Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

WAKESET: A Large-Scale, High-Reynolds Number Flow Dataset for Machine Learning of Turbulent Wake Dynamics

Zachary Cooper-Baldock, Paulo E. Santos, Russell S. A. Brinkworth, Karl Sammut

TL;DR

WAKESET tackles the data scarcity preventing robust ML for turbulent CFD by delivering a large-scale, high-Reynolds-number 3D wake dataset collected from an underwater AUV berthing within an XLUUV. It pairs a rigorous two-phase development (foundational hydro analysis and generalisation for ML) with expanded geometry, wide flow-conditions, and comprehensive data augmentation, culminating in 1,091 base simulations that become 4,364 augmented instances. The dataset is openly released with structured storage, loaders, and baseline GAN benchmarks for 2D and 3D flow field generation conditioned on input kinematics, highlighting both the practical value and the challenges of modelling high-Re wake dynamics. Overall, WAKESET provides a pivotal resource for training, benchmarking, and transferring ML models to real-world, high-Re CFD problems in autonomous underwater systems.

Abstract

Machine learning (ML) offers transformative potential for computational fluid dynamics (CFD), promising to accelerate simulations, improve turbulence modelling, and enable real-time flow prediction and control-capabilities that could fundamentally change how engineers approach fluid dynamics problems. However, the exploration of ML in fluid dynamics is critically hampered by the scarcity of large, diverse, and high-fidelity datasets suitable for training robust models. This limitation is particularly acute for highly turbulent flows, which dominate practical engineering applications yet remain computationally prohibitive to simulate at scale. High-Reynolds number turbulent datasets are essential for ML models to learn the complex, multi-scale physics characteristic of real-world flows, enabling generalisation beyond the simplified, low-Reynolds number regimes often represented in existing datasets. This paper introduces WAKESET, a novel, large-scale CFD dataset of highly turbulent flows, designed to address this critical gap. The dataset captures the complex hydrodynamic interactions during the underwater recovery of an autonomous underwater vehicle by a larger extra-large uncrewed underwater vehicle. It comprises 1,091 high-fidelity Reynolds-Averaged Navier-Stokes simulations, augmented to 4,364 instances, covering a wide operational envelope of speeds (up to Reynolds numbers of 1.09 x 10^8) and turning angles. This work details the motivation for this new dataset by reviewing existing resources, outlines the hydrodynamic modelling and validation underpinning its creation, and describes its structure. The dataset's focus on a practical engineering problem, its scale, and its high turbulence characteristics make it a valuable resource for developing and benchmarking ML models for flow field prediction, surrogate modelling, and autonomous navigation in complex underwater environments.

WAKESET: A Large-Scale, High-Reynolds Number Flow Dataset for Machine Learning of Turbulent Wake Dynamics

TL;DR

WAKESET tackles the data scarcity preventing robust ML for turbulent CFD by delivering a large-scale, high-Reynolds-number 3D wake dataset collected from an underwater AUV berthing within an XLUUV. It pairs a rigorous two-phase development (foundational hydro analysis and generalisation for ML) with expanded geometry, wide flow-conditions, and comprehensive data augmentation, culminating in 1,091 base simulations that become 4,364 augmented instances. The dataset is openly released with structured storage, loaders, and baseline GAN benchmarks for 2D and 3D flow field generation conditioned on input kinematics, highlighting both the practical value and the challenges of modelling high-Re wake dynamics. Overall, WAKESET provides a pivotal resource for training, benchmarking, and transferring ML models to real-world, high-Re CFD problems in autonomous underwater systems.

Abstract

Machine learning (ML) offers transformative potential for computational fluid dynamics (CFD), promising to accelerate simulations, improve turbulence modelling, and enable real-time flow prediction and control-capabilities that could fundamentally change how engineers approach fluid dynamics problems. However, the exploration of ML in fluid dynamics is critically hampered by the scarcity of large, diverse, and high-fidelity datasets suitable for training robust models. This limitation is particularly acute for highly turbulent flows, which dominate practical engineering applications yet remain computationally prohibitive to simulate at scale. High-Reynolds number turbulent datasets are essential for ML models to learn the complex, multi-scale physics characteristic of real-world flows, enabling generalisation beyond the simplified, low-Reynolds number regimes often represented in existing datasets. This paper introduces WAKESET, a novel, large-scale CFD dataset of highly turbulent flows, designed to address this critical gap. The dataset captures the complex hydrodynamic interactions during the underwater recovery of an autonomous underwater vehicle by a larger extra-large uncrewed underwater vehicle. It comprises 1,091 high-fidelity Reynolds-Averaged Navier-Stokes simulations, augmented to 4,364 instances, covering a wide operational envelope of speeds (up to Reynolds numbers of 1.09 x 10^8) and turning angles. This work details the motivation for this new dataset by reviewing existing resources, outlines the hydrodynamic modelling and validation underpinning its creation, and describes its structure. The dataset's focus on a practical engineering problem, its scale, and its high turbulence characteristics make it a valuable resource for developing and benchmarking ML models for flow field prediction, surrogate modelling, and autonomous navigation in complex underwater environments.
Paper Structure (29 sections, 17 equations, 9 figures, 4 tables)

This paper contains 29 sections, 17 equations, 9 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Motivation
  3. XLUUV Recovery: A Challenging and Relevant Case Study
  4. Methods
  5. Foundational Hydrodynamic Analysis
  6. Vehicle Geometries
  7. Approach Paths, Trajectories and Speeds
  8. Hydrodynamic Phenomenon of Interest
  9. CFD Model Setup and Validation
  10. Key Insights from the Foundational Analysis
  11. Generalisation to an ML Training Set
  12. Generalisation of the XLUUV Model
  13. Parametrisation: Expanding the Flow Regime
  14. CFD Model for Large-Scale Generation
  15. WAKESET
  16. ...and 14 more sections

Figures (9)

  • Figure 1: XLUUV hull structure in generalised form. All dimensions are provided in millimetres (mm). Payload bay geometry shown in the centre of the XLUUV body. X fin tail design with rear propeller appended to the centre of this structure as an actuator disk model within ANSYS Fluent.
  • Figure 2: XLUUV hull structure, vertical and horizontal plane data locations. Each individual point represents a CFD mesh vertex where the parameters have been recorded at. Tighter clustering is present in the boundary layer and wake region to adhere to the $y^+$ constraints for the turbulence model as detailed in Appendix \ref{['App:HydrodynamicPhenomenon']}.
  • Figure 3: Horizontal planes with velocity magnitude contour plots shown. Velocity magnitude is provided in meters per second (m/s). Moving from left to right, a cross flow of (a) 0 degrees, (b) 10 degrees, (c) 20 degrees and (d) 30 degrees is shown. The XLUUV hull geometry is located in the centre of the planar slice. The wake structure is shown centre bottom of every planar slice.
  • Figure 4: Horizontal planes with dynamic pressure contour plots shown. Dynamic pressure is provided in meters per second (Pa). Moving from left to right, a cross flow of (a) 0 degrees, (b) 10 degrees, (c) 20 degrees and (d) 30 degrees is shown. The XLUUV hull geometry is located in the centre of the planar slice. The wake structure is shown centre bottom of every planar slice.
  • Figure 5: Horizontal planes with turbulence intensity contour plots shown. Turbulence intensity is provided as a percentage (%). Moving from left to right, a cross flow of (a) 0 degrees, (b) 10 degrees, (c) 20 degrees and (d) 30 degrees is shown. The XLUUV hull geometry is located in the centre of the planar slice. The wake structure is shown centre bottom of every planar slice.
  • ...and 4 more figures