Section Leader, Engineering Computing. Manager, Experimental and Hatfield Test Site

Transcription

1 78-GT-193 '1! LIII Util1[I1 The Society shall not be responsible for statements or opinions advanced in papers or in discussion at meetings of the Society or of its Copyright 1978 by ASME Divisions or Sections, or printed in its publications. Discussion is printed only it the paper is published in an ASME journal or Proceedings. Released for general publication upon presentation. Full credit should be given to ASME, the Technical Division, and the $3.00 PER COPY author(s). $1.50 TO ASME MEMBERS 1.00 at Wembley Recent Experience on Centrifugal Compressors for Small Gas Turbines M. G. BEARD Section Leader, Engineering Computing C. M. PRATT Manager, Experimental and Hatfield Test Site P. H. TIMMIS Assistant Chief, Component Engineering - Aerodynamics Rolls-Royce Limited, Aero Engine Division, Leavesden, Watford, Herts, England The design philosophy and development of three recent Rolls-Royce centrifugal compressors, ranging in pressure ratio from 3.6 to 7, are described. The main geometrical and aerodynamic details are given, together with overall performance characteristics of the compressors. Contributed by the Gas Turbine Division of The American Society of Mechanical Engineers for presentation at the Gas Turbine Conference & Products Show, London, England, April 9-13, Manuscript received at ASME Headquarters January 5, Copies will be available until January 1, THE AMERICAN SOCIETY OF MECHANICAL ENGINEERS, UNITED ENGINEERING CENTER, 345 EAST 47th STREET, NEW YORK, N.Y

2 TA31E 1: MAST GEOMETRICAL DETAILS OF TI-P1 COMPRESSORS FEATURE 5:1 co"tpressor 3.6:1 COP+Ij'OEFSOR 7:1 I:v FESSOR Ini'. (cm.( 3.5 (8.89)..978 ( ) (5 %>) Incucer Shroud 1'iarneter in. (cm.) (20.13( 6.42 (16.307) 5.878* (14.93( pc 1 r TiI Li meter ice. (em (29.253( t.9'6 (25.339( 9.1 ( 1 r El wroth in. (cn.).14 (1.123) 225 (0.572(. S a o' L ller Vin-e Full 13 Splitter:> in A at ( DiiSu s -terap Are. m 2 ( 7.5 ' (4 5ta) 1 (19.419) 3.03 (19.548) in FrELOFf se Lerscli r1fa ,x) i4. (35.56) 548 (26.792) (28.204) Dot t Pific_cr Le^din 7dE_ ir_. (cm.) J. 79 (.53"i) i 254 (.645) er 't at D_Sfu e ira lrn, ;1 n. cni.( ' ;95)..7a 9j ) am ei * 'uac r rail F a in. (c_-.) I20. ^O.c) ^. 50.8) V DiLfu,.,r A ia,tio I c er d_;pnetea x.578" (11.16 cn. and tip r:idth " (.864 cm.) with axi-symmetric it take. sure ratio. 3 A 7:1 pressure ratio compressor of near optimum specific speed. A tabulation of the salient geometric parameters of the comrressors is given in Table 1. DESIGN METHOD The compressors were designed by optimizing the velocity triangles at the design point using a performance prediction method wh-ch models the various loss mechanisms within the compressors. The impeller geometry details were determined to meet manufacturing and aerodynamic requirements. Streamline curvature techniques were used to calculate the velocity distributions within the impellers. Radial diffusers and axial straighteners were designed using channel diffuser data l and axial cascade data. Each impeller featured swept back blading in the interests of stability and hence high working line efficiency. The impellers were produced by flank milling on five axis milling machines. Runstadler, P. W., Jr., and Dean, R. C., Jr., Straight Channel Diffuser Performance at High Inlet Mach Numbers, Creare. 2 TEST CONDITIONS AND TEST ANALYSIS The compressors were tested on rigs with throttling at the outlet. Inlet pressure and temperature were approximately the ambient values of 14.7 psi and R. Inlet total pressure to the compressors was measured with four wall static pressure tappings in a large pipe immediately upstream of the compressor. Inlet temperature was measured on a rake of six thermocouples. Downstream total pressures are calculated from continuity and static pressure measured on four inner wall and four outer wall tappings. The Mach number at this plane is very low (0.15 or less). Downstream total temperatures were measured on thermocouple rates which varied in number between the three compressor rigs. The 5:1 compressor and 7:1 compressor each had temperatures measured at 32 points -- various circumferential and radial positions. The 3.6:1 compressor rig used 7 measuring positions which were justified by previous testing with 20 measuring points which showed negligible radial temperature variation. The overall compressor adiabatic efficiency is calculated on a total-to-total basis and uses the measured temperatures directly in

3 Fig. 2 Impeller ( photograph ) 5:1 PRESSURE RATIO COMPRESSOR Fig. 1 5:1 Compressor annulus diagram the stage performance calculations. The impeller tip total pressure for the 5:1 impeller i s calculated from the scan of eight front wall and eight rear wall static pressure tappings, a radial velocity component calculated assuming no boundary layer blockage, a tangential velocity component calculated from the measured temperature rise with an allowance for an estimated windage temperature rise. This unit was designed as a research compressor for a duty appropriate to the first stage of a two-stage centrifugal compressor. Such a compressor is of high specific speed, since it is required to minimize compressor diameter and also to achieve an optimized compressor system with an h-p compressor which is of low specific speed if it is on the sane shaft. The project was restricted to investigating the performance of the impeller with large radius ratio vaned diffusers and it is Table 2 Aerodynamic Details of the Compressors at the Design Point FlFgdvmeSR 5:1 COS.'_FRESOR 3.6:1 CO;:RRESSOR 7:1 C01,1T?FESSOR 8.2.I1. at 288 K 2i1 e: _e qr cra tore Mac "r'1cwli.4 lb/eec. ('g./sec.) 9.5 (4.513) 2.6 (1.18) 5.25 (2.384) _-Iet Swirl ( Pressure eatie Adiabatic FIC_cieac Work Parameterri U= Tip Speed at 288 K Inlet Temperatere =t., (53`^u2 % ( ') Sip absolute,',:acr ito Di'_'aser Lc-acing Edge Mach Sic x at exit _'rom Axial Strai Iz oner _ rico Mace F.Fi % c=s. ^It.F.i r::. 3 ;ee. ) 1ei (4,.89) (27.9) 90. (3 9z) 1 Balje, O.E. A Study on Design Criteria and Hatching of Turbo Machine=. Fart B - Compressor and Fump Performance and Matching of Turbo Components A.S.M.E. Paper FO-WA-231 3

4 r 0% 2 AD DIFFUSER: THROAT AREA 6.75 IN DIFFUSER: THROAT AREA 7.50 IN 90% 0 % 80,o 7' IN ' X0.8` 0.80 ^ 0.8( 0.75 V ( 4H U 6 P P P3 P, 5 ll% 110%y \',, '" 130% 4-100% 4 70'% 800 / 134, 900 RPM SAD ^/ 90r 3 80% 2 70% c ^ 2 ^ % f pfs M j lb/sec Fig. 4 5:1 Compressor second diffuser matching Fig. 3 5:1 Research compressor impeller characteristics not suggested that these would be appropriate for an aircraft engine for which a more compact system would almost certainly be required. The compressor was designed for 5:1 pressure ratio and a mass flow of 9.5 lb/sec at 34,900 rpm. The annulus diagram of the compressor is shown in Fig. 1 and the impeller is illustrated in Fig. 2. The design point aerodynamic details are given in Table 2. In view of the high inducer tip 1,1ach number (1.22) it was expected that the impeller maximum efficiency would occur at a very small incidence and be very close to the inducer choke flow. To obtain good performance at this condition, very little camber was designed into the inducer and a very thin leading edge was used as can be seen in Fig. 2. The impeller was designed with 50 deg sweep back at the tip to give a good surge line and a high working line efficiency. A vaneless diffuser diffused the flow to a Mach number of about at the radial diffuser leading edge and this diffuser and an axial straightener completed the diffusing process. The compressor was tested both as a complete stage and with a vaneless diffuser to obtain the impeller performance over a wide range of flow. The impeller performance derived from tip static pressure measurements on the vaneless diffuser test is shown in Fig. 3. The results show two interesting features. The first is the very high efficiency achieved at the lower speeds where the diffusion in the impeller is reduced. The second point is that the impeller efficiency at design speed does not maximize close to choke but with approximately 10 percent choke margin. The consequence of the latter point was that the stage was not well matched to its design point. The compressor development thus entailed changes to the diffuser throat area to match the impeller. The characteristics for 4

5 TL Fig. 5 Rolls-Royce Gem engine two diffuser capacities are shown in Fig. 4. Both represent changes from the initial design and were achieved by reducing the diffuser blade height by contracting the vaneless space as shown in Fig. 1. Dimensions for each diffuser are included in Table 1. It will be seen that the second matching of the compressor achieved 5 :1 pressure ratio at design speed with 14 percent surge margin at 80 percent efficiency -- only 1.8 percent less than the peak efficiency at this speed. THE HIGH HUB:TIP RATIO COMPRESSOR OF 3.6 :1 PRESSURE RATIO Fig :1 High hub tip ratio compressor vaneless space configuration Combined axial/centrifugal compressors are an attractive configuration for the compressor of small gas turbines. The axial compressor is well suited to the high volumetric flow of the 1-p compressor stage and has high efficiency, small diameter and low weight Whereas the centrifugal compressor is a better choice for the hp than an axial compressor which.would require numerous small blades due to the low volumetric flows and high hub:tip ratio at the compressor outlet. A high hub:tip ratio may be imposed on an h-p centrifugal compressor if a severe swannecked duct between axial and centrifugal compressor is to be avoided. The compressor is also frequently of low specific speed, as even on a two spool engine the high pressure spool shaft speed is limited by mechanical considerations to less than the aerodynamic optimum. The centrifugal compressor of the Rolls- Royce Gem engine, Fig. 5, has high hub:tip ratio and low specific speed for these reasons. The high hub:tip ratio at inlet and the lo,i 5

6 I.? 14 1w ? ADIABATIC L H U metefic speed detract iron the efficiency of the compressor due to the high :vetted area: flow area and the fact that even on the impeller hub surface there is diffusion which is not the case with low hub:tin ratio design. The compressor pressure ratio at design speed has been increased duri.._- develcenent of the compressor from the design value of 3.5:1 to 3.6:1. This has been achieved through increasing the work done by the impeller by increasing the vane height at the impeller tip and hence reducing the relative velocity at impeller outlet. This led to an unusual feature in the vaneless space as the eatchin_r, Fig :1 High hub/tip compressor annulus P " ' "' RPM, ^ ^^ ' l P 16,/sec Fig s 9 5,00:1 High hub/tip characteristics of the increased tip width impeller to a standard diffuser required a contraction in the vaneless space which for mechanical reasons had to occur very close to the impeller tip. The resulting geometry is shown in!gig. 6. This rather unusual feature does n.t appear to cause any performance penalty and indeed, a small improvement to the overall compressor efficiency results from the combined effect of this feature and the increased tip width. The impeller of this compressor is illustrated in Fig. j and the compressor annulus is shoom in Fig. 8. The stage characteristics of the compressor are shown in Fig. 9. It will be observed that on this design, the peak efficiency is quite separate from the surge line. In the course of the development of the compressor, a change was made to the wear-away material on the impeller shroud. Shrouds had been produced finished both in porous aluminum 6

7 , 0.8' ,. \ G H AD i b 0.70 RUNNING CLEARANCE DATUM DATUM INCR ^0-ti5 DATUM INCH 0.75 H-5 P 3 P MICRO - INCH ALUMINIUM SHROUD MICRO INCH NICKEL GRAPHITE SHROUD AH ^' ^ t 0, j A RPM M et lb sec Fip, 10 Effect of roughness on compressor performance which had a smooth surface finish and also in a nickel/graphite compound which was much rougher. It was evident from engine performance that the two standards were giving different compressor performances, and so a comparative rig test was carried out in which the same impeller and diffuser wore tested with shrouds coated in each material. are was taken to ensure the shroud Profiles were identical. The results shown in ii g. 10 show the rougher shroud caused a loss of anproximately 2.5 percent efficiency due mainly to increased temperature rise attributed to increased windage against the rougher material. A isarther test carried out on this compressor varied the impeller tip clearance by axial movement of the impeller relative to the shroud. The results are shown in Fig. 11 and indicate a sensitivity of 1 percent efficiency for 2.7 percent clearance to blade height A ^^ 2.6 RPM R. P. M ", I (k se L M oe lb sec Fig. 11 Tip clearance investigation a project study on a small turboshaft enzine. The enfine was to have a free power turbine and the compressor speed was not limited by turbine mechanical considerations and a near optimum speed for this Pressure ratio was selected. 7:1 PRESSURE RATIO COMPRESSOR This compressor was developed as part of Fig. 12 7:1 Impeller (photograph) 7

8 Fig. 13 7:1 Compressor annulus diagrams (rig ano. engine) ADIABATIC EFFY ^ 50{ PRESSURE RATIO P, P, RIG INTAKE ENGINE INTAKE , 85% 3.0) 90 4 r 10% % 70 H '. 2.01r1^ I \ ^70`%' ra0 600 H 50% 100%=48,5000RPM P. e, Ibisec % 177:2. ornpoesso_, charracterrsticc (r i and engine) A reasonably low inlet hub diameter was also achieved desanite the engine having a front drive from the free norer turbine. Parametric studies had shown are-swirl to be a beneficial feature and the coraoressor was designed r,,ith 2.0 der, of inlet whirl. The reduced inlet relative -,zelocitf to the impeller in conjunction with the preferred amount of diffusion in the impeller, led to a high absolute flow angle at the impeller tip. To ensure vaneless diffuser stall was avoided, an entry contraction was designed into the vanele s diffuser rather similar to that on the Gem ho compressor, but applied on both front and rear walls. The impeller is illustrated in Fig. 12. The compress or was tested in teso standards. Initial rig testong was carried our using an ani oymmetric intake with pre-swirl vanes to induce the pre swirl. Later, rig testing was carried out with an offset ''chin'' intake simulating a turboprop -.-'stake. In this case the pre-swirl was induced 0... cambered 1^ )ered spookes rather than vanes. Detail Charles to the inducer and dif- _reer were also made, he two standards of ooroorescoo are illustrated in1g. 13, the l( er t,l.' '.el _g the initial andard. Chara ct e: c for both standards of a:essor are sicoaco on lop 1'1. In oreoari l this rarer, the authors have a ^ht to avoid making a detailed presentation of their droia rhiloscrh and merely to present data r_ o_.., which others imal draw their oc.:m con- 8

9 elusions and develop their own philosophies and design methods. However, there are a few points worthy of further comment. The significance of inducer and diffuser throat areas are of great importance to the matching of the compressor, and these may be varied by changing annulus lines quite signficantly without impairing the compressor efficiency. Shroud surface finish can easily be overlooked on a potential loss of performance but experiment has shown it to have a potent effect. The benefits of swept back impellers in minimizing efficiency penalties due to matching to achieve adequate surge margin are shown to apply on the compressors described in this paper. ACKNOWLEDGMENTS The authors wish to thank their colleagues who helped in the programs of work which provided the material for this paper. Particular thanks are due to the members of the Design, Instrumentation, Manufacturing and Testing Department. The authors also wish to thank Rolls- Royce Limited and the National Gas Turbine Establishment for permission to publish this paper; the views expressed, however, are those of the authors and not necessarily those of Rolls-Royce Limited. 9