In this study, a combined experimental and Large Eddy Simulation (LES) investigation is performed to identify the vortical structures, their dynamics, and interaction with a turbulent premixed flame in a swirl-stabilized combustor. Our non-reacting flow experiment shows the existence of large scale precessing motion, commonly observed for such flows. This off-axis precessing dynamics disappears with combustion but only above a critical equivalence ratio at which the flame attaches to the swirler centerbody and vortex breakdown changes from a cone to a bubble type. For compact flames stabilized along the inner shear layer (ISL), no precessing is seen, but large scale vortices along the ISL are observed; these structures interact with the ISL-stabilized flame and contribute to its wrinkling as revealed by laser-induced fluorescence data. After validating the LES results in terms of low order statistics and point temperature measurements in relevant areas of the flow, we show that it can capture the precessing motion in the non-reacting flow and its suppression with combustion. The simulations show that the ISL vortices in the reacting case originate from a vortex core that is formed at the swirler’s centerbody. This vortex core has a conical helical shape that interacts—as it winds out—with the ISL and the flame stabilized along it. The simulated helical vortex core (HVC) exists in both reacting and non-reacting flows; in the latter, it is dominated by the off-axis motion, whereas in the reacting case, that motion is damped and only remains the cork-screw type solid body rotation of the HVC.
A new mathematical analysis of particle and droplet clustering in a fluctuating vortex flow is presented. Two-dimensional particle and droplet dynamics equations are solved numerically and analytically, assuming an oscillating vortex flow field. Droplet grouping was found to occur in the vicinity of an oscillating vortex and new streak-like grouping patterns of droplets in a vortex are disclosed. The combination of circumferential and radial fluctuations produces the most significant effects on the droplet grouping, facilitating the construction and destruction of the streak patterns, depending on the chosen flow field and droplet properties. A new analytical formulation is derived using perturbation analyses to model the swirl fluctuations on the spray, providing a simple method to evaluate the effect of droplet dispersion and droplet clustering under fluctuating vortex flows. This approach was found to be in good agreement with the numerical results in cases of low Stokes numbers for both evaporating and nonevaporating droplets. It is shown that evaporation tends to increase the droplet fluctuation amplitude with time and therefore intensifies the grouping effect. The combination of oscillation and evaporation was found to have the most significant effect on droplet grouping, where the amplitude of the proximity curve rapidly increases in a nonlinear manner. Despite the simplicity of the current model its predictions provide insight into the driving mechanisms behind the much more complex turbulent spray-combustion regime, in which similar droplet grouping patterns occur.
Laws of large numbers guarantee that given a large enough sample from some population, the measure of any fixed sub-population is well-estimated by its frequency in the sample. We study laws of large numbers in sampling processes that can affect the environment they are acting upon and interact with it. Specifically, we consider the sequential sampling model proposed by Ben-Eliezer and Yogev (2020), and characterize the classes which admit a uniform law of large numbers in this model: these are exactly the classes that are online learnable. Our characterization may be interpreted as an online analogue to the equivalence between learnability and uniform convergence in statistical (PAC) learning. The sample-complexity bounds we obtain are tight for many parameter regimes, and as an application, we determine the optimal regret bounds in online learning, stated in terms of Littlestone’s dimension, thus resolving the main open question from Ben-David, Pál, and Shalev-Shwartz (2009), which was also posed by Rakhlin, Sridharan, and Tewari (2015).
1 December 2017 in International Journal of Multiphase Flow