M. Holmquist et al. Journal of the European Ceramic Society 20(2000)599-606 Table I as anisotropic but temperature invariant(T=1000.C hysical properties used in FE analysis of Alz O3/AlO3 composite tile The predictions gave a rough estimate of the tempera Temp(C)20 200 400 600 800 1000 1200 1400 ture distribution. The temperature gradients(both through thickness and in-plane) were somewhat under E1(GPa)15515415 7141133115100 estimated in the calculations, compared to the measure- 15515415 7141133115100 110110107104101948271 ments(Fig. The principal strain distribution is 2121 shown in Fig. Il. As can be seen, the maximum princi- 10 0.10 0.10 0.10 0.10 0.10 pal strain predicted for the tile is 0.329% occurring 0070.070070.070.07 around the air dilution hole. This strain has its major a1×10-65.306.707.507.858.108.308.508 5.30 6.70 7.50 7.85 8.10 8.30 8.50 8.70 component in the through thickness direction There are also high strains along the edges(oriented parallel to the i(W/mK)13.527.145.254.724864.955.015.01 edges) of the tile. At these locations the strains are 4.86 4.95 5.015.01 higher than the matrix cracking strain(0.059%)and 3. 13 6.93 5.10 4.58 4.72 4.81 4.864.86 localised matrix cracking is anticipated to occur as was p(/gK)0781021.141.201241.261.27128 also observed p(g/cm2)34 Fibre lay-up: [0/90]ss(i.e. 16 layers, symmetric, balanced 0.90 lay- up), assume orthotropic material properties 5. Conclusions In the future gas turbines will have to burn fuel more result in localised matrix cracking, which should relieve cleanly and efficiently for both economic and environ- stress. A non-linear (2 component linear)FE analysis mental reasons. In order to optimise the combustion was performed to allow a more accurate view of the process, minimising the generation of NOx, CO and damage accumulated by the component over a thermal unburned hydrocarbons, it will be necessary to avoid cycle. The physical properties in Table I were used with film cooling of the combustor walls. Ceramic matrix some additions. The material was modelled as linear- composites are potential materials for these applica- elastic to the matrix cracking strain (0.059%). Upon tions. All-oxide composites are inherently stable in oxi further loading the material was supposed to follow a dising environments at the required temperatures and linear behaviour but with a substantially reduced stiff- offer an advantage over currently commercially avail- ness(from 155 to 5.1 GPa). The matrix cracking strain able non-oxide ceramic composites was assumed to be temperature invariant. These The present investigation has shown that an alumina assumptions were based on results from tensile testing single crystal fibre reinforced alumina matrix is a viable as described previously. Due to restrictions in the concept when a porous zirconia coating allows fibre/ ANSYS code the material properties had to be modelled matrix interfacial decohesion and fibre sliding up matrix cracking. Microstructural observations of pulled-out fibres on fracture surfaces indicated that the load transfer mechanism is based on a wear effect of the porous zirconia interphase. In this process individual zirconia grains are formed which will act as ball-bear ings between the fibre and the matrix. A simple in-line slurry coating technique was used to make the zirconia interphase. It was incorporated into a tape casting pro cess to make green prepregs that were cut, stacked and Principal hot pressed. Both unidirectional and cross-ply compo- strain(%) site plates with dimensions up to 180x200 mm were made in this way. Cross-ply composites had ultimate 0005 ensile strengths of 110 MPa. The strain to failure (0.45%)is comparable to other conventional CMCs with large amounts of fibre pull-out. A loss of proper ties with increasing temperatures was observed; at 800C the strength decreased with 40% compared to the initial UTS. The Uts did not change between 1200 and 1400C with 50% of RT strength being retained The length of pulled-out fibres decreased as a function of the temperature in accordance with the decrease of the sap phire fibre strength Composite properties were retained Fig. 11. Predicted strain distribution for hot side of an AlO3/Al2O mposite tile.result in localised matrix cracking, which should relieve stress. A non-linear (2 component linear) FE analysis was performed to allow a more accurate view of the damage accumulated by the component over a thermal cycle. The physical properties in Table 1 were used with some additions. The material was modelled as linear￾elastic to the matrix cracking strain (0.059%). Upon further loading the material was supposed to follow a linear behaviour but with a substantially reduced sti€- ness (from 155 to 5.1 GPa). The matrix cracking strain was assumed to be temperature invariant. These assumptions were based on results from tensile testing as described previously. Due to restrictions in the ANSYS code the material properties had to be modelled as anisotropic but temperature invariant (T=1000C). The predictions gave a rough estimate of the tempera￾ture distribution. The temperature gradients (both through thickness and in-plane) were somewhat under￾estimated in the calculations, compared to the measure￾ments (Fig. 10). The principal strain distribution is shown in Fig. 11. As can be seen, the maximum princi￾pal strain predicted for the tile is 0.329% occurring around the air dilution hole. This strain has its major component in the through thickness direction. There are also high strains along the edges (oriented parallel to the edges) of the tile. At these locations the strains are higher than the matrix cracking strain (0.059%) and localised matrix cracking is anticipated to occur as was also observed. 5. Conclusions In the future, gas turbines will have to burn fuel more cleanly and eciently for both economic and environ￾mental reasons. In order to optimise the combustion process, minimising the generation of NOx, CO and unburned hydrocarbons, it will be necessary to avoid ®lm cooling of the combustor walls. Ceramic matrix composites are potential materials for these applica￾tions. All-oxide composites are inherently stable in oxi￾dising environments at the required temperatures and o€er an advantage over currently commercially avail￾able non-oxide ceramic composites. The present investigation has shown that an alumina single crystal ®bre reinforced alumina matrix is a viable concept when a porous zirconia coating allows ®bre/ matrix interfacial decohesion and ®bre sliding upon matrix cracking. Microstructural observations of pulled-out ®bres on fracture surfaces indicated that the load transfer mechanism is based on a wear e€ect of the porous zirconia interphase. In this process individual zirconia grains are formed which will act as ball-bear￾ings between the ®bre and the matrix. A simple in-line slurry coating technique was used to make the zirconia interphase. It was incorporated into a tape casting pro￾cess to make green prepregs that were cut, stacked and hot pressed. Both unidirectional and cross-ply compo￾site plates with dimensions up to 180200 mm2 were made in this way. Cross-ply composites had ultimate tensile strengths of 110 MPa. The strain to failure (0.45%) is comparable to other conventional CMCs with large amounts of ®bre pull-out. A loss of proper￾ties with increasing temperatures was observed; at 800C the strength decreased with 40% compared to the initial UTS. The UTS did not change between 1200 and 1400C with 50% of RT strength being retained. The length of pulled-out ®bres decreased as a function of the temperature in accordance with the decrease of the sap￾phire ®bre strength. Composite properties were retained Table 1 Physical properties used in FE analysis of Al2O3/Al2O3 composite tilea Temp (C) 20 200 400 600 800 1000 1200 1400 E1 (GPa) 155 154 151 147 141 133 115 100 E2 155 154 151 147 141 133 115 100 E3 110 110 107 104 101 94 82 71 1221 0.07 0.07 0.07 0.07 0.07 0.07 0.07 0.07 1323 0.10 0.10 0.10 0.10 0.10 0.10 0.10 0.10 3132 0.07 0.07 0.07 0.07 0.07 0.07 0.07 0.07 1 10ÿ6 5.30 6.70 7.50 7.85 8.10 8.30 8.50 8.70 2 5.30 6.70 7.50 7.85 8.10 8.30 8.50 8.70 3 5.30 6.70 7.50 7.85 8.10 8.30 8.50 8.70 l1 (W/mK) 13.52 7.14 5.25 4.72 4.86 4.95 5.01 5.01 l2 13.52 7.14 5.25 4.72 4.86 4.95 5.01 5.01 l3 13.13 6.93 5.10 4.58 4.72 4.81 4.86 4.86 cp (J/gK) 0.78 1.02 1.14 1.20 1.24 1.26 1.27 1.28  (g/cm3 ) 3.45 a Fibre lay-up: [0/90]8,s (i.e. 16 layers, symmetric, balanced 0.90 lay￾up), assume orthotropic material properties. Fig. 11. Predicted strain distribution for hot side of an Al2O3/Al2O3 composite tile. M. Holmquist et al. / Journal of the European Ceramic Society 20 (2000) 599±606 605