Análisis de sensibilidad paramétrica sobre un modelo computacional XFEM para la propagación de grietas en una probeta CT de acero de fase dual

The implementation of the extended finite element method (XFEM) has allowed modeling without increasing computational costs and minimizing the use of excessively fine meshes, discontinuities, complex geometric singularities, crack propagation processes, among others. Being a recent method, the effec...

Full description

Autores:: Romero Rodríguez, Daniel Steven

Tipo de recurso:: Work document

Fecha de publicación:: 2020

Institución:: Universidad Nacional de Colombia

Repositorio:: Universidad Nacional de Colombia

Idioma:: spa

Description
Summary:	The implementation of the extended finite element method (XFEM) has allowed modeling without increasing computational costs and minimizing the use of excessively fine meshes, discontinuities, complex geometric singularities, crack propagation processes, among others. Being a recent method, the effects of the input parameters on the response variables of the model are unknown. For this reason, the sensitivity analysis is relevant to determine the effect of various parameters involved in the modeling of crack propagation. The objective of this work is to carry out a parametric sensitivity analysis for an XFEM computational model of the propagation of cracks in a test piece for fracture toughness of the compact tension (CT) type made of DP600 steel, minimizing the number of nodes to 1000, for which five parameters are selected that vary in two conditions each. The selected parameters are the model dimensions, finite element size, material behavior, evolution, and damage tolerance in the material. As response variables, the Force vs. Displacement diagram and the crack trajectory were used. For statistical analysis, the Pareto graph, main effect graphs, and an analysis of variance were used, to determine the parameter with the most significant effect in the models. To quantify the error of the two model responses for the experimental data, a relative error was used between the strength of the computational model for the experimental results and a mean squared error (RMSE), totaling the error in only one for the Force vs. Displacement, in the case of two-dimensional models the average RMS errors were 46.52 %, while the three-dimensional models were closer to the experimental data with average RMS errors of 2.20 %. In the case of crack growth, RMS errors below 10 % were obtained. Through the statistical analyzes carried out, a high statistical significance was established in the parameter of the model type, being the main one for the decrease in the error calculated for experimental cases.

Análisis de sensibilidad paramétrica sobre un modelo computacional XFEM para la propagación de grietas en una probeta CT de acero de fase dual

Publicaciones similares