Martin G. Rose

Improving the Efficiency of the Trent 500 HP Turbine Using Non-Axisymmetric End Walls: Part 1 — Turbine Design

This paper describes the redesign of the HP turbine of the Rolls-Royce Trent 500 engine, making use of non-axisymmetric end walls. The original, datum turbine used conventional axisymmetric end walls, while the vane and (shrouded) rotor aerofoil profiles were nominally the same for the two designs. Previous research on the large scale, low speed linear cascade at Durham University, see Hartland et al [1], had already demonstrated significant potential for reducing turbine secondary losses using non-axisymmetric end walls - by about one third. This paper shows how a methodology was derived from the results of this research and applied to the design of the single stage Trent 500 HP turbine (model rig). In particular the application of a new linear design system for the parametric definition of these end wall shapes, described in Harvey et al [2], is discussed in detail.Copyright

NON-AXISYMMETRIC ENDWALL PROFILING IN THE HP NGV's OF AN AXIAL FLOW GAS TURBINE

The Nozzle Guide Vanes (NGVs) of an axial gas turbine generate circumferential non-uniformities in static pressure after their trailing edges. This paper aims to show that these non- uniformities can be removed locally by re-shaping the hub endwall. The ultimate objective is to improve engine performance by cutting the leakage of coolant from the disc rim seal. The technique is demonstrated theoretically using 3D viscous CFD. The required endwall shape is non-axisymmetric and is generated parametrically by altering a calculation grid for an existing NGV. The results show a reduction of the static pressure non- uniformities by 70%. Other favourable effects are also predicted and are commented upon. NOMENCLATURE

Non-Axisymmetric Turbine End Wall Design: Part II — Experimental Validation

The Durham Linear Cascade has been redesigned with the non-axisymmetric profiled end wall described in the first part of this paper, with the aim of reducing the effects of secondary flow. The design intent was to reduce the passage vortex strength and to produce a more uniform exit flow angle profile in the radial direction with less over turning at the wall. The new end wall has been tested in the linear cascade and a comprehensive set of measurements taken. These include traverses of the flow field at a number of axial planes and surface static pressure distributions on the end wall. Detailed comparisons have been made with the CFD design predictions, and also for the results with a planar end wall. In this way an improved understanding of the effects of end wall profiling has been obtained.The experimental results generally agree with the design predictions, showing a reduction in the strength of the secondary flow at the exit and a more uniform flow angle profile. In a turbine stage these effects would be expected to improve the performance of any downstream blade row. There is also a reduction in the overall loss, which was not given by the CFD design predictions. Areas where there are discrepancies between the CFD calculations and measurement are likely to be due to the turbulence model used.Conclusions for how the three-dimensional linear design system should be used to define end wall geometries for improved turbine performance are presented.Copyright

Non-Axisymmetric Turbine End Wall Design: Part I — Three-Dimensional Linear Design System

A linear design system, already in use for the forward and inverse design of three-dimensional turbine aerofoils, has been extended for the design of their end walls.This paper shows how this method has been applied to the design of a non-axisymmetric end wall for a turbine rotor blade in linear cascade. The calculations show that non-axisymmetric end wall profiling is a powerful tool for reducing secondary flows, in particular the secondary kinetic energy and exit angle deviations. Simple end wall profiling is shown to be at least as beneficial aerodynamically as the now standard techniques of differentially skewing aerofoil sections up the span, and (compound) leaning of the aerofoil. A design is presented which combines a number of end wall features aimed at reducing secondary loss and flow deviation. The experimental study of this geometry, aimed at validating the design method, is the subject of the second part of this paper.The effects of end wall perturbations on the flow field are calculated using a 3-D pressure correction based Reynolds Averaged Navier-Stokes CFD code. These calculations are normally performed overnight on a cluster of work stations. The design system then calculates the relationships between perturbations in the end wall and resulting changes in the flow field. With these available, linear superposition theory is used to enable the designer to investigate quickly the effect on the flow field of many combinations of end wall shapes (a matter of minutes for each shape).Copyright

Time Resolved HWA Measurements of the OTL Flow Field From a Shrouded Turbine HP Rotor Blade

Detailed time resolved Hot Wire Anemometry (HWA) measurement have been made of the over tip leakage (OTL) flow field, from a shrouded HP rotor blade. The objective of this study being to investigate the sensitivity of the flow conditions presented to the IP Nozzle Guide Vane (IPNGV), resulting from changes to the HP blade tip gap clearance and shroud geometry. An extensive experimental data set has been produced in the outer 25% of the annulus height. Post-processing of this data has allowed both the absolute & rotor relative velocity and whirl angle to be calculated, which has been presented in both a graphical and semi-animated form. A detailed analysis of the flow field has been undertaken, which has been compared with the exit flow field predicted by a comprehensive CFD simulation of the HP blade and shroud. Good agreement between the measured and the predicted flow field has been obtained. This has permitted the salient features, which influence the flow conditions presented to the IPNGV, to be identified. Advanced post-processing of the time-resolved data, has enabled some of the Reynolds and Deterministic stress components to be evaluated. A comparison of the magnitude of these components has been made, which the objective of assessing their relative importance in multistage steady CFD calculations.Copyright

Measurement and Calculation of Nozzle Guide Vane End Wall Heat Transfer

A 3-D steady viscous finite volume pressure correction method for the solution of the Reynolds averaged Navier-Stokes equations has been used to calculate the heat transfer rates on the end walls of a modern High Pressure Turbine first stage stator.Surface heat transfer rates have been calculated at three conditions and compared with measurements made on a model of the vane tested in annular cascade in the Isentropic Light Piston Facility at DERA, Pyestock. The NGV Mach numbers, Reynolds numbers and geometry are fully representative of engine conditions. Design condition data has previously been presented by Harvey and Jones (1990). Off-design data is presented here for the first time.In the areas of highest heat transfer the calculated heat transfer rates are shown to be within 20% of the measured values at all three conditions. Particular emphasis is placed on the use of wall functions in the calculations with which relatively coarse grids (of around 140,000 nodes) can be used to keep computational run times sufficiently low for engine design purposes.Copyright

Turbomachinery Wakes: Differential Work and Mixing Losses

In this paper the mixing of stator wakes in turbomachinery is considered. An extension is made to the existing model of Denton (1993), which addresses the effects of acceleration before mixing. Denton showed that if a total pressure wake was accelerated mixing loss diminished and vice versa. Here a total temperature wake is shown to exhibit a reverse trend. An attempt is also made to better understand the work transfer process between a stator wake and a rotor. The paper concentrates on axial turbines, but a brief look at compressors is included. It is argued that the freestream work is not the same as the wake work and the concept of ‘Differential Work’ is introduced. A simple steady velocity triangle based model is proposed to give an estimate of the ratio of wake work to freestream work (μ see later). The model is compared to an unsteady CFD result to offer some verification of the assumptions. It is concluded that the rotodynamic work process tends to reduce total pressure wake depths in turbines and compressors and therefore mixing losses. The mixing loss due to total temperature wakes is less strongly affected by the differential work process.Copyright