A5076511812- Fluid Dynamics and Turbulent Flows
- Model Reduction and Neural Networks
- Meteorological Phenomena and Simulations
- Probabilistic and Robust Engineering Design
- Wind and Air Flow Studies
- Fluid Dynamics and Vibration Analysis
- Fluid Dynamics Simulations and Interactions
- Nuclear Engineering Thermal-Hydraulics
- Computational Fluid Dynamics and Aerodynamics
- Granular flow and fluidized beds
- Coastal and Marine Dynamics
- Reservoir Engineering and Simulation Methods
- Seismic Imaging and Inversion Techniques
- Lattice Boltzmann Simulation Studies
- Aerodynamics and Acoustics in Jet Flows
- Advanced Power Amplifier Design
- Earthquake and Tsunami Effects
- Hydrology and Sediment Transport Processes
- Radio Frequency Integrated Circuit Design
- Geological formations and processes
- Aeolian processes and effects
- Gaussian Processes and Bayesian Inference
- Energy Load and Power Forecasting
- Aerodynamics and Fluid Dynamics Research
- earthquake and tectonic studies
Virginia Tech
2014-2023
University of Stuttgart
2023
Simulation Technologies (United States)
2023
Shandong University of Science and Technology
2022
First Institute of Oceanography
2022
Ministry of Natural Resources
2022
Qingdao University
2022
Institute of Mechanics
2021
ETH Zurich
2010-2012
Princeton University
2009
We show that the discrepancies in Reynolds-averaged Navier-Stokes (RANS) modeled Reynolds stresses can be explained by mean flow features. A physics-informed machine learning framework is proposed to improve predictive capabilities of RANS models leveraging existing direct numerical simulations databases.
Data from experiments and direct simulations of turbulence have historically been used to calibrate simple engineering models such as those based on the Reynolds-averaged Navier--Stokes (RANS) equations. In past few years, with availability large diverse datasets, researchers begun explore methods systematically inform data, goal quantifying reducing model uncertainties. This review surveys recent developments in bounding uncertainties RANS via physical constraints, adopting statistical...
Machine learning (ML) provides novel and powerful ways of accurately efficiently recognizing complex patterns, emulating nonlinear dynamics, predicting the spatio-temporal evolution weather climate processes. Off-the-shelf ML models, however, do not necessarily obey fundamental governing laws physical systems, nor they generalize well to scenarios on which have been trained. We survey systematic approaches incorporating physics domain knowledge into models distill these broad categories....
A study shows that when trained using channel flow data at just one Reynolds number and informed with known physics, the neural network works robustly as a wall model in large-eddy simulation (LES) of any number. It also outperforms equilibrium LES 3D boundary layer flow.
We present a comprehensive framework for augmenting turbulence models with physics-informed machine learning, illustrating complete workflow from identification of input/output to prediction mean velocities. The learned model has Galilean invariance and coordinate rotational invariance.
Reynolds-averaged Navier--Stokes (RANS) simulations with turbulence closure models continue to play important roles in industrial flow simulations. However, the commonly used linear eddy viscosity are intrinsically unable handle flows non-equilibrium turbulence. Reynolds stress models, on other hand, plagued by their lack of robustness. Recent studies plane channel found that even substituting stresses errors below 0.5% from direct numerical simulation (DNS) databases into RANS equations...
In this work, we propose using an ensemble Kalman method to learn a nonlinear eddy viscosity model, represented as tensor basis neural network, from velocity data. Data-driven turbulence models have emerged promising alternative traditional for providing closure mapping the mean velocities Reynolds stresses. Most data-driven in category need full-field stress data training, which not only places stringent demand on generation but also makes trained model ill-conditioned and lacks robustness....
Abstract A robust and efficient solver coupling computational fluid dynamics (CFD) with discrete element method (DEM) is developed to simulate particle-laden flows in various physical settings. An interpolation algorithm suitable for unstructured meshes proposed translate between mesh-based Eulerian fields particle-based La-grangian quantities. The scheme reduces the mesh-dependence of averaging procedures. In addition, fluid-particle interaction terms are treated semi-implicitly this...
Feature identification is an important task in many fluid dynamics applications and diverse methods have been developed for this purpose. These are based on a physical understanding of the underlying behavior flow vicinity feature. Particularly, they rely definition suitable criteria (i.e. point-based or neighborhood-based derived properties) proper selection thresholds. For instance, among other techniques, vortex can be done through computing Q-criterion by considering center looping...