Material Model Parameter Identification via Markov Chain Monte Carlo

C Knipprath, AA Skordos

Research output: Chapter in Book/Report/Conference proceedingConference Contribution (Conference Proceeding)

Abstract

This paper introduces a method to identify parameters for a material model by solving an inverse problem. In conventional methods the interpretation of the experiments is based on simple analytical formulae allowing the determination of material parameters [1] while the final value used in an FE simulation is determined as the average value obtained from the experimental results. The Markov Chain Monte Carlo (MCMC) method is a tool for the solution of inverse problems, which are often subject to ill-posedness. However, the MCMC method is conceptually able to overcome this problem. It was originally developed in the field of statistical physics [2] before it was successfully applied in other fields. For the demonstration of this method, experimental data was generated from a laminate under quasi- static conditions. The material models used in the context of finite element analysis are (a) a continuum damage model (based on a thermodynamic framework [3]) to cover the in-plane damage behaviour and (b) a cohesive interface model [4] for the representation of delamination effects in the interface. For comparison the identification of the material parameters was also undertaken using the conventional method as described in the model related literature. (continued)
Translated title of the contributionMaterial Model Parameter Identification via Markov Chain Monte Carlo
Original languageEnglish
Title of host publication5th International Conference on Composites Testing and Model Identification, Lausanne, Switzerland, February 14-16, 2011
Publication statusPublished - Feb 2011

Bibliographical note

Conference Organiser: Laboratory of Applied Mechanics and Reliability Analysis (LMAF)

Fingerprint

Dive into the research topics of 'Material Model Parameter Identification via Markov Chain Monte Carlo'. Together they form a unique fingerprint.

Cite this