A novel sensitivity index for analyzing the response of numerical models with interval inputs

Qi Chang, Changcong Zhou, Marcos A. Valdebenito, Hongwei Liu, Zhufeng Yue

Research output: Contribution to journalArticlepeer-review

Abstract

This study proposes a novel sensitivity index to provide essential insights into numerical models whose inputs are characterized by intervals. Based on the interval model and its normalized form, the interval processes are introduced to define a new sensitivity index. The index can represent the individual or joint influence of the interval inputs on the output of a considered model. A double-loop strategy, based on global metamodeling and optimization, is established to calculate the index. Subsequently, the proposed index is theoretically compared with two other existing indices, and it is experimentally applied to three numerical examples and a practical engineering problem of a honeycomb sandwich radome. The results indicate that the proposed index is an effective tool for interval sensitivity analysis.

Original languageEnglish
Article number115509
JournalComputer Methods in Applied Mechanics and Engineering
Volume400
DOIs
StatePublished - 1 Oct 2022
Externally publishedYes

Keywords

  • Global metamodeling
  • Interval
  • Model
  • Sensitivity analysis
  • Uncertainty

Fingerprint

Dive into the research topics of 'A novel sensitivity index for analyzing the response of numerical models with interval inputs'. Together they form a unique fingerprint.

Cite this