Hot gas ingestion through rim-seals (also commonly known as ingress) and into the wheel-space between rotating and stationary discs is a key concern and an important factor the engine designer should take into account when designing the turbine stage of an engine. The flow past the stationary vanes and rotating blades in the turbine annulus creates an unsteady 3D variation of pressure radially outwards the rim-seal. Ingress and egress occur through those parts of the seal clearance where the external pressure is higher and lower, respectively, than that in the wheel-space.Most computational studies of ingress have been conducted using unsteady flow. This report describes a computational study carried out to better understand ingress using a steady-state model which has the advantage of being faster and which requires significantly less computational resources. The ingested flow within a turbine stage was obtained by modelling the annulus alongside the wheel-space in a multi-domain simulation. Previous studies have shown that when modelling ingress via a steady model and using the blades in a fixed, or frozen position, an uncertainty arises as to where to position the blades relative to the vanes and how many positions are required to properly capture the ingested flow effect. In the present study, a non-bladed, 3D, steady CFD model was employed to eliminate the uncertainty and complexity of fixing the rotor blade at a given position, whilst also reducing the mesh size for the rotor domain.The simulations were carried out using the commercial CFD code, ANSYS CFX. Initially the code was used to calculate the flow structure of an axisymmetric rotor-stator system with no ingress, validating the fluid dynamics within the wheel-space against experimental data in terms of axial distributions of radial and tangential velocity. The CFD model was extended to include the annulus with only the stationary vane and excluding the rotating blade, in which the non-axisymmetric pressure variation in the annulus was seen to cause ingress. The pressure distribution in the annulus was analysed for different rotational Reynolds number computations, differing sealing flows and varying ratios of axial Reynolds number to rotational Reynolds number; in all the cases showed good agreement with experimentally measured data at the University of Bath.The axial-clearance seal was the main seal used to verify the non-bladed CFD method, which was extensively validated against experimental data. Validation was performed between the computed results and the measured data within the wheel-space in terms of radial distributions of core swirl ratio, and the sealing effectiveness and static pressures on the stator wall. The computations showed very good agreement with experiments for swirl ratio and pressure, while the sealing effectiveness showed reasonable agreement with experiments.The non-bladed CFD method was also used to perform computations for radial-clearance and double-clearance seals. Engine representative seals, known as double-clearance seals were also computed, where an outer wheel-space is created by the inner seal, which is seen to act as a damping chamber which holds and contains most of the ingested fluid, hence improving the sealing effectiveness and reducing the swirl within the inner wheel-space. The computed results for all the seals show a good prediction of the flow structure (swirl ratio) within the wheel-space, and reasonable agreement with experiments for the sealing effectiveness. In parallel with the computational programme, experiments were performed in the research facility which models hot-gas ingestion into the wheel-space of an axial turbine stage. The performance (i.e. sealing effectiveness) of three different radial-clearance seals was measured via CO2 gas concentration measurements. The radial-clearance seals are differentiated as follows: a baseline radial seal, a radial seal with a decreased axial overlap, and a radial seal with a tighter radial clearance. A non-dimensional sealing parameter, which combines the different effects of rotating flows inside the wheel-space into a single parameter, Φo is used for the presentation of the data. The measurements show that increasing the axial overlap and decreasing the radial clearance increases the effectiveness of the radial-clearance seal. The non-bladed CFD model is a useful 3D tool for a turbine design engineer. The lack of experimental facilities and/ or data would make this 3D model an indispensable CFD tool for quantitatively predicting the flow structure within the wheel-space, and qualitatively predicting the sealing effectiveness for any given rim-seal geometry.
|Date of Award
|12 Mar 2014
|Gary Lock (Supervisor) & Michael Wilson (Supervisor)