Computational Fluid Dynamics (CFD) is indispensable for analysing complex flowand heat transfer phenomena, yet its high computational cost restricts scalability and real-timeapplications. This study introduces a unified AI–CFD framework that integrates deep learning(DL) surrogate modelling, deep reinforcement learning (DRL) optimisation, andsuper-resolution (SR) visualisation to transform CFD into an adaptive, solver-agnosticframework. Using CFD datasets from nanofluid heat transfer simulations, convolutional neuralnetworks and transformer architectures were trained to predict velocity and temperature fieldswith high fidelity, achieving 25–30× speedup while maintaining accuracy within 5% errormargins. DRL agents, implemented via Proximal Policy Optimisation, dynamically tuned inletvelocity and nanoparticle fraction, yielding up to 18% improvement in heat transfer coefficientcompared to baseline CFD control strategies. In parallel, generative adversarial networksupscaled coarse CFD outputs, reconstructing fine-scale turbulence structures with PSNR> 32 dB and SSIM > 0.93, thereby enhancing interpretability without prohibitive computationalexpense. Together, these modules form a synergistic pipeline that accelerates CFD simulations,adaptively optimises thermal performance, and restores visual fidelity. Unlike prior studies thatinvestigated surrogates, RL agents, or SR methods separately, this work integrates all three intoa comprehensive framework, bridging fluid mechanics and artificial intelligence for scalable,real-time thermal system design. The framework was validated through mesh independence,solver benchmarking, and spectral fidelity checks across Fluent and OpenFOAM, confirmingrobustness and scalability across platforms.
Arjun Kozhikkatil Sunil (Thu,) studied this question.