| About this Abstract |
| Meeting |
TMS Specialty Congress 2025
|
| Symposium
|
8th World Congress on Integrated Computational Materials Engineering (ICME 2025)
|
| Presentation Title |
Enhancing Robust Design Using Improved Variance Estimation in Design Capability Index |
| Author(s) |
Pooja Mukundan, H M Dilshad Alam Digonta, Mathew Baby, Anand Balu Nellippallil |
| On-Site Speaker (Planned) |
Pooja Mukundan |
| Abstract Scope |
The simulation-based design of material systems involves uncertainties in design variables that must be managed. One approach for managing design variable uncertainties involves using the Design Capability Index (DCI) metric for Robust Design, which allows designers to identify robust solutions that are relatively insensitive to uncertainties. In DCI, the variance of responses due to uncertainties is computed using the first-order Taylor Series expansion, which fails to eliminate ‘uncertainty sensitive’ maxima/minima points from the robust solution space. First-order Taylor Series Expansion will also potentially result in errors for non-linear/multimodal response functions.
To address these drawbacks, we explore alternative variance estimation methods – the second derivative, multiple derivative, and multiple-point methods, in the DCI metric. We employ the hot rod rolling problem to explore the effect of these alternate methods on DCI. This exploration is facilitated using the machine learning-based visualization technique - interpretable Self-Organizing Map (iSOM). |
| Proceedings Inclusion? |
Undecided |