Accurate prediction of phase change fluid-flow phenomena represents a major engineering challenge in the design of energy-intensive power-production and chemical-process systems. Emerging enhanced heat transfer technologies employ complex geometries that cannot be modeled analytically and are challenging to instrument for experimental investigations. Recent advances in computing and interconnect systems enable direct simulation of these processes using HPC resources. While relatively mature tools are available for simulating adiabatic (no heat transfer) two-phase flows, methods for analyzing flows with phase change are still in their infancy.
In the present work, a highly parallelized first-principles-based phase-change model is developed, which forces interface-containing mesh cells to the equilibrium state. The operation on cells instead of complex interface surfaces enables the use of fast graph algorithms without geometric reconstruction. This simulation approach is validated for the canonical phase-change configurations of smooth and wavy falling-film condensation. Results are presented for a number of key applications, including the fundamental transport processes of dropwise condensation and nucleate boiling, distributed-heated bubble-pumps for passive absorption refrigeration, flow condensation in air-cooled power plant condensers, and direct-contact liquid-desiccant dehumidification. This simulation capability complements current thermal-fluid engineering approaches to guide development of next-generation energy systems.
