Mixed Helical Labyrinth Groove Seal Optimization Using Computational Fluid Dynamics

Abstract

Non-contacting annular seals are used in rotating machinery to reduce the flow of fluid across a pressure differential. Helical and labyrinth groove seals are two types of non-contacting annular seals frequently used between the impeller stages in a pump. Labyrinth seals have circumferential grooves cut into the surface of the rotor, the stator, or both. They function to reduce leakage by dissipating kinetic energy as fluid expands in the grooves and then is forced to contract in the jet stream region. Helical groove seals have continuously cut grooves in either or both of the rotor and stator surfaces. Like labyrinth seals, they reduce leakage through dissipation of kinetic energy, but have the added mechanism of functioning as a pump to push the fluid back towards the high pressure region as it tries to escape. Previous work has shown that both labyrinth and helical groove seals with grooves on both the rotor and the stator have lower leakage than seals with grooves on just one surface. The goal of this work is to analyze seals with helical grooves on one surface and labyrinth grooves on the other. Designs for both helical stator, labyrinth rotor and labyrinth stator, helical rotor will be simulated and the performance of each configuration will be compared. The primary variables considered for the designs of the seals include the width, depth, and the number of grooves for labyrinth seal and the width, depth, and the angle of the grooves for helical. The designs to simulate will be chosen using a Kennard-stone algorithm to optimally space them within the design space. Then, for both configurations, multi-factor quadratic regression models will be generated. Backward regression will be used to reduce the models to only statistically significant design parameters. From there, the response surfaces will be created to demonstrate the effects of each design parameter on the performance of the seal. Finally, optimal designs will be produced based on the regression models. These designs will be simulated to show the predictive power of the regression models. The simulations for this work will be run in ANSYS CFX for each seal type and configuration will be used to compare solutions for the two different types of designs to previous studies. The findings from this study is expected to show substantial decrease in leakage for a mixed helical-labyrinth seal in comparison to the seal with either helical or labyrinth grooves on both surfaces and thus will provide useful results needed to minimize amount of leakage and therefore improve the efficiency of the machine.