Article
Computer Science, Interdisciplinary Applications
Didier Lucor, Atul Agrawal, Anne Sergent
Summary: PINNs show promise as candidates for full fluid flow PDE modeling, but challenges remain in sustaining turbulence. By minimizing composite loss functions, surrogate modeling using PINNs for turbulent natural convection flows can reduce the need for large training datasets.
JOURNAL OF COMPUTATIONAL PHYSICS
(2022)
Article
Thermodynamics
Yuli Cao, Ruina Xu, Peixue Jiang
Summary: This paper presents an alternative approach to model the turbulent heat transfer of supercritical pressure fluid using deep neural networks. The modified ML-KTVT model shows good performance in predicting the convection heat transfer and reproducing turbulence development in heat transfer deterioration cases.
INTERNATIONAL JOURNAL OF HEAT AND MASS TRANSFER
(2023)
Article
Agricultural Engineering
Shaojun Ren, Shiliang Wu, Qihang Weng
Summary: Machine learning methods have shown a broad application prospect in biomass gasification modeling, but their physical interpretability is poor when relying on limited experimental data. In this study, a physics-informed neural network method (PINN) is developed to predict biomass gasification products, providing physically feasible predictions. The PINN models outperformed other machine learning methods in prediction capability and demonstrated better generalizability and scientific interpretability.
BIORESOURCE TECHNOLOGY
(2023)
Article
Energy & Fuels
Shanti Bhushan, Greg W. Burgreen, Wesley Brewer, Ian D. Dettwiller
Summary: The study shows that a machine learned turbulence model can provide grid convergent and smooth solutions, and converge to a correct solution from ill-posed flow conditions. The accuracy of the model relies on the choice of flow variables and training approach, with data clustering identified as a useful tool to avoid model skewness.
Article
Engineering, Mechanical
Michael Rom
Summary: This study proposes a method for solving the Reynolds equation with cavitation modeling using physics-informed neural networks (PINNs). The PINN solves multiple problems simultaneously and generalizes well by extending its inputs with parameters such as the relative eccentricity. Accurate pressure and liquid ratio predictions for further values of the relative eccentricity can be obtained by just evaluating the PINN, taking less than a second. Solutions for a journal bearing test case are compared with finite difference solutions.
TRIBOLOGY INTERNATIONAL
(2023)
Article
Computer Science, Interdisciplinary Applications
Wenqian Chen, Qian Wang, Jan S. Hesthaven, Chuhua Zhang
Summary: A reduced basis method based on a physics-informed machine learning framework, using neural networks to model parametrized partial differential equations, achieves more accurate and efficient results compared to traditional methods.
JOURNAL OF COMPUTATIONAL PHYSICS
(2021)
Article
Environmental Sciences
Amir H. Kohanpur, Siddharth Saksena, Sayan Dey, J. Michael Johnson, M. Sadegh Riasi, Lilit Yeghiazarian, Alexandre M. Tartakovsky
Summary: Estimating uncertainty in flood model predictions is crucial for various applications. This study focuses on uncertainty in physics-based urban flooding models, considering model complexity, uncertainty in input parameters, and the effects of rainfall intensity. The ICPR model is used to quantify floodwater depth prediction uncertainty, with results showing localized uncertainties. Model simplifications lead to overconfident predictions, while increasing model resolution reduces uncertainty but increases computational cost. The multilevel MC method is employed to reduce cost when estimating uncertainty in a high-resolution ICPR model. Utilizing ensemble estimates, the proposed framework improves flood depth forecasting accuracy compared to the ICPR model's mean prediction, even with limited measurements.
WATER RESOURCES RESEARCH
(2023)
Article
Chemistry, Physical
Mikkel L. Bodker, Mathieu Bauchy, Tao Du, John C. Mauro, Morten M. Smedskjaer
Summary: This study introduces a novel approach combining statistical mechanics and machine learning to accurately predict the structure of glasses, improving both interpolation and extrapolation abilities.
NPJ COMPUTATIONAL MATERIALS
(2022)
Article
Mathematics, Applied
Salvatore Cuomo, Mariapia De Rosa, Fabio Giampaolo, Stefano Izzo, Vincenzo Schiano Di Cola
Summary: In recent years, Scientific Machine Learning (SciML) methods, particularly Physics-Informed Neural Networks (PINNs), have become popular for solving non-linear partial differential equations (PDEs). This paper numerically tackles the groundwater flow equations using a PINN approach, approximating the Dirac distribution and analyzing its computational ability in higher-dimensional cases. The effectiveness of PINNs is demonstrated through numerical experiments in hydrological applications, comparing the results with the Finite Difference Method (FDM) and highlighting the advantages of PINNs in solving PDEs without discretization.
COMPUTERS & MATHEMATICS WITH APPLICATIONS
(2023)
Article
Engineering, Mechanical
J. Nathan Kutz, Steven L. Brunton
Summary: Data-driven modeling is enabled by modern machine learning algorithms and deep learning architectures. The goal is to generate models for prediction, characterization, and control of complex systems. In the context of physics and engineering, extrapolation and generalization play important roles, which can be supported through various forms of parsimony.
NONLINEAR DYNAMICS
(2022)
Article
Computer Science, Interdisciplinary Applications
Yang Chen, Yongfu Xu, Lei Wang, Tianyi Li
Summary: This study investigates the impacts of loss weights and random state on the performance of the physics informed neural network (PINN) in solving the Richardson Richards equation (RRE) and proposes possible solutions to mitigate these impacts. The results show that the baseline PINN's performance heavily relies on the configurations of loss weights and random states. While GN-PINN tends to ignore the train loss term and is not suitable for solving RRE problems, PLF-PINN can strike a good balance between loss terms and greatly enhance the robustness of PINN against loss weight initialization and random state.
COMPUTERS AND GEOTECHNICS
(2023)
Review
Chemistry, Multidisciplinary
Joseph Pateras, Pratip Rana, Preetam Ghosh
Summary: Physics-informed machine learning (PIML) is an emerging field that utilizes physically relevant prior information to extract physically relevant solutions from data lacking in quantity and veracity. This paper discusses recent advancements in PIML and highlights novel methods and applications of domain decomposition in physics-informed neural networks (PINNs). It also explores the use of neural operator learning to intuit relationships in physics systems traditionally modeled with complex governing equations and expensive differentiation techniques. Additionally, the paper discusses the limitations and applications of traditional physics-informed machine learning, and proposes a novel taxonomic structure to categorize PIML based on the derivation and injection of physics information into the machine learning process.
APPLIED SCIENCES-BASEL
(2023)
Article
Engineering, Mechanical
Junchuan Shi, Alexis Rivera, Dazhong Wu
Summary: This paper proposes a physics-informed machine learning method for accurate modeling and prediction of the remaining useful life (RUL) of Lithium-ion batteries. The method considers the impact of battery health and operating conditions on battery aging and combines a calendar and cycle aging model with an LSTM layer for modeling and prediction. Experimental results demonstrate that the proposed method can accurately model and predict the degradation behavior and RUL of Lithium-ion batteries under different operating conditions.
MECHANICAL SYSTEMS AND SIGNAL PROCESSING
(2022)
Article
Engineering, Civil
Pravin Bhasme, Jenil Vagadiya, Udit Bhatia
Summary: Current hydrology modeling approaches often rely on physics-based or data-science methods, each with its own limitations. While physics-based models offer better process understanding, machine learning algorithms exhibit superior predictive abilities. Hence, the development of a hybrid modeling approach that combines the strengths of both is necessary.
JOURNAL OF HYDROLOGY
(2022)
Article
Automation & Control Systems
Francesco Piccialli, Fabio Giampaolo, David Camacho, Gang Mei
Summary: Deep learning technology is driving the in-depth development of industrial automation. Wang et al. interpret the decision process of convolutional neural networks (CNNs) using a percolation model from a statistical physics perspective. They introduce the concept of differentiation degree and present an empirical formula for quantifying it.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2023)
Article
Engineering, Chemical
Paulo Roberto C. Mendes Junior, Ivan R. Siqueira, Roney L. Thompson, Marcio S. Carvalho
Summary: This study investigates planar extrudate swell flows of Newtonian liquids with a viscous liquid-gas interface. Results demonstrate that the extrudate swells dramatically as the interfacial viscosity grows, indicating a clear relationship between parameters.
Article
Mechanics
Vitor Figueiroa Ventura, Joao Felipe Mitre, Roney Leon Thompson
Summary: Waxy crude oils can turn into a gel at low temperatures, causing flow assurance issues during restart due to the higher pressure drop required. This study investigates the impact of cooling process after shutdown on the problem. The results show that the plastic number and the duration of cooling after shutdown play a role in the restart conditions, with higher plastic numbers and longer cooling times leading to more difficult restarts.
JOURNAL OF NON-NEWTONIAN FLUID MECHANICS
(2022)
Article
Mechanics
Hugo L. Franca, Cassio M. Oishi, Roney L. Thompson
Summary: This work presents a numerical study on the collision of viscoelastic drops under surface tension effects. Different constitutive models for non-Newtonian fluids are considered, and the outcomes of bouncing, coalescence, and separation are analyzed. The study also explores the effects of various parameters on the collision outcomes.
JOURNAL OF NON-NEWTONIAN FLUID MECHANICS
(2022)
Article
Mechanics
Bruno Jorge Macedo dos Santos, Felipe Warwar Murad, Angela Ourivio Nieckele, Luiz Eduardo Bittencourt Sampaio, Roney Leon Thompson
Summary: This study examines the application of nonlinear models to Reynolds Average Navier Stokes (RANS) models to improve turbulent flow prediction. By analyzing higher-order models and employing objective orthogonal tensors, two sets of nonlinear models are proposed and compared with DNS data.
MECHANICS RESEARCH COMMUNICATIONS
(2022)
Article
Energy & Fuels
Icaro Coelho, Edson J. Soares, Roney L. Thompson, Fabio de Assis Ressel Pereira, Adriana Teixeira, Leandro Saraiva Valim
Summary: The main flow assurance problem in the oil industry is the formation of gas hydrates in high pressure and low temperature conditions. Tetrahydrofuran (THF) is commonly used to mimic gas hydrates. This study investigates the rheological properties and behavior of THF, finding different outcomes for different mass concentrations and differing from previous literature.
Article
Engineering, Mechanical
Roney L. Thompson
Summary: This study examines the relationship between skew-symmetric tensors in Continuum Mechanics, specifically Motions With Constant Stretch History (MWCSH), the Zorawski condition, and Rivlin-Ericksen (R-E) tensors. Theoretical analysis reveals that the rate-of-rotation of R-E tensors in MWCSH is equal to the skew-symmetric tensor governing this type of motion, extending existing analytical results. The study also presents a compact form for the Zorawski condition, enabling theoretical analysis of inherently unsteady motions.
JOURNAL OF THE BRAZILIAN SOCIETY OF MECHANICAL SCIENCES AND ENGINEERING
(2022)
Article
Engineering, Mechanical
P. R. S. Costa, R. P. Barboza, R. A. C. Dias, J. L. Favero, A. O. M. Samel, M. A. Cruz, L. F. L. R. Silva, R. L. Thompson, M. P. Schwalbert
Summary: In this study, a model is proposed to predict the treatable size zones for acid reservoir stimulation, considering the coupling between well and reservoir using a 3D approach. The robust numerical code implemented using OpenFOAM(R) demonstrated that the 3D method provides qualitative and quantitative results that cannot be predicted by the simplified 2D model.
JOURNAL OF THE BRAZILIAN SOCIETY OF MECHANICAL SCIENCES AND ENGINEERING
(2022)
Article
Engineering, Chemical
Allan B. G. Motta, Roney L. Thompson, Mateus P. Schwalbert, Luiz F. L. R. Silva, Jovani L. Favero, Rodrigo A. C. Dias, Raphael J. Leitao
Summary: This paper proposes a flow model considering the non-uniform pore distribution in porous media, and through simulation and analysis, it is found that the non-uniform pore size distribution may be an important factor causing discrepancies between the viscosity functions obtained from rheometric devices and the Darcy equation.
TRANSPORT IN POROUS MEDIA
(2022)
Article
Mechanics
F. O. Silva, I. R. Siqueira, M. S. Carvalho, R. L. Thompson
Summary: We conducted a computational study on free surface flows in the film formation region of a slot coater, considering the rheologically complex interfaces. By solving the equations of motion for incompressible Newtonian liquids coupled with the Boussinesq-Scriven constitutive equation for viscous interfaces, we found that interfacial viscosity plays a significant role in the flow dynamics and operating limits of slot coating. The interfacial viscosity affects the stiffness of viscous interfaces, resulting in changes in stress jumps and retarding the film flow.
Article
Physics, Fluids & Plasmas
A. L. Guilherme, I. R. Siqueira, L. H. P. Cunha, R. L. Thompson, T. F. Oliveira
Summary: We study the impact of external magnetic fields on the dynamics of superparamagnetic ferrofluid droplets and the rheology of dilute ferrofluid emulsions in planar extensional flows. By altering the shape and magnetization of the ferrofluid droplets, the intensity and direction of uniform magnetic fields affect the planar extensional rheology of the emulsions. We observe unconventional stress-strain responses and introduce new extensional material functions to explain these rheological signatures.
PHYSICAL REVIEW FLUIDS
(2023)
Article
Mechanics
Joao P. Cunha, Paulo R. de Souza Mendes, Roney L. Thompson, Elias C. Rodrigues, Erick F. Quintella
Summary: This paper examines the mechanical behavior of microelements dispersed in the bulk flow in a four-roll mill and provides a comprehensive analysis of the flow type seen from the perspective of the microelement. The study takes into account material parameters to generalize a kinematic flow-type classification parameter. The numerical simulation using the finite element method shows how the local flow type varies under different flow conditions and analyzes the deformation of microelements positioned at the stagnation point.
JOURNAL OF NON-NEWTONIAN FLUID MECHANICS
(2023)
Article
Energy & Fuels
Andre S. Guimaraes, Thiago O. Marinho, Priamo A. Melo, Roney L. Thompson, Marcia C. K. de Oliveira, Marcio Nele
Summary: This study presents a novel coupled kinetic and rheological model for paraffinic wax oil gelation. The crystallization kinetics was investigated using differential scanning calorimetry, while the rheological properties were determined through dynamic oscillatory tests. The effect of cooling rate on fractal dimensional analysis was examined for different wax concentrations. The gathered data were successfully reconciled using scaling models, allowing for the discussion of microstructural characteristics. Additionally, simulations of waxy oil gelation during a pipeline shutdown were conducted based on the observed kinetic and rheological behavior.
GEOENERGY SCIENCE AND ENGINEERING
(2023)