New method improves computer modeling of multiphase flows

Understanding how two different fluids interact, like steam and water, is important for many industries. The motion of two or more immiscible fluids with different thermodynamic properties is known as multiphase flows. These flows are common in nature and play a critical role in several engineering areas. Accurately predicting the forces acting on both fluids helps engineers design safer reactors, better fuel systems, and more efficient waste treatment processes. 

Scientists use computer simulations to study how different fluids interact, but modeling two-phase flows remains a challenge. The interface separating fluids is extremely sharp and changes based on surface tension, pressure, and viscous forces. Capturing these effects accurately in a computer model is difficult because numerical errors can generate nonphysical oscillations and distort the interface.

Researchers at the U.S. Department of Energy’s (DOE) Argonne National Laboratory and their collaborators have developed a new approach to simulate multiphase flows in highly accurate computational fluid dynamics (CFD) codes. Their method combines advanced mathematics and computer science to accurately and robustly capture the moving interface between separate fluids. The key innovation, spectral vanishing viscosity, is a stabilization technique that eliminates any nonphysical oscillations while preserving important flow details, enabling accurate simulation of sharp interfaces.

The team’s method also uses a conservative level-set approach to represent the interface between fluids. This approach helps preserve the volume of each fluid throughout the simulation, preventing the artificial loss or gain of mass that commonly affects traditional interface-capturing techniques. To further improve accuracy, the method periodically reconstructs and corrects the interface geometry, ensuring that the evolving interface remains sharply defined.

They tested their method on a series of challenging two-phase problems. The approach demonstrated accuracy and robustness, successfully capturing complex interface motion and deformation. For example, it successfully modeled classic test cases like bubbles rising in water and water breaking through a dam. These are widely used benchmarks for validating two-phase flow simulations.

Why does this matter? More accurate and highly scalable computer simulations mean that scientists and engineers can better predict how fluids will behave in real-world systems and design safer and more efficient energy and heat transfer systems, such as nuclear reactors. The method is implemented in Nek5000 and NekRS, high-order simulation codes developed at Argonne, that run on the latest generation of supercomputers.

Next steps include extending the method to even more complex multiphase flows and leveraging high-performance computing platforms, such as the Argonne Leadership Computing Facility, to enable faster and larger multiphase simulations to guide the design of critical energy systems. This work, supported by DOE Office of Nuclear Energy, Nuclear Energy Advanced Modeling and Simulation program, could help unlock new insights into multiphase flows across a range of scientific and engineering fields.