The primary goal of this study was to develop and experimentally validate the methodology of labyrinth seals optimization concerning leakage. The problem was investigated using the ANSYS CFX commercial software.
This paper presents the methodology and results of the optimization of a straight-through labyrinth seal with two inclined fins against smooth-land. The optimization was performed using commercial tools implemented in the ANSYS Workbench environment, such as Goal-Driven Optimization (GDO). The response surfaces were created based on Latin hypercube samples found from CFD calculations. The CFD solver — Ansys CFX, using a steady-state scheme with the k-omega Shear Stress Transport turbulence model, was applied. The CFD model was previously validated concerning spatial discretisation and turbulence modelling. A screening algorithm was used to find the best candidates on the response surfaces. The objective function adopted in the labyrinth seal optimization was the minimization of the discharge coefficient value. A wide range of parameters of the fins position and shape, such as the angles, heights and widths, were taken into account, with physically justified degrees of freedom. The leakage reductions being the effect of the optimization were considerable. The cuts in the discharge coefficient significantly exceed the uncertainties of the CFD model and the test rig accuracy. The factors that have the strongest impact on the leakage reduction in are the inclination, thickness of the fin tips, and the distance between fins.
The optimization results were supported with the results of an in-house experiment performed on a stationary, linear test rig. The specimens tested experimentally were on the same scale (1:1) as the optimised ones. The test rig was fed by a high-capacity vacuum air blower, which made it possible to reach critical pressure ratios, with high-precision hot wire anemometry (HWA) mass flow evaluation. The measuring system also enabled assessment of the pressure distribution along the labyrinth structure. The experimental testing results were compared to the CFD calculations and the optimization effects, highlighting some specific tendencies in the labyrinth seal flow behaviour. Good agreement was obtained between the optimization results and the experimental data, which proves that the presented methodology is sufficient for the labyrinth seal optimization. The same methods will also be applied to more sophisticated sealing structures.