Indian Journal of Engineering & Material s Sciences Vo l. 9. December 2002, pp. 424-431

Experimental investigation of tip clearance effects on flow field in an annular turbine rotor cascade t

C Venkateswara Babu, N Sitaram* & M Govardhan

T herma l T urbomachines Laboratory, Department o f Mechanical En)neering Indian Institute o f Technology, C hennai 600036. India

Received I J Decelllber 2001: accepted II October 2002

The present paper reports resul ts of an experime nta l in vestiga ti on on the e ffects of tip c learance in a hi gh deflect io n annul ar turbine rotor cascade. The measurement s a re carried out at the inl et, in the blade passage and a t the cx it of the cascade using a fi ve- ho le probe and a seven-ho le probe. The results indi cate that the incom ing fl ow to the rotor cascade is

uniform and the inc idence is about ±2° When the c learance is increased , the s ize and the mag nitude of the tip c learancc loss region increase. The fl ow near the hub ex pe riences overturn ing, whe reas near the casing the fl ow subs tantiafl y undcrlul'llS due to tip c learance. The underturning increases w ith tip c learance. Secondary and tip c learance losses s tee ply in crcase dow nstream of the blade trailing edge indicating losses due to thc wake and c learance fl ow mi xing with the main flow.

Pri nc ipal internal losses occurring in a turbine are due [0 secondary fl ows, end wall flows and tip clearance flows. Tip clearance flow is due to the static pressure di fference across the blade surfaces, whereas, end wall flow is due to the presence of boundary layer on the endwall surfaces.There are two di stinct and equally important aspects of tip clearance fl ows . First, there is reduction in blade force , therefore the work done. Thi s occurs because the leakage fl ow passes over the blade tip essentia \(y witho ut being turned. As a conseq uence of viscous effects in the tip clearance, ent ropy is a lso produced . The second major aspect is the mi xing of the flow that passes through the tip c learance with main flow. Since the lift coefficient of a turb i ne rotor bl ade is much larger than that of a com pressor rotor blade, the vortex diameter is likely to be large r, thus influe ncing the flow over large reg ion near the tip . The tip clearance losses within the gap are re latively very less compared to overall tip clearance losses due to interaction with secondary and endwall flows. Analysis and understanding of tip c learance and secondary flows would be certainly helpful in improving the efficiency and aerodynamic performance of turbines.

Research o n secondary and tip clearance flows in ax ial turbines is receiving considerable attention, as

For Correspo ndence : E-mail: ns ita ram @iitm.ac.i n +Presented at the 28th Natio nal Conference o n Fluid Mechanics & Fluid Power. held at Punjab Engineering College, C handigarh . du ri ng 13- f5 Dece mber 2001.

ev idenced by a lecture seri es devoted to [hi s topic ' . Morphi s and Bindo n2 obtained tip static pressure di stribution on a turbine rotor blade at vari ous values of tip c learances. Yaras and Sjo lander3

. Sjolander and Amrud4 studi ed the influence of tip clearance on the downstream fl ow filed and the blade loading o f a linear turbine roto r cascade. Moore and Ti Iton5

studi ed the fl ow in the tip clearance space of linear turbine rotor cascade and deri ved a tlow model using the concepts of potenti a l flow and theory supplemented by flow mixing and boundary layer effects. Di shart and Moore6 in their in ves tiga ti o ns in a linear turbine rotor cascade have brought the pheno mena and sali en t features assoc iated with the loss generation, distribution and mechani sm of tip clearance fl ows. Sjo lander and Ca07 fo und multipl e vortices on the blade tip , which are probably [he reason for the burnout that some times occurs 011

turbine rotor blade tips near the pressure surface. Govardhan et al.8 stud ied ti p clearance Il ows wi thi n and ex it of a large defl ecti on linear rotor cascade. They found a strong horse-shoe vortex fo rmed for the zero clearance, which di sappears for 3% clearance indicating th at the pressure forces have dominating influence than the viscous forces for la rge c learance. The mechani sm of tip clearance fl ow is far from understood . The data available are not suffi c ient to test calcu lation methods o r loss prediction schemes. The main difficulty in obtaining deta iled measurements within the typical small d imensio ns of the tip gaps found in actual machines or c reated in the

VENKATESWA RA BABU el af.: ANNULAR TURBIN E ROTOR CASCADE .+25

laboratory. In additi on, the data ava il able on annular cascades. when the tip clearance is present, is very limited')·II). In the present in vestigati on, the incoming fl ow is non-uni form, so the tes t case corresponds to a realis ti c simul at ion of fl ow through an ax ial turbine rotor. Thus the results shoul d serve as a severe test case for va li dati ng commercial CFD codes.

Experimental Procedure The measurements were carried out in an annular

cascade tunnel available in Therm al Turbomachines Labora tory, Department of Mechanical Engineering, Indian In stitute of Technology. Madras.A schematic or facility is shown in Fig. I. The hub and the cas ing di ameters of the test secti on are 400 mm and 600 mm res pecti ve ly. The tunnel consists of 25 numbers of inl et sw irl vanes and outl et strai ghener vanes. The test cascade is l11 0unted in between these two vanes . Other detai ls of the tunnel are avai lable elsewhere ll

. The cascade blade profi Ie along wi th the measurement stati ons is shown in Fig. 2 and the major detail s are gi ven belo"v.

Chord. Ch:

Ax ial eh,ml. e:

Span. h:

r-\;,pee l rali o. h/Ch:

In lel blade angle. ~I h:

OUlkl blade angle. 132,,:

Camber angIe. 8:

Stager ang le. y:

Leading edge radiu s/ Chord:

Trailing edge thickness/ Chord :

Maximuill thi ckness/ Chord:

Pos iti () n of l11ax. thickness/ Chord:

No. of blades. Z:

95 111111

92.75 111 III

100111111

1.05

57.5 0

-62.5°

1200

12.50

0.25

0.05

0.3R

0.42 24

All angles are taken with reference to ax ial directi on. A five-hole probe at upstrea m and a seven-hole

probe within the rotor blade passage and dow nstrea m of the cascade are used for fl ow measurement s. Both probes have a head diameter of 2.6 mm only. Both probes are used in non-nulling mode of operati on fo llowing the method of Treaster and Y OCU lll

12 for the five- hole probe and the method of Vcnkateswara Babu el al. 1.1 for the seven-hole probe. The seven-ho le probe is calibrated in the hi gh angle range of ± ).+0 in both yaw and pitch planes . Both probes suffer many measurement errors ari sing from va ri ous sources. The followin g magnitudes of errors are es timated in the measurements obtained by the five-hole probe' -l .

Total and static pressures: ± I % of dy nam ic head Flow and radial angles: ± 1° Same magnitude of error may be ex pected in the

measurements by the seven-hole probe. In the

3 y

Inlet Plane

Fig. 2- Rotor blade passage showing l11easurel11ent po,;it ion,

~~---------moo--------~

- -+- 10

Stood All Dimensions in mm

Fig. 1- Schematic layout of annular cascade tunnel

426 INDIAN J. ENG. MATER. SCI.. DECEMBER 2002

boundary layer, wake and vortex regions , where the gradients in ve locity and pressure are large, the mag nitude of error increase substantially for both probes. T he detail s of the measurements are given in Table I.

Results and Discus:~ion

Inlet flow

Fig. 3 shows the non-dimensional total pressure at inle t to the inle t guide vanes . The boundary layer on

the hub is thicker (o"lh = 0.28) than that on the casing

(o)h = 0. 12), due to the fac t that flow com ing from the sett ling chamberc travels a longer di stance on the hub compared to that of the flow o n casing.

Rotor hlade inlet fl ow

The flow thai enters the IGY axia ll y is deflected by

abou t 57.5° Fig. 4 :;hows the distribution of the total press ure coeffi c ient, static pressure coefficient and flow angle at the inlet of the rotor blades (X = -O.34) .The X-axis represents the non-dimensional tangential d istance. Y = 0 represents the center of the measurement region. The measure ments are taken at

Table I - The details of the measure me nts

Grid size Ax ia l Station X (Span-w ise direction x Tangenti al direct ion)

-0.34 700 grid points (25 x 28)

~\ .;;;

° Li

0.20 168 grid poi nts (21 x 8) 0.40 147 grid points (21 x 7) 0.70 147gridpoints(2Ix7) 0.88 168 grid points (2 1 x 8) 1.04 693 grid points (2 I x 33) 1.29 693 grid points (2 1 x 33) 1.53 693 grid points (2 I x 33)

---------------~~----------------

D

-rICh = 0.00. T he total pressure is found to be uniform in the tangential d irect ion, indicating compl ete mi xing of the exi t flow from the IG Y (Fig. 4a).The total pressure losses are very low except near the hub and the casing.The static pressure increases towards the

casi ng to account for rad ial equi libri um of the fl ow (Fig . 4b). The maximum variation in flow ang le in the

reg ions away from the boundary layer is ± 2° from the

design value of 57.5°, so the incidence o n to the rotor

blades is about ± 2° (Fig. 4c). For the turbine blades. which have thick leading edges, the effect of this small incidence on the flow is negligib le.

Rotor blade passage flow

Contours of total pressure coefficien l in the blade passage at X = 0.4 and 0.88 are presented in Fi gs 5 and 6 for the three va lues of tip c learances . viz .. Tfell

1.0 r-----;-.------------,

I Q.

"-a: 0.9

c .. ~ o.a -.. o (,)

~ 0.7 .. .. ~

Q.

"Oo.s ~

I I

~I 0 1

1/1 .t:1

~ '0

1

0.5 0:--'--'---L-:2"='0...L..l..J.......L-:4

'-0....L....-J.......L-S

'-0 ...L...J'-.J......Ja 0'-.J...l..L..L....J1 00

Hub Peroentcige Span Casing

Fig . 3-DistribUli on of tota l pressure at inl et

'2' \ . ';;'

° Li

D

i 0 :.-o." 0.0 +0 .6 i °:'-0." 0 .0 . . +0.6 v

Non -Dimens iona l Tangential DIstance, I Non -Dim ensional fo ngentiol Distan ce, Y

a) Total Pressure b) Static Pressure c) Flow Angle

Fig. 4--Distribution o f total and static pressures and tlow ang le at inlet o f rotor blade

VENKATESWARA BABU el 01 .: ANNULAR TURBINE ROTOR CASCADE 427

Centre of tip clearance loss region

g' 1. g' 1 . . .., ' ;;'

0 0 u U

N N

L ~

Non - Dimensio nal Tangentia l +0.6

Non -Dime nsiona l Tangential Distance, Y :i °:"'0.6 +0. 6

Non-Dimensional Ta ngentia l Dis tance , Y

a) -rICh = 0 .00 b) -rICh = 0.01 c) -rICh = 0 .03

Fig. 5- Di stri but ion of total pressure in the passage of rotor blade (X=OAO)

" o (t lJ) ':,,0 Il.. o c " ~ '" '"

+ 0. 6

g' \. ';;' o u

,...

Centre of tip clearance loss region

I ! I

Non-Dimensional Tangen tial Distance, Y Non-Dimensiona l Tangential Non - Dimensiona l Tangen tiol

a) -rICh = 0 .00 b) -rICh = 0.01 c) -rICh = 0.03

Fig. 6- Distribution of tota l preSSlIrl' in the passage of rotor blade (X=O.88)

= 0.00, 0.0 I and 0.03 . Suction and pressure surfaces are marked as SS and PS. These contours clearly show the e ffect of tip cleara - i.e., decrease in total pressure in the clearance regi .-j increased extent of low-pressure region as the clea increases. The flow over the tip ro ll s into a vole K., , own as tip leakage vortex. The tip leakage voj'(ex causes large losses in the total pressure. Also the ex tent of low­pressure regIon increases with axial distance (Figs 5

and 6) . At X = 0.4 (Fig. 5), the boundary laye r on the suction surface is fo und to be thicker than that on the pressure surface due to the cross channel deflection of the boundary layer fluid from the pressure surface to the suction surface. At X = 0.88 (Fig. 6) , the ti p clearance loss region near the casing is sli ghtly pushed away from the suction surface. The boundary layer on the casing has become thin. This suggests movement of the low energy fluid of the boundary

-f28 INDIAN J. ENG. MATER. SCI.. DECEMBER 2002

1;1),(' r into thi s loss region. The losses are more conce ntrated and arc higher in their magni tude in the tip clearance reg ion . The centre of the tip clearance loss region at X=0.88 for L/CI/ = 0.003 is away from the casing <Ind towai'ds the sucti on surface co mpared to that for L/CI/ = 0.0 I. Thi s indicates a tip leakage vortex of larger diameter and of hi gher st rength . For the clcarance of -rICh = 0.00, no such loss reg ion is clearl y see n. Howner, passage vortex near the hub <Incl the cas ing can be seen at zero clearance as well as nO ll -zero clearances .

Rotol' blade exit flow

Figs 7 and 8 show the contours o /" total press ure coefficient at X = 1.04 and 1.53 for the values o/" -rICIt = 0.00, 0.0 I and O.OJ.As the trailing edge thidn esl., for the present cascade is thicker (about 5% o/" the blade chord), the wakes are broader ane! the losses ill the wake are considerable (Fig. 7). Although the blade tailing edge is rad ial , the wake is inclined towards sucti on surface due to cross channel press ure gradicllt. The wake center-lines are shown by dashed straight ­lines. However near the hub and the casing, the wakes

\-______________ Wake centreline

0 ' c: ';;'

'" u

0> 0> e: I . I .

i. e: 'Vi . ..,

N

0 0 u u

,... '"

~.D

i. ° ..... 0.& 0.0 +0.6 Non- Dime nsion a l Tange ntia l Distance, Y

a) T/Ch = 000 b) T/Ch = 0.01 c) T/Ch = 0.03

Fig. 7- Di stribution of tolQI pressure in the pJSSJgc o f rOlor blade (X= t.O'+ )

~o. o C a u

~ Q.

+0.6 Non-DimenS ional Tan gential Distance, Y

a) TiCh = 0.00

0> e: ';;' 0 u

I.

'"

+0.6 Non-Dimensiona l Ta ngential Dis ta nce.

b) -rICh = 0.01

m e: 'Vi

0 0

I.

, ... C-o

0.0 +0.5 Non- Dirnens ional Tongen l iol Distance . v

c) T/Ch = 0.03

Fig. 8- Distribution of totQ I pressure in the PJssQge of rotor blade (X= I S1)

VENKATESWA RA BABU el (II.: ANNULAR TURBINE ROTOR CASCA DE

interact with passage and tip leakage vortexes and the wak e eentrelines are 110 longer strai ght. The losses due to the hub and the casing passage vortexes are clearl y seen. The loss region due to hub passage vortex is found to be more and occupi es about 50% of the blade passage where as the loss region due to the cas ing passage vortex is found to occupy only 33 % of the passage. The total pressure loss reg ion near the hub is much broader in the tangential direction than near the casing. The loss region due to the hub vortex remains unal tered for all values of tip clearances. For r l CI! = D.O I (Fig. 7b) , the tip clearance vortex is

located nearer to the casi ng. When the clearance is further increased to r iCh = 0.03 (Fig. 7c), the size as we ll as magnitude of the tip clearance loss region increases. At X = 1.53 (Fi g. 8) . the wake mi xes with the main fl ow. At X = 1.53. the total press ure loss is much higher than that obtained at X = 1.04.The loss increases substantially with increase in tip clearance (Figs 8b and 8c).

(}itch-wise averaged flow angle

Fi g. 9 shows the spanwi se variati on of the pitch­wise averaged fl ow angle fro m X = 0.20 to 1.53 for

.,_ ;"·· .. , ~'~ 1 -6o J~_ - ~_ . ', _... l -I

" - e' - '"

X= 1 29 :-iI;: -80

""I

- GO i X = 1 .0 4 ----' .:- .~. :.: 0 ,. ' • .. 1

~~-- /"" ~-- ~ 'T!c'h I

r :;.g~h_ ... c. o o.oa I - 80 -' '«0 "'0 .0 1 I

I < '"" '" 0.03

- - Combe ' I l.J_._I-L..-1 , I_I I ~j---1A7_g, l~ o 20 40 60 80 100 Hub Pe rcenlag~ Span , Z Casing

Fig . lJ- Span-wise variati o n o r pitch-wise averaged !low ang le al difTcre nl ax ia l sta ti ons

0.3 -----------:-1 A .0

0 .0 0.2

Ys T/Ch

GEroeB 0 .00 o

[}{]OQO 0 .0 1 '".' D '

0.1 MM--6 0 .03Y p., /'

ii,' ::..----'" I

,~~~. i o (1 ~ ....J_~-I.~ L.. I I , I --L-L-LJ

·U .O 0 .5 1.0 1.5

X

0.3

0.0 0.2

0.1 -

~

SG

~ .4 .:1

a (l L -'- ill efil-r=' i I~- ~( .n- ..L...L...l · U.O 0 .5 1.0 1.5

X

Fi g. 10- Ax ial va riatio n o f' to tal. g ross secondary. net seconda ry and c learance losses

-1-30 INDIAN J. ENG. M ATER. SCI.. DECEMBER 2002

T/OI = 0.00, 0.0 1 and 0.03. At all ax ial stations, the variati on in the fl ow angle variati on is small near the hu b but substan tiaHy increases near the casing with increase in tip clearance. For TICh = 0.0 1 and 0.03, large va ri ati on in the fl ow angle is seen near the casing. Secondary flows cause fl ow overturning near the hu b and the casing. Away fro m the hub and the casi ng., fl ow overtuflls due to secondary fl ows. The effect of tip leakage flows is to cause fl ow undenurning. The n"lagnitude of underturning near the casi ng increases with increase in ti p clearance. For TICh = 0.0 I , tip clearance effects are pre dominant, as overturning associmed with secondary flows in not seen at all fo r X=0.70 and beyond .From the total pressu re contours and averaged fl ow angle distributi ons, it may be inferred that the tip leakage vortex is formed between X=0.20 and 0.40 and becomes stronger between X=0 .40 and 0.70. The pos ition of maxi mu m underturning does not co incide with the pos iti on of centre of loss region (Figs 7 and 8). Thi s Illeans that the point of lowest total pressure Illay not be acco mpanied by the larges t flow angle deviati on.

Loss coefficients

Fig. . 10 shows vanatl on of total loss coeffi cient, (Yr ). gross secondary loss coeffi cient , (Ysc), net seco ndary loss coefficient, ( Y.I) and ti p clearance loss coefficient , (YCI.) with ax ial distance fo r -rICh = 0.00, 0.0 I and 0.03.

They are defined as follows:

.1 Ii

f f (PIC1 - Po) Y.,. = -,-,0 -,,-0 __ =---:=-2-

0.5 p C2

Y\"(; = Yr - VI'

VI' = Profil e loss coefficient obtained from 2-D = 2

cascade tests= 4 % of 0.5 P C 2 downstream of

the cascade

Y.I = Ys(; - YIN

YIII' = Inlet boundz·ry layer loss coefficient = ?

= 6 % of 0.5 P c:2 Ya = Ys - ys(at -rICh = 0.00)

Fig. 10 shows that secondary and tip clearance losses increase steeply downstream of the bl ade trail ing edge. However, it must be remembered that

the loss coefficient in the passage does not include profile losses, as measurements cou ld no t be taken close to the bl ade surfaces. Even then. it Illay be in fe rred that tip clearance starts between X = 0.20 and 0.40, as there is a large increase in the magnitude of loss coefficient fro m X=0.20 [0 0.40. Also ti p clearance vortex is observed at X=0.40 as seen from the contours of to tal pressure coefficien t (Fig. 5) . At X=1.04, there is again a large increase in the tota l. secondary and tip clearance losses. It indi cates that the ti p clearance effect is less in the passage reg ion but increases downstream of the cascade. This is due to the interacti on of tip clearance flow with secondary and end wall fl ows.

Conclusions From the present in ves ti ga ti ons the fo llowing

conclusions are draw n: ( I ) The total pressure is found to be uniform in the

tangential directi on, indicating complete mi xing of the IGY exi t fl ow; (2) At X = 1.04, the loss reg ion due to hub passage vortex is found to be Illore and occupies about 50% of the bl ade passage where as the loss region due to the cas ing passage vortex is fou nd to occupy onl y 33% of the passage: (3) At X = 1.53 . wake mi xes with the ma in flow and losses are found to be hi gher than those obtai ned at X == 1.04; (4) When the clearance is increased, the size and the magnitude of the tip clearance loss region increase: (5) The tlow near the hub ex peri ences underturn ing fo ll owed by underturning, whereas near the cas ing. the now substanti all y underturll s due to tip clearance.The overturning increases with tip clearance : and (6)

Secondary and tip clearance losses steep ly illCJ"ease dow nstream of the blade traili ng edge indicating additi onal losses due to the wake and clearance flow mi xing with the main fl ow.

Nomenclature CII chord( m) e ax ia l chord (m) C veloc ity (m/s) C.. axia l ve loc ity (m/s) II blade span (m) Po to ta l pressure (N/m') P"., settling chamber pressure (NfJn ' ) .\ spac ing (m) .r, y and ;: = distances in ax ia l. ta ngenti al and spa nwise

d irectio ns (m) X no n-d ime nsio nal axia l d istance from black

lead ing edge. _x/e Y non-d imensiona l tangen ti al di stance . .I'll'

Yet tip c learance loss coefllc iel1l

VENKATESWARA BABU et al.: ANNULAR TURBINE ROTOR CASCADE -nl

z

P T

Ueferences

inlet boundary loss coeffi cient profi Ie loss coefficient net secondary loss coefficient gross secondary loss coefficient tota l loss coeffi cient di stance from centre of the blade passage along tangential direction (m) percen tage span. 100 (z - zhub)/h spa nwi se distance of hub (m) flow angle (de g) boundary layer thi ck ness on casing and hub respec tively (m) density (kg/m.1 ) tip clearance (m)

VK I Secondary and tip-clearance Il ows in axial turbines. VKI LS /997-0 / . 1997.

:2 Morphi s G & Bindon J p. ASME Paper 88-GT-256. 1988. J Yaras M I & Sjolander S A. ASME J Turbolll{/chinerv. III.

( 19R7) 276. -4 Sjolander SA & Amrud K K. ASME J Turbolllachinerv. 109.

237.

5 Moore J & Tilton J S. ASME J TuriJolllac/ull crr. 110 ( 19XX) 329.

6 Dishart P T & Moore J. ASME J Turlxl/lI{/chin£'rr. 11 :2 (1990) 599.

7 Sjolander S A & Cao K. ASME J TuriJolll{/chill crr. 109 (1995)237.

8 Govardhan M. Sastri S S S R K & Vishnubhot la V S . .I 771£'/7l1a/ Sci. 7 (1998) 149.

9 Casc iaro C. Sell M. Treiber M & Gyarmathy G. Vortex interacti on and breakdown phenomena in an axial turbine. ASME COl!/" COlllputation{/ 1 Techniques for F/(Ill-rrh £' mlll/l Structural and Chelllical Svstellls Applicmiolls. 1999.

10 Se ll M. Treiber M. Casc iaro C & Gyarmathy G. Tip clearance affec ted fl ow fields measu red in a turbine black row. Third Europe{/n COl!/" Turbolll{/chinerv. London. UK. 1999.

II Govardhan M. Seco/l{!arv .flows and losses in all (/II/I/d((r turbine c{/scade. Ph.D. Thesis. Indian Institute of Technology Madras. 1984.

12 Treaster A L & Yocum A M. ISA TrailS .. IR ( 1979) n . 13 Venkateswara Babu C. Govardhan M & Sitaram N.

Measurelllent Sci& Techllol .. 9 ( 1998) 468. 14 Sitaram N. Lakshm inarayana B & Rav indranath A. ASM!::.1

Eng Power. 103 ( 198 1 ) 406.