradients are most severe in regions exposed directly to the hot exhaust gases. They should be less of a concern near the edges of the flap insert where the surface is shielded by the adjacent seals. No evidence of delamination was observed in any of the CMC inserts engine tested. The analytical calculations used by John et al.6 indicated that the in-plane thermal stresses could exceed each material’s proportional limit. If this occurs, matrix cracking is expected along the edges of each CMC insert during ground testing. Because the DC material exhibited a higher proportional limit than the Kaiser Nicalon/SiNC, matrix cracks may or may not occur. Assuming they do, matrix cracks can allow oxygen and moisture to gain access to the composite interior. For the DC and DL composites, oxygen access could result in oxidation of the BN fiber coating and/or the Nicalon fibers. Water vapor, which comprises as much as 10%–15% of the exhaust gases, can react with the BN fiber coatings as well.18,19 Loss of the BN and/or oxidation of the fibers would eventually degrade the composite properties. Fortunately, because of the shielding by the adjacent seals, temperatures along the edges may remain too low for significant oxidation-related damage to occur. However, if buckling takes place because of the compressive stresses through the center of the insert, matrix cracking may be more widespread. The cyclic nature of the stresses caused by the intermittent AB lights and changing pressures of the nozzle create a fatigue loading environment for the Table IV. Summary of Tensile Results from Engine-Tested and As-Processed Specimens CMC Conditions Spec. orientation UTS (MPa) Elastic modulus (GPa) Prop. limit (MPa)† Failure strain (%) No. of specs tested DC Eng. test Axial 327.6 (33.7)‡ 95.8 (3.4) 127.6 (8.5) 0.55 (0.11) 45 Eng. test Trans. 407.8 (20.9) 93.9 (1.5) 133.3 (5.1) 0.67 (0.06) 4§ As-pro. Warp 350.0 (33.8) 101.6 (4.2) 120.1 (7.6) 0.57 (0.07) 7 As-pro. Fill 283.0 (8.5) 101.1 (8.9) 117.0 (5.6) 0.43 (0.02) 7 HT Eng. test Axial 178.1 (12.4) 55.0 (4.4) 44.6 (5.8) 0.46 (0.05) 45 Eng. test Trans. 191.8 (6.6) 66.2 (2.7) 57.0 (10.6) 0.43 (0.04) 4 As-pro. Axial 175.7 (7.5) 53.6 (6.4) 45.5 (7.9) 0.45 (0.08) 18 As-pro. Trans. 185.4 (5.2) 64.8 (9.6) 46.9 (11.4) 0.41 (0.05) 10 DL Eng. test Axial 79.6 (14.6) 143.4 (32.9) 57.8 (10.5) 0.10 (0.04) 34 Eng. test Trans. 79.3 (7.8) 151.2 (17.6) 52.3 (8.2) 0.12 (0.03) 10 As-pro. (witness)¶ n/a 103.5 (9.2) 138.5 (6.4) n/a 0.24 (0.13) 2 As-pro. (DARPA) n/a 190.7 (5.03) 193.3 (8.1) 67.0 (1.0) 0.54 (0.02) 3 GE Eng. test Axial 173.4 (11.4) 71.4 (6.4) 90.0 (12.1) 0.30 (0.04) 39 Eng. test Trans. 172.5 (5.4) 84.6 (4.2) 94.2 (18.7) 0.27 (0.03) 4 As-pro. n/a 172.0 (1.4) 64.9 (6.5) 73.5 (12.0) 0.33 (0.01) 2 † Based on a 0.01% strain offset. ‡ Numbers in parentheses are sample standard deviations. § Due to loss of strain output for one specimen, elastic modulus, proportional limit, and strain to failure based on three tests. ¶ Witness data provided by DuPont Lanxide. Fig. 10. SEM micrograph of a DL specimen fractured during roomtemperature residual strength testing. The material had experienced ;120 h of engine ground testing. Fig. 11. SEM micrograph of the localized damage observed in many of the Nicalon fibers of the DL insert. The damage originated during processing. 1736 Journal of the American Ceramic Society—Staehler and Zawada Vol. 83, No. 7

点击进入文档下载页（PDF格式）

共12页，试读已结束，阅读完整版请下载

点击下载（PDF格式）

浏览记录

《复合材料 Composites》课程教学资源（学习资料）第五章陶瓷基复合材料_j.1151-2916.2000.tb01457.x[1]

《复合材料 Composites》课程教学资源（学习资料）第五章 陶瓷基复合材料_j.1151-2916.2000.tb01457.x[1]

《复合材料 Composites》课程教学资源（学习资料）第五章陶瓷基复合材料_j.1151-2916.2000.tb01457.x[1]