Atomisation using liquid jets in crossflow is widely used and empirical models are relied on in many industries. A typical assumption in these models is that jet breakup takes place at a fixed location. This paper uses experimental and simulation methods together to investigate unsteady variation in the jet breakup location. Shadowgraphy data is presented for jets with momentum ratios from 4.4 to 95. Three of these cases (4.4, 8.8 and 63) are also simulated using a CLSVOF methodology. Good agreement is seen for jet penetration, surface instability wavelength, and qualitative agreement for phenomena such as bifurcation of the jet column. These validated simulations are then used to study the breakup. The predicted mean location of the end of the liquid column is approximately 9 jet diameters downstream, within the range of previously reported values. However, this location is seen to be highly unsteady, with the end point of the column moving by up to 50% of the mean. The process is shown to consist of instabilities leading to large, highly deformed, ”myriapodal” masses being released from the column which subsequently undergo further breakup. The release of these masses causes SMD downstream to fluctuate by ± 20 % . The masses have a strong interaction with the airflow, which affects the spray trajectory. Trajectories from this work are compared with correlations from the literature which are seen to vary in their agreement. The variation in position of the jet end point suggests that empirical models could be improved by including this phenomenon. • Unsteadiness of liquid jet in crossflow quantified using shadowgraphy and CLSVOF simulations. • Breakup point shown to vary by up to 50% of the mean value in the streamwise direction. • Revised description of breakup process given characterising the breakup based on 3 lengthscales. • “Myriapodal masses” released from the end of jet containing majority of liquid mass and momentum. • Temporal variations in SMD shown to be caused by jet instability that determines breakup process.
Wetherell et al. (Fri,) studied this question.