NHERI Computational Symposium

February 5-7, 2025

Software Spotlight: quoFEM - Enabling Easy and Efficient UQ Analysis for Simulation Models

Session 4D Hosts: Aakash Bangalore Satish & Sang-ri Yi
 

 


This session introduces SimCenter’s quoFEM application, a tool designed to enable Uncertainty Quantification (UQ) analyses for computational simulation models. Through its graphical interface, quoFEM guides users through four steps in setting up UQ analyses: selecting a UQ algorithm and options, specifying model scripts, defining input probability distributions, and identifying model outputs.

With quoFEM, users can propagate uncertainties in both model parameters and the models themselves, evaluating their impact on quantities of interest. If data is available, quoFEM also supports updating these uncertainty distributions. To handle the computational demands of UQ analyses, quoFEM enables seamless job execution on high-performance computing resources through DesignSafe, accessible directly from the user’s local computer.

This session will showcase a variety of UQ analysis options within quoFEM, including global sensitivity analysis, surrogate modeling, multi-fidelity Monte Carlo simulation, and surrogate-aided Bayesian calibration for computationally expensive models through examples to demonstrate opportunities for adapting uncertainty quantification into active research and proposal ideas.