Otras aproximaciones a la experiencia empírica

Two studies stand out in the literature surveyed. The first concerns the combination of axisymmetric endwalls with an airfoil designed to redistribute work in an attempt to decrease the overall loss in the blade row. The second study is focused on the same result but with lean, sweep and non-axisymmetric endwall contouring in combination.

Duden et al. (1999) investigated the use of work redistribution through thickening of the blade profiles at hub and tip in combination with axisymmetric endwall contouring of a tapered cascade typical of a low pressure turbine in the high speed cascade rig at the Bundeswehr University in Munich. A 26% reduction in secondary losses was achieved but this gain was largely eliminated by an increase in inlet boundary layer losses as a result of the thicker endwall profiles. The reduction in secondary losses resulted from a reduced span-wise pressure gradient at midspan, which restricted secondary losses to regions closer to the endwalls while the reduced load at hub and tip increasing the midspan load where a thin and efficient blade profile was employed. The effect of the contoured endwall was most felt in the reduction of over- and under-turning in the exit flow angle which was already improved with just the airfoil redesign. Duden and Fottner (1999) took this study one step further adding off design incidence (both negative and positive) to the test matrix.

This latter study showed that the secondary flow structures increased in radial extent for all blades with increasing incidence or load and that the lightly loaded off-design case saw little or no benefit, and were even slightly worse with the addition of first the airfoil improvements and then the contoured endwall. The three-dimensional designs did however act more favourably at high loads with most of the benefit resulting from the airfoil design rather than the endwall.

In probably one of the more comprehensive of such studies Bagshaw et al. (2005, 2008a &

b) investigated the use of first reverse compound lean and then combined this with forward sweep at the endwall leading edge of the blade before non-axisymmetric endwalls were added. The design intention was to reduce midspan losses through reverse compound lean, then reduce the suction side endwall loading with the application of sweep in preparation for profiled endwalls intended to further reduce secondary losses strengthened by the application of reverse compound lean initially. Two different profiled endwalls were implemented at the hub and tip respectively despite evidence that the design of the one impacted on the other. The outcome was to show that as much as 3% more loss reduction could be achieved through the application of all three of these techniques rather than pure endwall design and that the combination was 8% more effective than simply applying compound lean. Interestingly these gains were achieved despite the presence of a small separation on the endwall which was not predicted during the CFD design process.

A number of other studies such as that of Sharma et al. (2003) and the extensive cascade testing at Carleton University (Mahallati et al., 2007, Knezevici et al., 2008 & 2009, Zoric et al. 2007a & 2007b, and Praisner et al., 2007) also represent combination studies to some extent but are discussed in other sections of this survey for expediency.

2.5 Overview

The scale and structure of secondary flows in turbines have been discussed in some detail along with the state-of-the-art in terms of the means employed to reduce the effects of secondary flows.

From the literature it is clear that the majority of methods, particularly blade lean and sweep as well as axisymmetric endwall design are unsuccessful in reducing overall loss but instead redistribute the loss towards the midspan. Improvements may be realised from the

more uniform flows entering the downstream rows however. It is also clear that few of these methods have been extensively studied at off-design conditions or in the rotating frame.

Localised twist adjustments and non-axisymmetric endwall design appear to offer a solution which directly influences endwall secondary flows by attempting to reduce the endwall cross passage pressure gradient. Bagshaw et al. (2008b) have also shown that non-axisymmetric endwall contouring can be used in combination with other approaches to exploit the benefits of methods which reduce midspan loss but are known to increase endwall secondary flows and losses without suffering the full penalty of this technique.

This study addresses two gaps in the current knowledge base on the effect of non-axisymmetric endwalls, firstly the effect of rotation and the correct annular, three dimensional geometry and secondly the effect of off design flow conditions on the effectiveness of non-axisymmetric endwalls. A number of smaller geometric features are naturally present in these tests and are consistent with industrial turbines, these include tip gaps and split lines in the endwalls.

Furthermore this thesis uses generic endwall contours rather than customised profiles and is therefore a severe test of the robustness of the non-axisymmetric endwall approach.

Finally, this work represents a unique new test case for the validation of turbine endwall secondary flow models. The complete geometry is open for use and is given in the Appendices. Inherent in the analysis of the results therefore is an evaluation of the quantities used as objective functions in the design of endwall loss reduction geometries.

Pressure fields and flow streamlines are used to provide a physical understanding of the extent and nature of the effect of the introduction of the profiled endwalls on the flowfield with the rotor row.