journal article·2019

GEOMETRY OPTIMIZATION OF AIR-ASSISTED SWIRL NOZZLE BASED ON SURROGATE MODELS AND COMPUTATIONAL FLUID DYNAMICS

Zongli Yi, Li Hou, Qi Zhang, Yousheng Wang, Yunxia You

Atomization and Sprays· Volume 29· Issue 7
Publisher page

Abstract

Fuel atomization plays a vital role in aeroengine precombustion. In this study, we propose optimization models to improve the atomization performance of air-assisted swirl nozzles during startup phase under low pressure. We have investigated the effect of four crucial geometrical parameters: tilt angle of cyclone slot, ratio of swirl chamber diameters, diffuser expansion semi-angle, and airway helix angle, on the spray angle and liquid film thickness at exit. We selected 60 training samples and 21 test samples by optimal Latin hypercube sampling for this analysis. The spray angle and liquid film thickness were simulated using the volume of fluid method and Reynolds stress model. Based on the Kriging method optimized by the whale optimization algorithm, two surrogate models for spray angle and liquid film thickness were established. Subsequently, the optimal combination of geometrical parameters was obtained through a multiobjective evolutionary algorithm. We have proved that the optimized Kriging model can efficiently predict the impact of the four crucial parameters on spray angle and liquid film thickness. Compared to the characteristics of original air-assisted swirl nozzle, the optimized spray angle increased to a range of 85° to 91° and the liquid film thickness reduced by 40.7% (73 μm). The feasibility of the optimized structure of air-assisted swirl nozzle was validated by computational fluid dynamics.
Source: openalex_inverted_index · Quality: baseline

Research context

OpenAlex topics: Combustion and flame dynamics · Advanced Combustion Engine Technologies · Cyclone Separators and Fluid Dynamics

SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.

References

Reference metadata was not supplied for this record.