Session: 10-05-01: Multiphase Flows

Paper Number: 98995

98995 - Computational Protocol for Simulations of Spray Flows Including Primary Atomization 

Primary atomization is the key element in spray flow simulations.  We have, in our previous work, used and validated the integral form of the conservation equations, leading to the “quadratic formula” for determination of the drop size during spray atomization in various geometry.  A computational protocol has been developed where this formulation is adapted to existing computational frameworks for continuous and dispersed (droplet) liquid phase, for simulations of pressure-atomized sprays with and without swirl.  In principle, this protocol can be applied to any spray geometry, with appropriate modifications in the atomization criterion.  The pre-atomization continuous liquid motion (e.g. liquid column or sheet) is computed using volume-of-fluid (VOF) or similar methods, then the velocity data from this computation is input to the quadratic formula for determination of the local drop size.  This initial drop size, along with the local liquid velocities from VOF, is then used in a Lagrangian tracking algorithm for the post-atomization dispersed droplet calculations.  This protocol can be implemented on coarse-grid, time-averaged simulations of spray flows, and produces convincing results when compared with experimental data for pressure-atomized sprays with and without swirl.  This approach is general, and can be adapted in any spray geometries for complete and efficient computations of spray flows.

Existing algorithms for the continuous liquid (liquid jet column or sheet) and dispersed (droplet) phases are quite effective if initial and boundary conditions are fully specified.  Volume-of-fluid or level-set methods reproduce the liquid motion and deformation quite well at flow length scales.  However, small-scale liquid shearing leading droplet formation occurs at length scales mostly below the spatial resolution of large-eddy or turbulence model simulations.  Also, existing primary atomization modules require time-dependent tracking of the liquid-gas interface even for steady-state injection processes, straining the computational resources.  On the other side of the atomization, if the droplet initial size and velocities can be specified as some local initial conditions, then algorithms such as particle-in-cell (Eulerian-Lagrangian) methods work quite well in tracking the droplet trajectories.  Mass and energy transfer modules can also be added with good accuracy, to study evaporation and combustion processes in sprays.  The difficulty arises in computation of the droplet formation processes or specifications of the drop size.  The key component is the primary atomization module to link the pre- and post-atomization simulations. 

 

Here, the analytical results described in the introduction have several attributes that are ideal as the primary atomization module during computational simulations of spray flows.  First, the method is based on fundamental fluid physics of mass and energy conservation.  Secondly, it has been generalized and validated across all the major injection geometries.  The fact it requires liquid velocity as inputs to the current analytical results points to the ideal integration of this method with existing computational algorithms that can generate accurate velocity (momentum) data.  Also, time-averaged velocities can be used as inputs, so that steady-state simulations are sufficient for the pre-atomization VOF component.  For transient injection processes, unsteady VOF can be run.  Here, we only consider steady-state spray flows.  Therefore, we can run pre-atomization liquid-phase simulations using VOF or other methods to compute the continuous liquid velocity development from the injector exit to the atomization point.  The atomization location is specified using suitable physical criterion as described below.  Then, the liquid velocities from VOF are utilized to specify the local droplet size so that dispersed simulations can be initiated using this droplet size along with the local liquid velocity for the post-atomization part.  Some of the results in various injection geometries will be presented.

 

1. Run liquid-phase simulations (e.g. VOF) to obtain liquid volume fraction and velocities for the pre-atomization liquid flow field.

2. Apply the atomization criterion (described below).

3. Use the local liquid velocities in D₃₂ equation to find the local initial drop size.

4. Run the Lagrangian discrete particle simulations to track the droplets, using the above initial drop size (Step 3) and velocities (from Step 1).

 

 

Presenting Author: Taewoo Lee Arizona State Univ

Presenting Author Biography: A Ph.D. candidate in Mechanical and Aerospace Engineering, Arizona State University

Authors:

Taewoo Lee Arizona State Univ
j.e. Park Arizona State University