THE AMERICAN SOCIETY OF MECHANICAL ENGINEERS 345 E. 47th St., New York, N.Y. 10017 The Society shall not be responsible for statements or opinions advanced in papers or discussion at meetings of the Society or of its Divisions or Sections, 94-GT-456 or printed in its publications. Discussion Is printed only If the paper is pub- lished in an ASME Journal. Papers ere available from ASME for 15 months altar the meeting. Printed in U.S.A. Copyright © 1994 by ASME THE EFFECT OF REACTION ON AXIAL FLOW COMPRESSOR PERFORMANCE Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1994/78835/V001T01A143/2404462/v001t01a143-94-gt-456.pdf by guest on 02 October 2021 C. D. Farmakalides OPRA Hengelo, The Netherlands A. B. McKenzie and R. L. Elder School of Mechanical Engineering Cranfield University 111111111111111111111111 Bedford, United Kingdom ABSTRACT GREEK SYMBOLS This paper describes a study of the effects of design-point a Flow angle reaction on axial flow compressor performance. Particular am Mean flow angle attention is given to differences in stable operating range and Blade angle overall efficiency. Design of an 80% reaction, zero pre-whirl, 8 Blade deviation angle blading is presented together with a discussion on the applicability 0 Blade camber of currently available design correlations, mostly derived from 50% Blade stagger angle reaction binding tests, to such high reaction blading. Experimental a Blade solidity - 11(5/C) data obtained from tests carried out on a low speed 3-stage axial (1) Flow coefficient - C/U flow research facility, at Cranfteld Institute of Technology (now Cranfield University), using an existing 50% reaction blading and INTRODUCTION the new 80% reaction blading indicate that high reaction designs A stage velocity diagram may be specified by the choice of three can result in improved operating range at no loss of efficiency. parameters. Modem axial flow compressor blading differs only Tests carried out include performance measurements for each of slightly with respect to two of these parameters, namely work and the two bladings at various stator stagger settings and include inter- flow coefficients, so the diagram can be characterised by the third, row radial traversing at flow conditions near optimum and stall. the degree of stage reaction. For the last fifty years, the effect of reaction on compressor performance has been somewhat NOMENCLATURE controversial and in the early days in Britain the designs tended CO AVDR Axial velocity density ratio concentrate around 50% while German designs tended to favour Blade chord high reaction around 90% as used in the Jumo 004 engine. C, Axial component of absolute velocity Arguments may be found by Howell 11945]. Seippel 119401. which D.F Diffusion factor consider Mach number limitations to support 50% reaction. There Stagnation (total) enthalpy are arguments, however, Cumpsty [1989] and Casey [19871, Blade incidence angle suggesting that losses are mainly insensitive to reaction and Hans- Compressor inner diameter, see Davis [1973] Otto andHeinrich [1984] claims a larger stable operating range for Compressor outer diameter high reaction blading. Blade pitch (space) Industrial compressors an required to have the greatest possible Space-chord ratio range of mass flow at constant speed and pressure ratio. This Blade thickness range is usually achieved by using variable stagger stators. Tip clearance Variable geometry compressors, however, are considerably more Blade thickness to chord ratio expensive than fixed geometry compressors and there are clear Blade speed benefits to be achieved by minimising variable components and Presented at the International Gas Turbine and Aeroengine Congress and Exposition The Hague, Netherlands — June 13-16,1994 introducing high reaction blading if the penalties are small. low Mach number (<0.7) applications and is compatible with the The purpose of this study has been to evaluate the effects of existing design. reaction on compressor performance with particular attention to The work coefficient, niii/U 2 = 0.4, is considered to be typical of optimum efficiency and stable range of operation. Although this current trends in modem high designs (see for example the papers Downloaded from http://asmedigitalcollection.asme.org/GT/proceedings-pdf/GT1994/78835/V001T01A143/2404462/v001t01a143-94-gt-456.pdf by guest on 02 October 2021 area has attracted attention in the past, most of the work has been of Dong et al [1987] and Dring et al [1986] ). The flow unpublished and it is not clear whether properly designed high coefficient, CIO = 0.6, was selected from considerations of reaction blading can be competitive (on range and efficiency) with solidity, De Haller number, diffusion factor together with other the lower 50% reaction blading for which abundant design data is factors and is also believed to be representative of current designs. available in the open literature. The 80% reaction design was so chosen because this design has The work carried out during the study Farmakalides [1992], zero-whirl at stator exit for all radii (a well known form of design) involved the following areas which are discussed in the following and therefore has the practical advantage of requiring no I.G.V's sections. for a variable speed machine (and zero camber I.G.V's for variable (i) Design and manufacture of 80% reaction blading. stator constant speed applications) and the last stator requires no (ii) Experimental investigation of two kinds of blading with more deflection than the others. different stage reactions, each at various stator stagger settings. (iii) Evaluation of results BLADE LOADING CONSIDERATIONS The expression developed by Lieblein [1953] to evaluate the DESIGN OF BLADING diffusion factor (D.F.) is employed in its incompressible form. Reviews describing semi-empirical and computer aided Lieblein suggests values for D.F. not greater than 0.6 with typical approaches to axial flow compressor design have been published design values of 0.45 and the design met those recommendations. among others by Gostelow et al [1969], Howell [1945 and 1948], Horlock [1973], the NACA/NASA staff, Johnsen et al [1965], BLADE PROFILE DESIGN Serovy [1966] and AGARD [1981 and 1989]. These methods Reviews which describe semi-empirical methods of design have were generally adopted for the design of the 80% reaction blading been published by Carter [1950], Lieblein [1960], Horlock [1973], for this study (the 50% reaction blading existing previously). Serovy [1966] and many others. Tests carried out on single and The designs had the following parameters at a reference diameter multi-stage axial flow compressors have introduced empirical of 335.3 mm: factors which relate cascade performance to that of an actual compressor. In the case of McKenzie 11980 and 1988], a design Reaction 80% 50% method has been developed directly from compressor test data Work coefficient AN/U2 0.4 0326 Two methods of design were adopted and are referred to in this Flow coefficient Ci/U 0.6 0565 project as the McKenzie and NASA methods. The so called Whirl velocity at rotor inlet 0 30° NASA method does not follow exactly the recommendations of SP-36 [Johnsen et al 1965] but is in fact a combination of well 50% Reaction Bladinq established cascade correlations including Lieblein [1960], Carter The existing blading for this compressor is of free vortex design [1950] and Howell [1945]. with reaction varying from around 31% at the inner diameter The blade profile for the new designs is of C-4 thickness (ID.), to around 66% at the outer diameter (0.13.). The blades distribution on circular arc mean camber line with 10% maximum have a C-4 profile thickness distribution on circular arc camber thickness to chord ratio (t/C). To aid CAD-CAM manufacturing line with a maximum thickness to chord (t/C) ratio of 10%. Blade of the new blading the thickness distribution for C-4 is obtained chord is constant along the span at 3048 mm, to give an aspect from the relations given by Calvert [1988] and repeated in ratio (A.R.) of 2.0. The blades operate with a tip clearance (c) of Farmalcalides [1992]. 0.6 mm (0.15% of 0.D). Various aerodynamic and geometric design parameters are given S/C Ratio Selection in Table 1. Blade geometrical angles to meet the above conditions The starting point for blading design is the appropriate selection were calculated by the compressor designer using recommendations for space-chord ratio (S/C) which has been influenced by two of Roxbee-Cox [1946]. 'rules of thumb' as follows: (i) Over-blading is avoided by setting a minimum restriction 80% Reaction Bladinq on S/C ratio of 0.5. The new blading for the compressor is of free vortex design with (ii) Design point incidence is restricted to within ± 5°. In degrees of reaction varying from 71% at the inner diameter (I.D.) cases where incidence drops below -5°, the chord length to 86% at the outer diameter (OD.). The profile shape and tip is increased to reduce S/C and camber and move the clearance for this blading are as for the existing 50% reaction incidence to more positive values. build. The choice of free vortex blading is generally successful for 2 McKenzie Design Method Choice of S/C Ratio. Howell [1945] developed a blade loading McKenzie [1980] developed a design method based on tests of correlation in the form of curves of nominal fluid deflection LS vs a 4 stage low speed compressor. The relation for stagger (t) fluid outlet angle Cr., for various S/C ratios. The plot is used to suggested in his paper may be used to position the blade for indicate a SIC ratio selection compatible with an acceptable maximum efficiency. In order to increase stall margin, particularly loading to avoid boundary layer separation.