(428c) CFD Modelling of Multi-Regime Multiphase Flows

In many CFD simulations today, we would
specify the flow regime of the flow we are considering so that we can select
the appropriate modelling method and physics models for the simulation. For
examples, we would use the Volume of Fluid (VOF) model for free surface flows,
the Lagrangian particle tracking model for droplet and particle flows and the
Eulerian multiphase (EMP) model for bubbly flows. However, many real life flows
have a combination of different flow regimes that occur at different locations
and they often change with time. For example, a stratified gas-liquid flow
along a horizontal pipe, see Fig 1. As the gas velocity is increased, waves
start to form at the interface and grow into slug flow. At the front and back
of the slugs, gas is entrained into liquid to form bubbles and liquid into the
gas to form droplets. In this case, we have a combination of stratified flow
(ideally modelled using VOF), droplet and bubbly flows (using EMP). The slugs
can form at any location along the pipe at any time and once formed they travel
at a high speed along the pipe, so the flow regime changes with time and position.
We can find many flows in different industries that have multiple flow regimes,
for example boiling flows in a vertical channel, see Fig 2.

For these multi-regime flows, single
flow-regime models are obviously not sufficient. Also, it is usually not
possible to pre-describe the flow regime according to position as illustrated
by the slug example above. Ideally, we want our CFD model to determine the flow
regime and apply the appropriate models as part of the calculation. The development
of such a multi-regime multiphase flow model is the focus of this paper.

The generic nature of the Eulerian
multiphase (EMP) model allows it to represent different flow regimes in one CFD
model. What we need is to build into the EMP solution procedure a method that
can recognize and determine flow regime and with that select and assemble the
appropriate physics models (such as interphase forces, sharp interface treatments,
etc.) into the solution matrix for each computational cell. An important step
in developing the multi-regime model is the ability to recognize, determine and
therefore treat large scale interfaces (LSI) between two fluids. In the LSI
regions we can apply the “VOF” type of numerical treatment, bringing in surface
tension force model, including surface damping of turbulence, etc. The LSI
region also marks the transition between two different flow regimes on either
sides of it. In gas-liquid flows, it would usually be bubbly flow on the
liquid-dominated side and droplet flow on the gas-dominated side. For these
bubbly and droplet flow areas, we can apply the standard models (such as lift,
drag, turbulence dispersion and virtual mass forces) for these flows.

In this paper we present some key aspects
of large scale interface detection using the newly developed Adaptive Interface
Sharpening (ADIS) numerical scheme together with demonstration and validation
examples. Industrial flow examples are also provided to illustrate to what
level of details the current model can capture and in so doing pointing to
areas where further R&D work are needed.

Fig 1   Gas-liquid flow in a horizontal
pipe

Fig 2   Boiling flow in a vertical channel