Numerical Investigation of Swirl's Effects in the Outer Annulus of a Reverse-flow Gas Turbine Combustor Mostafa Ghanei Tayeblou and Kavous Ariafar Department of Civil Engineering Islamic Azad University, Fasa branch IRAN Javidan_ghanei@yahoo.com Abstract: In this present paper, flow characteristics in the outer annulus of a reverse-flow gas turbine combustor have been analyzed by using a finite volume method. Three different intensities of swirl have been selected to impose on the inlet flow and effects of them on velocity distribution and flow split through different holes on liner surface in the outer annulus have been investigated. The RSM turbulence model has been used for flow predictions. Comparisons of the present numerical predictions with the available experimental data are found to be in reasonable agreement. Key-Words: Reverse-flow gas turbine combustor, Outer annulus, Swirl, computational fluid dynamics Introduction In small gas turbine engines, there has been constant desire to reduce weight of the combustors resulting in the demand for shorter length of them. Annular reverse-flow gas turbine combustors are widely used for small engines to overcome high speed shaft whirling problem because they shorten the shaft length between turbine and compressor []. The annulus is an integral part of any combustor through which air is fed to the liner through different holes and play a significant role in air-fuel mixing in the liner. For this reason, flow study in this part of combustors is necessary. Complex geometry and restricted passage in this combustors, inhibits the internal flow mapping even with sophisticated measuring instruments. Due to the ability of Computational Fluid Dynamics (CFD) and computer simulation to explain to flow field in such combustors, modern trends for studying of flow in these combustors are assistance of computer simulations. Cadiou and Grienche [2] have conducted hot flow studies inside a liner with and without primary holes to assess the influence of primary holes on the efficiency of a reverse-flow combustor. They have reported that the combustion efficiency of the combustor was not significantly affected by these conditions, as the flow through the cooling holes was sufficient to meet the primary zone requirements. Riddlebaugh et al. studied dilution et traectory in a combustor turn-section. They also investigated the effect of spacing between the holes on the traectory [3]. A study on a Pratt and Whitney torroidal-vortex sector combustor model by Hu et al. has provided data on the mean velocity and turbulence quantities, under cold and hot flow conditions [4]. Bharani et al. have experimentally studied the effect of swirl on the flow characteristics in the outer annulus of a reverse-flow gas turbine combustor [5]. Flow characteristics have also reported by Mohan et al. [6] in the annuli of a two-dimensional axisymmetric reverse-flow combustor and predicted numerically the flow split in a simplified axisymmetric combustor model. Crocker and Smith [7] have predicted dilution zone flow behaviour for angular dilution ets. This present paper describes a numerical investigation of flow in the outer annulus of a model reverse-flow gas turbine combustor. The experimental data of Bharani et al. [5] has been used for comparisons. 2 Computational Model 2. Governing Equations The governing Reynolds-averaged Navier-Stokes equation for Steady, incompressible and constant properties flow, are continuity and momentum. These equations in tensor notations are as follows: Conservation of mass: ( U ) = () Conservation of momentum: ISSN: 79-595 262 ISBN: 978-96-474-58-8
( ) P U i ρ U iu = + μ ρuiu (2) i Where Ui and ui are the components of the mean and fluctuating velocities, P is the mean pressure, and ρ and μ are the fluid density (constant) and viscosity, respectively [8]. 2.2 Numerical Solution Procedure A finite volume, non-staggered grid approach has been employed to solve the governing equations in a three-dimensional, axisymmetric coordinate system. The pressure and mean velocity field coupling are resolved with the SIMPLE algorithm [8], and the convective terms are discretized with the secondorder upwind scheme. In the second-order approach, higher-order accuracy is achieved at cell faces through a Taylor series expansion of the cell-centred solution about the cell centre. The governing equations are solved numerically by using the fluent code [9], which has been extensively validated against various experimental data. 3 Geometry & Boundary Conditions The geometrical details of the prototype reverseflow combustor in experimental work are given in Table, and its schematic layout is shown in Fig.. Table I Geometric Details for Prototype Combustor Height of the inlet annular passage Height of the outer annular passage Height of the inner annular passage at the dome end Height of the inner annular passage at the turn-section end Height of the liner Diameter of dilution holes Pitch of dilution holes Diameter of primary holes Pitch of primary holes Diameter of liner cooling holes Pitch of liner cooling holes Diameter of turn-section cooling holes Pitch of turn-section cooling holes Height of outlet annular passage Outer casing diameter Liner diameter (outer) 2 mm 8.5 mm 8.5 mm 2.5 mm 36 mm 7 mm 28 mm 3.5 mm 4 mm 2 mm 7 mm.5 mm 4 mm 2 mm 28 mm 263 mm The swirl of different intensities has been imposed onto the inlet flow of the combustor with annular swirler which has 6 vanes each having thickness of.5 mm. Each vane fixed at 2, 3 and 4 with respect to the axis of the combustor giving swirl number (S) of.35,.55 and.84 respectively. The swirl numbers are based on geometry and have been calculated using the expression []: ( d / d2 ) ( d / d ) 3 S =.66 tanθ 2 (3) 2 Five axial planes, as shows in Fig., have been identified on the combustor for measurement. i) Inlet to the combustor (2 mm downstream of the inlet swirler), ii) Plane-A (upstream of the row of dilution holes), iii) Plane-B (downstream of the row of dilution holes and upstream of the row of primary holes), iv) Plane-C (downstream of the row of primary holes), v) Plane-E (at the outlet of the combustor). Axis of symmetry Fig. Schematic layout of prototype combustor Results have been presented for two locations at each plane. The first location is inline with the dilution hole and the second is not inline with it. For this present study, the working fluid is air and the mean inlet velocity is 28 m/s. The experimental work was carried out on full annular reverse-flow combustor. But due to the symmetric geometry with respect to the axis of the combustor, a 6 sector has been selected for simulation and at the both side of the sector the periodic boundary condition has been used. The other boundary conditions are velocity inlet and pressure outlet at the inlet and outlet of the combustor respectively. 4 Computational Grids Three grid sizes (as presented by table 2), are investigated. The original grid that has been produced by Gambit [] has been illustrated in figure 2. In this figure three swirler vanes are located on inlet of the combustor. ISSN: 79-595 263 ISBN: 978-96-474-58-8
Fig. 2 Calculation grid.5.9.6.3 Table II Numerical Grids Grid A 32 Grid B 485 Grid C 7 5 Results and Discussion Axial velocity profiles at Plane-I are depicted in figure 3. They are nearly flat in the centre of passage with marginally higher values towards the casing wall and lower close to the liner surface. Num. (S=.55) Exp. (S=.35) Exp. (S=.84) Fig. 3 Axial velocity profile at Plane-I This may be due to mild centrifugal force induced by the upstream annular swirler. Velocity profiles for S =.55 are nearly identical in nature and magnitude. Close to the liner surface, the profiles show defect below η =.2, where η = (R Ri) / (Ro Ri). The change in the nature of the axial velocity profiles can be attributed to the imposition of centrifugal force as observed in the case of S =.35. With further increase in the swirl number (S =.84) at the inlet, the nature of axial velocity profiles does not change much from the earlier case. Figure 4 (a, b) depicts axial velocity profiles at Plane-A located ust upstream of the row of dilution holes. The increasing in this magnitude over the inlet velocity is because of the reduction in passage area. 2.5.5.8.6.4.8.6.4.2 Num. (S=.55) Exp. (S=.35) a) Location Exp. (S=.35) Num. (S=.55) b) Location 2 Fig. 4 Axial velocity profile at Plane-A Axial velocity profiles along the annular passage for S =.55 is similar to the velocity distribution seen for S =.35 but with the lower velocities at location A (figure 4a) instead of A2 (figure 4b). The flow velocities at location A for S =.55 are lower as compared to S =.35 by about.5 Uavi. This may be because of relative shift of direction of flow at A and A2 vis-à-vis the downstream dilution hole with the increased swirl at inlet. For S =.84, the flow magnitudes are again reversed between A and A2 for the same reason. For 2 annular vane swirler (S =.35) the axial velocity profiles at location A is symmetric having mid-value of.6 Uavi, whereas the profile for A2 is nearly flat with magnitude being equal to the inlet average velocity. The difference in the nature at the two locations may be due to their respective position vis-à-vis the upstream swirler vanes and the downstream dilution holes. The location A is practically inline with the ISSN: 79-595 264 ISBN: 978-96-474-58-8
dilution hole but the follow direction at this location is not inline with the downstream dilution hole due to imposition of the swirl at the inlet. For no swirl at the inlet, Bharani et al. [5] have shown that the magnitude of flow velocities where the flow is inline with the downstream dilution holes are lower compared to the flow which is not inline. The flow velocities at A2 are lower because the flow direction is inline with the downstream dilution hole due to the imposed inlet swirl. The other factor responsible for this reduction could be the presence of the swirler vane wake. Axial velocity profiles at Plane- B are shown in figure 5. With the flow split at the dilution holes, the magnitude of axial velocities at Plane B reduces but the peak velocity moves towards the liner surface at both the locations. The nature of the axial velocity profiles at Plane C are similar to the distribution at Plane B with further reduction in magnitude due to the further flow split through the primary holes (figure 6)..6.4.8 Exp. (S=.35).6.4 Num. (S=.55).2 Fig. 5 Axial velocity profile at Plane-B at location.9.6.3 Exp. (S=.35) Num. (S=.55) higher than the inlet mean velocity as the combustor outlet duct has smaller area than the inlet flow passage..8.4 Num. (S=.55) Fig.7 Axial velocity profile at Plane-E For S =.55, the relative magnitudes of velocity increased at the three planes (A, B and C), whereas the values reduced at the exit Plane E, with the profiles coming closer at two locations at each plane. For S =.84, the relative magnitudes of velocity at the measurement locations are either less or nearly same at all planes, with increase in flatness in the profiles. Flow splits for swirling inlet flow with optimum dump-gap (2mm) are given in table 3. It is observed that with the imposition of swirl at the inlet, flow split through the turn-section and liner holes are affected depending upon the intensity of swirl. Imposition of swirl on the inlet flow increases the flow split at the turn-section. For S =.84, it is observed that flow split through dilution and adacent cooling holes increased. This increase may be due to impingement of inlet flow on the casing wall because of higher centrifugal force generated and the flow reflecting back to the liner surface, close to Plane-A. For S =.35 and S =.55, the flow split through dilution and adacent cooling holes are decreased respectively due to flow that moving away from the liner surface because of the centrifugal force. Flow split at primary and adacent cooling holes remains nearly constant for S =.35 and S =.55, While for S =.84, it reduces because of the increased flow split occurring at the row of dilution and turn-section cooling holes, as explained above. Fig. 6 Axial velocity profile at Plane-C at location Axial velocity profiles at the exit Plane E are nearly flat for both the locations. The magnitudes are ISSN: 79-595 265 ISBN: 978-96-474-58-8
Table III Flow split through the line holes for various swirl intensities Flow split, % of mean inlet velocity Type of holes Swirl number.35.55.84 Primary & cooling holes 7.6 7.7 8.5 6. Dilution & cooling holes 29.4 27.32 25.4 3.66 Turn-section cooling holes.6.66.87 3.6 5 Conclusions Effect of inlet swirl on flow characteristics of the outer annular passage for a reverse-flow gas turbine combustor with optimum dump-gap of 2 mm has been analyzed and following conclusions can be drawn:. Axial velocities at most of the measurement planes are lower near the liner surface and higher towards the casing wall. These also significantly affected by imposition of swirl. 2. Inlet swirl also affects the flow split at different liner holes. It is observed that the flow split at the turn-section and primary holes remained nearly constant up to S =.55, while the split through dilution holes decreased. With the imposition of higher intensity of swirl, flow split through the turnsection and dilution holes increase which subsequently reduces the flow split through primary holes. 3. The present computational work show that the Reynolds Stress Model (RSM) predicts the occurrence in outer annulus of a reverse-flow combustor in good agreement with the available experimental measurements. Publication Principles to allow readers [5] Bharani, S., Singh, S.N. and Agrawal, D.P., Effect of swirl on the flow characteristics in the outer annulus of a prototype reverse-flow gas turbine combustor, Journal of Experimental Thermal and Fluid Science 25, pp 337-347, 2. [6] Mohan, R., Singh, S.N. and Agrawal, D.P., Flow splits in a reverse-flow combustor, Proceedings of the 22nd National Conference on Fluid Mechanics and Fluid Power, Madras, India, Dec, 3-5, pp 82-86, 995. [6] Crocker, D.S. and Smith, C.E., Numerical investigation of enhanced dilution zone mixing in a reverse-flow gas turbine combustor, Journal of Engineering Gas turbine Power, 7, 272-28, 995. [7] Versteeg, H.K. and Malalasekera, W., An Introduction to Computational Fluid Dynamics, the Finite Volume Method, Longman Group Ltd, 995. [8] FLUENT User s Manual, Version 6..2, 2. [9] Sawyer, J.W., Sawyer's gas turbine engineering hand book, Turbo machinery international publications, Third edition, vol. : Theory and design. [] GAMBIT User s Manual, Version 2..4, 998. References: [] Lefebvre, A.H., Gas Turbine Combustion, McGraw-Hill, 983. [2] Cadiou, A. and Grienche, G., Experimental study of a reverse-flow combustor: influence of primary holes on combustion efficiency, ASME paper NO. 89-GT-249, 989. [3] Riddlebraugh, S.M., Lipshitz, A. and Greber, I., Dilution et behavior in the turn-section of a reverse-flow combustor. AIAA paper AIAA-82-92, 982. [4] B Hu, T.C.J., Cusworth,R. A., and Sislian, J.P., An experimental and computational investigation of an annular reverse-flow combustor, UTIAS report No. 338, Institute for Aerospace Studies, University of Toronto, 99. ISSN: 79-595 266 ISBN: 978-96-474-58-8