COMPOSITES SCIENCE AND TECHNOLOGY ELSEVIER Composites Science and Technology 59(1999)853-859 Fabrication and pressure testing of a gas-turbine component manufactured by a preceramic-polymer- impregnation method Kiyoshi Sato a ,*, Atsushi Tezuka, Osamu Funayama, Takeshi Isoda Yoshiharu Terada. ShinuI Kato Misao Iwata 6 a Corporate Research and Development Laboratory, TONEN Corporation, 1-3-1 Nishi-fsurugaoka, Ohi-lmachi, lruma-gun, Saitama 356-8505, Japan PNoritake Co, Limited, 300 Higashiyama, Miyoshi-cho, Nishikamo-gun, Aichi 470-02, Japan Received I July 1997; received in revised form 24 August 1998; accepted 8 January 1999 Abstract One of the components of the gas-turbine engine, the inner scroll support, was fabricated by a Si-C-O-fiber-reinforced Si-N-C composite of which the matrix was obtained from methylhydrosilazane. The part was shaped by laminating prepreg sheets, con- sisting of Si-C-O fiber cloth and methylhydrosilazane, and cured by using a vacuum-bagging method. Densification of the pro- duct was performed by polymer impregnation and pyrolysis(PIP). The fracture strength and stress/strain behaviour were measured by the use of a hydraulic internal pressurization test Hysteresis during cyclic loading and pseudo-elastic behavior were observed in the composite component. Thus behaviour was similar to that found in mechanical tests of small specimens. The fracture strength of the composite part was estimated to be 180 MPa, which was the same as the tensile strengths of test pieces. 1999 Elsevier Science Ltd. All rights reserved. Keywords: A. Ceramic-matrix composites; Strength test; Polysilazane; Silicon carbide fiber 1. Introduction There are a number of studies on the fabrication of structural parts by the PIP method [11, 12]. Regarding Ceramic-matrix composites which resist heat in an the evaluation of parts made of continuous-fiber-rein oxidising atmosphere have been developed as durable forced ceramics, some thermal exposure tests have been materials for mechanical uses. Many studies have been reported for combustors, but there are few reports of carried out on the fabrication process. Manufacturing a fracture tests [11, 12 fabrication of ceramic-matrix composites industrially 4 The purposes of this work are () to prove the fabri- product to near-net-shape is an important aspects of the because of costs and productivity. Several near-net- fiber-reinforced ceramics for which the matrix was shape processes have been developed; for example, chem- fabricated from methylhydrosilazane by a near-net-shape ical vapor impregnation(CVI), polymer impregnation process and ( ii) to evaluate the fracture strength of the and pyrolysis(PIP), reaction bonding(RB), and direct part by a hydraulic internal pressurization test and metal oxidation [1-4] compare the strength with that of a test piece PIP has considerable merit for shaping Shaping pro- cesses for fiber-reinforced plastics which are used ndustrially can be applied to Plp because both methods 2. Experimental procedure use a resin for the matrix. Many studies have recently been performed on PIP [5-8]. Silazanes synthesized by a 2. 1. M pyridine-adduct method were applied to the PIP method pieces ammufacture of the gas-turbine component and test by the authors [9, 10]. The silazanes have the distinctive features of low viscosity and high yield in converting The gas-turbine component, referred to as an inner ceramics by pyrolyzation. As a result, dense composites scroll support(Fig. 1), was selected because its simple can be fabricated shape makes it easy to fabricate and evaluate. This gas turbine was designed by the Japan Automobile esponding author. Research Institute (JARI) in the 100 kw Automotive 0266-3538/99/S- see front matter C 1999 Elsevier Science Ltd. All rights reserved. PlI:S0266-3538(99)00015
Fabrication and pressure testing of a gas-turbine component manufactured by a preceramic-polymer-impregnation method Kiyoshi Satoa,*, Atsushi Tezuka a , Osamu Funayama a , Takeshi Isoda a , Yoshiharu Terada b, Shinji Katob, Misao Iwata b a Corporate Research and Development Laboratory, TONEN Corporation, 1-3-1 Nishi-tsurugaoka, Ohi-machi, Iruma-gun, Saitama 356-8505, Japan bNoritake Co., Limited, 300 Higashiyama, Miyoshi-cho, Nishikamo-gun, Aichi 470-02, Japan Received 1 July 1997; received in revised form 24 August 1998; accepted 8 January 1999 Abstract One of the components of the gas-turbine engine, the inner scroll support, was fabricated by a SiÿCÿO-®ber-reinforced SiÿNÿC composite of which the matrix was obtained from methylhydrosilazane. The part was shaped by laminating prepreg sheets, consisting of SiÿCÿO ®ber cloth and methylhydrosilazane, and cured by using a vacuum-bagging method. Densi®cation of the product was performed by polymer impregnation and pyrolysis (PIP). The fracture strength and stress/strain behaviour were measured by the use of a hydraulic internal pressurization test. Hysteresis during cyclic loading and pseudo-elastic behavior were observed in the composite component. Thus behaviour was similar to that found in mechanical tests of small specimens. The fracture strength of the composite part was estimated to be 180 MPa, which was the same as the tensile strengths of test pieces. # 1999 Elsevier Science Ltd. All rights reserved. Keywords: A. Ceramic-matrix composites; Strength test; Polysilazane; Silicon carbide ®ber 1. Introduction Ceramic-matrix composites which resist heat in an oxidising atmosphere have been developed as durable materials for mechanical uses. Many studies have been carried out on the fabrication process. Manufacturing a product to near-net-shape is an important aspects of the fabrication of ceramic-matrix composites industrially because of costs and productivity. Several near-netshape processes have been developed; for example, chemical vapor impregnation (CVI), polymer impregnation and pyrolysis (PIP), reaction bonding (RB), and direct metal oxidation [1±4]. PIP has considerable merit for shaping. Shaping processes for ®ber-reinforced plastics which are used industrially can be applied to PIP because both methods use a resin for the matrix. Many studies have recently been performed on PIP [5±8]. Silazanes synthesized by a pyridine-adduct method were applied to the PIP method by the authors [9,10]. The silazanes have the distinctive features of low viscosity and high yield in converting ceramics by pyrolyzation. As a result, dense composites can be fabricated. There are a number of studies on the fabrication of structural parts by the PIP method [11,12]. Regarding the evaluation of parts made of continuous-®ber-reinforced ceramics, some thermal exposure tests have been reported for combustors, but there are few reports of fracture tests [11,12]. The purposes of this work are (i) to prove the fabrication of a gas-turbine component by using ceramic- ®ber-reinforced ceramics for which the matrix was fabricated from methylhydrosilazane by a near-net-shape process and (ii) to evaluate the fracture strength of the part by a hydraulic internal pressurization test and compare the strength with that of a test piece. 2. Experimental procedure 2.1. Manufacture of the gas-turbine component and test pieces The gas-turbine component, referred to as an inner scroll support (Fig. 1), was selected because its simple shape makes it easy to fabricate and evaluate. This gas turbine was designed by the Japan Automobile Research Institute (JARI) in the 100 kW Automotive Composites Science and Technology 59 (1999) 853±859 0266-3538/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved. PII: S0266-3538(99)00015-9 * Corresponding author
K Sato et al./ Composites Science and Technology 59(1999)853-859 58.8 rotor o寸o inner scroll support inner scroll outer scro (a (b) Fig. 1. 100 kW automotive ceramic gas-turbine engine(a) and the target component(b) Ceramic Gas Turbine(CGT) Development Project [13]. which was converted to amorphous Si-N-C by firing The inner scroll support holds an inner scroll which in a nitrogen atmosphere. Borosilazane(Bs), originally carries combustion gas at 1523-1623 K from a com- developed as a raw material for Si-B-O-N fiber, was bustor to a rotor. One side of the inner scroll support is produced by the reaction of perhydrosilazane and tri- heated by the inner-scroll up to 1523 K, and the other methylborate [14]. Bs had high chemical reactivity, so side is cooled by a metallic part. As a result, thermal the ceramic conversion yield of the MHs was increased stress is induced. In the case of an inner scroll support from 60 mass% to 70 mass by adding bS made from silicon nitride, the generation of a stress of Three types of MHS (MHS-l, MHS-2 and MHS-3) 83 MPa is estimated by finite-element analysis by JARL. and Bs were mixed to give adequate flexibility and A Si-C-O fiber, the surface of which was coated stickiness to the prepreg. The molecular weights and with carbon (Nippon Carbon in Japan, Nicalon, reactivity of MHS-1, 2 and 3 were controlled by varying NL607), was used as a reinforcing material. The matrix the synthesis conditions ( Table 1). The synthesis tem- consisted of amorphous Si-N-C, which was obtained perature of the MHS-I was high. As a result, MHS-I was by pyrolizing methylhydrosilazane. The fabrication a semi-solid resin with a high molecular weight. MHS-2 procedure for the inner scroll support is described in the contained a larger number of-(Si(CH3)HNH)-units next section. a plate was also fabricated by the same than either MHS-1 or MHS-3, and therefore was more process and testing coupons were cut from it. stable. MHS-3 was synthesized at a low temperature and showed low molecular weight and low viscosity 2.1.I. Fabrication of prepreg sheet The requirements for prepreg are (i) curing at 373- Methylhydrosilazane (MHS), the raw material for the 573 K; (ii) having flexibility and good adhesive qualities matrix, was synthesized by a pyridine-adduct method and (iii) stability for several hours during the stacking starting from Si(CH3)HCI,, SiH2 CI, and NH3. MHs process. After many tests, the composition of MHS-1/ was a random co-polymer, consisting of-(Si(CH3)NH)- MHS-2/MHS-3/BS=12 /12/ 6/1 was selected. Fig. 2(a) and-(SiH,NH)-units. MHS was a thermo-setting resin shows the change in the viscosity of the mixture by Table I Methylhydrosilazanes for matrix precursors Type MDCS/DCS mol ratio Molecular weight Elemental composition/mass N Total MHS-I 1/ MHS-2 MHS-3 1/
Ceramic Gas Turbine (CGT) Development Project [13]. The inner scroll support holds an inner scroll which carries combustion gas at 1523±1623 K from a combustor to a rotor. One side of the inner scroll support is heated by the inner-scroll up to 1523 K, and the other side is cooled by a metallic part. As a result, thermal stress is induced. In the case of an inner scroll support made from silicon nitride, the generation of a stress of 83 MPa is estimated by ®nite-element analysis by JARI. A SiÿCÿO ®ber, the surface of which was coated with carbon (Nippon Carbon in Japan, Nicalon, NL607), was used as a reinforcing material. The matrix consisted of amorphous SiÿNÿC, which was obtained by pyrolizing methylhydrosilazane. The fabrication procedure for the inner scroll support is described in the next section. A plate was also fabricated by the same process and testing coupons were cut from it. 2.1.1. Fabrication of prepreg sheet Methylhydrosilazane(MHS), the raw material for the matrix, was synthesized by a pyridine-adduct method starting from Si(CH3)HCl2, SiH2Cl2 and NH3. MHS was a random co-polymer, consisting of -(Si(CH3)NH)- and-(SiH2NH)-units. MHS was a thermo-setting resin which was converted to amorphous SiÿNÿC by ®ring in a nitrogen atmosphere. Borosilazane (BS), originally developed as a raw material for SiÿBÿOÿN ®ber, was produced by the reaction of perhydrosilazane and trimethylborate [14]. BS had high chemical reactivity, so the ceramic conversion yield of the MHS was increased from 60 mass% to 70 mass% by adding BS. Three types of MHS (MHS-1, MHS-2 and MHS-3) and BS were mixed to give adequate ¯exibility and stickiness to the prepreg. The molecular weights and reactivity of MHS-1,2 and 3 were controlled by varying the synthesis conditions (Table 1). The synthesis temperature of the MHS-1 was high. As a result, MHS-1 was a semi-solid resin with a high molecular weight. MHS-2 contained a larger number of -(Si(CH3)HNH)- units than either MHS-1 or MHS-3, and therefore was more stable. MHS-3 was synthesized at a low temperature and showed low molecular weight and low viscosity. The requirements for prepreg are (i) curing at 373± 573 K; (ii) having ¯exibility and good adhesive qualities; and (iii) stability for several hours during the stacking process. After many tests, the composition of MHS-1/ MHS-2/MHS-3/BS=12/12/6/1 was selected. Fig. 2(a) shows the change in the viscosity of the mixture by Fig. 1. 100 kW automotive ceramic gas-turbine engine (a) and the target component (b). Table 1 Methylhydrosilazanes for matrix precursors Type MDCS/DCS mol ratio Synthesis temperature/K Molecular weight Elemental composition/mass% Si N C O Total MHS-1 1/1 328 1400 55 23 11 0.6 89.6 MHS-2 2/1 323 920 51 21 13 1 86 MHS-3 1/1 268 770 59 22 13 1 95 854 K. Sato et al. / Composites Science and Technology 59 (1999) 853±859
K Sato et al./ Composites Science and Technology 59(1999)853-859 0.01 0.01 200300400500600 5 5 Temperature /K Time/ks Fig.2.(a)Viscosity change of a mixture of MHS-l, MHS.2, MHS.3 and BS with a mass ratio of 12/12/6/1 by heating a rate of 83x10-2K/s(b) Viscosity change of the mixture which the temperature was kept at 373 K. heating at a rate of 8.3x10-2K/s. The rate of decrease of the viscosity increased at about 350 K where self- bridge reaction started. Fig. 2(b) shows the change in product viscosity when the temperature was kept at 373 K. The viscosity was slowly increased. This suggests that by varying the duration of heat treatment at 373 K, the flexibility and adhesive qualities of the prepreg can be controlled The prepreg was manufactured as follows. Firstly, the plain-weave Si-C-O fibre cloth was cut to a size of 140x 530 mm at an angle of 45 to the main axis of the weave. To prevent any reaction with the silazanes, the poly-vinyl-alcohol which was coated on the fibers as a sizing agent, was removed by firing at 873 K in a nitrogen atmosphere. The mixture of the silazanes was coated onto a sheet of polytetrafluoroethylene(PTFE)by a coater at a rate of 0.15-0.2 kg/m The Si-C-O cloth framework vas put on the silazane- coated PTFE sheet and covered by another uncoated PTFE sheet. The laminate was rolled up and heated at 370 K for 10.8 ks in a nitrogen atmosphere. After the heat treatment, the laminate was unrolled and pressed. The thickness of the prepreg after Fig. 3. Structure of shaping jig. ne press was controlled, the volume fraction of the fiber was 0. 4. The manufactured prepreg had adequate adhe- sive qualities and flexibility, and it was, therefore, easy thick. Therefore, many sheets of prepreg were stacked to stack the prepreg onto a complex shaping jig on both ends. Prepreg sheets were cut out in the fol- lowing sizes and stacked as illustrated in Fig. 4:(A) 140×530mm,(B)20×530 mm and(C20×430mm. The shaping jig is shown in Fig. 3. The jig was Fourteen sheets were stacked in the center of the product. assembled by a framework, whereby the mold parts The laminate was shaped and cured by using were divided into three parts. Release oil was painted vacuum-bagging method. The autoclave pressure, bag onto the surface of the mold to allow for the easy pressure and product temperature during the process separation of the cured product are shown in Fig. 5. First, the air was removed from the The wall thickness of the inner scroll support was 3 mm bag in which the jig and the stacked sheets were packed in the center section, but each end was about 10 mm Next, the bag was heated to 400 K, which is just under
heating at a rate of 8.310ÿ2 K/s. The rate of decrease of the viscosity increased at about 350 K where selfbridge reaction started. Fig. 2(b) shows the change in viscosity when the temperature was kept at 373 K. The viscosity was slowly increased. This suggests that by varying the duration of heat treatment at 373 K, the ¯exibility and adhesive qualities of the prepreg can be controlled. The prepreg was manufactured as follows. Firstly, the plain-weave SiÿCÿO ®bre cloth was cut to a size of 140530 mm at an angle of 45 to the main axis of the weave. To prevent any reaction with the silazanes, the poly-vinyl-alcohol which was coated on the ®bers as a sizing agent, was removed by ®ring at 873 K in a nitrogen atmosphere. The mixture of the silazanes was coated onto a sheet of polytetra¯uoroethylene(PTFE) by a coater at a rate of 0.15±0.2 kg/m2 . The SiÿCÿO cloth was put on the silazane-coated PTFE sheet and covered by another uncoated PTFE sheet. The laminate was rolled up and heated at 370 K for 10.8 ks in a nitrogen atmosphere. After the heat treatment, the laminate was unrolled and pressed. The thickness of the prepreg after the press was controlled, the volume fraction of the ®ber was 0.4. The manufactured prepreg had adequate adhesive qualities and ¯exibility, and it was, therefore, easy to stack the prepreg onto a complex shaping jig. 2.1.2. Shaping and ®rst ®ring The shaping jig is shown in Fig. 3. The jig was assembled by a framework, whereby the mold parts were divided into three parts. Release oil was painted onto the surface of the mold to allow for the easy separation of the cured product. The wall thickness of the inner scroll support was 3 mm in the center section, but each end was about 10 mm thick. Therefore, many sheets of prepreg were stacked on both ends. Prepreg sheets were cut out in the following sizes and stacked as illustrated in Fig. 4: (A) 140530 mm, (B) 20530 mm and (C) 20430 mm. Fourteen sheets were stacked in the center of the product. The laminate was shaped and cured by using a vacuum-bagging method. The autoclave pressure, bag pressure and product temperature during the process are shown in Fig. 5. First, the air was removed from the bag in which the jig and the stacked sheets were packed. Next, the bag was heated to 400 K, which is just under Fig. 2. (a) Viscosity change of a mixture of MHS-1, MHS-2, MHS-3 and BS with a mass ratio of 12/12/6/1 by heating a rate of 8.310ÿ2 K/s. (b) Viscosity change of the mixture which the temperature was kept at 373 K. Fig. 3. Structure of shaping jig. K. Sato et al. / Composites Science and Technology 59 (1999) 853±859 855
K Sato et al./ Composites Science and Technology 59(1999)853-859 stacking mass ratio of 30/1.(ii) Recuring: in the first densification cycle, vacuum-bagging was performed using the same t procedure used for shaping. After the second cycle, the prepreg silazane soaked product was cured under a pressurized nitrogen atmosphere to 0.09 MPa at 523 K for 3.6 k to a third densified cycle, the firing condition was the same as that during he first firing. After the third cycle, the heating rate was mold reduced to prevent the product from fracturing, i.e. up to 473 K with the heating rate was 0. 166 K/s: to 773 K the rate was 0.005 K s; to 1073 K it was 0.017 K/s; and to 1623 K it was 0.025 k/ s prepreg The densified product was machined and finished to the target dimensions. Additionally, to adjust the prod- Fig. 4. Stacking structure of prepreg sheets. A number of sheets are uct to the testing jig, both ends of the product were tted on this illustration to clarify the structure coated with an epoxy resin and refinished 2.2. Testing method The fracture test was performed by applying an Bag pressure internal hydraulic pressure to the product by closing 700 0.10 both of the open ends as illustrated in Fig. 6(a). Strain gages were placed on the component to monitor the 600 0.08 c hoop strain during the test [Fig. 6(b) and(c)]. Two cyclic pressures, whose peak values were 4.9 and 5.5 500 :10.06 MPa, were applied before the fracture test. The increased rate of the pressure was 3.3 kPa/s on each test 0.04 A center section of the fractured product was cut out to measure porosity and the thickness of the product. 0.02 The apparent density was measured by the archimedes method for samples whose surfaces 0.00 paraffin. The true density was measured by picnometer 0510152025 from powder obtained by crushing a part of the pro- duct. Total porosity was calculated using the apparent Time /ks Fig. 5. Process condition haping and curing by vacuum bagging. fiber was calculated by the number of the stacked pre- preg sheets and the thickness taken from the cut section Tensile test pieces of dimensions 2x 12x1120 mm the minimum viscosity point of the silazane, [Fig. 2(b) were cut from the composite plate at angles of 45 or 0o in order to impregnate the polymer in to the cloth. Tensile tests were performed on the test pieces of the Finally, the pressure in the autoclave was increased to gage length of 40 mm with aluminum tab sat a cross 0.06 MPa and the temperature was elevated to 550K. head speed of 0.83 um/s. The porosity and the volum The cured product became a hard solid and was taken fraction of fiber were also measured in the test pieces from the mold. Pyrolyzing was carried out in a nitrogen atmosphere under the following conditions: first tem- perature was elevated to 1073 K at a rate of 0.083 K/s, 3. Results and discussions and then to 1623 K at 0. 1 66 K s. The maximum tem- perature was kept for 1. 8 ks Manufactured product is shown in Fig. 7. The gas- turbine component was successfully fabricated by the 2.1.3. Densification and machining PIP method After one firing the product had a high porosity The hoop stress generated in the inner scroll support because the precursor shrank. To obtain a dense pro- was calculated by correcting the FEM analysis results duct, densification was achieved through 5 cycles of which were performed by JARI, on a monolithic silicon reimpregnation, recuring and refiring (i) Reimpregna- nitride part. Fig. 8 illustrates the stress distribution of tion: the product was put in a vacuum vessel and the monolithic part when an internal pressure of 10 impregnated by a liquid mixture MHS-3 and bs with MPa is loaded
the minimum viscosity point of the silazane, [Fig. 2(b)] in order to impregnate the polymer in to the cloth. Finally, the pressure in the autoclave was increased to 0.06 MPa and the temperature was elevated to 550 K. The cured product became a hard solid and was taken from the mold. Pyrolyzing was carried out in a nitrogen atmosphere under the following conditions: ®rst temperature was elevated to 1073 K at a rate of 0.083 K/s, and then to 1623 K at 0.1 66 K/s. The maximum temperature was kept for 1.8 ks. 2.1.3. Densi®cation and machining After one ®ring the product had a high porosity because the precursor shrank. To obtain a dense product, densi®cation was achieved through 5 cycles of reimpregnation, recuring and re®ring (i) Reimpregnation: the product was put in a vacuum vessel and impregnated by a liquid mixture MHS-3 and BS with a mass ratio of 30/1. (ii) Recuring: in the ®rst densi®cation cycle, vacuum-bagging was performed using the same procedure used for shaping. After the second cycle, the silazane soaked product was cured under a pressurized nitrogen atmosphere to 0.09 MPa at 523 K for 3.6 ks without bagging. (iii) Re®ring: upto a third densi®cation cycle, the ®ring condition was the same as that during the ®rst ®ring. After the third cycle, the heating rate was reduced to prevent the product from fracturing, i.e. up to 473 K with the heating rate was 0.166 K/s; to 773 K the rate was 0.005 K/s; to 1073 K it was 0.017 K/s; and to 1623 K it was 0.025 k/s. The densi®ed product was machined and ®nished to the target dimensions. Additionally, to adjust the product to the testing jig, both ends of the product were coated with an epoxy resin and re®nished. 2.2. Testing method The fracture test was performed by applying an internal hydraulic pressure to the product by closing both of the open ends as illustrated in Fig. 6(a). Strain gages were placed on the component to monitor the hoop strain during the test [Fig. 6(b) and (c)]. Two cyclic pressures, whose peak values were 4.9 and 5.5 MPa, were applied before the fracture test. The increased rate of the pressure was 3.3 kPa/s on each test. A center section of the fractured product was cut out to measure porosity and the thickness of the product. The apparent density was measured by the Archimedes method for samples whose surfaces are coated with paran. The true density was measured by picnometry from powder obtained by crushing a part of the product. Total porosity was calculated using the apparent density and the true density. The volume fraction of ®ber was calculated by the number of the stacked prepreg sheets and the thickness taken from the cut section. Tensile test pieces of dimensions 2121120 mm were cut from the composite plate at angles of 45 or 0. Tensile tests were performed on the test pieces of the gage length of 40 mm with aluminum tab sat a crosshead speed of 0.83 mm/s. The porosity and the volume fraction of ®ber were also measured in the test pieces. 3. Results and discussions Manufactured product is shown in Fig. 7. The gasturbine component was successfully fabricated by the PIP method. The hoop stress generated in the inner scroll support was calculated by correcting the FEM analysis results, which were performed by JARI, on a monolithic silicon nitride part. Fig. 8 illustrates the stress distribution of the monolithic part when an internal pressure of 10 MPa is loaded. Fig. 4. Stacking structure of prepreg sheets. A number of sheets are omitted on this illustration to clarify the structure. Fig. 5. Process conditions for shaping and curing by vacuum bagging. 856 K. Sato et al. / Composites Science and Technology 59 (1999) 853±859
If the elastic properties of a composite are isotropic, sheets were stacked in the same orientation. In the pre- the FEM analysis result for the monolithic part is liminary calculation, the anisotropy was ignored. The applicable to a composite part. The composite of this thickness of the center section of the composite par study has anisotropic elasticity because the plain-weave and the monolithic part were different, 4 and 3 mm, 270° 90°H3 8 H4 strain gauges Fig. 6. Internal hydraulic pressure test of the inner scroll support. Apparatus (a) and gauge positions to measure strains(b);(c)is an enlarged 0.0020 0.0015 苏0.0010 0.0005 270° Fig. 7. Final product. Fig. 9. Hoop strain distribution measured on the composite part Oil pressure 10 MPa 200 150 切 Maximum stress 270 MPa pture test 000010.002000300040.005 ydraulic pressure test. The stress was calculated by FEM analysis of a monolithic silicon Fig. 10. Stress/strain curves from a hydraulic internal pressurization
If the elastic properties of a composite are isotropic, the FEM analysis result for the monolithic part is applicable to a composite part. The composite of this study has anisotropic elasticity because the plain-weave sheets were stacked in the same orientation. In the preliminary calculation, the anisotropy was ignored. The thickness of the center section of the composite part and the monolithic part were dierent, 4 and 3 mm, Fig. 8. Stress distribution generated by internal hydraulic pressure test. The stress was calculated by FEM analysis of a monolithic silicon nitride part. Fig. 7. Final product. Fig. 6. Internal hydraulic pressure test of the inner scroll support. Apparatus (a) and gauge positions to measure strains (b); (c) is an enlarged illustration of the section C in (b). Fig. 9. Hoop strain distribution measured on the composite part (loading pressure was 4.86 MPa). Fig. 10. Stress/strain curves from a hydraulic internal pressurization test. K. Sato et al. / Composites Science and Technology 59 (1999) 853±859 857
K Sato et al./ Composites Science and Technology 59(1999)853-859 Mechanical properties of inner scroll support and tensile test pieces Fiber content/vol% Apparent density/mg/m Porosity/vol% Strength/MPa Elastic modulus/ GPa Test piece(0° off axis) Test piece(45 off axis) 14 respectively. In general, if two shell structures have the same shape but differ in thickness, the stress generated by the inner pressure is inversely proportional to the thickness. As a result, only the thickness correction was performed on the FEM result An example of the measured hoop strain is shown in a Fig. 9. The product showed an 11% difference in the hoop strains around the latitudinal directions, when 4.9 MPa of hydraulic pressure was loaded. The reasons for he strain difference are (i the non-uniformity of the material; the distribution of fiber content and porosity and (ii) the thickness distribution of the product. The strain measured at 270 at position H4 was used for the ollowing analysis, because the generation of maximum b stress was estimated on the H4 position by the FEM result. and the measurement at 270 was a mean value as compared with that at 0, 90 and 180 The stress/strain curve for the test is shown in Fig. 10 This curve has the following characteristics: (i) hyster esis upon cyclic loading and (ii) pseudo-elasticity with a proportional limit in 20-30 MPa. These features have been observed in tests on many continuous fiber rein 10mm forced ceramics [15, 16]. This study found the behavior of a complex-shaped part during the inner hydraulic pressure test was the same as the tensile test of the test scroll support. On a sn diameter he weave st a longitudinal direction. On stretched to a latitudinal The strengths of the test coupons were 110 and 89 MPa for 0 and 45. off axis specimens, respectively, while 20 mm(b) Fig. 12. Fracture appearance of inner scroll support.(b)is an enlarged photograph of (a). Black arrows show a crack
respectively. In general, if two shell structures have the same shape but dier in thickness, the stress generated by the inner pressure is inversely proportional to the thickness. As a result, only the thickness correction was performed on the FEM result. An example of the measured hoop strain is shown in Fig. 9. The product showed an 11% dierence in the hoop strains around the latitudinal directions, when 4.9 MPa of hydraulic pressure was loaded. The reasons for the strain dierence are (i) the non-uniformity of the material; the distribution of ®ber content and porosity, and (ii) the thickness distribution of the product. The strain measured at 270 at position H4 was used for the following analysis, because the generation of maximum stress was estimated on the H4 position by the FEM result, and the measurement at 270 was a mean value as compared with that at 0, 90 and 180. The stress/strain curve for the test is shown in Fig. 10. This curve has the following characteristics: (i) hysteresis upon cyclic loading and (ii) pseudo-elasticity with a proportional limit in 20±30 MPa. These features have been observed in tests on many continuous ®ber reinforced ceramics [15,16]. This study found the behavior of a complex-shaped part during the inner hydraulic pressure test was the same as the tensile test of the test pieces. The strengths of the test coupons were 110 and 89 MPa for 0 and 45 o axis specimens, respectively, while Table 2 Mechanical properties of inner scroll support and tensile test pieces Fiber content/vol% Apparent density/mg/m3 Porosity/vol% Strength/MPa Elastic modulus/GPa Inner-scroll-support 39 2.33 11 180 ± Test piece (0 o axis) 43 2.34 10 110 122 Test piece (45 o axis) 43 2.34 10 89 114 Fig. 11. Distortion of weave on inner scroll support. On a small diameter part (a),the weave stretched to a longitudinal direction. On a large diameter part (b), the weave stretched to a latitudinal direction. Fig. 12. Fracture appearance of inner scroll support. (b) is an enlarged photograph of (a). Black arrows show a crack. 858 K. Sato et al. / Composites Science and Technology 59 (1999) 853±859
K. Sato et al./ Composites Science and Technology 59(1999)853-859 fiber-reinforced Si-N-C composite whose matrix was obtained from methylhydrosilazane. The fracture strength and the stress/ strain behaviour were measured by using a hydraulic internal pressurization test. Hys- teresis during cyclic loading and pseudo-elastic behavior were observed in the composite part, and similar beha vior was found during mechanical tests of small speci mens. The accuracy of the calculated strength of the part was relatively low. Nevertheless, the composite part showed the same level strength, 180 MPa, as the test cno Fig. 13. Fracture surface of inner scroll support This work was conducted by the Petroleum Energ Center with a financial support from the Ministry of the elastic moduli were 122 and 114 GPa, respectively. International Trade and Industry. The authors The fracture strength of the inner scroll support was 180 acknowledge Mr. Takao Izumi at the Japan Auto- MPa. The value was higher than the that of the 45 off mobile Research Institute for enforcement of the axis test piece, which had a similar fiber structure. The hydraulic internal pressurization test differences in porosity and fiber content between the part and the test pieces were small (Table 2). Tow pos- sible reasons for the strength difference between the test References coupons and the part are: (i) the detailed difference in the fiber structures (ii) error in the calculated stress on [] Naslain R, Langlais F Mater Sci Res 1986: 20: 14 ne part because the anisotropic elasticity of the compo- [2 Walker BE, Rice RE, Becher PE, Bender BA, Coblenz ws site was disregarded. By observing the fiber structure of eram Bull 1983: 62: 916 the inner scroll support, the extension of the direction 3 Corbin ND, Rossetti GA, Hartline SD. Ceram Eng Sci Proc 1986;7:958 close to the latitudinal direction was found, i. e. the angle (Newkirk MS, Lesher HD, White DR, Kennedy CR, Urgha between the fiber tows and the hoop direction changed Aw, Clarr TD. Ceram Eng Sci Proc 1987: 18: 879. from 45 to 40(Fig. 11). This extension was caused by Schwab ST, Page RA, Davidson DL, Graef RC, Tredway WK stacking the prepreg sheets onto the curved surface of eram Eng Sci Proc 1995: 16: 743. the shaping jig. The latitudinal orientation of the fiber [6 Miller DV, Pommel DL, Schiroky GH. Ceram Eng Sci Proc 1997;18:409 caused the increase in the fracture strength of the part [7 Inerrante Lv, Jacobs JM, Sherwood W, Whitmarsh Cw. Key The fractured composite part is shown in Fig. 12 and Eng mater I997;127-131:271 the fracture surface is shown in Fig. 13. If a monolithic [8 Duran A, Aparicio M, Rebstock K, Vogel WD. Key Eng Mater ceramic part is broken by the hydraulic internal pres- 97:127-131:287 9 Sato K, Suzuki T, Funayama O, Isoda T. Ceram Eng Sci Proc surization test, the test body is separated into more than 1992:13:614 2 pieces. The absence of separation after the test is a [10] Morozumi H, Sato K, Tezuka A, Kaya H, Isoda T.Ceramic distinctive feature of parts made with continuous fiber International 1997- 23: 179 reinforced ceramics. On the fractured surface, many [I Fohey WR, Battison JM, Nielsen TA, Hastings S Ceram Eng Sci long pulled-out fibers were observed, as also seen on the [12] Kochendorfer R, Krenkel W. High-temperature ceramic-matrix test pieces omposites I. Ceramic Transactions 1995: 57: 13 [ Kaya H, Izumi T. Proceeding of ASME TRUBO EXPO 96, Birmingham in UK, 1996. June 10-13. 96-GT-348 4. Conclusion [14 Funayama O, Kato T, Tashiro Y, Isoda T. J Am Ceram Soc 1993:76:717 [5 Aubard x, Lamon J, Allix O. J Am Ceram Soc 1994: 77: 218. One of the components of the gas-turbine, the inner [16] Mizuno M, Zuh S, Nagano Y, Sakaida Y,Kagawa Y,Watanabe scroll support, was successfully fabricated by a Si-C-O M. J Am Ceram Soc 1996: 7913: 3065
the elastic moduli were 122 and 114 GPa, respectively. The fracture strength of the inner scroll support was 180 MPa. The value was higher than the that of the 45 o axis test piece, which had a similar ®ber structure. The dierences in porosity and ®ber content between the part and the test pieces were small (Table 2). Tow possible reasons for the strength dierence between the test coupons and the part are: (i) the detailed dierence in the ®ber structures (ii) error in the calculated stress on the part because the anisotropic elasticity of the composite was disregarded. By observing the ®ber structure of the inner scroll support, the extension of the direction close to the latitudinal direction was found, i.e. the angle between the ®ber tows and the hoop direction changed from 45 to 40 (Fig. 11). This extension was caused by stacking the prepreg sheets onto the curved surface of the shaping jig. The latitudinal orientation of the ®ber caused the increase in the fracture strength of the part. The fractured composite part is shown in Fig. 12 and the fracture surface is shown in Fig. 13. If a monolithic ceramic part is broken by the hydraulic internal pressurization test, the test body is separated into more than 2 pieces. The absence of separation after the test is a distinctive feature of parts made with continuous ®berreinforced ceramics. On the fractured surface, many long pulled-out ®bers were observed, as also seen on the test pieces. 4. Conclusion One of the components of the gas-turbine, the inner scroll support, was successfully fabricated by a SiÿCÿO ®ber-reinforced SiÿNÿC composite whose matrix was obtained from methylhydrosilazane. The fracture strength and the stress/strain behaviour were measured by using a hydraulic internal pressurization test. Hysteresis during cyclic loading and pseudo-elastic behavior were observed in the composite part, and similar behavior was found during mechanical tests of small specimens. The accuracy of the calculated strength of the part was relatively low. Nevertheless, the composite part showed the same level strength, 180 MPa, as the test pieces. Acknowledgements This work was conducted by the Petroleum Energy Center with a ®nancial support from the Ministry of International Trade and Industry. The authors acknowledge Mr. Takao Izumi at the Japan Automobile Research Institute for enforcement of the hydraulic internal pressurization test. References [1] Naslain R, Langlais F. Mater Sci Res 1986;20:145. [2] Walker BE, Rice RE, Becher PE, Bender BA, Coblenz WS. Ceram Bull 1983;62:916. [3] Corbin ND, Rossetti GA, Hartline SD. Ceram Eng Sci Proc 1986;7:958. [4] Newkirk MS, Lesher HD, White DR, Kennedy CR, Urqhart AW, Clarr TD. Ceram Eng Sci Proc 1987;18:879. [5] Schwab ST, Page RA, Davidson DL, Graef RC, Tredway WK. Ceram Eng Sci Proc 1995;16:743. [6] Miller DV, Pommell DL, Schiroky GH. Ceram Eng Sci Proc 1997;18:409. [7] Inerrante LV, Jacobs JM, Sherwood W, Whitmarsh CW. Key Eng Mater 1997;127±131:271. [8] Duran A, Aparicio M, Rebstock K, Vogel WD. Key Eng Mater 1997;127±131:287. [9] Sato K, Suzuki T, Funayama O, Isoda T. Ceram Eng Sci Proc 1992;13:614. [10] Morozumi H, Sato K, Tezuka A, Kaya H, Isoda T. Ceramic International 1997;23:179. [11] Fohey WR, Battison JM, Nielsen TA, Hastings S. Ceram Eng Sci Proc 1995;16:459. [12] Kochendorfer R, Krenkel W. High-temperature ceramic-matrix composites I, Ceramic Transactions 1995;57:13. [13] Kaya H, Izumi T. Proceeding of ASME TRUBO EXPO `96, Birmingham in UK, 1996, June 10±13, 96-GT-348. [14] Funayama O, Kato T, Tashiro Y, Isoda T. J Am Ceram Soc 1993;76:717. [15] Aubard X, Lamon J, Allix O. J Am Ceram Soc 1994;77:218. [16] Mizuno M, Zuh S, Nagano Y, Sakaida Y, Kagawa Y, Watanabe M. J Am Ceram Soc 1996;7913:3065. Fig. 13. Fracture surface of inner scroll support. K. Sato et al. / Composites Science and Technology 59 (1999) 853±859 859