Numerical investigation of hot streaks in turbines

Abstract

A numerical method is developed for simulation of hot-streak redistribution in a two-dimensional model of a turbine rotor. The flow domain is divided into a viscous region near the blade where the Reynolds-averaged, thin shear-layer Navier-Stokes equations are solved using an implicit finite-volume technique, and an inviscid core region where the Euler equations are solved using an explicit finite-volume method. The computational mesh consists of an O-mesh and an H-mesh patched together smoothly to cover the domain of interest. Computations are performed using two different flow conditions. The first test case uses hot streaks with a temperature ratio of 1.2 and a tangential inflow angle of 40 deg, whereas the second test case is run with a temperature ratio of 2.0 and a tangential inflow angle of 45 deg. The computed solution from both test cases predicts a migration of hot gas to the pressure surface, which also has been observed experimentally. MODERN jet engine is designed to have an extremely high temperature gas leaving the combustor. The temper- ature tolerance of the guide vanes is usually based on an average value of the exit combustor temperature. Due to intro- duction of cooling air in the nozzle, the turbine entry temper- ature (TET) measured behind the guide vanes is lower than the combustor exit temperature. The temperature tolerance of the blades in the first rotor row is based on an averaged TET. Recent investigations, however, have shown that hot gas mi- grates to the pressure surface of the rotor blade. This can lead to peak temperatures on the rotor blade that might exceed acceptable metal temperatures and hence lead to blade fail- ures. This indicates that the TET, which is currently used, is too low an estimate for the rotor surface temperature. In an earlier work by Butler et al., 1 an experimental and analytical investigation of the redistribution process for an axial turbine stage was presented. In the experiment, a streak of hot air seeded with CO2 was introduced at one circumferen- tial location upstream of the inlet guide vane. The redistribu- tion of the hot streak was determined by measuring the con- centration of CO 2 inside the turbine stage. Measurements of CO2 taken on the rotor surface indicated that hot and cold gas had been segregated with the cold gas migrating to the suction surface and the hot gas to the pressure surface. In the same paper it was postulated that the segregation effect was due to the difference in rotor relative inlet angles of the hot and cold gases. The postulate is based on an observation by Kerrebrock and Mikolajczak2 in their work on wake transport in compres- sors. The observed segregation phenomenon1 has been investi- gated numerically by Rai and Dring.3 Their paper includes a comparison between Rai's computations and Butler's mea- surements. The computations did not give the same high tem- perature excess as was observed experimentally. The dis- crepancies between calculations and experiment were blamed on three basic differences in flow conditions: 1) the experi- ment was fully three-dimensional; 2) the temperature ratio between hot and cold gas was 1.2 in the calculations and 2.0 in