Professional Documents
Culture Documents
ASME Turbo Expo 2007: Power for Land, Sea and Air
May 14-17, 2007, Montreal, Canada
GT2007-27669
Andr Burdet
Thomas Behr
Reza S. Abhari
between pressure and suction side. The tip leakage flow enters
the tip clearance on the pressure side of the blade, travels
through the gap and enters again into the passage on the suction
side. As it mixes with the main flow from the passage the
leakage flow rolls up into the tip leakage vortex. Tip leakage
flow may be responsible for as much as one third of the
aerodynamic losses in the rotor blade row. Furthermore the hot
gas of the tip leakage flow presents the highest thermal load to
the blade tip, which burns away if it is insufficiently cooled.
Thus it is a limiting factor for the blade lifetime.
The detrimental effects of an increase in tip clearance are
well known. Nevertheless turbines often operate with tip
clearances that are larger than aerodynamically desirable. Two
main reasons may be mentioned for having these higher tip
clearance. On the one hand the design tip clearance may change
due to different thermal expansions of the rotor blade and the
casing wall during operations. On the other hand limitations in
manufacturing tolerances and the aim of avoiding rubbing set
the minimum tip clearance size. Consequently, there is a strong
motivation to look for means to relieve the stringent
requirement on tip clearance design and manage the tip
clearance flow to minimize its detrimental impact on
performance.
Tip clearance flow has been investigated with a number of
contributions in the open literature. A seminal contribution for
this type of flow refers to the work of Rains [1], who studied tip
clearance flow for axial flow pumps motivated by concerns of
cavitation. Moore and Tilton [2] investigated the tip leakage
flow both experimentally and analytically.
Bindon [3]
measured and investigated the tip clearance loss formation. He
could divide the total endwall loss into loss generated inside the
tip gap, mixing loss of the tip leakage vortex, and secondary
and endwall losses. He concluded that not only the tip leakage
mass flow is important for loss generation, but also the flow
structure inside the gap would play a significant role.
The influence of blade tip geometries on tip leakage flow
has also been recognized and studied. Bindon and Morphis [4]
investigated loss for three blade tip geometries. They found that
the overall loss remained unchanged although the gap loss
strongly varied between a baseline sharp edged flat tip and two
differently contoured blade tip edges. In fact, the sharp edge
case showed high loss inside the gap due to a strong separation
bubble at the pressure side lip. On the contrary, the contoured
cases showed no separation bubble on the gap inlet but in turn
an increased tip gap mass flow was found which ultimately
increased the mixing loss. In a study of different squealer tips
by Heyes et al. [5], it was also concluded that the separation
bubble with the associated vena contracta is effectively sealing
the gap and ultimately lowering the tip aerodynamic loss. A
computational study by Ameri [6] on heat transfer on the blade
tip showed that a flat tip with sharp edges performs best in
terms of efficiency and total pressure loss compared to a mean
camberline squealer tip and a flat tip with radiused blade tips.
In a recent work Mischo et al. [7] have shown that profiling of
recess cavity walls is a potential passive flow control. The
computational study showed an increase of the rotor blade row
efficiency for an optimized profiled recess cavity compared to a
base line flat tip. A related experimental study confirmed a rise
in overall turbine efficiency.
Another mean of potential success for turbine tip clearance
flow control would be to inject cooling fluid from the casing.
Traditionally, active tip clearance flow control studies have
been conducted for axial compressors. An overview of the
benefits resulting from active flow control was presented by
Lord et al. [8]. Bae et al. [9,10] explored control of compressor
tip clearance flows in a linear cascade using three types of
fluidic actuators. Impacts on tip clearance flow and compressor
stability were presented. It was found that actuations based on
mixing enhancement and/or leakage flow reduction are not
! as
effective as those based on streamwise momentum injection.
The limited literature on active flow control for turbine tip
clearance flow offers an opportunity for the investigation of
new concepts. In this context, this paper presents the numerical
study of the effects on turbine rotor tip clearance flow due to
flow injection from casing in the 1.5 stage research turbine
facility LISA [11]. A companion paper by Behr et al. [12]
presents the experimental investigation of the flow injection
from the casing used in this study. The common test
configuration of the experimental and computational parts
! of
this study allows a unique back-to-back comparison of
experimental data and computational results. The goal of this
!
paper is to gain insight of the tip clearance flow structure when
casing injection occurs, and the consequences on loss through
!
the rotor and on the overall turbine efficiency. A top-down
approach is deliberately chosen for this numerical paper. First
the computational procedure is shown, especially the handling
PT ,2 " P3
PT ,0 " P3
Cpt
P*
TT
cax
d
p
l
0
0
r
total temperature!
density
axial chord (cax = 50 mm)
!
hole diameter
hole to hole pitch distance
length
streamwise injection angle
compound injection angle
radius
normalized radius = r " rhub
rcas " rhub
cylindrical angle coordinate
r*
m
Nb
Nh
Nmod
DR
BR
"ht,a
is
"ht,a
"
Subscript / Superscript
!
abs
abs. frame of ref.
rel
rel.!frame of ref.
hub
hub
tip
blade tip
bas
upd
per
h2h
cle
inj
f
cas
0
1
2
3
turbine inlet
stator 1 outlet
rotor outlet
turbine outlet
COMPUTATIONAL METHOD
1. Numerical Solver
The CFD code used in this study, MULTI3, is solving the
unsteady compressible RANS equations. The solution method
is based on an explicit, finite-volume, node-based, Ni-LaxWendroff time marching algorithm developed by Ni [13]. The
finite-volume formulation uses a central cell vertex variable
location. To prevent high frequency oscillations and capture
shock waves, a combined second- and fourth-order numerical
smoothing is added, which is consistent with the second order
accuracy in both space and time of the scheme.
Several turbulence models are implemented in their lowReynolds form, such as the algebraic Baldwin-Lomax model,
the one-equation Spalart-Allmaras model and the two-equation
k- turbulence model. For the purpose of this study, the
algebraic turbulence model is chosen. In fact, the freestream
turbulence level is unknown but assumed to be relatively small.
In addition, the robust Baldwin-Lomax model as implemented
in the current code has been widely used and validated for
several turbomachinery flow cases. Eventually, the jet injection
model presented herein can currently only be enabled when
using this turbulence model.
The flow solution is obtained in the relative frame of
reference for rotor flow prediction. In this investigation, since
the casing injection occurs in the rotor area, the casing wall
with its associated injection holes are actually moving
backward, at the circumferential speed of -u = -rcas. This
velocity represents the non-slip boundary condition to be
imposed at this wall. An adiabatic wall boundary condition is
imposed at every wall. The inlet and outlet planes are treated
using a standard 1D non-reflecting boundary conditions
approach for subsonic flow.
2. Feature-based jet injection model
The jet injection model of Burdet et al. [14-17] is used here
for simulating the discrete casing injection. This jet injection
model consists of a comprehensive, experimentally anchored
modeling of the near-hole flow field see Fig. 1.
0
$ 1'
lcle
" inj # & ) *
2
%
(
2
u
*
inj sin + 0
2
2
" per
2* ,
=
1 " upd =
N
*
N
2
mod
b * N mod
" per * - * rtip 2 * , * rtip
2
=
2 l./ =
N mod
N b * N mod
3
(1)
(2)
COMPUTATIONAL PROCEDURE
1. Experimental conditions
The continuous loop 1.5 stages subsonic axial turbine
facility LISA briefly consists of a first stator row of 36 blades,
then a rotor row of Nb = 54 blades and finally an outlet stator
row of 36 blades (blade count ratio is 2/3). The nominal
rotational speed is = 2700 RPM, the turbine inlet total
temperature is TTabs
= 328 K, the turbine inlet total pressure is
,0
abs
0 = 12.13 kg/s for the
PT ,0 = 1.4 bar and the mass flow is m
entire annulus. The hub radius is rhub = 330 mm and the casing
radius!is rcas = 400 mm. The tip gap height lcle is 1% of the total
blade span. A more detailed description
of the turbine facility is
!
given by Behr et al. [12].
Measurements of the relative total pressure PTrel and the
relative yaw angle rel at the rotor inlet and the rotor outlet cross
planes are available for boundary conditions setup and
prediction validation. The computational inlet and outlet planes
!
are located 0.25 axial chord upstream of rotor leading edge and
0.50 axial chord downstream of rotor trailing edge respectively.
Only one rotor pitch has been simulated due to computer
limitation at the time of this study. In fact the blade count ratio
(2/3) should have lead to compute more than 8 passages or
more to get a realistic simulation of the entire axial turbine
facility. In order to be as close as possible to the experimental
flow conditions in the rotor, the inlet and outlet boundary
conditions have been imposed using the circumferentiallyaveraged profiles of the flow as given by the measurement data.
2. Injection configurations
Two different axial locations of the injection holes have
been simulated combined with two different injection mass
flows, see Fig. 4.
= 0.32
x inj
c ax
= 0.52
A1
B1
!
A2
B2
The tip clearance above the flat rotor blade tip is also fully
discretized using an O-block wrapped around an H-block. A
special mesh clustering has been applied in the region where
the injection holes lie. In order to fulfill the requirement of the
model use, about 4-5 and 6-8 grid nodes per hole diameter has
been ensured everywhere near the injection holes. In total, there
are 568120 grid nodes that cover the computational domain. In
particular, there are 15 cell layers from the blade tip to the
casing.
4. Computational history
A generic steady flow solution without casing injection has
been first calculated for baseline. In order to fulfill the quasisteady assumption criteria for the jet injection model, given by
Eqs. 1-2, the different time scales and the number of update of
the model per period have been set up as shown in Table 2.
per [s]
num [s]
Nmod [-]
upd [s]
inj [s]
l [mm]
ltip [mm]
As
4.1210-4
5.0110-9
100
4.1210-5
1.1710-6
4.8
16.5
Bs
4.1210-4
5.0110-9
100
4.1210-5
1.1710-6
4.8
18.9
1. Yaw angle
Measured and predicted relative flow yaw angles rel for the
baseline case are shown in Fig. 6. Six main flow features,
labeled A to F, are identified in both the experimental result
and the CFD prediction. Flow regions A and B should refer to
the wake and main flow. Flow regions C, D, E and F can be
identified to the tip clearance vortex system. While the CFD
results capture the same flow features present in the
experiment, two major discrepancies need to be noted. First, an
Figure 6: Measured (left) and predicted (right) time-averaged relative yaw angle rel on an axial plane 15%
axial chord downstream of rotor trailing edge.
over-prediction between 0.5% to 35% in over and under turning
can be seen in all flow features. Second, the spatial location and
extension of features related to vortical structures differ by 2%
span to 10% span from experiment to computation. In general,
it has to be underlined that the computed features related to
vortical structures are located on lower radial positions than the
experimental ones. Thus, the diffusion of the flow structures
appears to be under-predicted which leads to an over-prediction
of the downward radial migration of those ones. The
consequence on losses of this discrepancy is analyzed next.
3. Change of efficiency
The change of efficiency is given by
(3)
"# = # $ #bas
EXP
CFD
A1
0.55%
-0.07%
A2
0.41%
-0.42%
B1
-0.03%
-0.33%
B2
-0.12%
-0.91%
Table 3: Measured (EXP) and predicted (CFD) timeaveraged relative change of efficiency in the
rotor.
The predicted values of show a discrepancy with the
experimental data. Indeed, a decrease of predicted rotor
efficiencies in the range from -0.07% to -0.91% is observed
whereas the experimental data mainly concludes on an increase
m tip =
" m (x)dx
x LE
(4)
" h 2h =
2#
$ % N blade % N hole
(5)
10
11
ACKNOWLEDGEMENT
The authors would like to thank Dr. Olivier Byrde for his
precious help regarding the handling of the computation runs.
REFERENCES
[1] Rains, D.A., 1954, Tip Clearance Flows in Axial Flow
Compressors and Pumps, California Institute of
Technology, Hydrodynamics and Mechanical Engineering
Laboratories, Report No. 5, June 1954.
[2] Moore, J. & Tilton, J.S., 1988, Tip Leakage Flow in a
Linear Turbine Cascade, ASME Journal of
Turbomachinery, Vol 110, pp. 18-26.
[3] Bindon, J.P., 1989, The Measurement and Formation of
Tip Leakage Loss, ASME Journal of Turbomachinery,
Vol 111, pp. 257-263.
[4] Morphis, G. & Bindon, J.P., 1992, The Development of
Axial Turbine Leakage Loss for Two Profiled Tip
Geometries Using Linear Cascade Data, ASME Journal
of Turbomachinery, Vol 114, pp. 198-203.
[5] Heyes, F.J.G., Hodson, H.P., & Dailey, G.M., 1992, The
Effect of Blade Tip Geometry on the Tip Leakage Flow in
Axial Turbine Cascades, ASME Journal of
Turbomachinery, Vol 114, pp.643-651.
[6] Ameri Ali, A., 2001, Heat Transfer and Flow on the
Blade Tip of a Gas Turbine Equipped with a MeanCamberline Strip, ASME Paper 2001-GT-0156.
[7] Mischo, B. & Abhari, R.S., 2006, Flow physics and
profiling of recessed blade tips: Impact on performance
and heat load, ASME Paper 2006-GT-91074.
[8] Lord, W.K., MacMartin, D.G. & Tillman, T.G., 2000,
Flow control opportunities in gas turbine engines,
AIAA 2000-2234.
[9] Bae, J., Breuer, K.S. & Tan, C.S., 2003, Active control
of tip clearance flow in axial compressors, ASME paper
GT2003-38661.
[10] Bae, J., Breuer, K.S. & Tan, C.S., 2003, Control of tip
clearance flow in axial compressors, AIAA paper AIAA
2000-2233.
[11] Behr, T., Kalfas, A. I., Abhari, R. S., "Unsteady Flow
Physics and Performance of a One-and-1/2-Stage
Unshrouded High Work Turbine, ASME Paper No.
GT2006-90959.
[12] Behr, T., Kalfas, A.I. & Abhari, R.S., 2007, Control of
rotor tip leakage through cooling injection from casing in
a high-work turbine: Experimental investigation, ASME
paper GT2007-27269.
[13] Ni, R.H., 1981, A multiple grid scheme for solving the
Euler Equations, AIAA Journal, Vol.20, No. 11, pp.
1565-1571.
[14] Burdet A., Abhari R.S., Rose M.G., (2007). Modeling of
Film Cooling - Part II: Model for Use in 3D CFD,
GT2005-68780, to appear in ASME Journal of
Turbomachinery, Vol. 129, No. 2
[15] Burdet A., Abhari R.S., (2007). 3D Flow Prediction and
Improvement of Holes Arrangement of a Film-Cooled
12
13