The  design  of  new  nuclear  reactors,  and  the  safe,  efficient  operation  of  existing  reactors,  can benefit from fundamental understanding of the bubbly two­‐phase flows created as the water  boils.   The  most  accurate  technique  for  these  flows,  Direct  Numerical  Simulation  (DNS), captures all the length scales of turbulence in the flow.  DNS of turbulent two-phase flows  at  a  leadership  computing  facility  allows  achievement  of  unprecedented  level  of  detail and can answer fundamental questions about the interaction between the complex and evolving interfaces of the bubbles and droplets and turbulence in the flow.

The detailed simulation of all turbulent structures using DNS approach as well as interface evolution in turbulent flows (in form of bubbles, droplets, wavy interfaces between liquid and steam in various flow regimes) will allow the collection of statistical information about two-phase  flow  parameters  over  an  unprecedented  range  of  conditions.  Advanced  data  collection coupled with interface tracking will be used to process the large data sets.

Three subprojects will generate two-­phase flow results highly relevant to nuclear reactor flows: (i) bubbly flow through pressurized water nuclear reactor fuel bundles with complex geometry  spacers,  (ii)  complex  two-­phase  flow  regimes  (slug  and  churn  flow)  as  well  as  (iii) annular  flows  to  the  highest  level  of  detail.  Major  parameters  will  be  compared  to experimental  data  for  validation,  while  the  higher  order  statistics  produced  by  the simulations will bring new knowledge about two-­phase flows, such as spectral information about gas/liquid interaction, bubbles and droplets contribution to the turbulence.

In this project, the complex physical phenomena will be captured in unprecedented detail. The  level  set  interface  tracking  method  and  on-­the‐fly  mesh  adaptation  will  be  used  to  improve  locally  the  mesh  resolution  of  phase  interface  without  the  need  to  store intermediate  files.  In-­situ  co-processing  of  the  data  will  also  be  used  to  accelerate  the  analysis. Advanced analysis tools will allow tracking every bubble and collect information about  their  behavior.  This  approach  will  allow  partitioning  and  recognizing  different  patterns and phase interactions for two‐phase flows in complex geometries. 

This data processing will generate information about vapor/liquid interaction, enabling the creation of new multiphase fluid models and closure laws that can be used in engineering simulations, allowing reliable application of multiphase computational fluid dynamics (M-­CFD)  to  nuclear  reactor  systems.  This  will  improve  the  safety  margin  predictions  for existing reactors and facilitate the development and design of next-generation systems.