Turbulence remains one of the most persistent challenges in fluid dynamics and engineering. Even after more than a century of study, we lack a complete first-principles understanding that delivers reliable predictions across all scales and conditions. High-performance computing, together with improved numerical methods and data-driven approaches such as physics-informed neural networks (PINNs) and Fourier neural operators (FNOs), is now providing new ways to address these difficulties [1,2].
This Special Issue of Fluids on “Future Trends and Challenges in High-Performance Computing for Turbulence” brings together six papers that reflect current efforts in the field. Recent years have seen steady gains. DNS and LES at ever-higher Reynolds numbers have become more routine on modern HPC systems, while machine-learning surrogates help reduce the enormous computational cost of traditional solvers [3]. Real challenges remain, however. Models often fail to generalize well across different flow regimes, closure problems in RANS and LES persist, and coupling turbulence with other branches of physics, such as combustion or scalar transport, continues to be computationally demanding [1,4].
The collected papers approach these issues from several directions. They examine turbulence statistics and scaling laws using high-fidelity DNS (including channel flow studies), explore spectral features such as the bottleneck phenomenon in scalar dissipation across Schmidt numbers, and investigate how neural networks can learn nonlinear relationships in flow fields to serve as fast surrogates. One contribution in this issue, for example, optimizes PINNs with varying levels of physical constraints for turbulent flow over periodic hills [5]. Physics-informed approaches are tested on benchmark flows, while Fourier neural operators are applied to reaction–diffusion problems, demonstrating good accuracy and dramatic speed-ups once trained. A further contribution investigates flame stabilization in lifted dimethyl ether jet flames using a stochastic conditioning method within a RANS framework.
Taken as a whole, these papers show how high-fidelity simulations, careful numerical work, and machine-learning tools can be combined to extract fresh physical insight and make computations more practical. They move us a step closer to simulations that are both accurate and usable on today’s HPC systems.
Looking forward, several areas stand out. Hybrid methods that tightly integrate known physics with data-driven learning should improve robustness and generalization. Better algorithms for exascale machines, especially those that handle heterogeneous hardware efficiently, will be essential. Uncertainty quantification, rapid ensemble predictions, and extensions to more complex flows such as compressible, multiphase, or reacting flows with realistic boundaries also deserve strong attention [2,6,7,8,9].
We hope the work presented here encourages more researchers to explore the promising overlap between high-performance computing, turbulence physics, and modern data-driven techniques. Progress in this area has the potential to turn a long-standing scientific challenge into dependable tools for engineering and scientific applications.
Conflicts of Interest
The authors declare no conflict of interest.
References
- Duraisamy, K.; Iaccarino, G.; Xiao, H. Turbulence Modeling in the Age of Data. Annu. Rev. Fluid Mech. 2019, 51, 357–377. [Google Scholar] [CrossRef] [Scilit]
- Moin, P.; Mahesh, K. DIRECT NUMERICAL SIMULATION: A Tool in Turbulence Research. Annu. Rev. Fluid Mech. 1998, 30, 539–578. [Google Scholar] [CrossRef] [Scilit]
- Li, Z.; Peng, W.; Yuan, Z.; Wang, J. Fourier Neural Operator Approach to Large Eddy Simulation of Three-Dimensional Turbulence. Theor. Appl. Mech. Lett. 2022, 12, 100389. [Google Scholar] [CrossRef] [Scilit]
- Cleary, M.J.; Klimenko, A.Y. Multiple Mapping Conditioning: A New Modelling Framework for Turbulent Combustion. In Turbulent Combustion Modeling; Echekki, T., Mastorakos, E., Eds.; Fluid Mechanics and Its Applications; Springer: Dordrecht, The Netherlands, 2011; Volume 95, pp. 143–173. [Google Scholar]
- Fox, W.; Sharma, B.; Chen, J.; Castellani, M.; Espino, D.M. Optimising Physics-Informed Neural Network Solvers for Turbulence Modelling: A Study on Solver Constraints Against a Data-Driven Approach. Fluids 2024, 9, 279. [Google Scholar] [CrossRef] [Scilit]
- Li, Z.; Peng, W.; Yuan, Z.; Wang, J. Long-Term Predictions of Turbulence by Implicit U-Net Enhanced Fourier Neural Operator. Phys. Fluids 2023, 35, 075145. [Google Scholar] [CrossRef] [Scilit]
- Wang, Y.; Li, Z.; Yuan, Z.; Peng, W.; Liu, T.; Wang, J. Prediction of Turbulent Channel Flow Using Fourier Neural Operator-Based Machine-Learning Strategy. Phys. Rev. Fluids 2024, 9, 084604. [Google Scholar] [CrossRef] [Scilit]
- Patel, Y.; Mons, V.; Marquet, O.; Rigas, G. Turbulence Model Augmented Physics-Informed Neural Networks for Mean-Flow Reconstruction. Phys. Rev. Fluids 2024, 9, 034605. [Google Scholar] [CrossRef] [Scilit]
- Luo, T.; Li, Z.; Yuan, Z.; Peng, W.; Liu, T.; Wang, L.; Wang, J. Fourier Neural Operator for Large Eddy Simulation of Compressible Rayleigh–Taylor Turbulence. Phys. Fluids 2024, 36, 075165. [Google Scholar] [CrossRef] [Scilit]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2026 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license.