ENGINE COMBUSTION NETWORK: COMPARISON OF SPRAY DEVELOPMENT, VAPORIZATION, AND COMBUSTION IN DIFFERENT COMBUSTION VESSELS

Abstract

Development and mixing of Diesel sprays are long known to be key factors for combustion and pollutant
\nemissions but the related measurements in a real engine is not an easy task. This fact led researchers
\nto simulate engine conditions in special facilities that allow the use of high-fidelity diagnostics. The
\nEngine Combustion Network (ECN) has focused on overcoming the variability from one institution
\nto the next by testing nominally identical Diesel injectors in four different facilities for the first time,
\nincluding constant-pressure flow and constant-volume preburn chambers. Liquid- and vapor-phase
\npenetration, ignition delay, and lift-off length measurements are compared with similar experimental
\nsetups and processing methodologies. The consistency of the data obtained indicates a good level of repeatability
\nbetween the test rigs employed, and no deviation of the results can be associated with the
\nfacility type. Comparison of liquid length measurements via Mie scattering shows that this diagnostic
\nis sensitive to the orientation of the light source. For more repeatable results between facilities, diffused
\nback-illumination imaging is recommended. A novel image processing method has been employed to
\ndetect spray boundaries obtained in high-speed schlieren imaging: the method showed high accuracy
\nand robustness to the different schlieren setups employed by the institutions. High-speed broadband
\nchemiluminescence, as schlieren imaging, shows the onset of cool flame, and moreover when the combustion
\nis stabilized, it provides an important reference to define ignition delay and lift-off length. The
\nmethodology put in place by the ECN participants in this work allows an important step forward in
\ntwo directions. The first is to understand the repeatability related to experimental data in high-pressure,
\nhigh-temperature environments. The second is to advance the understanding of the different diagnostics
\napplied, thereby providing more quantitative measurements that yield to a more suitable datasets
\nfor computational fluid-dynamic model evaluation.