skip to main content
US FlagAn official website of the United States government
dot gov icon
Official websites use .gov
A .gov website belongs to an official government organization in the United States.
https lock icon
Secure .gov websites use HTTPS
A lock ( lock ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites.

Explore Research Products in the PAR
It may take a few hours for recently added research products to appear in PAR search results.


Title: Digital rheometer twins: Learning the hidden rheology of complex fluids through rheology-informed graph neural networks
Precise and reliable prediction of soft and structured materials’ behavior under flowing conditions is of great interest to academics and industrial researchers alike. The classical route to achieving this goal is to construct constitutive relations that, through simplifying assumptions, approximate the time- and rate-dependent stress response of a complex fluid to an imposed deformation. The parameters of these simplified models are then identified by suitable rheological testing. The accuracy of each model is limited by the assumptions made in its construction, and, to a lesser extent, the ability to determine numerical values of parameters from the experimental data. In this work, we leverage advances in machine learning methodologies to construct rheology-informed graph neural networks (RhiGNets) that are capable of learning the hidden rheology of a complex fluid through a limited number of experiments. A multifidelity approach is then taken to combine limited additional experimental data with the RhiGNet predictions to develop “digital rheometers” that can be used in place of a physical instrument.  more » « less
Award ID(s):
1847389
PAR ID:
10408787
Author(s) / Creator(s):
; ; ;
Date Published:
Journal Name:
Proceedings of the National Academy of Sciences
Volume:
119
Issue:
20
ISSN:
0027-8424
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. ; ; ; (, Proceedings of the National Academy of Sciences)
    The ability to concisely describe the dynamical behavior of soft materials through closed-form constitutive relations holds the key to accelerated and informed design of materials and processes. The conventional approach is to construct constitutive relations through simplifying assumptions and approximating the time- and rate-dependent stress response of a complex fluid to an imposed deformation. While traditional frameworks have been foundational to our current understanding of soft materials, they often face a twofold existential limitation: i) Constructed on ideal and generalized assumptions, precise recovery of material-specific details is usually serendipitous, if possible, and ii) inherent biases that are involved by making those assumptions commonly come at the cost of new physical insight. This work introduces an approach by leveraging recent advances in scientific machine learning methodologies to discover the governing constitutive equation from experimental data for complex fluids. Our rheology-informed neural network framework is found capable of learning the hidden rheology of a complex fluid through a limited number of experiments. This is followed by construction of an unbiased material-specific constitutive relation that accurately describes a wide range of bulk dynamical behavior of the material. While extremely efficient in closed-form model discovery for a real-world complex system, the model also provides insight into the underpinning physics of the material. 
    more » « less
  2. ; ; (, Journal of Geophysical Research: Solid Earth)
    Abstract Following the reanalysis of individual experimental runs of some widely cited studies (Jain et al., 2018,https://doi.org/10.1002/2017JB014847), we revisit the global data analysis of Korenaga and Karato (2008,https://doi.org/10.1029/2007JB005100) with a significantly improved version of their Markov chain Monte Carlo inversion. Their algorithm, previously corrected by Mullet et al. () to minimize potential parameter bias, is further modified here to estimate more efficiently interrun biases in global data sets. Using the refined Markov chain Monte Carlo inversion technique, we simultaneously analyze experimental data on the deformation of olivine aggregates compiled from different studies. Realistic composite rheological models, including both diffusion and dislocation creep, are adopted, and the role of dislocation‐accommodated grain boundary sliding is also investigated. Furthermore, the influence of interrun biases on inversion results is studied using experimental and synthetic data. Our analysis shows that existing data can tightly constrain the grain‐size exponent for diffusion creep at ∼2, which is different from the value commonly assumed (p= 3). Different data sets and model assumptions, however, yield nonoverlapping estimates on other flow‐law parameters, and the flow‐law parameters for grain boundary sliding are poorly resolved in most cases. We thus provide a few plausible candidate flow‐law models for olivine rheology to facilitate future geodynamic modeling. The availability of more data that explore a wider range of experimental conditions, especially higher pressures, is essential to improve our understanding of upper mantle rheology. 
    more » « less
  3. (, Rheologica Acta)
    Coussot, P. (Ed.)
    Developments in the last century, and especially in the last 50 years, have advanced understanding of suspension rheology greatly. Here, a limited review of suspension work over this period is presented, emphasizing advances over the last three decades in understanding of the particle pressure and strong shear thickening, which were motivated by crucial experimental observations, computational advances, and a critical review, all from the 1980s. This review serves as a preview to some outstanding challenges in suspension mechanics. This article considers primarily dispersions of spherical particles, which serve not only as a model material for understanding the rheology of more complex fluids of practical relevance, but also as a basic system for the study of nonequilibrium statistical physics. 
    more » « less
  4. ; ; (, Journal of Rheology)
    Empirical rules play a crucial role in industrial and experimental settings for efficiently determining the rheological properties of materials, thereby saving both time and resources. An example is the Cox–Merz rule, which equates the steady-shear viscosity with the magnitude of the complex viscosity obtained in oscillatory tests. This empirical rule provides access to the steady-shear viscosity that is useful for processing conditions without the instabilities associated with experiments at high shear rates. However, the Cox–Merz rule is empirical and has been shown to work in some cases and fail in others. The underlying connection between the different material functions remains phenomenological and the lack of a comprehensive understanding of the rheological physics allows for ambiguity to persist in the interpretation of material responses. In this work, we revisit the Cox–Merz rule using recovery rheology, which decomposes the strain into recoverable and unrecoverable components. When viewed through the lens of recovery rheology, it is clearly seen that the steady-shear viscosity comes from purely unrecoverable acquisition of strain, while the complex viscosity is defined in terms of contributions from both recoverable and unrecoverable components. With recovery tests in mind, we elucidate why the Cox–Merz rule works only in a limited set of conditions and present an approach that could allow for universal comparisons to be made. This work further highlights the significance of recovery rheology by showing how it is possible to extend beyond phenomenological approaches through clear rheophysical metrics obtained by decomposing the material response into recoverable and unrecoverable components. 
    more » « less
  5. ; ; ; (, Journal of Non-Newtonian Fluid Mechanics)
    Frigaard, Ian; Poole, Robert J (Ed.)
    We review a selection of models for wormlike micelles undergoing reptation and chain sequence rearrangement (e.g. reversible scission) and show that many different assumptions and approximations all produce similar predictions for linear rheology. Therefore, the inverse problem of extracting quantitative microscopic information from linear rheology data alone may be ill-posed without additional supporting data to specify the sequence rearrangement pathway. At the same time, qualitative parameter estimates can be obtained equally well from any of the models in question. Through our study, we also show that the Poisson renewal model can be reformulated as a differential constitutive equation on the tube survival prob- ability distribution function. Using this reformulation, we identify two previously overlooked inconsistencies with Poisson renewal and discuss how these can be resolved by re-interpreting what the model calls a `breaking time'. 
    more » « less
  • Citation Formats
  • MLA
  • APA
  • Chicago
  • BibTeX
  • Save / Share this Record
  • Send to Email