PSI - Issue 41

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

302

5

3. Monte Carlo simulations A Monte Carlo simulation procedure for the fatigue life assessment of a compressor stage was performed within a user-made MATLAB routine (MATLAB, 2021) implementing the logic illustrated in Figure 4. The routine considered as input data the variables presented in the previous section and provided as outputs the predictions on the residual fatigue lives estimated for the component. The procedure starts by reading the required input data, i.e. material parameters, experimental data, the K I vs. crack depth curves and the parameters defining the variations for geometrical tolerances and loading conditions. For each MC simulation, the routine starts by generating pseudorandom NASGRO curves until that a curve providing a good fit of the experimental data (Mom, et al., 1988) is obtained. The reader is referred to previous researches of some of the authors about a description of the procedure used to this aim. (Giannella, 2021a-b, 2022a-b) Consequently, each “good-fit” curve is associated with two random values of fillet radius r and rotational speed ω extracted according the corresponding probability distributions, see Figure 4. Finally, a fatigue life prediction is computed for each set of data and the procedure is re-iterated until the requested user-defined number of predictions to perform is reached. With such an approach, the distribution of the fatigue life predictions with allowance for these main sources of uncertainty was obtained, thus leading to a thorough assessment of the component life. For what concerning the fillet radius variability, a normal distribution was assumed having a nominal dimension of 1.5 mm as a mean value and ±1 mm as a potential range of variation, see Fig. 4. For what concerning the loading conditions variability, the probability distribution was defined by considering a normal distribution with the nominal speed as mean value (30000 rpm) and a ±15% of variation defined through a coefficient of variation of 0.06, see Figure 4. This distribution was set up according to literature, (Endres, 1992; Beretta et al., 2015) where overspeed of maximum 15% was defined for gas turbine engines.

Figure 4. Monte Carlo simulation strategy: each simulation consisting of a NASGRO curve that fits test data, a rotational speed value and a fillet radius value extracted from their normal distributions 4. Results and discussion The presented framework for fatigue life assessment of an aircraft engine component accounted for the uncertainty coming from material scattering, the main geometrical tolerances and the loading conditions of the component. After performing 1e5 Monte Carlo simulations, a probabilistic fatigue life assessment of the cracked