13th International Conference on Fracture June 16–21, 2013, Beijing, China -4- volve creep modelling of the bond coat. ܩ, ܭ ூ and ܭ ூூ have been computed for different h/L, different ߜ ୋ and continuously increasing crack lengths corresponding to damage Dm = a/L in the interval 0.0 < Dm < 1.0. See examples in Fig.5. Figure 5. Energy release rate G and stress intensity factors KI and KII computed for the TCF test. 4. Experimental details An experimental study has been conducted with the aim at giving quantitative data on crack propagation at a ceramic / metal interface. Evaluations are done on thermally cycled material and material exposed to engine conditions. The coatings evaluated are thin thermal barrier NiCoCrAlY coatings, with nominal thicknesses of top coat equal to 300µm and bond coat thickness of 150µm. Thermal cycling is performed in air with furnace heating and cooling by forced air. The furnace cycling consists of a standard cycle type where small coupons 30x50x5 mm are exposed to a thermal cycle with 60 minutes dwell time in the furnace at constant temperature. Heating and cooling parts of the cycle are 10 minutes respectively. Data have been retrieved for interrupted tests where the cycling has been stopped after approximately 30% of the expected fatigue. Ex-serviced material has been included with corresponding conditions, i.e. 30-40% of the expected service life. Coating infiltration by epoxy ensures that the exposed microstructure is retained and that the structure studied is not an artifact after sample preparation. After the epoxy infiltration, the material is cut and mounted for grinding and final polishing. All evaluations are done on as polished cross sections. Cracks at the top coat / bond coat interface are identified and measured. Measurements are registered for the general microstructure geometry in regions where isolated defects are detected, see Figure 6. In Figure 6 the typical microstructure TBC is shown (a). The interface is undulating and a least representative cell is not easily detected. Instead the morphology of individual undulations needs to be considered (b). The structure is necessarily not similar to the structure observed in the modeling approach (c). In general, the microstructure is assumed to be described with the height over width ratio as defined for modeling purposes.
RkJQdWJsaXNoZXIy MjM0NDE=