Studies on free stream turbulence as related to gas turbine heat transfer: A review of authors' past work and future implications

A review of the past work done on free stream turbulence (FST) as applied to gas turbine heat transfer and its implications for future studies are presented. It is a comprehensive approach to the results of many individual studies in order to derive the general conclusions that could be inferred from all rather than discussing the results of each individual study. Three experimental and four modeling studies are reviewed. The first study was on prediction of heat transfer for film cooled gas turbine blades. An injection model was devised and used along with a 2-D low Reynolds number k-ε model of turbulence for the calculations. Reasonable predictions of heat transfer coefficients were obtained for turbulence intensity levels up to 7%. Following this modeling study a series of experimental studies were undertaken. The objective of these studies was to gain a fundamental understanding of mechanisms through which FST augments the surface heat transfer. Experiments were carried out in the boundary layer and in the free stream downstream of a gas turbine combustor simulator, which produced initial FST levels of 25.7% and large length scales (About 5-10 cm for a boundary layer 4- 5 cm thick). This result showed that one possible mechanism through which FST caused an increase in heat transfer is by increasing the number of ejection events. In a number of modeling studies several well-known k-ε models were compared for their predictive capability of heat transfer and skin friction coefficients under moderate and high FST. Two data sets, one with moderate levels of FST (about 7%) and one with high levels of FST (about 25%) were used for this purpose. Although the models did fine in their predictions of cases with no FST (baseline cases) they failed one by one as FST levels were increased. Under high FST (25.7% initial intensity) predictions of Stanton number were between 35-100% in error compared to the measured values. Later a new additional production term indicating the interaction between the turbulent kinetic energy (TKE) and mean velocity gradients was introduced into the TKE equation. The predicted results of skin friction coefficient and Stanton number were excellent both in moderate and high FST cases. In fact these model also gave good predictions of TKE profiles whereas earlier unmodified models did not predict the correct TKE profiles even under moderate turbulence intensities. Although this new production term seems to achieve the purpose, it is the authors' belief that it is the diffusion term of the TKE equation, which needs to be modified in order to fit the physical events in high FST boundary layer flows. The results of these studies are currently being used to come up with new diffusion model for the TKE equation.