Professional Documents
Culture Documents
Nfgyju 70
Nfgyju 70
Proceedings of :
Power for Land, Sea, and Air
ASME TURBO EXPO 2003
June 16–19, 2003, Atlanta, Georgia, USA
June 16-19, 2003, Atlanta, Georgia, USA
GT2003-38770
GT-2003-38770
H Watanabe M Zangeneh
Advanced Design Technology Ltd Department of Mechanical Engineering
Monticello House, 45 Russell Square University College London
London WC1B 4JP, U.K. London WC1E 7JE, U.K.
ABSTRACT showed that sweep reduced the corner stall and the secondary
The application of sweep in the design of transonic fans has flow at the endwall region.
been shown to be an effective method of controlling the From the analogy to the swept wing theory, swept blading in
strength and position of the shock wave at the tip of transonic the transonic fan or compressor were aiming to reduce the
fan rotors, and the control of corner separations in stators. In shock losses. Wadia et al, (1998) investigated the effects of
rotors sweep can extend the range significantly. However, using forward and backward sweep on the shock formation and
sweep in conventional design practice can also result in a performance of fan rotor experimentally and analytically. They
change in specific work and therefore pressure ratio. As a result, reported the forward swept rotor showed better results than
laborious iterations are required in order to recover the correct backward swept one in efficiency and stall margin by the
specific work and pressure ratio. reduced shock boundary layer interaction. Blaha et al, (2000)
In this paper, the blade geometry of a transonic fan is designed also showed there was no improvement in performance and in
with sweep using a 3D inverse design method in which the stall margin with aft-swept compressor stage.
blade geometry is computed for a specified distribution of blade Denton and Xu, (2002) investigated the genuine effects of
loading. By comparing the resulting flow field in the sweep or lean on the flow and the performance of the fan by
conventionally and inversely designed swept rotors, it is shown using the identical profiles at each spanwise position in each
that it is possible to apply sweep without the need to iterate to swept, lean and datum fan. They concluded there was very
maintain pressure ratio and specific work when using the small benefit in fan efficiency by using sweep but significant
inverse method. influence on the stall point of the fan. As in the case of Wadia et
al, (1997), they showed the forward sweep increased the stall
margin of the fan, while it decreased the choke margin at the
INTRODUCTION same time. The improvement in the stall margin by applying
The recent progress in design procedure on the turbomachinery forward sweep was explained by the fact that stall tends to
blading aided by CFD and some geometry optimization occur soon after the pressure side shock reaches the leading
techniques provide an opportunity to the designers to edge at the tip and the forward sweep made this occur at a lower
investigate three-dimensional effects in detail and to improve mass flow rate.
the performance. Considering the benefit in the stall margin, forward sweep
The blade sweep is one of the most usual three-dimensional application seems to be more preferable in the transonic fan
blading design techniques applied in the axial fan and design. At the same time, the choke margin is decreased by the
compressor design recently. Simon et al, (2002) showed that forward sweep, and moreover, the specific work of the blade
sweep was effective at reducing the cross passage flow near the and therefore pressure ratio will be changed. To recover these
hub and interaction between the hub endwall and profile defects of forward sweep, a designer must be forced to modify
boundary layers in axial compressor rotors. Sasaki and another design function such as chord length, blade angles etc
Breugelmans, (1997) investigated the sweep and dihedral in conventional design practice. As a result, laborious iterations
effects on compressor cascade performance experimentally and are required in order to optimize the sweep fan geometry in
order to achieve the correct specific work and pressure ratio.
R [m]
∂ f − θ t
∂( f − θ )
Lower surface: V − 2 Vθ− − 2 (5)
z = − Vr 0.16
Trailing
∂z r ∂r
edge
o
Applying the equations to an initial camber f and adding (4)
0.12
and (5) gives,
o ∂f
o
1 ∂t rV o o ∂f
o
1 ∂t
V zbl + ∆V zo θ = θbl − Vrbl − ∆Vro θ (6)
0.08
∂z 4 ∂z 2 ∂r 4 ∂r
r -0.04 0.00 0.04 0.08 0.12
+ −
where V = V + V and ∆V = V + − V − . Z [m]
bl
2
To modify the blade camber line shape so it achieves the target 0.100
~ ~* ~o
rVθ , the correction term [ rVθ − rV ] is introduced into the
θ
Original
equation to give a first approximation of the new camber 0.075
Tip profile
SW1
1
geometry, f ,
~ ~
∂f 1 1 ∂t rV o + [ rVθ* − rVθo ] o ∂f
1 ∂t
− ∆Vro θ (7)
1 0.050
o
V zbl + ∆V zo θ = θbl − Vrbl
∂z 4 ∂z 2 ∂r 4 ∂r
r
rθ [m]
rV θ 2 [m 2/s]
Pressure ratio
36
Peak efficiency point
1.6 33
Original 30
1.5 SW1 27
Original (fine mesh) 0.0 0.2 0.4 0.6 0.8 1.0
1.4
Span
31 32 33 34 35
42
Mass flow rate [kg/s] Original (33.58 kg/s)
39
0.94
SW1 (33.08 kg/s)
rV θ 2 [m 2/s]
36
0.92 Peak efficiency point
33
Efficiency
0.90
30
0.88 27
Original
0.0 0.2 0.4 0.6 0.8 1.0
0.86 SW1
Span
Original (fine mesh)
0.84 Fig.3 Spanwise distribution of the blade
31 32 33 34 35 work (rVθ) and pressure ratio at the trailing
Mass flow rate [kg/s] edge plane, comparison at the peak
0.32 efficiency point for the each fan
42
Original (33.58 kg/s)
39
0.28
rV θ 2/RtUt
36
0.24 Original 33
SW1 30
Original (fine mesh)
0.20 27
31 32 33 34 35 0.0 0.2 0.4 0.6 0.8 1.0
Span
Mass flow rate [kg/s]
1.80
1.70
in the axial direction in which the maximum shift dz is at the tip. 1.65
In the cylindrical (z-rθ) plane, profiles at each spanwise 1.60
Original (33.58 kg/s)
position move along the local camber, so that it is composed of
1.55 SW1 (33.62 kg/s)
the axial shift dz and the tangential shift d(rθ). This sweep
1.50
definition is basically the same as was used by Denton et al.
0.0 0.2 0.4 0.6 0.8 1.0
(2002) and produce no additional blade stress in the radial
Span
direction if the blade profiles are not modified.
Forward sweep is chosen in this study for its advantage in the Fig.4 Spanwise distribution of the blade
characteristic in the lower flow rate region and stall margin. The work (rVθ) and pressure ratio at the trailing
sweep is applied at the half of the blade above the midspan and edge plane, comparison at the same flow
the maximum axial shift at the tip is 22% of the axial chord at rate condition
the tip and 0% at the midspan. Firstly, there is no change in the Fig.2 shows the performance of the original NASA67 fan
profile at each spanwise position to investigate only the sweep geometry (Original) and sweep geometry (SW1) which were
effects. analyzed using TURBOdesign-2 Euler code with viscous
modelling. The mass averaged value of total pressure and total
Ps /Pt0
0.8
0.6
Tip 0.4 : Original (33.58 kq/s)
0.2 : SW1 (33.08 kg/s) Tip
Peak efficiency point
0
0 0.2 0.4 0.6 0.8 1
Meridional distance
1.6
1.4
1.2
1
Ps /Pt0
Midspan 0.8
0.6 Midspan
0.4 : Original (33.58 kq/s)
: SW1 (33.08 kg/s)
0.2
Peak efficiency point
0
0 0.2 0.4 0.6 0.8 1
Meridional distance
1.6
Hub
1.4
1.2
1
Ps/Pt0
0.8
(a) Original, m=33.58 kg/s (b) SW1, m=33.08 kg/s 0.6
: Original (33.58 kq/s) Hub
0.4
Fig.5 Blade-to-blade Mach number contours 0.2
: SW1 (33.08 kg/s)
Peak efficiency point
0
temperature at the inlet and exit boundary plane were used to 0 0.2 0.4 0.6 0.8 1
calculate the pressure ratio and the efficiency. The inlet Meridional Distance
boundary locates at about 70% of the hub axial chord upstream
from the leading edge plane. The exit boundary locates at one
axial chord of the hub downstream from the trailing edge plane. Fig.6 Surface pressure distribution
The performance curve of the SW1 indicates a shift towards the total pressure between 0% to 60% span were slightly increased
lower flow rate from the Original fan blade. That is, the choke in SW1 but total pressure decreased near the tip. These figures
flow is reduced and the efficiency at the lower flow is increased
from the Original fan. The peak efficiency of the SW1 is show rVθ and total pressure were decreased especially at the
slightly worse than Original fan. The blade work for the SW1 tip side by the sweep.
fan is 5.3% lower than that of Original at the peak efficiency Fig.5 shows the blade-to-blade Mach number contours near the
& = 33.58 kg/s).
point for the Original ( m hub and at the midspan and near the tip, at the peak efficiency
Also shown in Fig.2 is the predicted characteristic of the point of the Original and SW1 fan. In both cases strong passage
Original fan using finer mesh (454005 grid points). The analysis shock takes place near the tip section. The shock location in
results performed with a 118581 (29×141×29) points shows SW1 moved toward the trailing edge as compared to the
very good agreement with the results by fine grid. From these Original case.
results, 118581 points grid was used in the inverse design In the midspan, the leading edge shock becomes stronger in
process and in performance analysis. SW1 than Original. This change in shock structure at the
midspan is observed in the Original fan if the flow rate is
Fig.3 shows the pitchwise averaged rVθ and total pressure slightly reduced from the peak efficiency point. In other words,
distribution at the trailing edge plane of Original and SW1 fan although the flow rate of the peak efficiency point is decreased
at the peak efficiency points for each fan. Fig.4 is the from the 33.58 kg/s in the Original case to 33.08 kg/s for SW1
comparison of same variables of each fan at the peak efficiency case, the characteristics of the shock structure at the tip section
point of Original fan. At the peak efficiency point, rVθ and
1600
50% span
1400 SW1 fan with strong shock
1200
1% span
d(rV )/dm [m/s]
1000
Specified smooth
800 loading distribution
to suppress shock
Fig.11 Comparison of the blade profiles
600
(SW1IV2 fan)
400 tip. This figure shows quite a large modification in the blade
profile of SW1IV2 as compared with SW1 design from 74% to
200 Abrupt increase
(effect of strong shock) 100% span. An interesting feature of the SW1IV2 design at
0 around 74% span, is the presence of a so-called “de-camber”
0.0 0.2 0.4 0.6 0.8 1.0 region which helps to reduce the shock strength near the suction
Span surface.
Fig.9 Loading distribution at the 93.3% span
50% span
Suction side Pressure side Fig.13 Blade-to-blade Mach number contours of SW1IV2 fan
(b) SW1IV2, m=33.37 kg/s formation of above mentioned stronger shock-boundary layer
Fig.12 Mach contour near the suction and the pressure surfaces interaction than SW1, more loss is generated in SW1IV2.
In fig.12, the Mach number contour near the suction and the Comparison of the fan performance
pressure surfaces of the SW1IV2 and the SW1 fan geometry are Fig.16 shows the performance comparison between SW1IV1,
compared. From the figures on the suction side, one can find the SW1IV2, SW1 and Original fan geometry. Although the work
shock wave developed in the SW1 become very mild in of the SW1IV1 at the peak efficiency point of the Original
SW1IV2 from the midspan to the tip. This can be observed near become the same as Original, efficiency drops from the SW1
the pressure surface above 75% span. and there is no improvement in the choke flow rate.
Fig.13 and fig.14 show the Mach number contour and the Denton et al. (2002) described in his paper on the detailed
surface pressure distributions at the 50%, 74% and 99% span of investigation on the characteristic of the sweep and lean fan, the
the SW1IV2 design. The Mach number at the suction side change in choke mass flow of the sweep or lean blades was
changes very smoothly from the throat towards the trailing edge merely caused by the change of the total throat area. In the case
at the 50% and 74% span. The de-camber profile at the 74% of forward sweep applied at the tip, annulus area is decreased
span effectively works to disperse the compression wave of the around the tip section and this reduces the throat area around
leading edge shock on the suction surface. the tip. Thus, the choke mass flow reduces. To compensate this
At the 99% span, shock wave near the pressure surface become decrease in throat area around the tip, throat area at the other
mild, however the maximum Mach number near the suction section could be increased. For example, shifting the blade
surface just upstream of the shock increased above 1.55. This profiles in some area above the hub side to increase the annulus
increases the loss concerned to the shock-boundary layer area around that region could recover the total mass flow when
interaction on the suction surface, so the wake region develops the forward sweep is adopted around the tip region.
further than SW1. The reason for no improvement in choke mass flow rate of
In fig.15 the spanwise enthalpy loss distribution at the trailing SW1IV1 can be explained that the throat area does not change
edge plane is presented. The figure shows that the predicted and efficiency is not improved although the pressure ratio is
loss at the trailing edge of SW1IV2 has improved as compared increased.
to the SW1 case and Original case from 20% to 80% of span. The peak efficiency of the SW1IV2 is improved by 0.5 point
The reason for no improvement in loss near the hub is because from the original and the characteristic in the choke flow side
similar loading as SW1 is used for the SW1IV2 design. Near also recovered much from SW1IV1 for its improvement in
the tip region, as a result of using higher loading and the efficiency, though it is not complete compared with the Original
Loss ( ∆ H/U2)
1 0.08 SW1IV2
Ps/Pt0
0.8 0.06
0.6 0.04
0.4 : SW1 (33.08 kq/s)
: SW1IV2 (33.37 kg/s) 0.02
0.2 Peak efficiency point
0.00
0
0.0 0.2 0.4 0.6 0.8 1.0
0.0 0.2 0.4 0.6 0.8 1.0
Span
Meridional distance
0.8 1.7
Pressure ratio
0.6
0.4 : SW1 (33.08 kq/s) 1.6
Original
: SW1IV2 (33.37 kg/s)
0.2 SW1
Peak efficiency point 1.5 SW1IV1
0 SW1IV2
0.0 0.2 0.4 0.6 0.8 1.0
1.4
Meridional distance 31 32 33 34 35
Mass flow rate [kg/s]
1.6 0.94
1.4 50% span
0.92
1.2
Efficiency
1 0.90
Ps/Pt0
0.0 0.2 0.4 0.6 0.8 1.0 Mass flow rate [kg/s]
REFERENCES
Adamczyk,J.J., Celestina, M.L., Greitzer, E.M., “The Role of
Tip Clearance in High-Speed Fan Stall”, Journal of
Turbomachinery, Vol. 115, pp 28-39, 1993
AGARD-AR-275, “Test Cases for Computation of Internal
Flows in Aero Engine Components”, AGARD Advisory Report
No.275, 1985.
Blaha,C., Kablitz, S., Hennecke, D.K., Schmidt-Eisenlohr, U.,
Pirker, K., Haselhoff, S., “Numerical Investigation of The Flow
in an Aft-Swept Transonic Compressor Rotor”, ASME Paper
No. 2000-GT-0490, 2000.
Denton, J.D., “The Calculation of Three Dimensional Viscous
Flow Through Multistage Turbomachines”, ASME Paper No.
90-GT-19, 1990.
Denton, J.D., Xu, L., “The Effect of Lean and Sweep on
Transonic Fan Performance”, ASME Paper No. GT-2002-
30327, 2002.