Factors of Influence on the Leakage of Brush Seals

The continuous improvement of overall efficiency of thermal turbomachinery as well as the efficiency of their sub-systems is a crucial task. One of these sub-systems is the secondary flow system. There is a large variety of challenges to be faced that relate to this system: supplying certain points with fluid, blocking it from others, controlling the axial loads of the turbomachinery’s shaft(s), active clearance control, and anti-icing of jet engines.
    
    All these tasks demand reliable fluid transfer, which is usually supported by fluid seals. A very promising seal in terms of efficiency improvements is the so-called brush seal, which mainly consists of a great number of individual wires with the ability to reversibly deform under the attack of external forces like the ingress of shafts. A main driver for the provided fluid blockage is the movement of the wires in operation and, depending on the movement, their geometrical position in relation to the surfaces of stator and rotor.
    
    To lay the foundation for a deeper understanding of brush seals, a bottom-up strategy, gaining an empirical validated numerical fluid-structure interaction (FSI) model, is required. The first cornerstone of this strategy – the generation of a sub-system model of the bristle package – is to be accomplished by creating a parameterized computational fluid dynamic (CFD) model and analyzing the flow through the bristle package without its movement. Validating this model, experimental results from literature and industrial partners are used.

    Previous research from the authors investigated the influence of the chaotic inner arrangement of the bristle package as well as the influence of different numerical and domain settings on the leakage mass flow rate. Based on the outcome of these studies, an investigation regarding the influence of various geometrical and numerical parameters on the leakage mass flow rate is presented in the final paper. The effects of varying the sealing gap, the backing plate and/or cover ring gap, free bristle height, considering or neglecting of housing components, as well as different dimensional domain settings are discussed. Furthermore, preliminary results regarding the FSI model of a brush seal are presented briefly in the final article.
    
    The results are gained by a MathWorks MATLAB tool – which generates an ANSYS ICEM CFD input script – and a commercial simulation tool (e. g. ANSYS Fluent) to solve the flow field. For the purpose of quantifying the numerical uncertainties the grid convergence index (GCI) is used. Structural simulations are performed using ANSYS APDL and/or a literature based and by the authors enhanced analytical model.
    
    The numerical results and experimental data will be compared and discussed by comparing the mass flow rate, pressure drop and geometrical scale of the system.