STOCHASTIC MODELING OF FATIGUE CRACK GROWTH IN 2024T-351 ALUMINUM ALLOY
Adriano Francisco Siqueira1; Cauê Pettermann Carvalho2; Viktor Pastoukhov2; Carlos Antonio R.P. Baptista2; Marcelo Augusto Santos Torres3
1 EEL/USP - Depto Ciências Básicas e Ambientais; 2 EEL/USP - Depto Engenharia de Materiais; 3 FEG/UNESP Depto de Mecânica
doi:10.20906/CPS/COB-2015-0934
Resumo
In this study, the stochastic simulation of fatigue crack propagation was performed for specimens made of a widely used structural aluminum alloy, specified as AA 2024 T351. The proposed modelling tool uses irregularity of crack advances observed in empirical studies as a key feature of the process. This approach is aimed at prediction of crack propagation curves scattering, usually observed in repeated tests under similar conditions. The model is based at stochastic differential equation that includes two terms. The first one represents the predominant trend in the process, and the second one simulates the "noise" of a random nature. The fitting constants obtained using only one experimental curve were applied for generation of fifty predictions of crack growth from the same initial length and under same cyclic loading. The simulated scatter was compared with two sets of experimental data collected from technical literature. One set consists of 30 specimens tested under same constant-amplitude loading and other of 25 specimens tested under random variable-amplitude loading. In order to generate the simulations, one experimental curve representing the average crack behavior was chosen, and from these data the model was able to reproduce well the real data scattering. However, random choice of basic experimental curve may produce less reliable simulations, including non-conservative predictions. Therefore, a discussion about minimum number of randomly chosen data necessary for realistic scattering prediction is included in this work. The development of stochastic methods of prediction helps to increase the efficiency of the experimental characterization of resistance to crack propagation, by reducing the number of repetitive tests.
Palavras-chave: Fatigue Crack Growth; Modeling; Stochastic Differential Equation; Aluminum Alloy