 |
Accurate prediction of admixture distribution in space and in size is important for spray modelling in applications ranging from medical aerosols to fuel sprays in internal combustion engines. Modelling polydisperse sprays with inertial particles/droplets is particularly challenging as these types of flows may have trajectory crossing and caustics. The family of Eulerian-Eulerian methods are shown to be computationally expensive for these types of flow. The conventional approach to simulate admixture with low volume fraction is Lagrangian particle tracking, which implies direct calculation of the number of particles/droplets in a computational cell. The focus of this study is investigation of droplet dynamics and evaporation based on the Fully Lagrangian Approach (FLA). According to this method, particles/droplets are treated as a continuum (or a set of continua), which makes it possible to calculate particle/droplet number density from the continuity equation for the dispersed phase along chosen particle/droplet trajectory.
The study is focussed on the implementation of two-way coupled FLA into open source CFD software OpenFOAM and generalisation of FLA for polydisperse flows. Both strands aiming to develop the FLA for wider use in engineering applications.
The FLA has been implemented as part of OpenFOAM v6 Lagrangian library, allowing it to be used with all OpenFOAM solvers. A new steady-state two-way coupled 2D OpenFOAM solver has been developed and applied to test the FLA implementation in the case of gas-particle flow in a backward-facing step. The effect of mass loading of particles on the size of the recirculation zone has been studied. The predictions of two-way coupled FLA implementations have been compared to the predictions of conventional Eulerian-Lagrangian method (Discrete Parcel Method).
To take into account the effect of polydispersity of droplets, the continuity equation in the FLA has been generalised by introducing a new particle distribution function (PDF). This function represents the distribution of droplets over space, time and sizes. The set of Lagrangian variables in this case increases to include initial droplet size. This approach has been applied to a number of 1D and 2D flows of evaporating droplets with and without trajectory crossing.
