Data‐Driven Prediction of Turbulent Flow Statistics Past Bridge Piers in Large‐Scale Rivers Using Convolutional Neural Networks

Abstract

High-fidelity numerical simulations provide a crucial tool to study flood flow dynamics and sediment transport of natural waterways (

Review of Challenges and Opportunities in Turbulence Modeling: A Comparative Analysis of Data-Driven Machine Learning Approaches

“…The CNN was developed to minimize the error between the LES results and CNN predictions in the training method without physical constraints. It remains to be seen whether the CNN algorithm can be generalized to other rivers with different geometries and boundary conditions as it was trained using the LES results from a training testbed river [49].…”

On the Morphodynamics of a Wide Class of Large‐Scale Meandering Rivers: Insights Gained by Coupling LES With Sediment‐Dynamics

“…For example, Zhang et al. (2022) employed the flow field of a large‐scale meandering river, produced using large‐eddy simulation (LES), to develop a machine‐learning algorithm that allowed for the prediction of flood‐induced flow fields in other large‐scale meandering rivers. Their study was limited to flood flow predictions without the ability to generate the equilibrium bed morphology of the rivers.…”

“…As a result, previous studies using high-fidelity flow models, including large-eddy simulations and direct numerical simulations, have mostly focused on highly simplified geometries, for example, 90° bends (Ruther & Olsen, 2005). To circumvent the high cost of high-fidelity modeling, researchers have recently attempted to develop machine-learning algorithms to predict flood flow fields (Mosavi et al, 2018;Qian et al, 2019;Zhang et al, 2022). Despite their potential, such machine-learning approaches are in their infancy and require more research to enable the prediction of flood-induced flow fields and bed deformations in large-scale rivers.…”

“…Moreover, such methods need extensive data sets for a variety of meandering rivers for training purposes. For example, Zhang et al (2022) employed the flow field of a large-scale meandering river, produced using large-eddy simulation (LES), to develop a This study aims to develop the first database of high-fidelity morphodynamic simulations for a wide range of meandering channel geometries representative in shape and scale of those encountered in nature. We developed simulations of 3D flow fields under bed morphology and bed shear stress at flood discharges.…”

Fast Prediction of Solute Concentration Field in Rotationally Influenced Fluids Using a Parameter-Based Field Reconstruction Convolutional Neural Network

“…A significant advantage of the CNN algorithm compared to traditional ML algorithms (such as artificial neural network and genetic programming) is that CNNs can better extract local features in image-like data and eliminate unnecessary computational costs thanks to the convolutional and pooling layers [44]. Examples of successful applications of CNNs in ocean and water-related problems include the classification of water pixels in satellite images [50], prediction of sea surface temperature [51], velocity field estimation on density-driven solute transport [44], modeling of two-dimensional (2D) unsteady flows around a circular cylinder [52], downscaling of daily precipitation and temperature [53], automatic water stage measurements [54], prediction of turbulent flow statistics past bridge piers in large-scale rivers [55], etc. Although these previous studies based on CNNs have made significant contributions, they have not focused on concentration field modeling of effluent-driven solute transport in rotating fluids.…”