Skip to content

MultiFlow Research Group

Computational Modelling of Multiphase Flows

Understanding and predicting the behaviour of flows involving multiple phases — liquids, gases, bubbles, droplets and particles — is one of the central challenges in fluid mechanics and engineering. Such flows are ubiquitous in nature and industry, from the breakup of fuel sprays in combustion engines and the rise of bubbles in chemical reactors, to sediment transport in rivers and the dynamics of aerosols in the atmosphere. Accurate numerical simulation of these phenomena is essential for advancing our fundamental understanding and for the design of efficient industrial processes.

The MultiFlow Framework

MultiFlow is an in-house finite-volume framework for the high-fidelity simulation of multiphase flows in complex geometries, developed continuously since 2000. It is built on a fully coupled, pressure-based algorithm that solves the governing flow equations simultaneously as a single linear system — delivering superior robustness and convergence compared to traditional segregated approaches, particularly in the presence of large density contrasts, strong interfacial forces, and complex source terms.

The framework is capable of simulating a broad range of flow configurations and regimes:

  • Single- and multiphase flows - from creeping Stokes flows to turbulent regimes
  • Incompressible and compressible flows - across the full Mach number range, from the incompressible limit to supersonic and hypersonic conditions
  • Interfacial flows - using a balanced-force Volume-of-Fluid (VOF) approach or Front-Tracking (FT) approach with accurate surface tension treatment, including implicit methods that overcome the capillary time-step restriction
  • Dispersed flows - via Eulerian-Lagrangian methods for particle- and droplet-laden flows
  • Flows in complex geometries - the code supports any type of control volume element, including tetrahedral, hexahedral, polyhedral and mixed meshes on arbitrarily complex domains
  • Immersed boundary method - for tracking moving boundaries, enabling particle-resolved direct numerical simulation (PR-DNS) of flows with freely moving solid bodies
  • Discrete Element Models (DEM) - state-of-the-art collision models for both spherical and non-spherical particles, enabling coupled fluid-particle simulations at the grain scale
  • Viscoelastic flows - a fully coupled solver for non-Newtonian viscoelastic fluids, capturing the complex rheological behaviour of polymer solutions and melts
  • Flows on adaptive meshes - with octree-based adaptive mesh refinement for efficient resolution of multiscale phenomena

MultiFlow relies on the PETSc library for the parallel solution of the linear equation systems, enabling efficient simulations on large-scale distributed-memory HPC clusters.

Research Group

The MultiFlow group is based at the Chair of Mechanical Process Engineering at Otto-von-Guericke University Magdeburg. Our work spans fundamental algorithm development, high-performance computing, and applications in chemical engineering, energy systems, and environmental flows.