Digital rheometer twins: Learning the hidden rheology of complex fluids through rheology-informed graph neural networks

Mahmoudabadbozchelou, Mohammadamin; Kamani, Krutarth; Rogers, Simon A.; Jamali, Safa

doi:10.1073/pnas.2202234119

Cited by 26 publications

(11 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…More recently, physics-based ML algorithms, the so-called Physics-Informed Neural Network 23 (PINN), have been developed to reduce to diminish the need for big data sets by including governing physical laws in the ANN framework. This approach has then been extended to rheology by the Rheology-Informed Neural Networks (RhINNs) 24,25 which enables an accurate modeling of the rheological properties of a fluid with a limited number of experiments.…”

Section: Rheological Identification Of the Fluidmentioning

confidence: 99%

Rheological identification of jetted fluid using machine learning

et al. 2022

View full text Add to dashboard Cite

The understanding of flowing properties of fluids and the knowledge of the related rheological properties are crucial from both a research and industrial point of view. To determine the complex rheological properties of fluids many devices have thus been developed, the so-called rheometers.The main objective of the present proposal is to develop a new kind of rheometer usingboth large datasets and the Continuous Inkjet Printing (CIJ) process.When ejecting a fluid, the CIJ ejection process competes between several forces: inertial, viscous, surface tension and elasticity, which affect the morphology of the resulting jet.Also, under certain conditions, the morphology of the jet is unique and directly related to the rheological properties of the fluid.We want to use this uniqueness to identify the fluid among a large dataset of known fluid jet morphologies to be compared with, to obtain its rheological properties.Using a large numerically-generated dataset of Newtonian fluid jets, we show in this article that the identification of the viscosity using Neural Network (NN) is not only feasible, but proves to be very accurate with an average error of less than 1% for a large range of viscosities.

show abstract

Section: Rheological Identification Of the Fluidmentioning

confidence: 99%

Rheological identification of jetted fluid using machine learning

et al. 2022

View full text Add to dashboard Cite

show abstract

“…Specifically, in all cases for τ xx , τ yy , and τ zz , the a-Lasso has failed to identify the constant term in Eqs. ( 24) and (25), which is a possible source of larger errors compared to the STRidge. In the case of the STRidge, we confirmed that by choosing the appropriate α (α = 1 × 10 −1 ), nearly correct differential equations can be obtained, as shown in the upper part of Fig.…”

Section: Fene-p Dumbbell Modelmentioning

confidence: 99%

“…Some applications have successfully identified constitutive relations of complex fluids or governing equations to predict the dynamics of fluids with knowledge of rheology. These studies have employed neural networks (NN), including deep NN [24], graph NN [25], recurrent NN [26], physics-informed NN [27][28][29], multi-fidelity NN [30], and tensor basis NN [31]. Gaussian processes (GP) have also been employed, for example, for strain-rate dependent viscosity [32] or for viscoelastic properties [33][34][35][36].…”

Section: Introductionmentioning

confidence: 99%

“…Inspired by these successes, symbolic regression methods have recently started to be used in the field of rheology as well. For example, Mohammadamin and coworkers [38] relied on SINDy for flexibly identifying the constitutive equations of an elasto-visco-plastic fluid. Nevertheless, although there are several attempts along this line, a comprehensive study to test SINDy for rheological data has not yet been conducted.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Response: Naoko Saito, Finding as Founding: Rejoinder to René Arcilla’s Review, Naoko Saito, Associate Professor of Education at Kyoto University, Japan. Kyoto University, Yoshidahonmachi, Sakyo-ku, Kyoto 606-8501

Saito

2020

Stud Philos Educ

View full text Add to dashboard Cite

In , arXiv:1909), I reported on a double outburst and rebrightenings in 2018 in V544 Her. Such a phenomenon is usually observed in WZ Sge stars which evolved after the period bounce and the colors of V544 Her in quiescence apparently exclude this possibility. Although this phenomenon was considered to be rare, I detected almost exactly the same one in 2021 using ZTF, ATLAS and ASAS-SN public data. I also detected a phenomenon very similar to this in ASASSN-19yt in 2022. The same object showed a different type of outburst in 2019 whose morphology looked like that of an SS Cyg star. If ASASSN-19yt is an SU UMa star, the morphology of the 2019 outburst would challenge our knowledge in SU UMa stars. If this object, or V544 Her, is an SS Cyg star, what causes a double outburst and rebrightenings would become an unsolved problem in dwarf novae.

show abstract

“…Now, many scholars have used machine learning methods to analyze or identify the structure of amorphous materials. [18][19][20][21][22] For example, they use neural networks with certain training to classify or predict non-crystalline structures. [23][24][25] These machine learning methods improved material identification quality and identification efficiency when applied to a variety of complex systems.…”

Section: Introductionmentioning

confidence: 99%

Classification and structural characteristics of amorphous materials based on interpretable deep learning

Cui

Zhao

et al. 2023

Chinese Phys. B

View full text Add to dashboard Cite

Defining the structure characteristics of amorphous materials is one of the fundamental problems that need to be solved urgently in complex materials because of their complex structure and long-range disorder. In this study, we develop an interpretable deep learning model capable of accurately classifying amorphous configurations and characterizing their structural properties. The results demonstrate that the multi-dimensional hybrid convolutional neural network can classify the 2D liquids and amorphous solids of molecular dynamics simulation. The classification process does not make a priori assumptions on the amorphous particle environment, and the accuracy is 92.75%, which is better than other convolutional neural networks. Moreover, our model utilizes the gradient-weighted activation-like mapping method, which generates activation-like heat maps that can precisely identify important structures in the amorphous configuration maps. We obtain an order parameter from the heatmap and conduct finite scale analysis of this parameter. Our findings demonstrate that the order parameter effectively captures the amorphous phase transition process across various systems. These results hold significant scientific implications for the study of amorphous structural characteristics via deep learning.

show abstract

Digital rheometer twins: Learning the hidden rheology of complex fluids through rheology-informed graph neural networks

Cited by 26 publications

References 39 publications

Rheological identification of jetted fluid using machine learning

Rheological identification of jetted fluid using machine learning

Response: Naoko Saito, Finding as Founding: Rejoinder to René Arcilla’s Review, Naoko Saito, Associate Professor of Education at Kyoto University, Japan. Kyoto University, Yoshidahonmachi, Sakyo-ku, Kyoto 606-8501

Classification and structural characteristics of amorphous materials based on interpretable deep learning

Contact Info

Product

Resources

About