Artificial Intelligence for Advanced Materials Modeling in Thermal, Energy, and Aerodynamic Systems: A Systematic Review

Artificial Intelligence (AI) is increasingly transforming the development and analysis of advanced materials and engineering systems by enabling data-driven prediction, materials discovery, simulation acceleration, and optimization. However, the rapidly expanding literature remains fragmented across materials science, thermofluid modeling, Computational Fluid Dynamics (CFD), physics-informed learning, reduced-order modeling, and digital twins. This study presents a systematic review of AI applications in advanced materials, thermal and energy systems, and aerodynamic engineering using a PRISMA-based methodology. A systematic search was conducted in the Scopus database for peer-reviewed studies published between 2019 and 2026. From 1468 records initially identified through Scopus, 122 studies met the predefined eligibility criteria and were included from the database-search pathway, while six additional studies identified through other methods were assessed separately against the substantive eligibility requirements and retained, resulting in a final corpus of 128 studies that were subsequently subjected to methodological quality assessment. The review identifies major applications of Machine Learning (ML), Deep Learning (DL), Graph Neural Networks (GNNs), generative AI, and materials informatics in materials property prediction, characterization, screening, and design. It further examines AI-assisted CFD and thermofluid modeling, including heat and mass transfer, turbulence modeling, flow prediction, and aerodynamic optimization. The analysis also highlights the growing role of physics-informed learning, reduced-order and surrogate models, optimization algorithms, and digital twins in improving computational efficiency and supporting intelligent engineering decision-making. Despite these advances, data availability, generalization, uncertainty quantification, validation, computational cost, scalability, and interoperability remain important barriers to practical implementation. Overall, the literature indicates a transition from isolated AI-based prediction toward integrated physics–data-driven engineering frameworks that connect materials development, simulation, optimization, and real-time system operation.