Generative Understanding of Ultrafast Fluid Dynamics

Cel

Fluids are fundamental to a wide spectrum of natural phenomena and technological applications. Spectacular manifestations of patterns and order arise at small spatiotemporal scales from ultra-fast fluid dynamics (UFD). Initiated by focused deposition of high thermal energy, causing local disruption of equilibrium, violent competing fluid dynamics with extreme local state variations characterize the subsequent relaxation. UFD offer a unique potential for exploration in micro-manufacturing and energy conversion. How can this potential be leveraged/harnessed? What mechanisms and inherent properties determine UFD in complex manufacturing environments? How can UFD be controlled to deliver nanoparticles or micro-structured surfaces with predictable properties? What is the potential of employing UFD to control breakup of phase interfaces? Our objective is to answer these questions by decisive advances in generative computational predictions and high-fidelity numerical experiments. Data-driven multi-fidelity models break the curse-of-dimensionality barrier. Identification of interpretable low-dimensional manifolds of UFD dynamics generates understanding from data. Transformer models produce UFD realizations without solving equations. Harnessing efficient sampling from these methodologies enables us to optimize UFD processes, targeting applications in precision engineering. Developed paradigms, methodologies, and computational tools will be delivered to the scientific and engineering community. Our group has strong foundations in complex-fluid physics and advanced computational methods, and a strong record of successfully integrating fundamental research and technical applications. Our goal is to provide unprecedented insight into UFD in complex environments and to unravel the path to technical solutions. Leveraging the hidden potential of UFD gives access to breakthrough innovations and high-impact technologies in micro-manufacturing and energy conversion.