Sunteți pe pagina 1din 3

Journal of Turbomachinery

Technical Brief
2 The Smith Theorem
The argument starts from a theorem originally stated by Smith 2 : As the ow passes through one USF stage all streamlines will experience the same increase in static pressure; this must be so because the same radial and circumferential distribution of velocities exist and hence radial and circumferential pressure gradients must be the same according to the momentum equation. Furthermore all streamlines experience the same increase in total pressure, which is the same as the static pressure because the dynamic pressure is the same at stage inlet and outlet from each streamline.

Flow Through a Repeating Stage in an Axial Compressor: A Reconsideration


J. H. Horlock
Whittle Laboratory, Cambridge University, Madingley Road, Cambridge CB3 0DY, UK
A simple thermo-uid-dynamic analysis is presented for the ultimate steady ow in a repeating stage of an axial-ow compressor. DOI: 10.1115/1.1791282

Introduction

Analysis

Many papers in the literature have been devoted to study of the boundary layers on the annulus walls of axial ow turbomachines. But it is now generally agreed that conventional boundary layer theory is inappropriate for describing these ows, as they are three dimensional, often with areas of separation, and full CFD calculations of the whole ow are generally required. However the concept of the ultimate steady ow USF in a repeating stage, rst described by Howell 1 and later by Smith 2 that through an embedded stage in a multi-stage compressor or turbine where the ow changes little between entry and exithas proved useful and enlightening. Smith reported on the testing of a twelve-stage General Electric research compressor with high-aspect-ratio blading as follows: The ow does not actually deteriorate, but remains reasonably well behaved. Of course the ow details may be quite complex, particularly near the walls. Evidence that losses are larger near the walls than at mid-passage is given . . . . The excess total temperature near the blade ends is seen to increase as the ow proceeds through the compressor: the uid near the walls is the receptacle of the energy dissipated by these losses. In a recent paper 3 , Horlock and Denton reported CFD calculations of the ow through a Rolls Royce research compressor, and these were compared with experimental measurements obtained by Howard et al. 4 at Craneld University. It was indeed shown that the axial velocity and stagnation pressure proles repeated quite well from stage to stage, after the rst. Further the work input per unit radius repeated, its magnitude increasing towards the walls, and Smiths observation that the difference between the stagnation temperature near the walls and that at midspan increased continually through the machine was partly conrmed. In this Technical Brief, an attempt is made to present simply and analytically the thermo-uid-dynamic implications of this phenomenon.
Contributed by the Turbomachinery Division of THE AMERICAN SOCIETY OF MECHANICAL ENGINEERS for publication in the JOURNAL OF TURBOMACHINERY. Manuscript received and accepted by the ASME Turbomachinery Division, May 6, 2004. Associate Editor: D. C. Wisler.

A statement of the Second Law in terms of radial gradients of stagnation enthalpy H, stagnation pressure P, stagnation temperature T and entropy s is dH ds T dr dr 1 dP . dr (1)

We shall consider ow of a perfect gas of constant specic heat c p ) at low Mach number through the repeating stage of a compressor; the ow then approaches one of constant density , so that P may be written as P p 1/2 c 2 , (2)

where c is the velocity. From Smiths theorem, the complete velocity distribution repeats from entry station 1 to exit station 3 of the stage under study. Thus for the component velocities in the x axial , u tangential and r radial directions: c x1 r c x3 r ; c u1 r c u3 r ; c r1 r c r3 r ; the ow angles also repeat. Note however that the velocity proles change from stator exit to rotor exit state 2 , i.e. c x1 r c x2 r .] It follows then from the radial equation of motion that the static pressure gradient p 1 / r at stage entry is equal to p 3 / r at stage exit but not equal to p 2 / r at the rotor exit . Since both static pressure p and kinetic energy c 2 /2 repeat, the stagnation pressure P must also repeat, i.e., P3 / r and P3 P1 / r P / r 0. (4) P 1 / r, (3)

Thus the stagnation pressure rise, P ( P 3 P 1 ), is constant through the end-wall region and equal to P in the mainstream. P is also equal to the constant static pressure rise p. But note P 1 / r. again that P 2 / r If it is now assumed that the ow is in radial equilibrium with c r 0, the partial differentials become full differentials. This assumption will be valid near the annulus walls in a stage with long blades, and indeed approximately true across the whole stream for a compressor of high hub-tip ratio. OCTOBER 2004, Vol. 126 677

Journal of Turbomachinery

Copyright 2004 by ASME

As previously argued, the work transfer per unit mass ow, w H c p T, repeats from stage to stage, since the velocity triangles will be the same in each stage. It then follows from Eqs. 1 and 3 that d w dr dH 3 dr dH 1 dr d H dr ds 3 ds 1 T3 T1 dr dr dL , (5) dr

* x

r HUB

1
r HUB

cx /

cx

MS

dr

(9)

and the force decit thickness


r HUB

1/F uMS
r HUB r HUB

F uMS F u dr F uMSU F u U dr.


r HUB

where L T m s is the loss or lost work, in which T m is a mean temperature. To summarize this discussion of the ow in a repeating stage, following from Smiths theorem: i there are gradients in H and P with r; ii there is a gradient in work input w and therefore in H) with r, usually an increase toward the annulus walls; iii there is no gradient in sure; P, the increase of stagnation pres-

1/F uMSU It then followed that Q P l

(10)

* / F uMSU l

(11)

where l is the blade length and Q is the ow for the annulus running full, for which the stage efciency would be
MS

iv there is a gradient in loss L T m s, and it is equal to the gradient of the work input, H. Analytically these results may be written as d H dr P d H dr dL . dr (6)

P/F uMSUl.

(12)

Thus the stage efciency of the USF stage is given by


MS

1 1

* /l /l

(13)

Hence if subscript MS refers to the mainstream and subscript EW to a location within the end-wall region, Eq. 6 may be integrated to give H EW or H EW H MS X r , (7) H MS L EW L MS X r ,

L EW L MS X r . Thus the extra work input X(r) at any radius in the end-wall regions is equal to the extra loss there. In the absence of any mixing the stagnation temperature will distort progressively through the compressor within the end-wall region, as will the entropy; both will go on rising. But mixing will spread the increases in these extensive properties into the main stream due to secondary ow and/or turbulent diffusion, as Adkins and Smith 5 and Gallimore and Cumpsty 6 , Wisler et al. 7 have explained. Cumpsty 8 has given an excellent account of this phenomenon, together with an illustration of a calculation of the radial redistribution of the losses, without dissipation, in which the total stage loss remained unchanged.

which is essentially Smiths nal result. It can be elaborated to allow for end-wall regions on separate hub and casing walls, for lower hub-tip ratio stages, and variations in blade speed with radius. Smith gave experimental correlations of ( * /g) where g is the staggered spacing of the blading, and of the ratio / * ; these correlations depended on the stage geometry, including tip clearance. It is of interest that Smiths equation 13 does not involve the extra loss explicitly. 4.2 Alternative Analysis. Smiths approach has the virtue of simplicity but to some extent it hides the basic understanding of the ow that follows from the result derived in Sec. 3that the extra work in the end-wall region is equal to the extra loss there. In an alternative approach, Smiths result can be derived using the analysis of Sec. 3, as follows. It may be shown from Eq. 7 that
r HUB

L EW,TOTAL
r HUB

c x X dr

cx

MS .

(14)

The total loss in the USF stage is


r HUB

The Stage Efciency

L STAGE

c xL c xL

MS

l
r HUB r HUB

c x L MS X dr

Smith expressed the efciency of the whole repeating stage in terms of the mainstream efciency or the efciency of the stage running full, MS), the end-wall displacement thicknesses and blade force decit thicknesses. He used the fact that the pressure rise static or stagnation in the end-wall region was the same as that in the mainstream. Here we initially repeat Smiths approach and then examine it in the light of the analysis derived above. 4.1 Smiths Analysis. For simplicity here we assume that near the inner casing at r r HUB) there is just one thin end-wall region of thickness , through which the blade speed U changes little. Smiths approach was simply to state the stage efciency as cx P P dr c x dr cx H dr (8)

MS

*
r HUB

c x X dr.

(15)

Similarly the total work input may be written as


r HUB

W STAGE so that 1 1

cx

MS

*
r HUB

c x X dr,

(16)

L STAGE / W STAGE c xL cx H
MS MS

l l

r HUB r HUB

c x X dr c x X dr . (17)

r HUB r HUB

F u U dr,

Using Eq. 14 for L EW,TOTAL , and


MS

where F u is the local tangential force on the blade, per unit length. The integrals refer to the total ow, but Smith restated them using the ow displacement thickness 678 Vol. 126, OCTOBER 2004

L MS / H MS

(18)

it follows from Eq. 17 that Transactions of the ASME

1 l

MS

l *

* *
MS

1 1

* /l , /l (19)

which is Smiths expression for the stage efciency.

Discussion

An earlier attempt to calculate in the latter way was made in the preparation of Ref. 9 , but as the results did not correlate well with Smiths data for / * this work was not included in the published paper. However, even had this method proved more successful, there would still be an inconsistency in this second approach, as in the Smith method of Sec. 4.1, in that no direct calculations would be made of the viscous effects.

This note has made reference to the extra loss and the extra work in the end-wall ows of repeating stages, and shown that these extra quantities are equal to X(r) in the notation of this paper . This would imply that if the extra work could be determined then the equal extra loss would also be known. Attention has already been drawn to the fact that Smiths expression for stage efciency does not explicitly involve the extra losses in the end-wall regions. But Eq. 14 shows that if and * can each be determined for example from Smiths correlations then the total extra loss across the end-wall region (L EW,TOTAL) is indeed known. It is therefore included within the Smith efciency expression, if in a hidden way. An alternative way to enter this logistical circle is suggested from the analysis of this note. It would be to use Smiths correlation for the blockage * and to nd the extra work from inviscid secondary ow analysis. Such an attempt to calculate the ultimate secondary ow in repeating stages was made earlier Horlock 9 . The angle perturbations through the end-wall regions in Howards compressor 4 were calculated from the mainstream deections and the observed velocity distributions. With the effects of tip clearance also included, the general trend of the experimental data was conrmed. Thus the prediction of the USF stage efciency would involve the following steps: a Find the displacement thicknesses * in the USF stage from Smiths original results for stator exit, and from Horlock 10 for rotor exit and make estimates of the corresponding velocity proles; b calculate the angle distributions and the velocity triangles in the end-wall region; c either calculate the total extra work and the total extra loss, L EW,TOTAL , and hence the force decit thickness from Eq. 14 ; or calculate the value of more directly from F u (r).

Conclusion

Thus the conclusion must be drawn that the concept of the ultimate steady ow in a repeating stage is really not more than an experimental observation, and that the Smith efciency expression is therefore essentially an ingenious development of this observed phenomenon.

References
1 Howell, A. R., 1947, Fluid Dynamics of Axial Compressors, Inst. Mech. Engrs. War Emergency Issue 12; American Edition ASME January 1947, p. 450. 2 Smith, L. H., 1969, Casing Boundary Layers in Multi-Stage Axial-Flow Compressors, Flow Research in Blading, L. S. Dzung, ed., Elsevier Publishing Company, New York. 3 Horlock, J. H., and Denton, J. D., 2003, A Review of Early Design Practice Using Modern CFD, ASME Paper No. 2003 GT-38973. 4 Howard, M. A., Ivey, P. C., Barton, J. P., and Young, K. F., 1994, End Wall Effects at Two Tip Clearances in a Multi-Stage Axial Flow Compressor With Controlled Diffusion Blading, ASME J. Turbomach., 116, pp. 635 647. 5 Adkins, G. G., and Smith, L. H., 1982, Span-Wise Mixing in Axial-Flow Turbomachines, ASME J. Eng. Power, 104, pp. 97100. 6 Gallimore, S. J., and Cumpsty, N. A., 1986, Span-Wise Mixing in MultiStage Axial-Flow Compressors, ASME J. Turbomach., 108, pp. 216. 7 Wisler, D. C., Bauer, R. C., and Okiishi, T. H., 1987, Secondary Flow, Turbulent Diffusion and Mixing in Axial-Flow Compressors, ASME J. Turbomach., 109, pp. 455 482. 8 Cumpsty, N. A., 1989, Compressor Aerodynamics, Longman, UK. 9 Horlock, J. H., 1995, Secondary Flow in the Repeating Stages of Axial Turbomachines, Proc. Inst. Mech. Eng., 209, pp. 101110. 10 Horlock, J. H., 2000, The Determination of End-Wall Blockage in Axial Flow CompressorsA Comparison Between Various Approaches, ASME J. Turbomach., 122, pp. 218 224.

Journal of Turbomachinery

OCTOBER 2004, Vol. 126 679

S-ar putea să vă placă și