Flash-boiling

Pressurised liquids undergoing rapid phase change and intense bubble nucleation, strongly affecting atomisation, spray structure, and overall system performance in injectors, propulsion systems, and energy applications

Technical capabilities in MPflow

MPflow provides a dedicated pressure–velocity coupling algorithm specifically designed for flashing flows in internal geometries and nozzle-like configurations. The solver is fully compatible with OpenFOAM libraries and enables unified modelling of internal flow and spray formation.

Key capabilities include:

  • Specialised pressure–velocity coupling for internal multiphase flows
  • Fully coupled internal flow + atomisation + spray modelling
  • Compatibility with OpenFOAM-based workflows and libraries
  • Embedded ML-accelerated turbulence models (RANS & LES) for improved efficiency and scalability
  • Robust modelling of phase change, metastable states, and flashing dynamics

The modelling framework captures the full transition from liquid to spray, including:

  • Bubble nucleation & growth under rapid depressurisation
  • Flow regime transition inside nozzles
  • Primary and secondary atomisation processes
  • Dense and dilute spray behaviour

Our approach CFD

  • High-fidelity simulation of flashing flows using a unified Eulerian–Lagrangian framework

  • Coupled modelling of internal flow and downstream spray evolution

  • Implementation of advanced atomisation physics including:

    • Primary and secondary break-up

    • Droplet collisions and evaporation

    • Thermal non-equilibrium effects

  • Use of surface density (Σ-equation) for accurate tracking of liquid structures

  • Avoidance of simplified assumptions (e.g. pure liquid at nozzle exit)

Physics-based ML acceleration

  • Embedded ML-accelerated RANS and LES turbulence models

  • Reduction of computational cost for highly transient flashing flows

  • Rapid evaluation of multiple operating conditions (pressure, superheat, geometry)

  • Scalable workflows enabling design space exploration and optimisation

Why ML-CFD matters for flash boiling

  • Up to 500-1000× faster predictions, enabling rapid evaluation of injector and nozzle designs
  • 5-7x faster simulations with on-the-fly deployment
  • High accuracy (<5% deviation) in key flow features and spray characteristics
  • Efficient modelling of strongly transient and nonlinear phase-change phenomena
  • Scalable analysis across: Different fuels, Operating pressures and superheat conditions, complex geometries and injector configurations