Machine Learning for Time-Dependent Multiphysics Systems: Transient Coupling, Unsteady Flows and Structural Dynamics

Transient analysis and time-dependent multiphysics systems are central to many engineering applications, including energy systems, aerospace structures, offshore engineering, advanced materials design, and thermo-fluid systems. In these applications, the physical response is often governed by the interaction between unsteady fluid flow, structural deformation, dynamic loading, fatigue evolution, and material behavior over time. Traditional numerical methods, such as computational fluid dynamics (CFD), finite element analysis (FEA), and coupled multiphysics solvers, have long underpinned the modeling of such systems. However, they continue to face significant limitations related to computational cost, scalability, long transient simulations, high-dimensional parameter spaces, and the treatment of strongly coupled physical processes.Machine learning has emerged as a powerful complement to conventional computational methods by enabling faster prediction, reduced-order representation, and improved integration of experimental and simulation data. Techniques such as physics-informed neural networks, operator learning, neural surrogates, reduced-order models, and hybrid data–physics frameworks are increasingly being applied to unsteady flows, fluid–structure interaction, moving-boundary problems, fatigue and damage evolution, and coupled thermo-mechanical processes. These approaches are reshaping how complex engineering systems are modeled, particularly in cases where transient behavior, dynamic coupling, and real-time or repeated prediction are important.The primary goal of this Research Topic is to advance the understanding and application of machine learning techniques for transient CFD and time-dependent multiphysics systems. While significant progress has been made in individual areas such as computational fluid dynamics, structural mechanics, fatigue analysis, and materials modeling, integrating these fields remains challenging due to nonlinear interactions, strong coupling, dynamic loading, moving or deforming boundaries, and computational limitations.This collection aims to address these challenges by promoting research that combines data-driven and physics-based approaches to improve predictive accuracy, computational efficiency, temporal stability, and generalization capability. Particular emphasis will be placed on developing models that can capture time-dependent behavior, complex geometries, long-horizon dynamics, fatigue evolution, and real-world engineering constraints.Ultimately, this Research Topic seeks to establish a focused platform for interdisciplinary collaboration between computational mechanics, fluid dynamics, materials science, and artificial intelligence. The collection aims to support the development of reliable AI-based modeling tools for engineering systems where transient behavior and dynamic coupling are central.This Research Topic welcomes original research articles, review papers, and perspective contributions that address machine learning applications in transient CFD and time-dependent multiphysics systems. Topics of interest include, but are not limited to:o Machine learning for unsteady and thermally coupled fluid flows, including convective heat transfer, buoyancy-driven flows, conjugate heat transfer, and multi-phase thermal systemso Data-driven and physics-informed modeling of fluid–structure interaction, moving-boundary problems, and deforming geometrieso Surrogate, reduced-order, and operator learning methods — including Fourier neural operators and DeepONet — applied to unsteady fluid flows, fluid–structure interaction, and structural dynamicso Physics-informed neural networks and hybrid data–physics frameworks for transient CFD, thermo-fluid systems, and coupled structural problemso Machine learning for dynamic loading, fatigue evolution, and damage accumulationo Uncertainty quantification and model reliability for transient CFD, FSI, and structural dynamics applicationso Digital twins and real-time simulation of dynamically coupled engineering systemso Integration of experimental and simulation data for transient CFD, thermo-fluid, and FSI applicationsSubmissions focusing on both methodological innovation and engineering applications are encouraged. Contributions that demonstrate improved physical consistency, interpretability, temporal stability, and generalization of machine learning models for transient CFD and dynamically coupled multiphysics systems are particularly welcome.This collection aligns with the mini-symposium "Data-Driven and Physics-Informed Computational Methods for Thermal-Fluid Flows", organized by R. S. Ransing (Swansea University, United Kingdom), R. K. Pandey (Indian Institute of Technology (BHU), India) and D. Tripathi (National Institute of Technology, Uttarakhand, India) at the 24th IACM Computational Fluids Conference (March 2027, Yokohama, Japan). The thermo-fluid and unsteady CFD threads of this Research Topic directly complement the mini-symposium scope. Authors whose work falls within these areas and who submit to this collection may be invited to present at the mini-symposium, subject to standard conference registration. Topic editors: Haris Hameed Mian, Rajesh S. Ransing, Hari Ponnamma Rani, Rajesh Pandey, Khazar Hayat Source: https://www.frontiersin.org/research-topics/82332/machine-learning-for-time-dependent-multiphysics-systems-transient-coupling-unsteady-flows-and-structural-dynamics