GT DRAFT DEVELOPMENT OF NEW HIGH AN 2 LAST LP STAGE TURBINE & EXHAUST SYSTEMS A COST EFFECTIVE SOLUTION FOR THE 21 ST CENTURY

Transcription

1 Proceedings of TURBO EXPO 2005: ASME Turbo Expo: Land, Sea & Air 2005 June 6-9, 2005, Reno-Tahoe,Nevada, USA GT DRAFT DEVELOPMENT OF NEW HIGH AN 2 LAST LP STAGE TURBINE & EXHAUST SYSTEMS A COST EFFECTIVE SOLUTION FOR THE 21 ST CENTURY Brian Haller Siemens Industrial Turbomachinery, Ruston House, PO Box 1,Waterside South, Lincoln LN5 7FD, UK brian.haller@industrial-turbines.siemens.com ABSTRACT A new high AN 2 last LP stage turbine has been developed to provide leading performance for turbomachines in the 21 st Century. It required a multi-disciplinary design approach involving aerodynamics, new materials (forged γ- Titanium Aluminide alloy), mechanics (stress and vibration) and new manufacturing technologies. The objective of the design was to achieve around 5% gain in last stage total-to-static efficiency,relative to the current competitive datum, for a very compact machine with few parts count ie reduced cost. It will be shown that the new material allows the optimum aerodynamic design for the stage to be achieved which cannot be done with conventional nickel-based alloys. The paper will present details of the novel approaches used for the design including preliminary optimization, blading design and results from multistage 3D viscous predictions. The new stage has been tested in a Warm Air Turbine test rig at full scale engine representative conditions. The loop was closed by comparing the detailed test measurements with 3D viscous analyses. This gives high confidence in the new approaches used. Additionally, the development of new compact high performance axial-radial and axial exhaust systems,which were designed to operate downstream of the new last LP stage, are described. The new LP technologies have already been scaled and cloned to a conventional material design and sold by the Company. Nomenclature A exhaust annulus area D diameter (h=hub, t=tip) LE leaving energy m mass flow rate N shaft rotational speed P01 inlet stagnation pressure T01 inlet stagnation temperature U mean blade speed Va axial velocity x exhaust loss coefficient H stage total-to-total heat drop Introduction The purpose of this Technology project was to exploit the material properties of new forged γ-tial alloy (PX5-300 ie TNB V5) to produce a new high efficiency/cost effective LP stage design. Forged TiAl has around ½ the density of conventional nickelbased alloys with higher specific properties. Therefore much longer LP stage designs can be used with higher efficiency. The material also allows the optimum aerodynamic designs to be used eg pitch/width ratio and number of blades etc. Its elongation at room temperature is satisfactory (around 1%) and it can be ground/5-axis milled by specialist techniques. This is the same material which has revolutionized the performance of F1 racing cars and is also finding applications in turbochargers, other IC engine components etc. Initially cast TiAl material was investigated but this had problems in the casting (cracks, cavities..) and had around half the specific strength of the forged material. Also casting requires a thicker blade trailing edge and other thickening compromises which increases weight. Specifically the objectives were : - develop new high AN 2 /high load TiAl LP stage to provide stage total-to-static efficiency gain of 5% at a total-to-total pressure ratio of around 4:1 cf. current production design - achieve better off-design performance characteristic and explore extending to even higher load conditions. 1 Copyright 2005 by ASME

2 - achieve life and reliability first time (34,000 hour, 99% availability, 2400 normal starts) - halve the cost for LP rotor blades - (also the new material can be used to reduce weight in aero engines and to reduce disc stresses for the same blade height). Note that a) the design is fully shrouded for improved vibration control and to reduce overtip leakage losses and b) the design inlet relative stagnation temperature to the rotor is around 650 deg C at mid-height. Note that the TiAl material is currently limited to around 700 deg C at the tip. This new LP stage was designed to operate downstream of the new HP stage described in Ref. 1 in a two-stage overhung turbine configuration. A pressure ratio of around 16:1 and above can be achieved with only 2 turbine stages. This gives large advantages in terms of : - cost/parts count - performance - length and weight - being able to achieve the optimum downstream axial exhaust performance - assembly - on-site maintainability etc... New stage design The new LP stage and exhaust systems detailed in this paper is patent applied for and pending (Ref. 2 and 3) and has the following overall design characteristics: Dt/Dh=1.72 AN 2 =60*10 12 mm2 rpm2 (with solid LP rotor ) Va/U=0.47 Loading H/U 2 =1.55 m& T Inlet swallowing capacity 01 = Kg/s K 1/2 bar -1 P 01 This is an increase in exhaust area of around 25% cf conventional design ie 50% lower leaving energy. The amount of disc/root blade cooling is very small (this affects AN2 which can be achieved).the design pressure ratio of the stage is around 4:1 but it has the capability to operate efficiently to higher load conditions. It was originally designed for a 5 MW class single-shaft industrial gas turbine but the new methods and technologies are generic and it can be scaled/cloned to any application. The methods and philosophies used to design the stage and exhaust systems are detailed in the following sections. Stage layout and throughflow design Optimising the stage design involved aerodynamic and mechanical disciplines with a number of iterations to establish the LP rotor base radius and exhaust area. The next step was to optimise the stage layout. For the last stage the important performance parameter is the overall total-to-static efficiency of the LP stage and exhaust system This involves maximising the LP stage total-to-total efficiency and minimising the exhaust loss parameter (x.le/ isentropic stage total-to-total heat drop). A Controlled Flow design philosophy (see Ref. 1) was used for the LP NGV to : - reduce secondary loss and give the highest total-to-total efficiency for the stage - match the flow at outlet from the HP stage, ensure a consistent design philosophy throughout the turbine and maintain a smooth streamline pattern. Three alternative types of LP rotor design philosophy were studied via a streamline curvature throughflow code : - constant relative outlet angle, beta2 - beta2 increasing towards the tip (relative to the axial direction ) to reduce the leaving energy from the LP stage. Differing amounts of twist were studied. - Controlled Flow vortex distribution (this is the first time it has been tried for a last LP stage ). The Controlled Flow rotor was selected as this gave : - best flow structure and efficiency - consistent design philosophy and streamline pattern throughout the whole turbine - ideal stagnation pressure distribution into the exhaust system to minimise the losses (see later section). The layout of the new stage installed in the Warm Air Turbine test Rig is shown on Fig.1. In this case for the first test there is an Inlet Guide Vane which simulates the flow (swirl) delivered from the upstream HP stage. Fig.1 Layout of the new LP stage and axial-radial exhaust in the Warm Air Turbine Test Facility U 3 optimisation A U 3 optimisation study (for method see Ref. 4 and 5 ) was carried out for the LP rotor and LP NGV using the velocity 2 Copyright 2005 by ASME

3 triangles from the throughflow calculations. This a physicallybased method. Space precludes the publication of all the U 3 results as around 1000 alternative sections were calculated for the NGV and rotor. This gives the optimum pitch/width for the blades and optimum profile designs at root, mean and tip ie ideal velocity and turning distributions through the passages. It is not possible to do this optimisation with conventional correlations or Navier Stokes codes.fig.2 shows the results for the rotor where the stage efficiency debit due to profile and secondary losses for the rotor is plotted against pitch/axial width at the root. Stage Efficiency debit(tial rotor) Variation of stage efficiency debit with (s/w)root (s/w)root TiAl rotor Fig.2 Summary of results from U 3 optimisation study on LP rotor Conventional shrouded nickel-based designs have to use much lower values of pitch/width at the root which increases the profile losses and lowers the stage total-to-total efficiency (see Fig.3 ). There were only 36 LP NGV s and 53 LP rotor blades. This lowers the transonic trailing edge blockage losses. Previously 52 LP NGV s were used and a nickel-based LP rotor design typically has around 80 blades. Thus the changes are very substantial. (pitch/axial width)root Diameter ratio=dt/dh TiAl GT 1(unshrouded) GT 2 GT3 ST 1 ST 2 ST 3 Fig.3 Comparison of root (pitch/axial width) for shrouded LP blade designs (GT= gas turbine, ST=steam turbine). All the designs are shrouded apart from GT1. 3D design and analysis using multistage 3D RANS code Having decided on the skeletal parameters for the stage, the blading was designed using the in-house XDESIGN system. This is a full interactive Q3D transonic design system with cloning, inverse, stacking options etc. It includes a finite element potential flow solver, MISES (with base drag losses) and integral/differential boundary layer methods. It is very fast and well validated over 10 years. The blade profiles are defined by Nutbourne parametric cubic splines with continuous smooth curvature Fig. 4 and 5 show the 3D views of LP NGV and LP rotor. Both blades are Controlled Flow. The LP rotor was stacked on the centroid. The LP NGV has : Tangential lean of +10 degrees (pressure surface pointing radially inwards) Taper Tilt All the above were to optimize the flow structure and maximize the efficiency. The performance of the stage was then analysed using a number of 3D multistage viscous methods - Stage3d (adapted from Dawes), MULTALL (Denton), TF3D (Li He) by 3 independent specialists. All results showed very good flow structure and performance. 3 Copyright 2005 by ASME

4 Fig.5 Isometric view of high AN2 LP rotor The grid for the Stage3d calculation engine configuration is shown on Fig.6.This case used a standard H-grid but a multiblock curvilinear grid is also available. The Baldwin- Lomax turbulence model was used. The relative Mach numbers contours for the 2 stages are shown on Fig.7. It is clear that all the blades are transonic /high load/high lift designs. As an example of the level of checking, the relative velocity vectors onto the LP rotor at mid-height are shown on Fig.8 It can be seen that the stagnation point is correctly aligned with no incidence onto the blade. All the sections were checked in this manner. Fig.4 Isometric view of LP NGV Fig.6 Meridional view of grid for HP and LP stages for multistage 3D RANS calculation Fig.7 Relative Mach number contours at mid-height 4 Copyright 2005 by ASME

5 New compact high performance axial-radial and axial exhaust systems In the axial-radial exhaust the flow is diffused and turned through 90 degrees. The flow on the lip governs the exhaust performance since this is where the velocity is lowest by simple streamline curvature considerations. The dynamic head is lost in the large exhaust plenum and so the stagnation pressure at outlet pipe discharge is simply equal to the static pressure at the separation point on the lip. Therefore for the new axial-radial exhaust diffuser, the peak velocity on the lip was minimized in order to provide higher recovery. A conventional circular arc lip design will accelerate the inlet flow due to its curvature, give a higher peak velocity on the lip and therefore always give separation at a higher velocity (see Fig. 10). Fig.8 Relative velocity vectors onto LP rotor leading edge at mid-height The root reaction for the stage was optimum at around 20% (healthy and positive otherwise large root separation losses can occur). In terms of reaction, the shocks losses were balanced in the NGV and rotor to obtain the highest stage efficiency. The radial distribution of stagnation pressure delivered by the LP stage into the exhaust is shown on Fig. 9. It is ideal for high exhaust performance being very flat with a slight increase at the tip. Fig.10 Static pressure distributions on hub and casing of original circular arc axial-radial diffuser and new design (lip are lower 2 curves and hub are higher 2 curves) Fig.9 Radial distribution of stagnation pressure delivered into exhaust In the new axial exhaust a higher performance can be achieved since the flow is not turned 90 degree. Again, improvements can be achieved via an understanding of the boundary layer physics eg see Ref..6. Initially when the starting turbulent boundary layer is thin and has small momentum thickness it can be diffused more rapidly without danger of separation. Gradually as the surface boundary layer thickens the diffusion is reduced. This enables diffusion down to a lower exit velocity (ie higher pressure recovery ) and the ability to achieve the recovery in a smaller axial distance. Initially the turbulent boundary layer shape factor is around 1.4, is diffused up to around 2.0 and held constant at this value along the surface length of the diffuser on both hub and tip. This provides margin against any danger of separation.. The diffuser was designed using an axisymmetric streamline curvature throughflow method coupled to an axisymmetric integral boundary layer method.it was then finally checked in a 5 Copyright 2005 by ASME

6 multistage viscous calculation with the upstream turbine stages. This also enabled the strut angles to be optimized. The principles are similar to those used for the design of Controlled Diffusion compressor blades. On Fig. 11 the distinctive shape of the new axial exhaust is shown together with the streamline pattern. The hub is a simple cone and the outer wall shape is designed to be concave to give the Optimised Diffusion velocity distribution. The inner cone has to flare outwards to avoid the swirl from the turbine being magnified. It can be seen that the area increases more rapidly initially. Fig. 12 shows the the velocity distributions on the hub and tip of the Optimised Diffusion diffuser. It is also an ideal distribution along the surface of the hub of the diffuser because it is mainly controlled by the 1D area distribution. The static pressure distributions on the hub and tip are shown on Fig.13 and the recovery can be seen. The outer wall shape is easily made by spinning. The new axial exhaust gave very high predicted recovery and is currently being tested on 2 engines. Fig.11 Streamlines for new Optimised Diffusion axial exhaust system 160 Fig.13 Static pressure distributions on hub and tip of new Optimised Diffusion axial exhaust Test results The new LP stage and axial-radial exhaust has been tested in a full scale Warm Air Turbine Test Rig at full scale and correct Reynolds numbers /Mach numbers (see Fig.1).In the rig the rotor blade was made from Titanium. The power produced by the turbine is absorbed by a hydraulic dynamometer and full area traversing carried out at LP stage exit and IGV exit using a 5-hole probe. The rig is extensively instrumented. Tests were made over a range of speed and pressure ratio with two tip clearance levels. Typical conditions are around : stagnation pressure at inlet = 5 bar stagnation temperature at inlet = 650K mass flow rate = 35 Kg/s rotational speed = rpm shaft power = 4.4 MW ABS.CON. velocity (m/s) x (mm) Tip Hub The picture of the balanced LP rotor assembly is shown on Fig.14 The quality of the test hardware is evident. The rig measurements confirmed the original design predictions and that the targets had been surpassed. Presently the loop is being closed by comparing all the detailed rig measurements (swallowing capacity, efficiency, radial distributions of flow parameters, off-design performance, exhaust recovery) with 3D analysis predictions. Fig.12 Optimised diffusion velocity distributions on hub and tip of new axial exhaust 6 Copyright 2005 by ASME

7 The detailed material test results will be published in a specialized materials paper. New manufacturing technologies and processes Fig.14 Picture of LP rotor assembly for Warm Air Turbine Rig Test Material test results Extensive material tests have been carried out over a number of years including : Tensile Creep Low cycle fatigue High cycle fatigue Pre-exposure tests at 400, 550, 700 deg C Coating tests Advanced heat treatment Wear and fretting tests for shroud for shroud interlock face Residual stress measurements caused by the manufacturing processes Strength (MPa) % PS (MPa) TS (MPa) 0.2% PS (MPa) Cast 4722XD (Mean) TS (MPa) Cast 4722XD (Mean) Poly. (TS (MPa)) Poly. (0.2% PS (MPa) Min) Temperature ( C) Poly. (0.2% PS (MPa)) Poly. (TS (MPa) Min) Fig.15 Tensile strength of PX5-300, extruded material 14:1 As an example, Fig.15 shows the tensile strength of the forged material compared to 4722XD cast TiAl material plotted against temperature up to 800 deg C. The much higher strength of the forged material is obvious. Fig.16 Picture of forged/ground/machined TiAl last LP rotor blade (this is conventional blade / material substitution design for spin-rig and engine mechanical strain-gauge tests) Fig.16 shows a picture of a finished blade. The manufacturing processes are : - Raw billet production at foundry - Forging and heat treatment - Grinding of blade roots - 5-axis NC machining of aerofoil (possibly ECM could also be used) - Grinding of shroud On Fig. 16 some waviness on the profile caused by the milling is evident. This was within the manufacturing tolerances for the blade but the waviness could be halved at a small additional cost. The trailing edge thickess is 0.7 mm and no additional thickening relative to the conventional cast nickel-based design was required. No problems were experienced with the blade manufacture. However forged TiAl is a different material and therefore requires different sealed processes for all aspects in its life (manufacture, handling, assembly, testing etc). Conclusions A new design philosophy for low pressure turbines, utilising Controlled Flow / U 3 optimisation /high AN 2 has been demonstrated. New compact high performance axial-radial and Optimised Diffusion axial exhaust systems have been 7 Copyright 2005 by ASME

8 presented. The aerodynamics of the new LP stage and axialradial exhaust was validated via a Warm Air Turbine Rig Test. The new axial-radial and axial exhaust systems are also presently being validated on engines. Very significant step changes have been obtained and the original aggressive design targets exceeded. The loop is being fully closed by comparing all the detailed rig and engine measurements with the detailed test information.. Acknowledgements The author wishes to thank UK DTI New and Renewables Energy Programme for funding this work. Gordon McColvin, Materials Technology Manager provided Fig.15. References 1. Haller B.R. Anderson J. Development of new high load/high lift transonic shrouded HP gas turbine stage design a new approach for turbomachinery, ASME paper GT Haller B.R. Gas turbine low pressure stage, UK Patent Application GB A, 23/7/ Haller B.R. A diffuser for diffusing the exhaust gas produced by an engine, UK Patent Application 2004P01249GB, July Denton J.D. Loss mechanisms in turbomachines, ASME 93-GT Haller B.R. VKI Lecture Series on Secondary and Tip Clearance Flows in Axial Turbines, Full 3D turbine blade design, February Stratford B.S. An experimental flow with zero skin friction throughout its region of pressure rise, J. Fluid Mech, Vol. 5, 17-35, Copyright 2005 by ASME