A NOVEL EXPERIMENTAL TECHNIQUE FOR INVESTIGATING NATURAL CONVECTIVE HEAT TRANSFER IN A GAS TURBINE ANNULUS

When a gas turbine is shut down it cools asymmetrically due to natural convection. After some time, a thermal gradient develops across the rotor drum as the top section cools slower than the bottom. This induces a thermal stress that causes the shaft to deflect a phenomenon known as rotor or thermal bow. In some cases, this can lead to compressor tip rubs. All engines are affected by rotor bow to some degree. However, the mechanisms of natural convective heat transfer and impact of design features on rotor bow is not well understood; thus, the issue is not identified until late in the design stage, hampering an engines development. A novel experimental technique and facility has been developed that allows for accurate, detailed measurements of natural convective heat transfer in a gas turbine annulus to be made. The apparatus has the capacity to test circumferential and axial variations in temperature distribution, in addition to an isothermal wall boundary. An IR camera looking directly at the rotor surface is used to provide high-resolution heat transfer measurements. The test piece for this study consisted of a cylinder section and a conical section. Results from the isothermal cylinder test cases agreed quite well with previously published data, which gives confidence to the method that has been employed. The addition of a conical section had a profound effect on the results; however, more work is required before a method can be developed which will accurately predict the heat transfer coefficients based on initial temperature distribution. A 2D concentric cylinder methodology has also been developed that gives results in strong agreement with previously published data. Going forward, this new facility and the CFD method will be further developed to help quantify the effect of various design features on the natural convective heat transfer performance of a large civil gas turbine.