Sachin Shanbhag
Dept. of Scientific Computing,
Florida State University
"Accelerated Discovery of Constitutive Models for Soft Matter using Spectral Methods"
Wednesday, Dec 4, 2024
Abstract:
Constitutive models (CMs) mathematically describe the mechanical response of soft materials like polymers, colloids, gels, etc., to deformation. These CMs can be incorporated into computational fluid dynamics software to predict flows in complex geometries, thereby accelerating the design and scale-up of new processes and products. In this talk, a data-driven algorithm to discover parsimonious CMs from oscillatory shear rheology is proposed and applied to synthetic data generated from popular nonlinear CMs for polymers. The algorithm relies on three key elements: (i) a spectral method called FLASH which efficiently computes the oscillatory shear response of arbitrary CMs, (ii) representation of unknown nonlinear terms using tensor basis functions which incorporates physical constraints of symmetry and frame-invariance by design, and (iii) sparse polynomial regression which encodes our preference for simple interpretable models. Implications for experiment design, algorithmic improvement, and non-uniqueness of inferred CMs are discussed.
| File | Description | File size |
|---|---|---|
 Sachin Shanbhag Research.png | Research | 77 kB |
