Indian Journal o f Engineering & Materials Sciences Vol. 9, December 2002, pp. 472-479

Effect of inlet swirl and dump-gap on the wall pressure distribution of a model can-combustort

Abdur Rahim", S V Veeravallih & S N Singhh

"Department of Mechanical Eng ineering, Faculty of Eng ineering & Technology, Jamia Millia Is lamia, New De lhi I 10025, India

hDepartment of Applied Mechanics, Indi an Institute of Technology, New Delhi 11001 6, India

Received 13 December 200 I .. accepted 4 October 2002

Wall stati c pressure distributions alo ng the casing and liner wall have been measured using a Scanivalve digital manometer, for a model can-combustor under isothermal now conditions for non-swirling and sw irling now at inlet wi th d ifferent positions of the liner (dump-gap). It is observed that swirl reduces the size of wall recirculation zone and permits earli er llow development whereas, dump-gap alters the reattachment length for weak swi rling nows only. It is al so observcd that liner wall pressurc (spec ially dome reg ion) has a strong dependence on dump-gap for strong swirl ing flows clue to fo rmation of a central recircu lat ion zone.

Single can-combustors are used for low power output in small gas turbine engines. These are widely utili zed in the industry and in vehicles because of their compact size. A can-combusto r with dump diffuser system is preferred over faired diffuser system due to it s wide range o f applicability. Swirlers are pravJided at the inlet for better fuel/air mixing and flame stabili zati on. The overa ll performance o f the combustor depends on the mixing process and the air flow split occurring at various zones of the combustor liner. Further, for aircraft eng ines, the req uire ments of low size and weight are of paramount importance. Flow through can-combustor is both turbulent and swi rling at inlet with dump diffusion. Wall stati c pressure vari ation along casing and liner walls is estab li shed to g ive a reasonable approximat ion for the s ize of recirculation zone and flow uniformity in the annu lar region of combustor. Many researchers have carried o ut in vesti gations of fl ow through combustors with emphasis on the line r flow characteristics l

-4.

Bicen and Jones l have ex perimentally investigated the ve locity characteristics under isothermal and combusti ng flow conditio ns in a model can­combustor. They have observed that more flow enters the combustor th rough the primary holes at low Rey nolds number. Koutmos and McGuirk2 have in ves ti gated in detail the vari ation of the flow split between the primary and the dilution holes o n the

t presented at the 2X'h National Conference on Fluid Mechanics & Fluid Power, held at Punjab Engineering College, Chandigarh , duri ng 13- 15 December 2001

flow pattern in the primary zone of a mode l can­combustor usi ng water as the fluid . McGuirk and Palma3 have predicted the mean velocity and turbulence kinetic energy inside a model can-

'combustor and compared them with measurements. .; They observed large discrepanc ies between

measurements and the predictions in the primary region which may be due to higher level of momentum diffusion. Annular dump di ffu ser system has been investigated by Fishenden and Steven4 for monitoring the overa ll system performance while varying the pre-diffuser area ratio, dump gap and mass flow split to the inner and o uter annuli . They have conc luded that the principal determinants of total pressure loss in such systems are the amount of diffusion being attempted and the radius of curvature undertaken by the flow as it passes around the flame tube/liner. Carrotte et at. 5 experi menta lly investigated the fl ow characteristics and aerodynamics performance of a modern gas turbine dump diffuser and found that stagna tion pressure loss is a strong function of dump-gap. Bharan i6 experimenta lly invest igated the flow characteristics in the annuli of a reverse fl ow gas turbine combustor and concluded that changes in the dump-gap do not signi ficant ly influence the nature of velocity profile. However, velocity magnitudes reduce with the increase in the dump-gap. Ahmed and Nejad7 have investigated the effect of inle t swi rl on the characteristics of a co-axial dump combustor model in an isothermal envi ronment lI si ng LDY. Two sets of experiments were carri ed out to examine the effect of swirl s trength on the fl ow

RA HIM ('/ (/1 .: WA LL PRESSURE DISTRIBUTION OF A MODEL CAN-COMBUSTOR 473

Combustor model

Fig. I a - Schematic layou t of the ex peri mental set-up

field characteri stics and the effect of inlet swirl pro fil e and its evolution th roughout the combustor. The resu lts indicate that inlet swi rling motion alters the tlowfield and reduces the corner recircul ati on region signi ficantly. Agrawal and Singh8 have experimen tall y investi gated the in tl uence of sw irl on the flow development of co-ax ial jets in a dump co nfinement. They observed th at impos iti on of swirl on the fl ow improves mixing and fl ow development. They also observed that the impos ition of swirl in co­swirl mode coul d lead to tlame stabili zati on. Based on the literature review, it is observed that the emphasis of research on combustors has been to es tab li sh the flow characteristics in the liner of the combustors and there are some studies in a combustor model without the li ner. Flow through the annulus is crucial as it feed s air to the liner through va rious holes on the liner wall. Flow split through the annulus to the li ne r is affec ted by the wa ll pressu re of ai r cas ing and line r. Pressure vari ation along the air casing and liner and its dependency on in let condition and dump-gap have not been in ves ti gated thoroughly up to now. The present study is a systemat ic attempt to fill thi s void.

Experimental Procedure The experimental set-up used in the present

investigati on is shown in Fig. I a. The apparatu s is composed of a blower. rectangul ar di ffuser, the set (I i ng chamber wi th a set of screens fo r red uci ng the turbulence level and mak ing flow uniform, a be ll mouth ent ry, in let pi pe, and the model can-combustor. Air fro m the blower passes th rough the rectangul ar diffuser, se ttling chamber and enters the model can­combustor th rough the inlet pipe. Prov isions have been made in the inl et pipe to meas ure the inl et veloc ity profil e. A U- tu be manometer is connected between the settl ing chamber and inl et pipe fo r con ti nuous monitori ng of the mass tlow rate. Wall static press ure taps are fixed on the air cas ing wa ll as well as on the dome/liner surface to measure the stati c pressure distri bution on these units.

* Mea 5 l.,O''' ME' f'I t LOLQ t lon s

Fig. I b - Can-combustor mode l

Vane swirlers were prov ided between the inlet pipe and model can-combustor with the help of coupling to prov ide swirling tlow at inlet. These swirlers were installed 250 mm upstream of the dump expansion so as to ensure that the wake of the swirler did not interfe re with the tlow characteri sti cs in the model can-combustor. Vane swirlers were des igned as per the procedure given by Mathur9

. Each vane swi rler had eight vanes made of 0.5 mm thi ck brass sheet.

Model can-combustor The model can-combustor used in the present study

is shown in Fig. I b. It consists of a cy lindrical ai r cas ing having diameter of 152.4 mm and length of 457.2 mm attached to the inlet pipe hav ing diameter of 54 mm. Wall stati c pressure taps at eight locations along the ax ial length of the air cas ing at equal distances are fixed for measurement of wa ll pressure. The wall stat ic pressure at any given ax ial location is measured as an average of three pressure taps placed at equal angles along the circum ference. To model the fl ame tube, a circul ar pipe of di ameter 76.2 mm with hemispherical dome at one end is placed co-ax iall y with the air cas ing. Seven pressure taps are provided on the dome surface and sixteen on the liner surface. Pressure taps are closer on the dome surface as compared to the liner surface . The other end of liner is attached to the support system, as shown in Fig. 1 a, resting on a platform by whi ch dump-gap can be va ri ed.

Determination of parameters Wall pressure tappings of casing and liner were

connected to the Scani valve Digital Interface Uni t. The Scani valve Digital system was ca li brated against a Betz mi cromanometer to obtain pressure in mm of water column. To ensure the geometri cal symmetry of the liner in the air casing, a circul ar disk hav ing ex tern al diameter equal to casing and internal diameter equal to li ner diameter was fabricated to check the placement of liner regularly. To keep the

474 INDI A J. ENG. M ATER. SCI., DECEMBER 2002

liner in its pos ition , three screws were provided at 120° apart through the air cas ing. Inlet conditions were changed by providing different vane swirlers at inl et. Vane swirl ers with angles of 0°, 15°, 30°, and 45° were selected for the present study. The corresponding swirl numbers for the swirlers are 0, 0. 18, 0.38 and 0.67. For each inlet condition, meas urements were taken at three dump-gaps, viz. DG = 0.5, 1.0 and 1.5 at a Reynolds number of 1.2x I 05

. The dump-gap denotes the distance of liner dome from the dump inlet plane. Velocity and turbulence intensity were measured by three-hole probe and hot-wire probe respectively at different axial locati ons. Measurements by three-hole probe were taken using the null technique.

Results and Discussion The co-ordinate system is as shown in Fig. I b

where the di stances are normalized with the diameter of casing and liner, and the origin is located at the dump inlet plane. The inlet conditions have been measured in the pipe at a di stance of 200 mm downstream of the swirler (50 mm upstream of the dump expansion).

Inlet conditions

The velocity pro fil e at inl et has been measured from wall-to-wall. However, only the data for the upper half are shown here with its mirror image in the lower half. In ac tual measurements, the discrepancy between the upper half velocity profile and lower half is in the range of 3-4%. This deviation is as a result of the blockage effect of the probe. Figs 2-5 present the va ri ati on of inlet axial velocity, inlet tangential veloc ity, stati c pressure and turbulence intensity for non-swirling and swirling inlet conditions. Fig. 2 shows inl et profiles for non-swirling flow . It is seen that axial veloc ity is tl at in the middle and falls grad ually towards the wall showing the growth of boundary layer. The tangen ti al velocity is nearly zero. Static pressure is constant across the pipe cross sect ion. Turbulence intensity variation shows the ex istence of strong shear layer close to the wall, which is expected due to the growth of boundary layer. In the central portion, the value is constant and is around I %. In the fi gure, it appears that the peak value is approximately 7%. However, if measurements much closer to the wall are made, it is expected that the intensity could ri se to values close to 15%, which is typical of standard boundary layers.

Fig. 3 shows the inlet: conditions fo r 15° swirl er. It is seen that the axial velocity on the central line drops and the fl ow is forced outwards, the peak velocity lying somewhere midway between the centre and the wall on both sides . The tangenti al velocity di stribution shows a forced vortex nature close to the centre over a very small region. For the remaining cross section , the tangential velocity is constant. Static pressure variation across the pipe cross section is still nearly constant and the value is also same as that of

a. Axial velOOty 15

~, I 1.0 · ' '\. .s: .. \ :: " ::0

.~

0.5 ·

0.0 -1 0

:: j

\! I

-0.5 0.0 05 ' 0

R / ~

b. TcrgertiaI velocity

:::J 0.0 .............. .. ... 05 ! -5

-05

-, 0

-1.5 1-1 --_--____ _

-1.0 -05 00 05 , 0

R / ~

C. Stciic p'esstJ'e 05 r----

00 i ~ -0.

51 ...... --.

-;;.

" I

•• I •••• ••

-'5 1-- --~--1.0 -0.5 0.0 05 1.0

R/~

d. T lrtUence iriensity :,-' 20 i

~ 15

-1.0 -0.5 0.0 05 1.0

R/ ~

Fig. 2 - Inlet profiles for non-swirli ng flow

RAHIM el 01.: WALL PRESSURE DISTRIBUTION OF A MODEL CAN-COMBUSTOR 475

.~

:=: ro

::l

.~

::l

:5

~ F

a. Axial velocity 1.S 1

1.0

0. 5

0.0 .--__ -_--_- - -< _1,0 --as 0.0 0.5 1.0

RI R;

1. 5 b. Tangential velocity

lD

0.5

no

--<l5

- 1.0

-1.5 -1.0 -<l.5 0.0 0.5 1.0

RI R;

c. Static pressure 0 5,-----------'--·- -------,

0.0

- 10

- 15 -1 0 -<l 5 0 0 0 5 10

RI R;

d. Turbulence intensity l:)

25

20

15

10

- 10 -<)5 00 05 10

RI R;

Fi g. 3 - Inlet pro fi les for 15 0. swirli ng fl ow

non-swi rli ng flow . The turbulence intensity variation is a lso simi lar to the non-swirling case with slight increase in magnitude.

Fi g . 4 depicts the in le t conditions for the

30° swirl er. It is seen that the ax ia l ve locity on the centre line drops significantly and the peak velocity is forced outwards. The tangential velocity d istribution shows a forced vortex nature from the center axis up to the midway between centre line and wal l and then

a. Axial velocity 1 5 I

I 10

.~

::l .. ::J

0.5 v 00

-1.0 -0.5 0.0 0.5 1.0

R I Ri

b. Tangential velocity 1.5

10

0.5

." .. ::J 0.0

s

"I .:: -

-1.0 -0.5 0.0 0.5 1 0

RI Rj

c. Static pressure 0.5 ~------'---------,

-2.0

·2.5

·3.0 -1 .0 -0.5 00 0 5 10

RI Rj

d. Turbulence intensity 30

25

20

~ 15 ;:::

10

-1.0 -0.5 0.0 0.5 1.0

RIRj

Fig . 4 - Inlet profiles for 300. swirling fl ow

drops down towards the wall showing a free vortex nature. The static pressure variation is similar to the axial velocity variation, i.e., minimum pressure on the centerline that increases towards the wall. Thi s is due to the setting of rad ial pressure grad ients because of centrifugal force . The turbu lence intensity variation shows a presence of strong shear at the centre and ex istence of shear layer c lose to the w:.1ll s as in other cases.

476 INDI AN J. ENG. MATER. SCI.. DECEMBER 2002

.

a. Axial velocity 15,------------

, 0

'" 0 5

oo ~--__ --__ wrlr~ __ -__4

- 1 0 -0 5 0.0 0.5 1 0

R / Rj

b. Tangential velocity 15

10

0 5

~ '" 0 0

5 -" 5

_1 0

_15 .!--_____ - __ - _--_I

_ 1 0 -0 5 00 05 1 0

c. Static pressure 05~-----------

00

-0 ,

c: -1 0 ~ ~ -1 5 Q.

- 2 0

_25

~o.!----__ - __ -_-- _I

-1,0 -0. 5 0,0 0.5 1. 0

R/ Rj

d. Turbulence intensity 30,--------~--_,

20

,0

Fig. 5 -- Inlet profiles for 45° swi rling tlow

Fig. 5 hi ghli ghts the inle t conditions for

45° swirl er. It is seen that the variati on of ax ial velocity, tangential ve locity, static pressure and

tu rbu lence intensity are simil a r to the 30° sw irle r conditi ons (Fig . 4) with slight increase in magnitude which is expected with the excepti ons that turbulence in tensity reduces sl ig htly. The reduction in the turbul ence intensity is due to the increase in the size of the centra l shear regio n with a corresponding decrease in the grad ient.

Air casing wall pressure variation

Fig. 6a shows the cas ing wall pressure variat ion witho ut the liner fo r a ll the inlet conditio ns and Figs 6b, 6c and 6d show the wall pressure vari ation with liner at three dump-gaps for the same inlet conditio ns . The wall static pressure at any g iven axia l location was measured as an average o f three press ure taps placed at equ al ang les a long the circumference.

Wall pressure variati o n depicted in Fig . 6a for no liner shows the ex istence of wall recirculation zone extending beyond the po ints of measurement of pressure for 0° and I S° sw irl. For 0° sw irl , the pressures variation is seen to be nearl y constant up to half-length of the casing and then increases g radua lly .

For IS° swirl , variatio n is a lmost simil ar to the

0° swirl with slig ht increase in velocity and pressure

due to the centrifugal force. For 30° swirl , the flow

attaches to the wall around X/Dc =l.25 . For 45° sw irl , the flow is forced outwards due to large centrifugal force resulting in sudde n increase of pressure at the wa ll . The po int of attachment of flow is seen to be close to X/D,=0.5 but the fl ow resep, rates and o nl y at taches to the cas ing wall around X/D,= 1.0 same as 30° swirl. The va ri atio n of pressure is seen to be identical to the flow observed in sudden ex pans io ns. From the pressure variation, it can be concluded that the wall rec irculation zone is the small est for 45° swirl.

Fi gs 6b, 6c and 6d show the wall pressure va riation a lo ng the air casing length with liner at three dump­gaps (0.5, 1.0, and 1.5). From the press ure variatio n, it is clearly seen that the point of reattachment moves c loser compared to the case of no liner. This is due to deflection of fl ow towards the wall as a result of the liner pl acement. The width of the wall rec ircul at io n zone is a lso affected. For dump gap o f 0.5 , the reattachment points fo r the fl ow on the wall a re X/ Dc =1.75 for 0°, 1.5 for IS°, 0.75 for 30° and 1.25 for 45°. The pressure vari ation fo r 45° shows th at the flow attaches faster but reseparates to rea ttach only by X/D,= 1.25. The fl ow reseparates from the wall due to sudden pressure created due to accelera ti on o f fl ow at the dome of the liner, which does not happen for other swirl conditions . The wall pressure variation for other dump gaps has similar variation (Figs 6c and 6d) except in the sh ift of attachment point for the fl ow. For dump gap of 1.0, the attachment points are 1.75 for 0° swirl, 1.5 for IS° swirl , 0.75 for 30° swirl and

1.0 for 45° sw irl. In this case, the press ure va ri ati on

for 45° sw irl sti ll shows the reseparati on of flow at the

c: ~

<l. -~

<l.

RAHIM el (/1.: WALL PR ESS URE DISTRIBUTION OF A MODEL CAN-COMI3USTOR

.S

.,;

I -- sw= 0° 0.2

1 0.1

0.0

:: j -0.31 -0.4 L ____ ~----~----~----~----~---__l

0 .0

0.2

0 .1

0.0

-0 .1

-0 .2

-0 .3

-0 .4 0.0

0.2

0.1

0.0

-0 .1

0.5 1.0 1.5

Xci DC

2.0 2 .5

Fig. 6a - Waif pressure variation along the casing w ithout liner

0 .5 1.0 1.5 2.0 2 .5

Fig. 6b - Wall pressure along the cas ing wi th dump-gap = O.S

3.0

3.0

<l.

~ <l.

-0 .2

-0.3

-0 4 0.0 0.5 1.0 1.5 2.0 2.5 3.0

Fig. 6c - Wall pressure along the casing with dump-gap = 1.0

0 .2 -

0 .1

0 .0 c: ~

<l. -0 .1 -~ . 1.

-0 .2

-0 .3

-0.4 0 .0 0 .5 1.0 1.5 2 .0 2 .5 3 .0

Fig_ 6d - Waif pressure along the casing w ith dump-gap = I.S

477

478 IND IAN J. ENG. MATER. SC I. , DECEMB ER 2002

SW=4U 1 . 0~--------~----------------------------------------------------------,

0.5

c

~ !:l-- 0.0

3 !:l-

·'.). 5

-1 .0~--------~----------.-----------~----------r-----------r---------~ o 3 4 5

Fig . 7a -- Wall pressure variat ion along the liner with dUIllP-gap = 0.5

/.0 ,----------. --------------------------------------------------

0.5

c ~

!:l- ao -3

!:l-

-0. 5

-1.0 ~--------,_--------~--------_.----------,_--------~--------~

o 2 3 4 5

Fig . 7b -- Wall pressure variation along the li ner with dUlllp-gap = 1.0

.: 0 .---------~--------------------------------------------------------_.

0.5

c ~

!:l- 00 -3 !:l-

-0. 5

-1.0 0 2 3 4 5

X/1 O,

Fig. 7c -- Wa ll pressure variati on along the liner with dump-gap = 1.5

RAHIM et 01.: WALL PRESSURE DISTRIBUTION OF A MODEL CAN-COMBUSTOR 479

wall. For 1.5 dump gap, the corresponding lengths are 2.5, 1.75, 1.0 and 1.0, respectively.

From the wall pressure variation, it can be

concluded that 30° sw irl is the optimum swirl for dump gap 0.5 in terms of flow attachment and size of

recirculation zone whereas it is 30° and 45° swirl for other two dump gaps.

Liner wall pressure variation

Figs 7a, 7b and 7c show the wall static pressure variation on dome and liner for three dump-gaps 0.5 , 1.0 and 1.5, respectively , for the same inlet condi tions.

Fig. 7a shows that the wall pressure at the dome is

the hi ghes t at the centre for 0° swirl , which reduces drastically over the dome resulting in negative pressure over the liner due to acceleration of flow on the do me. The pressure starts to recover in the liner and reaches the zero pressure value at X/Of =2.5.

Similar variation is seen for 15° swirl. For 30° and 45° swirl, the pressure variation is significantly different in the dome region from non-sw irling flow. The flow is defl ec ted towards the casing wall side creating a partial vac uum at the centre depending upon the intensity of swirl. As swirl intensity increases, pressure at the centre of dome goes on decreas ing. The pressure variation on the liner wall is almost similar for all the inlet conditions. The negative pressure on the liner wall is indicative of nearly a separated flow on the liner. The pressure on the liner

is constant beyond X/Of=2.5 for 0° and I S° swirl and

X/ Of = 1.5 fo r 30° and 45° swirl. The variation of pressure on the liner for other dump gaps is almost similar to dump gap equal to 0.5 with slight shift in uniform pressure on the liner. For dump gap equal to 1.0, the uniform pressure on the liner is achieved by

X/Of =2 .0 for 0° and I S° swirl and 0.75 for 30° and 45° swirl (Fi g.7 b). For dump gap equal to 1.5,

(Fi g. 7c), these values are 1.5 for 0° and IS ° and 0 .5

for 30° and 45° swirl. From the pressure variation on the liner wall , it can be concluded that 30° and 45° swirl at inlet is the proper swirl for dump gap of 1.0 and 1.5.

Conclusions Swirling inlet flow and positioning of the liner

(dump-gap) have been identified as important parameters , which influence the flow development and uniformity in can-combustors. Based on the wall

pressure measurements on the casing and liner, following conclusions have been drawn: (i) Swirling flow reduces the size of wall recirculation zone both in terms of length and width , resulting in possibility to improve the flow mlxlIlg in the annulus; (ii) Attachment length increases with increas ing dump-gap for weak swirling flows only; (iii) Dump­gap has no marked effect on the variation of liner wall pressure for a weak swirling flow, whereas for strong swirling flow , the liner wall pressure is nearly constant for the entire annulus length for dump-gaps of 1.0 and 1.5, a desirable feature for flow split into the liner through different liner holes provided on the liner wall; and (iv) The optimum combination for achieving better mixing and flow development

appears to be 30° swirl at inlet for dump gap= 1.0.

Nomenclature D, = combustor air casing diameter, m D[ = liner diameter, m DG = dump-gap, VDI

L = di stance of dome head from the dump inlet plane, m P = static pressure, N/ml Pd.;" = inlet dynamic pressure, N/m2 Ph' = wall pressure, N/m 2 R = radial distance o f measurement locati on from the ax is o f

sy mmetry. m R; = inle t pipe radius, m T/ = turbulence inte nsi ty , ( % ) Ua = axial ve locity, m/s U",.; = mass averaged inle t ve loc ity, m/s U, = tangenti al velocity, rnls Xc = distance of axial locati on on casing from dump inle t plane,

m XI = di stance of ax ial location on dome/liner from dome head,

m

References I Bicen A F & Jo nes W P, Combllst Sci Tecllllol. 49 ( 1986)

I-IS .

2 Koutmos P & McG uirk J J, Trans ASME, J Eng Gas Tll rbines Power, III ( 1989) 3 10-3 17.

3 McG uirk J J & Palma J M L M, Trans ASME, J Eng Gas Tllrbin es POIVer, 115 ( 1993) 594-602.

4 Fishcnden C R & Stevens S J, J Aircraft, 14 (I) ( 1977) 60-67.

5 Carro tte J F, Bailey D W & Frodsham C W, Trans IISME. J Eng Gas Tllrbin es Power, 117 (1995) 679-685.

6 Bharani S, Isothermal flo w sllldies in a reverse flo\\' gas IlIrbin e combllslor, Ph.D. thesis, Dept. of Applied Mechanics, liT De lhi , India, 1997.

7 Ahmed S A & Nejad A S, J ProplIls Power, 8 (2) ( 1992) 339-334.

8 Agrawal D P & Singh S N, Arabian J Sci Eng, 20 ( 1995 ) 66 1-675.

9 Mathur M L. J Insl Eng (India) , 55 ( 1974) 93-96.