A novel isolation curtain to reduce turbine ingress heating and an advanced model for honeycomb labyrinth seals

摘要

A combination of 3-D and 2-D computational fluid dynamics (CFD) modelingas well as experimental testing of the labyrinth seal with hexagonal honeycomb cells onthe stator wall was performed. For the 3-D and 2-D CFD models, the hexagonalhoneycomb structure was modeled using the concept of the baffle (zero-thickness wall)and the simplified 2-D fin, respectively. The 3-D model showed that even a small axialchange of the tooth (or honeycomb wall) location, or a small circumferential change ofthe honeycomb wall location significantly affected the flow patterns and leakagecharacteristics especially for small tooth tip clearance. Also, the local details of the flowfield were investigated.The seven basic procedural steps to develop a 2-D axisymmetric honeycomblabyrinth seal leakage model were shown. Clearly demonstrated for varying testconditions was the 2-D model capability to predict the 3-D honeycomb labyrinth flowthat had been measured at different operating conditions from that used in developing the2-D model. Specifically, the 2-D model showed very close agreement with measurements. In addition, the 2-D model greatly reduced the computer resourcerequirement needed to obtain a solution of the 3-D honeycomb labyrinth seal leakage.The novel and advanced strategy to reduce the turbine ingress heating, and thusthe coolant requirement, by injecting a ??coolant isolation curtain?? was developednumerically using a 3-D CFD model. The coolant isolation curtain was applied under thenozzle guide vane platform for the forward cavity of a turbine stage. Specifically, theisolation curtain serves to isolate the hot mainstream gas from the turbine outer region.The effect of the geometry change, the outer cavity axial gap clearance, thecircumferential location of the injection curtain slot and the injection fluid angle on theingress heating was investigated. Adding the chamfer to the baseline design gave asimilar or higher maximum temperature T*max than did the baseline design withoutchamfer, but implementation of the injection curtain slot reduced substantially T*max ofthe outer region. In addition, a more desirable uniform adiabatic wall temperaturedistribution along the outer rotor and stator surfaces was observed due to the presence ofthe isolation curtain.