Articles

High-Performance Computing (HPC) is indispensable for large-scale scientific simulations, but achieving optimal performance on modern supercomputers is increasingly challenging. As HPC systems scale toward exascale, they face escalating complexity in hardware, software, and workloads. Traditional optimization methods (manual tuning and heuristic algorithms) struggle to cope with dynamic workloads and intricate system behaviors. Artificial Intelligence (AI) techniques offer a promising approach to address these challenges. This article provides an overview of AI-powered optimization in HPC, focusing on how machine learning and related AI methods enhance performance, efficiency, and scalability of scientific simulations. We survey key AI techniques applied to HPC optimization including machine learning for performance modeling, deep reinforcement learning for resource management, and AI-driven surrogate models for accelerating simulations and illustrate their impact through case studies in domains such as job scheduling and fluid dynamics. We discuss the practical applications of these techniques, highlighting reported performance gains (e.g., substantial reductions in simulation run time and improved resource utilization). We also examine the challenges in integrating AI with HPC (such as training overhead, data movement, and reliability concerns) and outline future directions for research. The convergence of AI and HPC is poised to produce “smart” simulation workflows that intelligently adapt and optimize in real time, pushing the frontiers of scientific computing.

Pulsatile blood flow in a subject-specific human aortic arch and its major branches is studied computationally for a peak Reynolds number of 1553 and a Womersley number of 22.74. The aortic geometry is constructed from the CT-scan images of a subject. The aorta has out-of-plane curvature and significant area variation along the flow direction. A physiologically representative pulsatile velocity waveform is applied as boundary condition at the inlet of the aorta. The primary velocity profiles are skewed towards the inner wall of the ascending aorta during the entire cardiac cycle. In the decelerating phase, reverse flow is noted along the inner wall and the magnitude of maximum velocity is about 50 % of the peak flow condition. Flow separation is observed in the inner wall of the ascending aorta during the decelerating and reverse flow phases of the cardiac cycle. In the accelerating phase, however, flow separation does not occur. The major observation of the present work is the existence of complex and asymmetrical vortical flow structures which are not observed either in simple curved pipes or in idealized aortic arch computational studies. The relative strength of the secondary flow with respect to the primary flow is quantified by means of Relative Secondary Kinetic Energy whose highest value is evaluated to be 1.202 occurring near the entrance of the right carotid artery during the maximum reverse flow condition. High values of wall shear stress is observed at distal of the left and right subclavian arteries, the bifurcation of brachiocephalic artery between right subclavian artery and right carotid artery, and proximal inner wall of descending aorta during the cardiac cycle. The wall shear stress at the bifurcations of the branches are low and oscillatory and generally correlates with the preferential sites for atherosclerosis. The flow structures on the aorta wall are explicitly highlighted by the limiting streamlines. The application of limiting streamlines to clearly elucidate the complex on-wall flow structures is one of the key contributions of the present study. During the decelerating and reverse flow phases several critical points are observed on the aortic wall. These complex flow structures vanish during the accelerating phase. The observations made in the present study will be helpful in creating accurate and clinically useful computational models.

Many attempts have been made to improve heat transfer for thermally integrated microchannel reforming reactors. However, the mechanisms for the effects of design factors on heat transfer characteristics are still not fully understood. This study relates to a thermochemical process for producing hydrogen by the catalytic endothermic reaction of methanol with steam in a thermally integrated microchannel reforming reactor. Computational fluid dynamics simulations are conducted to better understand the consumption, generation, and exchange of thermal energy between endothermic and exothermic processes in the reactor. The effects of wall heat conduction properties and channel dimensions on heat transfer characteristics and reactor performance are investigated. Thermodynamic analysis is performed based on specific enthalpy to better understand the evolution of thermal energy in the reactor. The results indicate that the thermal conductivity of the channel walls is fundamentally important. Materials with high thermal conductivity are preferred for the channel walls. Thermally conductive ceramics and metals are well-suited. Wall materials with poor heat conduction properties degrade the reactor performance. Reaction heat flux profiles are considerably affected by channel dimensions. The peak reaction heat flux increases with the channel dimensions while maintaining the flow rates. The change in specific enthalpy is positive for the exothermic reaction and negative for the endothermic reaction. The change in specific sensible enthalpy is always positive. Design recommendations are made to improve thermal performance for the reactor.