Activity Number:
|
522
- Contributed Poster Presentations: Section on Physical and Engineering Sciences
|
Type:
|
Contributed
|
Date/Time:
|
Wednesday, August 2, 2017 : 10:30 AM to 12:20 PM
|
Sponsor:
|
Section on Physical and Engineering Sciences
|
Abstract #323193
|
|
Title:
|
Mixed Effects Material Strength Models with Model Discrepancy
|
Author(s):
|
Jason Bernstein* and Kathleeen Schmidt and Ana Kupresanin and Jeff Florando and Nathan Barton
|
Companies:
|
and Lawrence Livermore National Laboratory and Lawrence Livermore National Laboratory and Lawrence Livermore National Laboratory and Lawrence Livermore National Laboratory
|
Keywords:
|
uncertainty quantification ;
mixed effects model ;
Gaussian process ;
materials science ;
engineering ;
model discrepancy
|
Abstract:
|
Engineering efforts often require knowledge of how a material deforms under application of an external force. In a compressive experiment, a sample of a material is subject to different levels of compressive force and the resulting change in length of the sample is measured. This procedure is generally applied to multiple samples under different experimental conditions, e.g., various combinations of temperature and strain rate, and conclusions made from fitting a fixed-effects material strength model to the complete data set. I will discuss an alternative strength model that allows for variability of the model parameters between samples with mixed effects and represents model discrepancy with a Gaussian process. A Markov Chain Monte Carlo (MCMC) algorithm is employed to compute the posterior distribution of the parameters in the proposed model. The two models are compared in a simulation study.
|
Authors who are presenting talks have a * after their name.