Application of big data analysis technique on high-velocity airblast atomization: Searching for optimum probability density function

In this paper, the droplet size distributions of high-velocity airblast atomization were analyzed. The spray measurement was performed by a Phase-Doppler anemometer at several points and different diameters across the spray for diesel oil, light heating oil, crude rapeseed oil, and water. The atomizing gauge pressure and the liquid preheating temperature varied from 0.3 to 2.4 bar and 25 to 100 °C, respectively. Approximately 400 million individual droplets were recorded; therefore, a big data evaluation technique was applied. 18 of the most commonly used probability density functions (PDF) were fitted to the histogram of each measuring point and evaluated by their relative log-likelihood. Among the three-parameter PDFs, Generalized Extreme Value and Burr PDFs provided the most desirable result to describe a complete drop size distribution. With restriction to two-parameter PDFs, the Nakagami PDF unexpectedly outperformed all the others, including Weibull (Rosin-Rammler) PDF, which is commonly used in atomization. However, if the spray is characterized by a single value, such as the Sauter Mean Diameter, i.e. an expected value-like parameter is of primary importance over the distribution, Gamma PDF is the best option, used in several papers of the atomization literature.

Keywords

Big data, Airblast, Rapeseed oil, PDA, Probability density function, Likelihood

Citation

FUEL. 2020, vol. 273, issue 1, p. 1-12.
https://www.sciencedirect.com/science/article/pii/S0016236120307870

Document type

Peer-reviewed

Document version

Published version

Language of document

en

Document licence

Creative Commons Attribution 4.0 International
http://creativecommons.org/licenses/by/4.0/