Data-Driven Fluid Mechanics: Combining First Principles and Machine Learning

Data-driven methods have become an essential part of fluid dynamicists' methodological portfolio, motivating students and practitioners to gather practical knowledge from computer science, statistics, optimization, signal processing, pattern recognition, nonlinear dynamics, and control. This book presents an overview and pedagogical treatment of data-driven and machine learning tools that are leading research advancements in model-order reduction, system identification, flow control, and data-driven turbulence closures.

Format: Hardback
Length: 468 pages
Publication date: 02 February 2023
Publisher: Cambridge University Press

Data-driven methods have emerged as a crucial component in the methodological toolkit of fluid dynamicists, fostering a drive among students and practitioners to acquire practical insights from a diverse array of disciplines. These fields encompass computer science, statistics, optimization, signal processing, pattern recognition, nonlinear dynamics, and control. Fluid mechanics, with its vast volumes of data, has historically served as a fertile ground for the development and application of data-driven techniques, offering valuable shortcuts, constraints, and interpretations rooted in fundamental physics. Consequently, hybrid approaches that combine data-driven and fundamental principles are at the forefront of ongoing and dynamic research endeavors.

This book, originating from a one-week lecture series course offered by the von Karman Institute for Fluid Dynamics, provides a comprehensive overview and pedagogical treatment of key data-driven and machine learning tools that are driving advancements in model-order reduction, system identification, flow control, and data-driven turbulence closures.

Model-order reduction is a crucial aspect of fluid dynamics, as it involves reducing the complexity of a mathematical model while preserving its essential features. Data-driven methods, such as ensemble modeling and machine learning algorithms, have shown promising results in reducing the computational cost and improving the accuracy of model predictions.

System identification is another area where data-driven methods have made significant contributions. By analyzing experimental data, these methods can identify the parameters and structure of a system, enabling the development of more accurate models and the optimization of control strategies.

Flow control is another application where data-driven methods have shown great promise. By analyzing flow data, these methods can identify patterns and optimize control strategies to improve energy efficiency, reduce emissions, and enhance system performance.

Data-driven turbulence closures are another area where data-driven methods have made significant strides. By analyzing high-resolution numerical simulations, these methods can improve the accuracy of turbulence models, enabling more realistic predictions of flow behavior and its impact on various applications.

In conclusion, data-driven methods have revolutionized the field of fluid dynamics, providing students and practitioners with powerful tools to gather practical knowledge from a diverse range of disciplines. By leveraging the vast volumes of data available in fluid mechanics, researchers can develop more accurate models, optimize control strategies, and improve our understanding of complex flow phenomena. As the field continues to evolve, hybrid approaches that combine data-driven and fundamental principles will undoubtedly play a pivotal role in driving further advancements and shaping the future of fluid dynamics.

Weight: 1024g
Dimension: 176 x 252 x 27 (mm)
ISBN-13: 9781108842143