58338 - Effect of Droplet Starting Conditions on the Spray Dispersion Resulting From a Swirl Cup Injector 

In order to reduce the environmental pollution, it is essential to improve the efficiency and to reduce
the emissions of jet engines. In this context the design of the combustion chamber is crucial. Hence, it
is important to predict the resulting pollutant emissions precisely during an early design phase. The
fuel atomization process is essential for the local air-fuel ratios and thus the formation of emissions.
Up to now models for predicting the atomization process based on first principles are not industry
standard. Consequently, the analysis of the pollutant formation requires detailed experimental investigations
for each operating point. Experiments in this context are difficult to perform due to
extreme thermodynamic conditions in combustion chambers and the limited optical accessibility of
dense spray. Although high-fidelity simulations of primary atomization are nowadays possible, simulating
realistic combustion chamber geometries with a sufficient number of breakup events, however,
would considerably exceed current computing capacities.

In most cases Euler-Lagrange simulations are used for predicting the spray propagation and dispersion
neglecting primary atomization. In this work the spray dispersion resulting from Euler-Lagrangian
simulations of an annular prefilming airblast atomizer (swirlcup design) is presented. The dispersion of
the spray is mainly driven by the gas flow including the recirculation zone and the precessing vortex
core. But also the droplet starting conditions imposed at the fuel spray nozzle are affecting the spray
placement.

Consequently, realistic droplet starting conditions are needed to reproduce the spray characteristics.
However, the most numerical studies don't account for stochastical processes involved in primary
atomization and these statistics are not reflected by droplet starting conditions. Typically only the
statistic of the initial droplet size is included and all statistics concerning initial velocities and positions are neglected [3, 4]. Such droplet characteristics may be provided by primary atomization models. These droplet breakup models use analytical and experimental knowledge to predict droplet starting characteristics based on the conditions of the gaseous flow.

In the present work a numerical investigation of the sensitivities of the spray dispersion to different
droplet starting parameters in a realistic three-dimensional fuel injector geometry is investigated.
The simulations are carried out using the Euler-Lagrange method. The primary atomization model
PAMELA [2] is used to predict the droplet diameter. In addition user-defined normal distributions
for the droplet starting position and velocities are imposed.

The simulated results allow the comparison of drop size and velocity distributions with PDA data
[1]. In the study droplet starting conditions were determined which provide optimal agreement to the
experimental data. Thereby, a strong influence of the initial droplet velocities, the recirculation zone
and the precessing vortex core as well as the turbulence modelling approach on the spray dispersion
was identified.

[1] E. Bärow et al. "Effect of the Precessing Vortex Core on Primary Atomization". In: Zeitschrift für Physikalische Chemie 229.6 (Jan. 2015). doi: 10.1515/zpch-2014-0619

[2] G. Chaussonnet et al. "A new phenomenological model to predict drop size distribution in Large-Eddy Simulations of airblast atomizers". In: International Journal of Multiphase Flow 80 (Apr. 2016), pp. 29-42. doi: 10.1016/j.ijmultiphaseflow.2015.10.014.

[3] J. Keller et al. "Numerical and experimental investigation on droplet dynamics and dispersion of a jet engine injector". In: International Journal of Multiphase Flow 75 (Oct. 2015), pp. 144-162. doi: 10.1016/j.ijmultiphaseflow.2015.05.004.

[4] J. Senoner et al. "Eulerian and Lagrangian Large-Eddy Simulations of an evaporating two-phase flow". In: Comptes Rendus Mécanique 337.6-7 (June 2009), pp. 458-468. doi: 10.1016/j.crme.2009.06.002.