ASME Journal of Turbomachinery, vol. 120, No. 4., Oct. 1998, pp. 753-759.

EFFECT OF SQUEALER TIP ON ROTOR HEAT TRANSFER AND EFFICIENCY A. A. Ameri

E. Steinthorsson
Institute for Computational Mechanics in Propulsion (ICOMP) NASA Lewis Research Center

David L. Rigby
NYMA, Inc. NASA Lewis Group

ABSTRACT Calculations were performed to simulate the tip ow and heat transfer on the GE-E3 rst stage turbine, which represents a modern gas turbine blade geometry. Cases considered were a smooth tip, 2% recess, and 3% recess. In addition a two-dimensional cavity problem was calculated. Good agreement with experimental results was obtained for the cavity calculations, demonstrating that the k- turbulence model used is capable of representing ows of the present type. In the rotor calculations, two dominant ow structures were shown to exist within the recess. Also areas of large heat transfer rate were identied on the blade tip and the mechanisms of heat transfer enhancement were discussed. No signicant difference in adiabatic efciency was observed for the three tip treatments investigated. INTRODUCTION The tips of turbine blades in gas turbines experience large thermal loads which can lead to tip burnout. The large thermal load on blade tips is due to hot gases owing through the gap between the blade tip and the shroud. The ow accelerates due to pressure difference between the pressure and suction sides, causing thin boundary layers and high heat transfer rates. The ow across the tip is also undesirable from the perspective of efciency since it increases the losses in the ow. A common strategy to reduce the ow on the tip is to use a recessed tip, also known as a squealer tip. By using a squealer tip, the tip gap can be made smaller without the unacceptable risk of catastrophic failure should the tip rub against the shroud in the course of turbine operation. The smaller gap reduces the ow rate through the tip clearance, which leads to smaller losses and lower heat transfer. It is believed that the tip recess also acts to increase the resistance to the ow (Metzger et al. ,1989). In order to study the effects of a squealer tip geometry on heat transfer and losses, Metzger et al. (1989), and Chyu et al. (1989) performed experiments using cavities of varying depth to width and gap to width ratios. As a result of these experiments Metzger et al. made a number of important conclusions. In particular, they concluded that for a given pressure difference across the gap there is an optimum value of depth to width ratio beyond which no further ow reduction will occur. They also concluded that 1

although the rate of heat transfer on the cavity oor is lower than that on a at tip, the reduction in heat transfer is offset by the high heat transfer in the redeveloping ow on the downstream gap and by the additional heat transfer area created on the side walls. Thus, they recommended that shallow cavities are preferred if overall heat transfer reduction on the cavity wall is the goal. The experimental studies cited above provide insight into the nature of the ow eld around the squealer tip. However, actual turbine blades differ greatly from the idealized experimental setup. The ow in the blade tip and gap are highly three-dimensional. The relevant parameters, namely, depth to width and clearance gap to width ratios of the recess vary widely along the gap. The pressure difference across the tip also varies widely along the blade. Thus it is unlikely that a simple model will provide designer of a turbine blade with the information needed to understand the ow in the tip region. At the present time neither experimental data nor numerical simulations are available in open literature that shed light on the details of the ow and heat transfer on a squealer tip ow. Also, with the exception of the heat transfer data of Yang and Diller (1995) which was taken at a single point on the cavity oor of a rotor blade, there are no data available on heat transfer on squealer tips. In this paper we conduct a numerical study of heat transfer on a squealer tip of a generic modern gas turbine blade. This numerical simulation also allows insight into the nature of the ow within the cavity and its effects on efciency. The blade chosen for this study is the General Electric E3 design detailed in two NASA reports (Halila et al., 1982 and Timko1982). The numerical simulations of the heat transfer within the simplied squealer tip model used by Metzger et al. (1989) are rst performed with a view to demonstrate the ability of the numerical model to accurately predict the ow and heat transfer as well as to forewarn us as to the possible limitations of the analysis. In the section to follow we will give a brief description of the numerical method used in the simulations, the turbulence model and the numerical boundary conditions. Afterwards we will present the results of heat transfer predictions and

comparison with experimental data on the walls and rim of transverse grooves used in the experiments cited above to simulate the squealer tip. We will subsequently show the heat transfer results obtained for the GE-E3 turbine blade for a at tip and two recessed tip cases and discuss the results. Finally we will present the calculation of efciency and close by presenting a summary and the conclusions. COMPUTATIONAL METHOD The simulations performed in this study were done using a multi-block computer code called TRAF3D.MB (Steinthorsson et al. 1993). This code is a general purpose ow solver designed for simulations of ows in complicated geometries. The code is based on TRAF3D, an efcient computer code designed for simulations of steady ows in turbine cascades (Arnone et al. 1994). The TRAF3D.MB code solves the full compressible Navier-Stokes equations. It uses the nite volume method to discretize the equations. The code uses central differencing together with articial dissipation to discretize the convective terms. The overall accuracy of the code is second order. The TRAF3D.MB code was described in detail by Steinthorsson et al. (1993). The present version of the code which employs a twoequation model has been used in connection with an internal ow calculation of heat transfer as described by Rigby et al. (1996) Turbulence and Transition Models Algebraic model of Baldwin and Lomax (1978) has been quite successful in predicting the rate of heat transfer to turbine blades. Combining this turbulence model with other models simulating laminar-turbulent transition and models simulating leading edge heat transfer enhancement have been shown to produce accurate results (Ameri and Arnone 1996, Boyle(1991). When using a multiblock approach in connection with complex geometries involving many no-slip surfaces, it is advantageous to use a turbulence model that does not require the computation of the dimensionless distance to the wall (y+) as is done in BaldwinLomax zero equation model and many two-equation models. Therefore, for the present computations, it was decided to use the k- turbulence model developed by Wilcox (1994a,1994b) with subsequent modications by Menter (1993). The model integrates to the walls. Chima (1996) incorporated the latter model in a Navier-Stokes solver and presented results of its application to turbomachinery ows and heat transfer. Below we present the equations describing the turbulence in tensor notation.
1 ( s i ) ,t + ( s i u j + q ij ) ,j = -- ( P D ) t q i, j = + ---- s i, j k

P -- =

Re 2 ----------- t u i, i
2 2 -- u i, i 3

(3)

where is the magnitude of vorticity. The destruction terms, D, are given by

(4)

The coefcients appearing in the model are as follows: =2.0, =3/40, *=0.09F, =(5/9)(F/F), and *=F, where
5 18 + ( Re T R ) F = ---------------------------------------------4 1 + ( Re T R ) 0 + ( Re T R ) F = ------------------------------------1 + ( Re T R ) 0 + ( Re T R ) F = ---------------------------------------1 + ( Re T R ) k Re T = ------
4

(5)

(6)

(7)

(8)

0=0.1, 0*=0.025,R=8, R=2.7 and Rk=6. The turbulent thermal diffusivity is computed from:
t t = ---------Pr t

(9)

(1)

i=1,2

(2)

where s1=k and s2= also t=* ----- . The production source terms, P, of equation (1) are dened as:

where is density. A constant value of 0.9 is used for turbulent Prandtl number Prt. Boundary Conditions The types of boundary conditions encountered are as follows: 1) Inlet: The inlet boundary condition for axially subsonic ows is treated by specifying the inlet total temperature and inlet total pressure as well as the inlet angle profiles. The outgoing Riemann invariant is extrapolated to the inlet from within. The total temperature and total pressure profiles are determined to match the law of the wall for the specified hydrodynamic and thermal boundary layer thicknesses on the hub and/or the shroud. For the turbulence quantities for the blade row, an inlet turbulence intensity of 8% and a length scale of 10% of axial chord is used. These values are estimated and believed to be representative of t h e c o n d i t i o n s ex i s t i n g a t t h e i n l e t o f t h e b l a d e r ow. 2) Exit: At the exit boundary, for the subsonic axial ow, the pressure is specied and all the other conditions (including the turbulence quantities) are extrapolated from within. The pressure at the exit plane for the turbine blade considered here is computed

Ameri et al.

C
3 1 2

w y
3

11 12

x
4 5 6 7 8 9 10 Fig. 2 Three-block grid used for the discretization of the ow.

Fig. 1 Diagram and nomenclature of the cavity used in the experiments of Metzger et. al. (1989)
150.0

SR =

50 ------ 2, K < 25 R K R 100 --------, K R 25 KR

Fig. 3 Nusselt number distribution along the rim, side walls and the bottom of the cavity from Metzger et. al.(1989). D/W= 0.5, C/W=0.1, Re=15,000. (11)
150.0

0.02

D/W

Fig. 5 Measured Pressure coefcient (closed symbols) of Metzger et.al.(1989) and present computations (open symbols). Ameri et al.

0.00 0.00

0.20

0.40

F


0.04


Cavity The geometry and the data chosen for this test were taken from Metzger et al. (1989). The sketch of the geometry used in the experiment and some of the nomenclature are shown in Fig. 1. Two cases with different cavity depth to width ratios (D/W) of 0.2 and 0.5 were considered. For both cases heat transfer data for the rim as well as the bottom of the cavity are available. For the deeper cavity the heat transfer data on the side walls were also taken. For both cases considered, the gap to width ratio (C/W) is 0.1 and the Reynolds number based on gap height and average inlet velocity is 15,000. Fig. 2 shows a three-block grid used to discretize the ow.(the grid is coarsened for clarity) The blocks contain 81x65, (81x41 for the shallower case), 129x41 and 25x41 grid points respectively. Coarser grid was also used to calculate the ow and the results presented

Fig. 4 Nusselt number distribution along the rim, and the bottom of the cavity from Metzger et. al.(1989). D/W=0.2, C/W=0.1, Re=15,000.
0.06

C/W=0.05 C/W=0.10 C/W=0.15 C/W=0.10

0.60

RESULTS AND DISCUSSION We rst compute the rate of heat transfer on the surfaces of a cavity and determine the pressure losses due to such a ow. This is done in order to verify the ability of the numerical procedure and in particular the turbulence model, to handle the similar problem of predicting the heat transfer on the rim and within the cavity on the blade tip for which no measured data is available.

Nu

KR is the equivalent sand roughness height. Here KR=5 corresponding to an hydraulic smooth surface is used. 4) Periodic boundaries are computed as interior points.

Present Analysis Data,Bottom and Rim, Metzger et al., 1989 100.0

50.0

0.0 0.0

5.0

where

10.0

= S R u y

(10)
wall

by integration of the radial equilibrium equation with specied hub endwall pressure at the exit. 3) Walls: At the walls, the normal pressure gradient is set to zero, the temperature is specied, and the no-slip condition is enforced. The density and total energy are computed from the pressure and the temperature. Also the kinetic energy of turbulence (k) is set to zero while the specic dissipation () is calculated using the following boundary conditions.

100.0 Nu

Sides Rim Bottom Present Analysis

50.0

0.0 0.0

5.0

10.0 String distance

15.0

20.0

15.0

herein are spatially converged. Note that in the computations, the grid in the exit region was extended to allow complete ow attachment very near the rim before exiting the computational domain. Figure 3 presents the rate of heat transfer as computed and measured for the deeper cavity. The abscissa is the string distance measured along the rim and walls of the cavity, starting from x=0, y=4. The ordinate is the Nusselt number dened as: hc Nu = ----(12) k In Eqn. 11, h is the heat transfer coefcient based on the inlet temperature, c is the inlet gap height and k is thermal conductivity of the bulk inlet ow. The gure shows generally good agreement between the experimental results and the analysis. However a large rise in heat transfer is predicted on the upper portion of the downstream side wall. A rise in the rate of heat transfer in that region is expected to exist due to ow stagnation. The data however do not indicate as large a rise as predicted by the analysis. In Fig. 4 the comparison between analysis and experimental data for the shallow cavity is presented. The experimental data were taken on the rim and the bottom of the cavity but not on the side walls. Again it is observed that the agreement in general is good. Finally, in Fig. 5 the measured pressure loss coefcients dened as: (13) are plotted as a function of cavity depth ratio for three families of gap clearance ratios. In the above denition, L is the gap length in ow direction and and V are the bulk density and velocity of the inlet ow and p is the pressure difference across the gap. The calculated values of the pressure loss coefcient for the two cases considered here are superimposed on this plot using open symbols. The agreement is very good. The two computed cases of ow in a cavity, demonstrate the capability of the present turbulence model to produce results of reasonable accuracy for this particular type of ow. Although this ow is essentially two-dimensional while the ow on a blade tip is three-dimensional, these results lend credence to the predictions presented in the following section. The turbulence model has been shown to perform quite well for ow and heat transfer predictions on turbine blades as was demonstrated by Chima(1996). Blade Tip Geometry and the grid. The geometry of the GE-E3 blade with the simulated squealer tip is shown in Fig. 6a. The blades have a constant chord length of 2.87 cm and an aspect ratio of 1.39. A squealer tip thickness of 0.030 in. (0.77 mm) was chosen for this study. Three geometrical cases were considered, a at tip, and two typical tip recesses of 2% and a 3%. The tip gap clearance was 1% for all the three cases. Figures 6b and 6c show the grid on the blade, the hub and the tip surfaces. Note that every other grid line has been eliminated for clarity. The grid topology is essentially the same as given in Ameri and Steinthorsson (1996) with the exception of the grid in the tip. For the present geometry, the tip grid is constructed using two blocks. One block covers the entire tip clearance excluding the cavity while the second block covers the space within the cavity. Grid is rened close to all no-slip surfaces such that the distance of the cell centers adjacent to solid walls, measured in wall units (y+), is close to unity. This aspect of the grid construction is
f = ( pC ) ( 2LV )
2

Fig. 6 Geometry and grid distribution for the E3 blade. Alternate grid lines are eliminated for clarity.

crucial to both the solution accuracy and to convergence. The dimension of the C grid covering the blade from hub to shroud is 193x49x99. In the tip clearance the grid dimension is 129x57x33. In the cavity there are 129x 33x 41 grid points for both the 2% and the 3% cases. The at tip case was run with the same grid in the tip as the recessed cases(129x 57x 33) to allow direct comparison of heat transfer devoid of resolution differences. A single block of 9x9x99 grid points covers the entire inlet upstream of the blade. Floweld. The conditions used for the numerical simulations are the same as experimental conditions of the warm rig used by Timko (1982) for the GE-E3 turbine and are as listed in Table 1. The inlet angle variation is taken

Ameri et al.

Pressure side 20%

Suction side

40%

60% Fig. 7 View of the velocity vectors in and around the tip at various axial locations for the case of 3% recess.

of reference, hence the nonzero velocity on the shroud. The tip flow is highly three-dimensional. The variation in flow patterns in the tip clearance is apparent from examining the vector plots. For the 20% to 60% axial distance the plots show a complex system of vortices in the tip clearance. Figure 8 shows the streamline patterns in the tip. In that figure at least two distinct vortices can be discerned to exist within the cavity. One vortex is a separation vortex, generated as the incoming flow separates off the inner edge of the pressure side rim. This vortex hugs the pressure side, sidewall and spills out of the cavity near the trailing edge of the blade. The second vortex which is apparently also a separation vortex, runs from the stagnation region to the suction side of the blade. These vortices are generated in addition to the separation vortex along the pressure side rim, suction side rim and the blade suction side tip flow vortex. This vortex system apparently offers additional blockage to the flow through the gap. The flow rate through the tip gap for the flat tip case was calculated to be 1.98% of the mass flow through the blade row. The 2% and 3% recessed blades had leakage mass flow rates of 1.8% and 1.7% of the total mass flow rate equivalent to 90% and 86% of that of the flat tip case. Heat Transfer The rate of heat transfer is presented in terms of Stanton number dened as:
h St = --------------------------- ref V ref C p

(14)

Fig. 8 Streamlines showing the blade tip ow patterns. TABLE 1. Run conditions Absolute pressure Ratio across the Blade Row Absolute inlet angle 0.44 69 Hub 74.5 Mid Span 78.5 Shroud Rotation Rate Tw/Tt 8450 RPM 0.7

from measurements but the wall to total temperature ratio was set at a typical value. In Figure 7 plots of computed velocity vectors in the tip region projected on planes normal to the machine axis for various axial locations are shown. The flow is shown in the rotating blade frame

where h is the heat transfer coefcient based on the absolute inlet total temperature. The reference velocity is the average relative velocity at a location 20% of the axial chord upstream of the blade. The reference temperature is the average absolute total temperature and the reference density is the average density at that location. (The inlet relative total temperature at the midspan is 0.879 times the absolute inlet total temperature.) Heat transfer rates on the blade surface at three spanwise locations are given in Fig. 9. This gure serves as a guide as to the magnitude of the tip heat transfer as compared to the blade surface heat transfer. The heat transfer results on the tip of the blade are given in Fig. 10a-c. Fig. 10a shows the Stanton number distribution on the at tip surface. The patterns of heat transfer contours on the blade tip are as expected and have been seen before in connection with our previous studies (Ameri & Steinthorsson 1995 and 1996). Those patterns include, sharp entrance effect on the pressure side of the tip surface where the rate of heat transfer reaches a maximum due to ow reattachment; the ensuing drop in heat transfer downstream of that location; a large rate of heat transfer around the corner from the blade stagnation point and the large increase in heat transfer on the suction side of the blade (near the crown). The average level of heat transfer on the at tip is similar in magnitude to the heat transfer rate on the blade leading edge. For the two recessed cases the heat transfer distribution is shown using one gure which shows the heat transfer on the bottom of the cavity and the rim surface of the squealer tip and a separate Fig. 11 to show the heat transfer on the side-wall of the recess. Fig. 10b and c show the heat transfer on the surfaces of the 2% and 3% cavity, respectively. It is observed that the rate of heat transfer on the bottom of the cavity reaches higher values than is seen on the at tip. On the rim the rate of heat transfer on the pressure side is comparable to the at case but is somewhat higher

Ameri et al.

8.0 10% Span 50% Span 90% Span

Vortex impact regions

6.0

Stx1000

4.0

15 3
  

3


2.0

Fig. 9 Surface heat transfer at 10%, 50% and 90% span on the at tipped blade.
3


2
 

3 4
1 8
 "

6 1

a
No recess
3 4 8

Pressure side

L.E.

Suction side

Fig. 11 Heat transfer distribution on the side wall of the cavity as unwrapped about the minimum x location in terms of 1000xSt.
1 4 5
8.0 6.0 4.0 2.0


7 7 2

b
2% recess
3 6 71 3 2 6 8 4 5 4 2 4

Stx1000

Flat tip 2% Recess 3% Recess

Percent Chord

Fig. 12 Tangentially averaged shroud heat transfer on the suction side. The large rate of heat transfer on the bottom of the cavity is due to ow impingement containing hot gas. This impingement can best be seen in Fig. 7 for the 20% axial distance.
4 2 3 1

c
3% recess

It is worth mentioning that in both the experimental investigations cited above, the rate of heat transfer on the bottom of the cavity was always observed to be less than that on the rim. However, in their numerical calculations, Chyu et al. (1987) do show instances where the bottom of the cavity has substantially higher heat transfer coefcient compared to the rims. Squealer side wall heat transfer is presented in Fig. 11. The inside wall surface is unwrapped around the minimum axial location. The ordinate is exaggerated for clarity. The abscissa is the axial chord distance. Areas of large heat transfer rates are generated as a result of the vortical action inside the cavity. The vortical action increases due to ow acceleration and stretching of these vortices. The highest rate of heat transfer is observed to exist near the trailing edge of the blade on the suction side wall. The cause of this increase was found to be the impact of the vortex spillage with the side wall. As such this rise in heat transfer, although present may be exaggerated as was discussed in connection with the two-dimensional cavity.

5 2 4 4 3

Fig. 10a-c Heat transfer distribution (1000x Stanton number) on the cavity oor and rim for no recess as well as 2 and 3% tip recess.

Ameri et al.

0.0 0.0

50.0

100.0

&

-0.5 Pressure side

0.0 Normalized distance

0.5 Suction Side

1.0

0.0 -1.0

8 1

2%

14

3%

150.0

The shroud heat transfer is presented in Fig. 12. As can be seen from this gure the rate of heat transfer on the shroud is higher for the at tip case as compared to the two recessed cases. Finally the total tip heat transfer for the three cases was compared. It was found that the 2% and 3% recess cases, had an increase of 80% and 90% respectively in total heat addition to the tip compared to the at tip case of the same tip clearance. Efciency As a by-product of our heat transfer computations, it is possible to compute the efciency and the possible effect of recess on efciency. The expression for the adiabatic efciency is
T in T ex = ----------------------------------------------------------------- p ex ( 1 ) T in 1 ---------- p in

dominant vortical structures were identied to exist in the recess region. The heat transfer rate on the recess surfaces were found to be strongly affected by these ow structures. Especially high heat transfer rates were observed where the vortices exit the recess region. No signicant effect due to the recessed tip on efciency was detected although the mass ow rate through the tip gap was found to be smaller by as much as 14% for the squealer tip as compared to at tip of the same clearance height. ACKNOWLEDGEMENT The authors wish to express their gratitude to Dr. Raymond Gaugler, Chief of the Turbine Branch, as well as to Dr. Louis Povinelli, Chief of the Turbomachinery and Propulsion Systems Division of NASA Lewis Research Center and Director of ICOMP for their support and encouragement of this work. Thanks are also due to Mr. Robert Boyle for his suggestions and his guidance. The computations were performed on the CRAY-C90 of NAS at NASA Ames Research Center. REFERENCES Ameri, Ali A., and Steinthorrson, E., 1995, Prediction of Unshrouded Rotor Blade Tip Heat Transfer, ASME 95-GT-142. Ameri, Ali A., Arnone, A., 1996, Transition Modeling Effects on Turbine Rotor Blade Heat Transfer Predictions, J. of Turbomachinery, Vol. 118, pp. 307-313. Ameri, Ali A., and Steinthorrson, E., 1996, Analysis of Gas Turbine Rotor Blade Tip and Shroud Heat Transfer, IGTI 96-GT189. Arnone, A., Liou, M. S., and Povinelli, L. A., 1994, ''Viscous Analysis of Three-Dimensional Rotor Flow Using a Multigrid Method, J. of Turbomachinery, Vol. 116, pp. 435-445. Baldwin, B. S. and Lomax, H., 1978, Thin Layer Approximation and Algebraic Model for Separated Turbulent Flows, AIAA paper 78-0257. Boyle, R. J., 1991, Navier-Stokes Analysis of Turbine Blade Heat Transfer, J. of Turbomachinery, Vol. 113, pp. 392-403. Chima, R. V. 1996, A k- Turbulence Model for QuasiThree-Dimensional Turbomachinery Flows, AIAA paper 96-0248 Chyu, M.K., Moon, H.K. and Metzger, D.E., 1989, Heat Transfer in the Tip Region of Grooved Turbine Blades, J. of Turbomachinery, Vol. 111, pp. 131-138 Chyu, M.K., Metzger, D.E. and Hwan, C. L., 1987,Heat Transfer in Shrouded Rectangular Cavities, J. of Thermophysics, Vol. 1, No.3, July 1987, pp 247-252. Halila, E. E. and Lenahan, D. T., and Thomas, L. L., 1982, Energy Efcient Engine, High Pressure Turbine Test Hardware Detailed Design Report,, NASA CR-167955. Menter, Florian R., 1993, Zonal Two-Equation k- Turbulence Models for Aerodynamic Flows, AIAA-93-2906. Metzger, D. E., Bunker, R. S. and Chyu, M. K., 1989, Cavity Heat Transfer on a Transverse Grooved Wall in a Narrow Channel, J. of Heat Transfer, Vol. 111, pp. 73-79. Rigby David, L., Ameri Ali, A. and Steinthorsson E., 1996, Internal Passage Heat Transfer Prediction Using Multiblock Grids and k-w Turbulence Model, IGTI paper 96-GT-188. Steinthorsson, E., Liou, M. S., and Povinelli, L.A., 1993, Development of an Explicit Multiblock/Multigrid Flow Solver for Viscous Flows in Complex Geometries, AIAA-93-2380. 7

(15)

In Eqn. 14 the primes signify relative total values, subscript in and ex, signify inlet and exit mass averaged values and is the specic heat ratio. The calculated efciency is based on the assumption of an ideal stator upstream and the average inlet values are computed at 20% axial chord upstream of the rotor. The exit values are computed at 50% axial chord downstream of the rotor. Because the blade surface temperature was set at a constant temperature of 0.7xthe inlet total temperature, in order to recover adiabatic conditions, the gas total temperature at the exit was modied to reect the loss of energy through heat transfer. Failure to do so leads to an error of 5 points in the efciecy for the present computations. The change in mass averaged normalized total temperature at the exit of the blade row can be computed as follows: (16) In the above equation is dimensionless viscosity and n is the normalized distance to the wall m is the normalized mass ow rate through the blade passage and A is the surface area. The integration is performed over all of the heat transfer surfaces. The mass averaged value of the exit total temperature is thus computed as: T' ex = T' unmodified + T' ex (17) The efciency for the row of blades was computed to be 91.3%, 91.4% and 91.4% for the 0%, 2% and 3% recess cases, in spite of the fact that there was a 10 to 14% reduction in gap ow. SUMMARY AND CONCLUSIONS In this paper, results are presented from three-dimensional simulations of ow and heat transfer over a turbine blade with a squealer tip. The simulations were carried out using a second order accurate nite volume scheme on a multi-block grid system containing 1.2 million grid points. Effects of turbulence on the ow eld were modeled using a k- two-equation turbulence model. The ability of the turbulence model to predict the heat transfer in ows of the type considered here was veried by simulating the ow in a two-dimensional model of squealer tips for which experimental data is available. For the threedimensional case an actual modern gas turbine geometry, namely the GE-E3 rst stage turbine blade was used. Two

A ( T n w ) dA T' ex = -----------------------------------------------------m RePr

Timko, L.P., 1982, Energy Efcient Engine High Pressure Turbine Component Test Performance Report, NASA CR-168289. Wilcox, D. C., 1994a, Turbulence Modeling for CFD, DCW industries, Inc. La Canda, CA. Wilcox, D. C., 1994b, Simulation of Transition with a Two-Equation Turbulence Model, AIAA Journal, Vol. 32, No.2, pp.247-255. Yang, Timothy, T. and Diller, Thomas, E., Heat Transfer and Flow for a Grooved Turbine Blade Tip in a Transonic Cascade, ASME paper No. 95-WA/HT-29.