Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{198088,
        author = {Swapnil Pratap Wadkar and Sagar Shinde},
        title = {Performance Evaluation of a Physics-Informed AIML Turbulence Closure on Benchmark Turbulent Flows},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {12},
        number = {11},
        pages = {9289-9303},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=198088},
        abstract = {This paper evaluates the predictive performance of a physics-informed artificial intelligence and machine learning (AIML) turbulence closure model for Reynolds-Averaged Navier–Stokes (RANS) simulations across representative benchmark turbulent flows.(Riccius et al., 2023) The proposed closure framework integrates invariant feature representations and physically constrained learning to ensure realizability, symmetry and numerical robustness. The model is assessed on canonical and complex turbulent flow configurations including fully developed channel flow, backward-facing step flow, turbulent flow over a two-dimensional airfoil and turbulent jet flow. The predictive capability of the AIML closure is systematically compared with conventional RANS turbulence models in terms of velocity distributions, pressure fields and turbulence quantities. The results demonstrate consistent improvement in capturing separation behaviour, reattachment length, turbulence anisotropy and mean-flow structures. The study confirms the practical applicability of physics-informed AIML turbulence closures for industrial RANS simulations. Furthermore, this integration of domain knowledge into machine learning algorithms enhances data efficiency and prediction stability, providing opportunities to augment or even replace high-fidelity numerical simulations which are often computationally expensive (Sharma et al., 2023). This approach, which embeds fundamental fluid dynamics equations within deep learning architectures, facilitates efficient turbulence modeling and reduces computational time without sacrificing precision (Sahibzada et al., 2025).},
        keywords = {Benchmark turbulent flows; RANS validation; physics-informed machine learning; turbulence closure; CFD.},
        month = {April},
        }