Experiments were carried out in a model air turbine stage to study the influence of rotor-stator rim cavity configuration on the ingestion of mainstream gas into the cavity. The three rim cavity configurations differed in their aspect ratio (height/width); the rim seal geometry remained the same. The aspect ratio was changed from the baseline ratio by installing an inner shell on the stator at an appropriate radius; this effectively introduced an axial-gap seal between the rim cavity and the cavity radially inboard. The initial step in each experiment was the measurement of time-average static pressure distribution in the turbine stage to ascertain that proper flow condition had been established. Subsequently, tracer gas concentration and particle image velocimetry techniques were employed to measure the time-average but spatially local main gas ingestion and the instantaneous velocity field in the rim cavity. At low purge air flow, regions of ingestion and egress could be identified by inspecting the instantaneous radial velocity distribution near the rim seal obtained from cavity gas velocity maps close to the stator. While the tangential velocity tended to be slightly larger for the so determined ingested gas, a more clear-cut indicator of ingestion was the strong inward gas radial velocity. Information provided by ensemble-average velocity maps was not sufficient for identifying ingestion because the averaging smeared out flow details, which varied from instant to instant. Velocity fields obtained from three-dimensional, time-dependent numerical simulation of a rim seal-cavity sector with similar dimensions qualitatively showed similar characteristics in the outer part of the cavity and provided insight into the complex flow in the seal region.

The ingestion of mainstream gas into turbine rotor-stator disk cavities and simultaneously, the egress of cavity gas into the main gas path are consequences of the prevailing unsteady, three-dimensional flow field. To understand these processes, we are carrying out a study that combines experiments in a model single-stage axial turbine with computational fluid dynamic (CFD) simulations. The turbine stage features vanes, blades, and axially overlapping radial clearance rim seal. In this paper, we present time-resolved velocity maps, obtained by particle image velocimetry, of the flow in the disk cavity at four experimental conditions as defined by the main air flow rate, rotor speed, and purge air flow rate. Time-averaged but spatially local measurement of main air ingestion is also presented. Significant ingestion occurred at two of the four experimental conditions where the purge air flow rate was low — it is found that high tangential (swirl) velocity fluid intersperses with lower tangential velocity fluid in the rim region of the cavity. It is argued that the high tangential velocity fluid is comprised of the ingested main air, while the lower tangential velocity fluid is the indigenous cavity air. This interpretation is corroborated by the results of the unsteady, three-dimensional CFD simulation. When the purge flow rate was high, no ingestion occurred as expected; also, large-scale structures that were unsteady appeared in the cavity flow giving rise to large velocity fluctuations. It is necessary to obtain time-resolved information from experiments and computation in such a flow because even when the vane-blade relative position is matched during a particular experiment, the instantaneous flow field does not necessarily remain the same. As such, some of the flow patterns will be smeared out if the interrogation time scale is large.

Experiments were carried out in a model air turbine stage to study the influence of rotor-stator rim cavity configuration on the ingestion of mainstream gas into the cavity. The three rim cavity configurations differed in their aspect ratio (height / width), the rim seal geometry remained the same. The aspect ratio was changed from the baseline ratio by installing an inner shell on the stator at an appropriate radius; this effectively introduced an axial-gap seal between the rim cavity and the cavity radially inboard. The initial step in each experiment was the measurement of time-average static pressure distribution in the turbine stage to ascertain that proper flow condition had been established. Subsequently, tracer gas concentration and particle image velocimetry techniques were employed to measure, respectively, the time-average but spatially local main gas ingestion and the instantaneous velocity field in the rim cavity. At low purge air flow, regions of ingestion and egress could be identified by inspecting the radial velocity distribution near the rim seal obtained from cavity gas velocity maps close to the stator. While the tangential velocity tended to be slightly larger for the ingested gas, a more clear-cut indicator of ingestion was the strong inward gas radial velocity. Information provided by ensemble-average velocity maps was not sufficient for identifying ingestion because the averaging smeared out flow details which varied from instant to instant. Velocity fields obtained from three-dimensional, time-dependent numerical simulation of a rim seal–cavity with the same dimensions qualitatively showed similar characteristics in the outer part of the cavity and provided insight into the complex flow in the seal region.