Applied Thermal Engineering 126 (2017) 53–69

Contents lists available at ScienceDirect

Applied Thermal Engineering

journal homepage: www.elsevier.com/locate/apthermeng

Research Paper

Investigation on axial effect of slot casing treatment in a transonic

compressor
Xiaoyong Zhou, Qingjun Zhao ⇑, Weiwei Cui, Jianzhong Xu
Institute of Engineering Thermophysics, Chinese Academy of Sciences, Beijing 100190, China

h i g h l i g h t s

The relative importance of the mechanisms of slot casing treatment is obtained.

The bleeding and injecting effect is the most important mechanism for the slots.

Effective slots should cover the tip leakage vortex and boundary separation zone.

a r t i c l e i n f o a b s t r a c t

Article history: This paper reports the effect of the axial position of the slot casing treatment on the performance of tran-
Received 18 May 2016 sonic compressor NASA Rotor 67 by unsteady numerical simulation. The interaction of the recirculation
Revised 21 July 2017 in the slots and flow near the blade tip is analyzed to understand the flow mechanisms. The relative
Accepted 21 July 2017
importance of the mechanisms of stall margin improvement due to the slot casing treatment is evaluated
Available online 22 July 2017
with the relative weight method. The results show that the bleeding and injecting effect caused by the
recirculation is the most important factor that affects the blockage in the blade tip region, which deter-
Keywords:
mines the stall margin improvement. When the slots cover the initial position of the tip leakage vortex
Transonic compressor
Casing treatment
(TLV) and the boundary layer separation zone downstream the shock, the recirculation is stronger due to
Axial skewed slot the larger pressure difference between the front and rear part of the slots. Consequently, the slot casing
Tip leakage vortex treatment can reduce the blockage near the casing more effectively, which results in a larger stall margin
improvement. Shifting the slots upstream or downstream will reduce the driving force for the recircula-
tion and the effect extent on the low energy fluid of the slots, which decreases the stall margin
improvement.
Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction
The aero engines develop towards the direction of high thrust-
weight ratio, low fuel consumption and wide stable operating
range. This requires the compressor to have high stage loading, high
efficiency and high stall margin. For highly loaded transonic com-
pressors, the complex flow structures near the blade tip including
the tip leakage flow, shock and boundary layer separation can
induce flow losses and instability [1–3], which have significant
effect on the performance of compressors. It has been indicated that
the leakage vortex is a key factor for the flow instability if the blade
loading does not exceed the stability limit near stall. The interaction
between the tip leakage vortex and the shock can generate a low
⇑ Corresponding author at: Key Laboratory of Light-Duty Gas-Turbine, Chinese
Academy of Sciences, China.
E-mail addresses: xyzhou510@sina.com (X. Zhou), zhaoqingjun@iet.cn (Q. Zhao),
cuiweiwei@iet.cn (W. Cui), xjz@iet.cn (J. Xu).
momentum region downstream the shock. The tip leakage vortex
breakdown can occur if the intensity of the interaction exceeds
the vortex breakdown limit [4–6]. As the mass flow rate decreases,
the tip leakage vortex moves upstream and causes large blockage
region near the blade tip, which leads to the stall of the compressor
[7]. If the blade loading exceeds the stability limit near stall, bound-
ary layer separation occurs at the suction side. Then the boundary
layer separation causes blockage in the blade passage and con-
tributes to the stall of the compressor [8].
In order to extend the stable operating range of compressors,
the researchers have proposed a variety of casing treatments [9–
11]. One of the various casing treatments is the configuration
which consists of a discrete number of slots. The slot casing treat-
ment is capable to improve the stall margin of compressors of sig-
nificantly [12,13]. Many investigations have been conducted to
reveal the effect mechanisms of slot casing treatments on the com-
pressor performance and achieve proper slot geometries.

http://dx.doi.org/10.1016/j.applthermaleng.2017.07.165
1359-4311/Ó 2017 Elsevier Ltd. All rights reserved.
54 X. Zhou et al. / Applied Thermal Engineering 126 (2017) 53–69
Nomenclature
A area (m2)
B blockage for outer 80% span
CT casing treatment
D position of the shock
L normalized circumferential length
Lu difference between the inlet and outlet tangential com-
ponent of relative velocity
M meridional direction
Mr normalized recirculating mass flow of the slots
m_c mass flow of the compressor (kg/s)
m_r recirculating mass flow of the slots
m_s bleeding or injecting mass flow of the slot
NS near stall
P averaged pressure difference at 98% span (Pa)
PR total pressure ratio
PS pressure surface
R radial direction
SC smooth casing
SM stall margin
Wilke et al. [14] numerically investigated the impact of the slot
casing treatments on the performance of a transonic compressor.
The vortex flow in the slots bled fluid out of the blade passage
and injected it back into the main flow upstream. This effect could
suppress the tip leakage vortex and was responsible for the stall
margin improvement. Two configurations with different axial posi-
tions were studied. Configuration 1 is positioned from 7.5% to 92.5%
chord. Configuration 2 is shifted upstream and covered only the
upstream 25% chord. The two configurations improved the stall
margin identically. Whereas, configuration 2 had less effect on
the flow field and shock than configuration 1, so configuration 2
obtained higher efficiency than configuration 1. Another investiga-
tion [15] also indicated that the effectiveness and efficiency of the
slot casing treatments relied much on the position of the slots.
The downstream positioned slots obtained larger stall margin
improvement but induced higher efficiency penalty than the down-
stream positioned slots. Lu et al. [16] numerically investigated the
slot casing treatments with two different axial positions on a tran-
sonic compressor stage. It indicated that the proper position for
stall margin improvement should have its forward end cover the
tip leakage vortex initial position and its aft end cover the high pres-
sure and stagnation region downstream of the passage shock.
Because the high energy flow after the shock can be used to re-
energize the tip leakage vortex. Zhu et al. [17] and Lu et al. [18]
investigated the effect of slot casing treatments with several differ-
ent axial positions on the performance of axial compressors. The
main mechanism for delaying stall was attribute to the tip leakage
vortex towards the trailing edge. The best configuration in terms of
stall margin improvement and efficiency was obtained. But the cou-
pled flow features through the blade passage and the different slot
casing treatments were not compared. Brignole et al. [19] and Hem-
bera et al. [20] conducted parametric investigations on the slot cas-
ing treatments for transonic compressors. Some parameters such as
entering mass flow coefficient, geometrical efficiency and rothalpy
rise coefficient were proposed to evaluate the slot casing treat-
ments. These parameters helped to quantitatively compare the flow
in the slots and the blade tip flow for different casing treatment
configurations. Streit et al. [21,22] and Brandstetter et al. [23] inves-
tigated the effect of an axial-slot casing treatment in the rotor’s
leading edge region. Both numerical simulation and experiment
showed that the slots could remove the blockage region caused
by the interaction of the shock and tip leakage vortex, resulting in
SS suction surface
T tangential direction
TLV tip leakage vortex
V absolute velocity
W relative velocity
Wm meridional component of relative velocity
Wu tangential component of relative velocity
Wz axial velocity
X vertical radial direction
Y horizontal radial direction
Z axial direction
a angle between the tip leakage vortex trajectory and the
axial direction (°)
b relative flow angle (°)
ei relative weight
fn absolute vorticity coefficient
q density
x rotor angular velocity
the downstream repositioning of the shock, which improve the sta-
bility. Ma et al. [24] studied the effect of axial-slot casing treatment
on the peak efficiency of the compressors numerically. The slots
covered the whole tip chord. An equation derived from the Denton’s
leakage mixing model is used to evaluate the tip leakage loss. Based
on this analysis, the slot casing treatments that can improve the
peak efficiency of the compressors were designed.
In general, the blade tip flow structures change significantly
along the blade passage downstream in transonic compressors.
The axial position of the slots has great effect on the interaction
between the slots flow and the near casing flow. But the guidelines
that help to design the axial position of slot casing treatments for
stall margin or efficiency are still not clear. The optimum relative
position between the slots and the blade tip flow structures includ-
ing the shock, tip leakage vortex and boundary layer separation
remains to be explored. And further investigation on the variation
of the coupling flow through the blade passage and slots with dif-
ferent axial positions is necessary.
For these reasons, a study on the effect of the axial skewed slot
casing treatments with four typical axial positions on the perfor-
mance of a transonic compressor NASA Rotor 67 is performed in
this paper. The investigation is implemented with unsteady
numerical simulations.
2. Investigated compressor and numerical method
2.1. Investigated compressor
The transonic axial compressor NASA Rotor 67 was adopted in
this investigation. The detail geometry and performance parame-
ters of NASA Rotor 67 can be found in Ref. [25]. Some main design
parameters are listed in Table 1. The rotating stall of NASA Rotor 67
occurs in the blade tip region, which is related to the tip leakage
vortex. Thus the rotor is suitable for the investigation of casing
treatment.
2.2. Numerical method
The commercial CFD solver NUMECA/EURANUS was used for
the numerical simulation. The simulation was based on the turbu-
lent Favre Reynolds averaged Navier-Stokes equations in the rotat-
 X. Zhou et al. / Applied Thermal Engineering 126 (2017) 53–69
Table 1
Main design parameters of NASA Rotor 67.
Rotational speed (rpm) 16042.8
Number of rotor blades 22
Design mass flow rate (kg/s) 33.25
Adiabatic efficiency 0.896
Total pressure ratio 1.629
Tip relative inlet Mach number 1.379
Tip clearance/Tip chord 1.1%
Tip speed (m/s) 429
ing relative frame. The cell centered control volume approach was
used for the space discretization of the equation. The turbulence
was taken into account with the Spalart-Allmaras model. The
steady solution was achieved with the four-stage Runge-Kutta
time integration scheme. The unsteady calculation was conducted
with the implicit dual time stepping technique. For per blade pas-
sage, 40 physical time steps were performed with 20 pseudo time
iterations. The calculation for the smooth casing configuration and
the casing treatment configurations were all run with the unsteady
setup.
Fig. 1 shows the grid topology of the rotor blade passage and the
casing treatment. The rotor passage was meshed in HOH topology
and the semi-circular shape slots were meshed in butterfly topol-
ogy. The rotor passage mesh included 1,000,000 grid points and
the one slot mesh included 85,000 grid points. For the rotor pas-
sage mesh, the minimum skewed angle is 23.2°, the maximum
aspect ratio is 1704.5 and the maximum expansion ratio is 6.3.
For the slot mesh, the minimum skewed angle is 43.7°, the maxi-
mum aspect ratio is 319.0 and the maximum expansion ratio is
1.79. The grid density was increased towards the solid boundaries
to keep y + <2 at the walls for accurate capturing the viscous flux-
ing. The non-rotating slot blocks were mounted on a very thin H
block (10% of the tip clearance) and connected to the rotating blade
passage block with a rotor-stator interface at the casing. The frozen
rotor method was applied on the rotor-stator interface for the
steady simulations, because the frozen rotor method can simulate
the pitch-wise change of the flow, needs small amount of calcula-
tion and has good robustness, The domain scaling method was
used for the unsteady simulations, because this method allows sin-
gle rotor passage computation with acceptable time.
During the numerical simulation, adiabatic and no-slip bound-
ary conditions were applied at the solid boundaries. Total pressure,
total temperature and axial flow condition were set at the inlet.
Static back pressure was set at the outlet, and radial equilibrium
equation was used to get the outlet static pressure distribution.
Fig. 1. Grid topology of the blade passage and the slot casing treatment.
55
The stall limit was detected by gradually increasing the static back
pressure. The unsteady computation was initialized with the
steady solution or the unsteady solution of lower back pressure.
Numerical stall appears as the continuous decrease of mass flow,
adiabatic efficiency and total pressure ratio. Near the peak effi-
ciency point, the back pressure was raised in 1000 Pa increments.
Near the stall point, the back pressure was raised in 500 Pa
increments.
2.3. Validation of numerical method
The numerical method is validated with the experimental data
which comes from Ref. [26]. The numerical and experimental over-
all performance of NASA Rotor 67 is compared in Fig. 2. The mass
flow is normalized by the choked mass flow. The numerical simu-
lation predicts the overall trend of the experimental results. The
calculated normalized near stall mass flow is 0.922 and the exper-
imental data is 0.921. The numerical simulation predicts the near
stall mass flow correctly. Besides, the calculated results follow
the same trend with the experimental results, although there is
some difference. The calculated total pressure ratio has an error
within 4% of the experimental data, while the calculated adiabatic
efficiency has an error within 3% of the experimental data. This dif-
ference can also been found in other investigations on NASA Rotor
67 [27,28]. The span-wise profiles of total pressure ratio and total
temperature ratio at the peak efficiency condition are shown in
Fig. 3. The computed profile shape can match the experimental
results well. Besides, the relative Mach contours at 90% span at
the peak efficiency condition are shown in Fig. 4. The numerical
simulation can predict the flow field including the shock position
well. Overall, the numerical simulation can predict the main flow
features of the transonic rotor.
3. Axial skewed slot casing treatment configurations
The structure of the axial skewed slot casing treatment is shown
in Fig. 5. Each blade passage corresponds to four semi-circular
shape axial skewed slots. The slots are inclined by 45° against a
meridian plane in the direction of blade rotation. The slots occupy
half of the area in the circumferential direction. The meridional
length of the slots is 80% blade tip axial chord length. The simu-
lated four slot casing treatment configurations are shown in
Fig. 6. The distance from the downstream end of the slots to the
leading edge plane is 20%, 50%, 80% and 110% axial chord length,
which is denoted with CT_20, CT_50, CT_80 and CT_110,
respectively.
56 X. Zhou et al. / Applied Thermal Engineering 126 (2017) 53–69

Fig. 2. Numerical and experimental overall performance of NASA Rotor 67 at design speed.

Fig. 3. Numerical and experimental span-wise profiles of total pressure ratio and total temperature ratio at the peak efficiency condition.

(a) Simulation (b) Experiment

Fig. 4. Numerical and experimental relative Mach contours at 90% span at the peak efficiency condition.
 X. Zhou et al. / Applied Thermal Engineering 126 (2017) 53–69 57

Fig. 5. The structure of the axial skewed slot casing treatment.

(a) CT_20 (b) CT_50

(c) CT_80 (d) CT_110

Fig. 6. The simulated slot casing treatment configurations.

4. Numerical results all increase the stall margin of the compressor, but decrease the
peak efficiency. The CT_20, CT_50, CT_80 and CT_110 configuration
4.1. Compressor overall performance improves the stall margin by 7.07%, 19.14%, 24.70%, 18.92% com-
pared to the smooth casing, respectively. The stall margin is calcu-
Fig. 7 shows the effect of the four axial skewed slot casing treat- lated by the following equation:
ments on the overall performance of the transonic compressor at

PRstall m_ peak
design speed. The total pressure ratio is normalized by the total SM ¼   1  100%
PRpeak m _ stall
pressure ratio of the smooth casing at the peak efficiency point,
and the adiabatic efficiency is normalized by the peak efficiency where PRpeak and PRstall reprensent total pressure ratio at the peak
of the smooth casing. The four axial skewed slot casing treatments _ peak and m
efficiency and near stall condition. m _ stall reprensent mass
58 X. Zhou et al. / Applied Thermal Engineering 126 (2017) 53–69
Fig. 7. Overall performance of the transonic compressor at design speed.
flow at the peak efficiency and near stall condition. Meanwhile, the
four slot configurations all reduce the peak efficiency of the com-
pressor, and the absolute discrepancy of the peak efficiency
between the smooth casing and the CT_20, CT_50, CT_80, CT_110
configuration is 0.71%, 0.48, 0.70%, 0.96%, respectively. The CT_80
configuration generates the largest stall margin improvement,
while the CT_50 configuration induces the minimum reduction in
peak efficiency. In general, the axial slot casing treatments can
improve the stall margin up to absolute discrepancy of 30%, and
reduce the efficiency up to absolute discrepancy 4% [13]. The
change of the stall margin and peak efficiency caused by the slot
casing treatments in this paper is within the normal range and this
is expected.
4.2. Analysis of the internal flow field
4.2.1. Analysis of the stall mechanisms of the compressor
Fig. 8 shows the relative Mach number distributions on several
sections perpendicular to the blade tip chord and the streamlines
of the tip leakage flow. It can be seen that the tip leakage flow rolls
up and forms the tip leakage vortex after interacting with the main
flow. The vortex core is composed of the leakage flow near the
blade tip leading edge, and the outer part of the vortex is composed
of the leakage flow downstream the tip chord. At the peak effi-
ciency point, the interaction between the tip leakage vortex and
the passage shock causes a low speed region near the blade tip.
At the near stall point, the tip leakage vortex is strengthened and
spreads over a larger area due to the higher blade loading. The
shock is detached from the blade tip leading edge. A larger area
of low speed region compared to the peak efficiency condition is
formed after the shock near the leading edge of the pressure side.
The low speed region leads to critical flow blockage in the blade tip
region. Fig. 9 shows the circumferential averaged absolute vorticity
coefficient fn distributions. The definition of absolute vorticity
coefficient fn is as follows:
!
jr  V j
fn ¼
2x
!
where V is the absolute velocity vector and x is the rotor angular
velocity. In order to quantitatively evaluate the intensity of the tip
leakage vortex, the integral of the circumferential averaged absolute
vorticity coefficient fn at the meridional (Z, R) plane (denoted with
H) is calculated. The value of H at the peak efficiency condition and
near stall condition is 0.1169 and 0.1183, respectively. As shown in
Fig. 10, the high absolute vorticity region of the tip leakage vortex
gradually mixes with the main flow and eventually disappears.
The tip leakage vortex does not suddenly expand and there is no
stagnation region in the vortex core, so the vortex breakdown does
not occur in this compressor. As the stall condition is approached,
the interface of the leakage flow and incoming flow reaches the
leading edge plane near the blade tip. This is coincident with the
criteria of the stall of compressor proposed by Vo et al. [29].
As shown in Fig. 11, the low speed region is reduced in the case
of the slot casing treatments. And the intensity of the tip leakage
vortex is reduced by the slot casing treatments. The value of H
for the CT_20, CT_50, CT_80 and CT_110 configuration is 0.1135,
0.1104, 0.1098 and 0.1096, respectively. These values are all smal-
ler than the value of 0.1183 for the SC configuration. For the CT_20
configuration, the intensity of the tip leakage vortex is relatively
large. There is still a large area of low speed region near the leading
edge of the pressure surface, which causes some flow blockage. The
angle between the axial direction and the leakage vortex trajectory
before the shock at 98% span is reduced by 1.0° compared to the SC.
For the CT_50 and CT_80 configuration, the low speed region is sig-
nificantly eliminated. And the angle between the axial direction
and the leakage vortex trajectory before the shock at 98% span
decreases by 4.6° and 6.5° compared to the SC. The tip leakage vor-
tex is repositioned significantly towards the suction surface of the
blade. For the CT_110 configuration, the slots do not interact with
the blade tip flow field within the extent of upstream 30% blade tip
chord. Tip leakage flow near the leading edge causes a low speed
region near the leading edge of the blade pressure surface. The
angle between the axial direction and the leakage vortex trajectory
before the shock at 98% span increases by 1.9° compared to SC. The
tip leakage vortex moves further from the suction surface of the
blade. In general, compared to other configurations, the CT_80 con-
figuration is more effective in reducing the blockage caused by the
tip leakage flow near the blade tip, and the trajectory of the tip
leakage vortex for the CT_80 configuration is closer to the blade
suction surface.
For the slot casing treatment cases, as the compressor’s mass
flow further decreases, the intensity of the tip leakage vortex
increases. The value of H at the near stall condition for CT_20,
CT_50, CT_80 and CT_110 configuration is 0.1152, 0.1153, 0.1143
and 0.1126, respectively. The value of H at the near stall mass flow
for each configuration is larger than that at the near stall mass flow
of SC. As shown in Fig. 11, the CT_50, CT_80 and CT_110 configura-
 X. Zhou et al. / Applied Thermal Engineering 126 (2017) 53–69
(a) peak efficiency point
Fig. 8. Mach number distributions and tip leakage flow for the smooth casing.
(a) peak efficiency point
Fig. 9. Circumferential averaged absolute vorticity coefficient distributions for the smooth casing.
tions are more effective in reducing the intensity of tip leakage vor-
tex and the blockage due to tip leakage vortex compared to the
CT_20 configuration. Thus, the stall mass flow of these configura-
tions is lower than that of CT_20 configuration. As shown in
Fig. 12, For the CT_20 configuration, the tip leakage vortex leads
to the blockage near the blade tip. For the other configurations,
the boundary layer separation occurs at the blade suction side,
which also contributes to the blockage near the blade tip. Fig. 13
shows the difference between time-averaged relative flow angle
of CT_80 and that of SC at near stall condition along the leading
edge circumferential line at 98% span. And Fig. 14 shows the differ-
ence between time-averaged inlet axial velocity of CT_80 and that
of SC at near stall condition. The location of the leading edge cir-
cumferential line and the location corresponding to L = 0.0 and
1.0 are shown in Fig. 15. For the CT_80 configuration, the inlet rel-
ative flow angle at the near stall mass flow is larger than that at the
SC near stall mass flow. As a result, the boundary layer separation
occurs near the blade tip due to the high angle of attack at the
CT_80 near stall mass flow. Besides, the inlet axial velocity at the
59
(b) near stall point
(b) near stall point
CT_80 near stall mass flow is lower than that at the SC near stall
mass flow. Thus, the axial momentum of the inflow is low and is
not enough to suppress the low speed region caused by the leakage
vortex. The low speed region of the leakage vortex in turn deflects
the inflow to the circumferential direction. This effect also causes
the increase of the angle of attack and boundary layer separation
at the blade suction side. The blockage caused by the tip leakage
vortex and boundary layer separation finally leads to the stall of
the compressor.
Fig. 16 shows the flow blockage value at the outer 80% span as a
function of the compressor’s mass flow. The blockage is calculated
with the definition proposed by Khalid [30]. The blockage is calcu-
lated with the following equation:
RR  
Ab 1  qqWWme dA
B¼ ¼ ð1Þ
e
At At
where Wm is the velocity component in the through flow direction.
qe and We are the density and velocity along the edge of the block-
60 X. Zhou et al. / Applied Thermal Engineering 126 (2017) 53–69

(a) peak efficiency point (b) near stall point

Fig. 10. Absolute vorticity coefficient distributions for the smooth casing.

(a) CT_20 (b) CT_50

(c) CT_80 (d) CT_110

Fig. 11. Mach number distributions and trajectory of the tip leakage vortex for different casing treatment configurations at the SC near stall mass flow.
 X. Zhou et al. / Applied Thermal Engineering 126 (2017) 53–69 61

(a) CT_20 (b) CT_50

(c) CT_80 (d) CT_110

Fig. 12. Mach number distributions and trajectory of the tip leakage vortex at the near stall condition of each casing treatment configuration.

Fig. 13. Circumferential distributions of the difference between time-averaged inlet Fig. 14. Circumferential distributions of the difference between time-averaged inlet
relative flow angle of CT_80 and that of SC at near stall condition at 98% span. axial velocity of CT_80 and that of SC at near stall condition at 98% span.
62 X. Zhou et al. / Applied Thermal Engineering 126 (2017) 53–69
Fig. 15. Leading edge circumferential line of the rotor at 98% span.
Fig. 16. Comparison of the blockage growth of the smooth casing and the four
casing treatment configurations.
age region. The blockage B represents the ratio of the blockage area
Ab to the total flow area At just downstream the blade trailing edge.
In the smooth casing case, the blockage near the blade tip gradually
increases as the compressor’s mass flow approaches the stall point.
With the CT_50, CT_80 and CT_110 configuration, the blockage
value is reduced within the mass flow range of the smooth casing
case. But the CT_20 configuration behaves differently than the other
configurations. The CT_20 configuration is obviously less effective
in reducing the blockage near the casing than the other configura-
tions. The reason is that the recirculation in the slots for the
CT_20 configuration is weaker than the other configurations, as
shown in Fig. 21. The weaker recirculation is determined by the
smaller pressure difference between the front and rear part of the
CT_20 slots which is mainly located in front of the shock, as shown
in Figs. 22 and 23. At high mass flow rate, the CT_20 configuration
even increases the blockage value compared to the SC configuration,
because the recirculation in the slots is so small that it has little
effect on the tip leakage flow and the bleeding effect of the slots
reduces the axial velocity near the leading edge. At the near stall
mass flow of the smooth casing, the blockage value for the smooth
casing is 0.053, and the blockage value for the CT_20, CT_50, CT_80
and CT_110 configuration is 0.047, 0.035, 0.032 and 0.035, respec-
tively. At identical compressor’s mass flow, the blockage value for
the CT_80 configuration is lower compared to the other configura-
tions. This results in a higher stall margin improvement of the
CT_80 configuration.
Based on the stall mechanisms of the compressor with and
without slot casing treatments, it can be concluded that effective
slot casing treatment configurations should have good ability in
suppressing the tip leakage vortex and the boundary layer
separation.
4.2.2. Effect of the slot casing treatments on the flow stability
Fig. 17 shows the instantaneous flow in a slot for the CT_50 con-
figuration during one blade passing period at the near stall mass
flow of SC. The static pressure distributions near the blade tip
and in the slot middle plane are also shown. Driven by the pressure
difference between the upstream and downstream region in the
blade passage, a recirculation is formed inside the slots. The recir-
culation varies with the relative positions between the slot and the
blade passage, as shown in Fig. 18. When the rear part of the slot is
located in the high pressure region downstream the shock and near
the pressure side, the bleeding mass flow is relatively large and the
injecting mass flow is relatively small. When the slot is located in
the low pressure region upstream the shock, the bleeding mass
flow is relatively small and the injecting mass flow is relatively
large. Firstly, the slot casing treatment can bleed the tip leakage
flow into the slots or deflect the tip leakage flow radially inwards
near the blade tip. Besides, as shown in Fig. 19, the slot casing
treatment changes the blade loading near the casing, which affects
the driving force of the tip leakage flow. It can be seen that the
pressure near the leading edge of the pressure side increases with
the CT_80 and CT_110 configuration, because the low pressure
region near the pressure side caused by the tip leakage vortex is
weakened. For the CT_20 and CT_50 configuration, although the
low pressure region near the pressure side caused by the tip leak-
age vortex is also reduced, the bleeding effect of the slots can
decrease the pressure near the leading edge of the pressure side.
The combined effect is that the pressure near the leading edge of
the pressure side is increased with the CT_20 configuration, while
decreased with the CT_50 configuration. And for the CT_80 config-
uration, the flow accelerates from the high pressure region near the
leading edge of the pressure side to the low pressure region down-
stream, which causes the kink at around 20–30% normalized chord.
Meanwhile, the shock moves downstream due to the reduction of
the blockage for the slot casing treatment cases. The combined
effect of the radial flow and blade loading alteration determines
the intensity of the tip leakage vortex. As shown in Fig. 20, the
bleeding effect of the slots reduces the tip leakage flow velocity
at 0–50% normalized chord for the CT_50 configuration, which is
beneficial to reduce the intensity of the tip leakage vortex.
Whereas at the immediate downstream of the slots, the velocity
of the tip leakage flow increases, because the pressure difference
across the blade tip increases due to the downstream moving of
the shock. Besides, because the relative velocity near the suction
side decreases with the reduction of the blockage of the tip leakage
vortex near the pressure side, the pressure near the suction side
increases. As a result, the velocity of the tip leakage flow near
the trailing edge decreases with the CT_50 configuration. Secondly,
the low speed fluid in the blade passage near the tip can be bled
into the slots and transported upstream. Then the fluid in the slots
is re-injected into the main flow, which can energize the flow in
the upstream region and deflect the tip leakage vortex trajectory
towards the blade suction side.
Fig. 21 shows the time-averaged normalized slot recirculating
mass flow for the four casing treatment configurations. The time-
averaged slot recirculating mass flow varies with the axial posi-
tions of the slots and the compressor’s mass flow. Fig. 22 shows
the Azimuthal averaged pressure at 98% span and the axial extent
for different slot treatment configurations. And Fig. 23 shows the
instantaneous radial flow at the slots entrance with static pressure
distributions at 98% span at the near stall point of the SC. The fluid
enters the slots at the red region of the slots entrance, and the fluid
leaves the slots at the blue region of the slots entrance. As
 X. Zhou et al. / Applied Thermal Engineering 126 (2017) 53–69 63

(a) 1/4T (b) 2/4T

(c) 3/4T (d) 4/4T

Fig. 17. Instantaneous streamlines in the middle plane of a slot for the CT_50 configuration during one blade passing period at the SC near stall mass flow.

Fig. 18. The normalized bleeding and injecting mass flow of a slot for the CT_50
configuration during one blade passing period.
Fig. 19. Time-averaged static pressure distributions at 98% span at a normalized
mass flow close to the stall mass flow of SC.
indicated in Figs. 21–23, the slots are mainly located in the low
pressure area in front of the shock for the CT_20 configuration.
The driving force for the recirculation is small, so the recirculation downstream the shock near the pressure side. The driving force
is weak. For the CT_50 configuration, with the slots shifted down- for the recirculation is strengthened. So the recirculation of CT 50
stream, the rear part of the slots can pass the high pressure area configuration is stronger than that of the CT_20 configuration.
64 X. Zhou et al. / Applied Thermal Engineering 126 (2017) 53–69
Fig. 20. Time-averaged Normal velocity distributions at the suction side of the tip
gap.
Fig. 21. Time-averaged normalized slot recirculating mass flow for different slots.
Fig. 22. Azimuthal averaged pressure at 98% span and axial extent for different
slots.
For the CT_80 configuration, with the slots shifted further down-
stream, the rear part of the slots can be located in the high pressure
area downstream the shock for a longer time. The driving force for
the recirculation is further strengthened. Hence, the recirculating
mass flow of the CT_80 configuration is larger than that of the
CT_50 configuration. For the CT_110 configuration, the slots are
mainly located in the downstream of the shock. The driving force
for the recirculation is weakened. This results in the decrease of
the recirculating mass flow for the CT_110 configuration compared
to the CT_80 configuration. A strong recirculation in the slots helps
to reduce the blockage near the blade tip. At low compressor’s
mass flow, the recirculation in the slots of the CT_80 configuration
is stronger compared to other configurations. This contributes to a
higher stall margin improvement of the CT_80 configuration. For
the CT_50 and CT_80 configuration, as the compressor mass flow
decreases, the slot recirculating mass flow increases. Because the
blading loading increases with the reduction of the compressor
mass flow and the rear part of the slots can be located in the high
pressure area downstream the shock for a longer time as the shock
moves upstream, which increases the driving force and thus the
intensity of the recirculation in the slots. For the CT_110 configura-
tion, the recirculating mass flow remains almost the same at most
of the compressor mass flow range and decreases towards stall.
Because although the blading loading increases with the reduction
of the compressor mass flow, the forward part of the slots can be
located in the low pressure area upstream the shock for a shorter
time as the shock moves upstream. The CT_110 slots are mainly
located in the downstream of the shock near stall, which decreases
the driving force and thus the recirculating mass flow.
The axial position of the slots not only affects the intensity of
the recirculation but also the axial extent of the interaction
between the slots and the flow field near the blade tip. The
CT_20 slots can only interact with the blade tip flow field within
the extent of the upstream 20% blade chord. The extent of their
impact on the tip leakage flow and the resulting low speed region
is small. The CT_50 slots interact with the tip leakage flow and the
resulting low speed fluid within a larger extent of the blade chord
than the CT_20 slots. This is advantageous to reduce the blockage
due to the tip leakage vortex. For the CT_80 configuration, the rear
part of the slots can pass over the boundary layer separation zone
at the blade suction surface, so the low speed fluid from the bound-
ary layer separation at low compressor’s mass flow can be bled into
the slots, as shown in Fig. 24. For the CT_110 configuration, the
slots cannot effectively suppress the tip leakage flow within the
extent of the upstream 30% blade chord, as shown in Fig. 11. And
low speed region near the leading edge of the pressure side cannot
be energized by the injecting effect at the front part of the slots. In
general, the CT_80 configuration can suppress the tip leakage flow
and the resulting low speed fluid within a large extent of the blade
chord, bleed the boundary layer separation zone at the blade suc-
tion surface and energize the low speed fluid near the leading edge
by the injecting effect at the front part of the slots. This also con-
tributes to a higher stall margin improvement of the CT_80
configuration.
Above all, the slot casing treatments can reduce the blockage
near the blade tip, reposition the tip leakage vortex further from
the pressure side and reposition the shock towards the trailing
edge. These mechanisms are responsible for the stall margin
improvement due to the slot casing treatments. Table 2 shows
the stall margin improvement (DSM), the reduction of the blockage
near the blade tip (DB), the movement distance of the shock
towards the trailing edge (DD) and the reduction of the angle
between the tip leakage vortex trajectory and the axial direction
(Da) for different slot casing treatment configurations. The values
of DB, DD and Da are obtained at the near stall mass flow of the
smooth casing. Based on the correlation coefficients shown in
Table 3, the predictors (DB, DD and Da) are correlated with the
dependent value (DSM) and the predictors themselves are also cor-
related. The correlation coefficient is calculated as follows:
 X. Zhou et al. / Applied Thermal Engineering 126 (2017) 53–69 65

(a) CT_20 (b) CT_50

(c) CT_80 (d) CT_110

Fig. 23. Instantaneous radial flow at slots entrance with static pressure distributions at 98% span at the SC near stall mass flow.

(a) CT_50 (b) CT_80

Fig. 24. Instantaneous Mach distributions near the blade tip and flow in the mid-plane of a slot at normalized mass flow of 0.85.
66 X. Zhou et al. / Applied Thermal Engineering 126 (2017) 53–69

Table 2 b ¼ ðZT ZÞ1 ZT y

Parameter indexes for different slot treatment configurations.
where y is an N  1 vector of scores on the dependent variable, and
CT_20 CT_50 CT_80 CT_110
b is a K  1 vector of regression weights linking the orthogonal vari-
DSM 0.0707 0.1914 0.2470 0.1892
ables to the dependent variable. And a K  K matrix of regression
DB 0.0060 0.0182 0.0210 0.0177
DD 0.024 0.095 0.095 0.024 coefficients K linking the uncorrelated variables and the original
Da 1.0 4.6 6.5 1.9 correlated variables can be obtained by

K ¼ ðZT ZÞ1 ZT X

Table 3 Finally, the relative weights of the original variables can be

Relative importance of the predictors for the stall margin improvement. obtained with the following equation:
Predictors Correlation coefficient Relative weight ei e ¼ K2 b2
DB DD Da DSM DSM
where e ¼ ðe1 ;    ; ei ;    ; ek Þ, and ei is the relative weight of the orig-
DB 1 0.6735 0.4316 0.9886 0.7347 inal predictor. The relative weight ei describes the proportion of
DD 0.6735 1 0.9257 0.6941 0.1775
Da 0.4316 0.9257 1 0.5000 0.0878
variance in the dependent variable associated with the original
predictor.
The relative weights obtained by calculation are listed in
Table 4 Table 3. The relative weight (ei ) of DB, DD and Da for DSM is
Parameter indexes for different slot treatment configurations. 0.7347, 0.1775 and 0.0878, respectively. A higher value of relative
CT_20 CT_50 CT_80 CT_110 weight indicates a larger contribution to the stall margin improve-
ment. Thus, the most important mechanism of stall margin
Mr 0.1900 1.6413 2.5174 2.1223
DLu 15.03 24.67 69.55 83.75 improvement is that the slot casing treatments reduce the block-
DWz 6.03 16.45 0.55 9.35 age near the blade tip. The movement of the shock towards the
trailing edge is of second importance. The movement of the tip
leakage vortex further from the pressure side is the least
important.
Table 5
Three factors have effect on the blockage near the blade tip, the
Relative importance of the predictors for different dependent variables.
position of the shock and the trajectory of the tip leakage vortex.
Relative weight ei Firstly, the slots have bleeding and injecting effect on the low
DB DD Da DSM speed fluid near the blade tip by recirculation. Secondly, the slots
Mr 0.7213 0.6317 0.4640 0.6828 change the blade loading near the casing. Thirdly, the slots ener-
DLu 0.2612 0.3580 0.4305 0.2932 gize the inflow near the casing. Table 4 shows the time-averaged
DWz 0.0175 0.0103 0.1055 0.0239 recirculating mass flow in the slots (Mr ¼ m _ r =m
_ c ), the reduction
Pn of the difference between the inlet and outlet tangential compo-

i¼1 ðxi  xÞðyi  yÞ
 nent of relative velocity (DLu = (DWu)SC  (DWu)CT) and the
r ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Pn qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Pn ffi
  increase of inlet axial velocity (DWz) at 98% span for different slot
i¼1 ðxi  xÞ i¼1 ðyi  yÞ
casing treatment configurations. The values of Mr, DLu and DWz are
where x is the average value of the variable xi and y
 is the average obtained at the near stall mass flow of the smooth casing. The rel-
value of the variable yi. ative weight analysis is then adopted to quantify the relative
In order to quantify the relative importance of the different importance of the correlated predictors (Mr, DLu and DWz) for
mechanisms of the casing treatment more definitely, the relative the dependent variables (DB, DD and Da). As shown in Table 5,
weight analysis [31] is the conducted. The relative weight analysis for all the three dependent variables (DB, DD and Da), the
can consider both the independent effect of each predictor and its sequence of the relative weight is Mr > DLu > DWz. Besides, the rel-
effect when combined with the other predictors. The relative ative weights of Mr, DLu and DWz for DSM are calculated by the lin-
weight analysis is a procedure to estimate the relative importance ear weighted sum of the relative weights of DB, DD and Da for
of the correlated predictors in multiple regression models. The rel- DSM. And the linear weighted coefficients are the relative weights
ative weight is calculated as follows: of Mr, DLu and DWz for DB, DD and Da. Then the proportion of vari-
X is assumed to be an N  K matrix of the predictors in standard ance in DSM associated with Mr, DLu and DW through DB, DD and
score form. K is the number of the predictors and N is the number Da is obtained. As shown in Table 5, the sequence of the relative
of the subjects. Then, a singular value decomposition of X is importance of Mr, DLu and DWz for DSM is Mr > DLu > DWz. It indi-
conducted. cates the bleeding and injecting effect is the most important mech-
anism that determines the blockage near the blade tip, the position
X ¼ PRQ T of the shock and the trajectory of the tip leakage vortex. As a result,
the intensity of the recirculation is the most important mechanism
where P is an N  K matrix of eigenvectors associated with XXT , Q is
for stall margin improvement. The reduction of the blade loading is
a K  K matrix of eigenvectors associated with XT X, and R is a K  K of second importance and the increase of the inlet axial velocity is
diagonal matrix containing the singular values of X that are equal to of third importance.
the square roots of the eigenvalues of XXT and XT X.
Then an orthogonal approximation of X can be obtained by 4.2.3. Effect of the slot casing treatments on the efficiency
T The four slot casing treatments all decrease the peak efficiency
Z ¼ PQ
of the compressor, and the discrepancy of the peak efficiency
Z contains a new set of variables that are uncorrelated with one between the smooth casing and the CT_20, CT_50, CT_80, CT_110
another but are related to the original set of correlated variables configuration is 0.71%, 0.48%, 0.70%, 0.96%, respectively. Among
(X). This new set of orthogonal variables can then be used to gen- the four different slot casing treatment configurations, the CT_50
erate regression coefficients by regressing y on Z: configuration causes the minimum losses in peak efficiency. As
 X. Zhou et al. / Applied Thermal Engineering 126 (2017) 53–69 67

Fig. 25. Spanwise distributions of adiabatic efficiency at the peak efficiency Fig. 27. Spanwise distributions of adiabatic efficiency at the near stall mass flow of
condition. the SC.

shown in Fig. 25, at the peak efficiency condition, the CT_50 config-
uration reduces the efficiency at 94–100% span. As shown in the
above sections, the slot casing treatment can suppress the tip leak-
age vortex, so the losses caused by the tip leakage vortex can be
reduced. Meanwhile, the recirculating vortex flow in the slots
and the interaction between the slots flow and the blade passage
flow generate additional losses near the blade tip. As indicated in
Fig. 26, the losses due to the bleeding and injecting effect exceed
the losses reduction due to the suppression of tip leakage vortex.
Thus, the CT_50 configuration decreases the peak efficiency of
the compressor.
At the near stall mass flow of the SC, the CT_50 configuration
increases the efficiency by 0.76%, while the CT_20, CT_80 and
CT_110 configuration reduces the efficiency by 0.04%, 0.07% and
0.39%, respectively. As shown in Fig. 27, the CT_50 configuration
decreases the efficiency slightly at 95–100% span and increases
the efficiency significantly at 65–95% span. As a result, the CT_50
configuration increases the efficiency compared to the SC. As
shown in Figs. 28 and 29, the pressure near the shock rises more Fig. 28. Time-averaged static pressure distributions at 85% span at the near stall
slowly with the CT_50 configuration than the other configurations, mass flow of SC.
and the boundary layer separation region downstream the shock at
high span can be reduced effectively with the CT_50 configuration.
As a result, the CT_50 configuration improves the efficiency of the tion, although the shock intensity and boundary separation region
compressor. For the CT_20 and CT_110 configuration, the boundary at high span is reduced, additional losses generate near the casing
layer separation region downstream the shock at high span is due to the mixing of the slots flow and the blade passage flow, thus
enlarged, which decreases the efficiency. For the CT_80 configura- the efficiency of the compressor is reduced.

(a) SC (b) CT_50

Fig. 26. Time and circumferential averaged entropy contours at each peak efficiency condition.
68 X. Zhou et al. / Applied Thermal Engineering 126 (2017) 53–69
(a) SC (b) CT_20 (c) CT_50
Fig. 29. Time-averaged static pressure distribution and limiting streamline on the suction side at the near stall mass flow of SC.
5. Conclusions
The effect of the axial skewed slot casing treatments with four
different axial positions on the performance of NASA Rotor 67
was investigated by unsteady numerical simulations. The conclu-
sions are summarized as follows:
(1) When the front part of the slots covers the initial position of
the tip leakage vortex and the rear part of the slots covers
the boundary layer separation zone downstream the shock,
the recirculating mass flow in the slots is stronger due to
the larger driving force. Thus, the slots can reduce the block-
age due to the tip leakage vortex and the boundary layer
separation more effectively, which causes a larger stall mar-
gin improvement. When the slots are shifted upstream com-
pared to the optimal configuration, the recirculation is
weakened and the blade suction boundary separation can-
not be suppressed effectively. When the slots are shifted
downstream compared to the optimal configuration, the
recirculation is also weakened and the tip leakage vortex
cannot be suppressed effectively. As a result, the stall margin
improvement is reduced with the slots shifted upstream or
downstream.
(2) The analysis of the interaction mechanisms between the
slots and the flow near the blade tip indicate that the inten-
sity of the recirculation in the slots, the blade loading and
the inlet axial velocity have effect on the blockage, the shock
position and the trajectory of the tip leakage vortex near the
blade tip, which determines the stall margin improvement
due to the slot casing treatments. Based on this, the relative
importance of the factors that affect the effectiveness of the
slots on the stall margin improvement is obtained with the
relative weight method. The results show that the sequence
of relative importance of the factors in the first level is the
intensity of the recirculation in the slots, the blade loading,
and the inlet axial velocity near the blade tip. And the
sequence of relative importance of the factors in the second
level is the blockage, the shock position, the inlet axial veloc-
ity near the blade tip.
Acknowledgments
This work has been carried out with the support of National
Natural Science Foundation of China (No. 51406195) and National
Key Technology Research and Development Program of China (No.
2016YFB0901402). The authors sincerely thank it for funding this
work.
(d) CT_80 (e) CT_110
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.applthermaleng.
2017.07.165.
References
[1] K.L. Suder, M.L. Celestina, Experimental and computational investigation of the
tip clearance flow in a transonic axial compressor rotor, ASME J. TurboMach.
118 (2) (1996) 218–229.
[2] J.D. Denton, L. Xu, The exploitation of three-dimensional flow in
turbomachinery design, Proc. Inst. Mech. Eng., Part C: J. Mech. Eng. Sci. 213
(2) (1999) 125–137.
[3] R. Biollo, E. Benini, Shock/boundary-layer/tip-clearance interaction in a
transonic rotor blade, AIAA J. Propul. Power 25 (3) (2009) 668–677.
[4] W. Hofmann, J. Ballmann, Tip Clearance Vortex Development and Shock-
Vortex-Interaction in a Transonic Axial Compressor Rotor, AIAA 2002-0083,
2002.
[5] C. Hah, D.C. Rabe, A.R. Wadia, Role of Tip-Leakage Vortices and Passage Shock
in Stall Inception in a Swept Transonic Compressor Rotor, ASME Paper GT2004-
53867.
[6] K. Yamada, K. Funazaki, M. Furukawa, The Behavior of Tip Clearance Flow at
Near-Stall Condition in a Transonic Axial Compressor Rotor, ASME paper
GT2007-27725, 2007.
[7] C. Hah, J. Bergner, H.-P. Schiffer, Short Length-scale Rotating Stall Inception in a
Transonic Axial Compressor-criteria and Mechanisms, ASME Paper GT2006-
90045, 2006.
[8] E.M. Greitzer, J.P. Nikkanen, D.El Haddad, A fundamental criterion for the
application of rotor casing treatment, J. Fluids Eng. 101 (2) (1979) 237–243.
[9] M.D. Hathaway, Passive Endwall Treatments for Enhancing Stability, NASA
TM-214409, 2007.
[10] V. Lyengar, L. Sankar, S. Niazi, Assessment of the Self-Recirculating Casing
Treatment Concept Applied to Axial Compressors, AIAA 2005-632, 2005.
[11] J.H. Kim, K.J. Choi, K.Y. Kim, Aerodynamic analysis and optimization of a
transonic axial compressor with casing grooves to improve operating stability,
Aerosp. Sci. Technol. 29 (1) (2013) 81–91.
[12] M. Hembera, H.-P. Kau, E. Johann, Simulation of casing treatments of a
transonic compressor stage, Int. J. Rotat. Mach. 2008 (2008), 10 657202.
[13] M.W. Muller, H.-P. Kau, M. Voges, et al., Investigation of Passage Flow Features
in a Transonic Compressor Rotor with Casing Treatments, ASME Paper
GT2011-45364, 2011.
[14] I. Wilke, H.-P. Kau, A Numerical Investigation of the Influence of Casing
Treatments on the Tip Leakage Flow in a HPC Front Stage, ASME Paper
GT2002-30642, 2002.
[15] I. Wilke, H.-P. Kau, G. Brignole, Numerically Aided Design of a High-efficient
Casing Treatment for a Transonic Compressor, ASME Paper GT2005-68993,
2005.
[16] Jialing Lu, Wuli Chu, Yanhui Wu, Investigation of Skewed Slot Casing on
Transonic Axial-flow Fan Stage, ASME Paper GT2009-59417, 2009.
[17] Junqiang Zhu, Wuli Chu, The Effects of Bend Skewed Groove Casing Treatment
on Performance and Flow Field Near Endwall of an Axial Compressor, AIAA
2005-809, 2005.
[18] Xingen Lu, Wuli Chu, Juqiang Zhu, et al., Experimental and Numerical
Investigation of a Subsonic Compressor with Bend Skewed Slot Casing
Treatment, ASME Paper GT2006-90026, 2006.
[19] G.A. Brignole, F.C.T. Danner, H.-P. Kau, Time Resolved Simulation and
Experimental Validation of the Flow in Axial Slot Casing Treatments for
Transonic Axial Compressors, ASME Paper GT2008-50593, 2008.
 X. Zhou et al. / Applied Thermal Engineering 126 (2017) 53–69
[20] M. Hembera, F.C.T. Danner, H.-P. Kau, et al., Numerical Design and
Optimization of Casing Treatments for Transonic Axial Compressors, AIAA
2008-5063, 2008.
[21] J.A. Streit, C. Brandstetter, F. Heinichen, et al., An advanced axial-slot casing
treatment on a transonic compressor: a close look with computational fluid
dynamics and experimental validation, Proc. IMechE Part A: J. Power Energy
227 (6) (2013) 683–691.
[22] J.A. Streit, F. Heinichen, H.-P. Kau, Axial-Slot Casing Treatments Improve the
Efficiency of Axial Flow Compressors: Aerodynamic Effects of a Rotor Redesign,
ASME Paper GT2013-94975, 2013.
[23] C. Brandstetter, F. Wartzek, J. Werner, et al., Unsteady Measurements of
Periodic Effects in a Transonic Compressor with Casing Treatments, ASME
Paper GT2015-42394, 2015.
[24] Ma Ning, Nan Xi, Lin Feng, Numerical Study on Effects of Axial-Slot Casing
Treatment on Peak Efficiency of Axial Compressors, ASME Paper GT2016-
56554, 2016.
69
[25] S.C. Walter, S. William, C.U. Donald, Design and Performance of a 427-Meter-
Per-Second-Tip-Speed Two-Stage Fan Having a 2.40 Pressure Ratio, NASA
Technical Paper 1314, 1978.
[26] A.J. Strazisar, J.R. Wood, M.D. Hathaway, et al., Laser Anemometer
Measurements in a Transonic Axial-Flow Fan Rotor, NASA Technical Paper
2879, 1989.
[27] Du Juan, Lin Feng, Zhang Hongwu, et al., Numerical investigation on the self-
induced unsteadiness in tip leakage flow for a transonic fan rotor, ASME J.
Turbomach. 132 (021017) (2010) 1–9.
[28] J.J. Adamczyk, M.L. Celestina, E.M. Greitzer, et al., The role of tip clearance in
high-speed fan stall, ASME J. Turbomach. 115 (1) (1993) 28–39.
[29] H.D. Vo, C.S. Tan, E.M. Greitzer, Criteria for Spike Initiated Rotating Stall, ASME
Paper GT2005-47306, 2005.
[30] S.A. Khalid, A.S. Khalsa, I.A. Waitz, et al., Endwall blockage in axial compressors
499 ASME J. Turbomach. 121 (3) (1999), -229.
[31] S. Tonidandel, J.M. LeBreton, J.W. Johnson, Determining the statistical
significance of relative weights, Psychol. Meth. 4 (1) (2009) 387–399.