PSI - Issue 41

Venanzio Giannella et al. / Procedia Structural Integrity 41 (2022) 298–304 V. Giannella / Structural Integrity Procedia 00 (2022) 000–000

299

2

83444; 1971). By safely adding damage tolerant lifing to the component design process, substantial cost savings can be realized in the form of extended component lives. Namely, where Non-Destructive Testing (NDT) inspections reveals the presence of flaws, different options can be applied based on the combinations of material, fabrication procedure, stress, environmental factors, etc. As instance, when the flaws do not exceed the quality control levels in the appropriate application standard, no further actions can be even required. Rotor discs for aircraft engines are heavy components usually designed on Low-Cycle Fatigue (LCF) (Corran et al.; 2007; Muehle et al., 1990) adopting design curves with suitable probabilistic margins. (Beretta et al., 2014; Zhu et al., 2018) As a matter of fact, when the accuracy of the input information employed (e.g. stress levels, material properties at the appropriate temperature, flaw sizes information, etc.) is in question, appropriate additional safety factors need to be agreed. (Giannella, 2021a-b). This option of using appropriate safety factors has been incorporated or is inherent throughout all standards nowadays. To these aims, it is generally worth adopting probabilistic approaches (Armentani et al., 2020) in order to avoid any unnecessary design conservatism, for describing the whole component life under the different sources of scatter and for obtaining reliable failure probabilities. The aim of this paper is to present the application of a straightforward procedure for the damage tolerance assessment of an aircraft compressor under input data uncertainty. Such a procedure is based on the numerical simulation, to assess the fracture behaviour of the component during the advancing of a crack, whose outputs were inputted to a Monte Carlo simulation procedure, to eventually predict the component life in a statistical way. A schematic of this procedure was reported in Fig. 1. The main sources of uncertainty/variability that were considered through this procedure were: i) the most relevant geometrical tolerances, ii) the variability of the actual loading conditions, iii) the inherent scattering of material data. These are commonly considered as the primary contributes affecting a structural integrity assessment (Zhu et al., 2018; Niu et al., 2021; Patriarca et al., 2019; Giannella, 2022a). Firstly, the geometrical tolerance that was considered as having the most relevant influence on the crack propagation was considered as being the fillet radius between blades and disk. Multiple crack propagation simulations were computed with the commercial code FRANC3D (FRANC3D, 2021) by varying the fillet radius from the nominal value for each propagation simulation. Secondly, material scattering was considered through the application of the numerical framework already proposed in other works by some of the Authors (Giannella, 2021a b, 2022a-b). Namely, material scattering was simulated through a series of NASGRO crack-growth laws that were derived based on the experimental data (Mom, et al., 1988) so as to cover the whole range of variability shown by test data. Thirdly, the variability of the actual loading conditions was considered according to Eq. 1, i.e. K values were linearly increased/decreased according to � and a range of variability suggested by literature (Endres, 1992; Beretta et al., 2015). Finally, all the input parameters were used to feed a Monte Carlo (MC) simulation procedure to derive the final life predictions for the component by considering these three sources of uncertainty. Statistical evaluations were made and the contributes of all the considered uncertainty sources were compared and classified. 2. Numerical simulations A one-ninth section of the whole compressor stage was modelled by FEM. The corresponding CAD model was shown in Figure 1. Boundary conditions consisted of cyclic symmetry conditions at the cutting surfaces, whereas a rotational body load was applied to the whole component to simulate the rotational speed (nominal working speed was set to 30000 rpm). Nearly 80k quadratic tetrahedral elements of type C3D10 were used for the modelling, even though this number increased/decreased depending on dimensions of crack and fillet radius. Multiple models were built up by considering different fillet radii at the interconnection between blades and disk, see Figure 2. An example of the resulting maximum principal stress fields was reported in Figure 2, where it can be noticed the position of the most stressed area, subsequently considered as the crack initiation point. It is worth noting that the size of the fillet radius turned out to produce a relevant effect on the maximum principal stresses reached at the root of the blades. In ref K  K    2 ref        (1)