Identification of the parameters contained in a cyclic cohesive zone model for fatigue crack propagation

Cyclic cohesive zone models provide a useful tool to describe fatigue driven crack propagation, covering a wide range of engineering applications. For a proper use of these models, particular attention must be devoted to the correct calibration of the parameters contained, considering that some of these can be characterized by a large variability and/or by the absence of a precise physical meaning so that they are not amenable to a direct measurement. This paper proposes a robust inverse analysis procedure, to investigate the identifiability of the model parameters governing the fatigue induced damage evolution, in a recently proposed cyclic cohesive zone model. A novel control for compact test specimens providing more meaningful experimental information is proposed. The identification problem is formulated by considering fatigue crack propagation curve and deformations measurements in a discrete number of points of the specimen surface as input data of the inverse algorithm. The finite element operator, adopted to simulate the experimental tests, has been substituted by a proper calibrated meta model to reduce the computational cost of the forward operator and, thus, to solve the inverse problem in a stochastic context through Monte Carlo like procedures. Representative results, obtained starting from virtual data affected by different levels of noise, are reported to highlight the identifiability of the model parameters on the basis of the experimental data adopted. Indications regarding the minimum number of measurements needed to make the inverse problem well-posed are also provided, supporting possible planning of measurements setups for laboratory investigations.

Automated summary

Cyclic cohesive zone models are useful for describing fatigue driven crack propagation in various engineering applications. This paper proposes a robust inverse analysis procedure to investigate the identifiability of the model parameters governing fatigue induced damage evolution. The procedure utilizes a calibrated meta model to reduce computational cost and solves the inverse problem using Monte Carlo like procedures. The results highlight the identifiability of the model parameters and provide indications for measuring setups in laboratory investigations.