Recent Advances in Computational Fluid Dynamics Using Artificial Intelligence, Machine Learning, and Deep Learning: A Comprehensive Review
The integration of Artificial Intelligence (AI), Machine Learning (ML), and Deep Learning (DL) techniques with Computational Fluid Dynamics (CFD) represents a paradigm shift in fluid mechanics research and engineering applications. This comprehensive review examines recent advances from 2015 to 2025 in applying AI/ML/DL methodologies to enhance traditional CFD workflows across various domains, including turbulence modelling, heat transfer prediction, surrogate modelling, inverse design, and real-time flow control. The paper systematically analyses the fundamental concepts underlying these technologies, their specific applications in fluid dynamics problems, and the challenges associated with their implementation. Key findings indicate that physics-informed neural networks, convolutional neural networks, and reinforcement learning algorithms have demonstrated significant potential in accelerating CFD simulations while maintaining accuracy. However, challenges related to data quality, model interpretability, generalisation capabilities, and integration with high-performance computing platforms remain significant barriers to widespread adoption. The review concludes by emphasising future research directions, highlighting the need for interdisciplinary collaboration to develop robust, interpretable, and physically consistent AI-enhanced CFD frameworks that can revolutionise engineering design and optimisation processes. Major Findings: (1) AI/ML/DL methods are transforming CFD by enabling efficient surrogate modelling, turbulence prediction, and inverse design with reduced computational costs. (2) Physics-informed and hybrid models ensure physical consistency, improving generalisation and trust in predictions. (3) Integration with digital twins and real-time systems paves the way for adaptive, intelligent control of fluid systems.