Naslov
Stochastic identification of parameters in lattice fracture model subjected to mechanical loads

Autori
Nikolić, Mijo ; Stanić, Andjelka ; Friedman, Noemi ; Matthies, Hermann G.

Vrsta, podvrsta i kategorija rada
Sažeci sa skupova, sažetak, znanstveni

Izvornik
ESMC 2022, 11th European Solid Mechanics Conference / - , 2022, 1-1

Skup
ESMC 2022, 11th European Solid Mechanics Conference

Mjesto i datum
Galway, Irska, 04.07.2022. - 08.07.2022

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
lattice model ; stochastic identification ; MCMC

Sažetak
Identification of unknown parameters in lattice fracture model is performed with the stochastic Bayesian inverse method [1]. This is important in computations when we do not have reliable input parameters or they cannot be obtained directly from the experiments. Some model parameters can also be efficiently identified. In this work we use the stochastic Markov Chain Monte Carlo (MCMC) method for identification of mechanical and model parameters of the lattice fracture model [2], where quadrilateral Q6 element from [3] is used as a reference solution. Polynomial chaos expansion is used for creating a surrogate model in this work. Lattice fracture model is based on the embedded strong discontinuity which is positioned inside the lattice elements to represent progressive propagation of discontinuities and material failure. The main strength of the failure model is efficiency in terms of computational time and correct representation of the discontinuity with fracture energy as input parameter. The embedded strong discontinuity model provides a mesh independent response in material softening which leads to fracture [4]. Due to this, we can identify the fracture energy parameter, which is not dependent on the mesh. Computational efficiency of the model allows us to compute multiple simulations required for the stochastic MCMC method.

Izvorni jezik
Engleski

Znanstvena područja
Građevinarstvo

POVEZANOST RADA