effects of blade row interactions on unsteady stator

11
1 1 ISABE-2015-20104 Effects of Blade Row Interactions on Unsteady Stator Surface Pressures in an Embedded Compressor Stage Natalie R. Smith and Nicole L. Key Purdue University West Lafayette, IN 47907 USA Abstract Blade row interactions drive unsteady blade forces in compressors. This paper presents a perspective on understanding how the pitchwise variations in the flow at the exit of the rotor affect the surface pressures on the downstream vane. The rotor wakes and tip leakage flows are the primary unsteady flow features that drive unsteady lift on the vane. However, those rotor flow features are affected by their interaction with the wakes from upstream vanes. Thus, the interaction with Stator 1 and Rotor 2 must be understood to adequately characterize the interaction between Rotor 2 and Stator 2. This paper utilizes vane clocking, or the circumferential shift in successive vane rows of similar counts, to illuminate these blade row interaction effects on the downstream vane surface pressure distribution. To accomplish this, experiments were performed in a three-stage axial compressor where high-frequency response pressure transducers were flush-mounted in the Stator 2 pressure and suction surfaces at 50% and 80%span. Results show that Rotor 1 Rotor 2 interactions contribute significantly to the changes in unsteady stator surface pressure over the course of a rotor revolution. Also, the Rotor Rotor 2 interaction levels change in the pitchwise direction downstream of the rotor, and thus, the changes in surface pressure are affected by vane clocking. Furthermore, the rotor tip leakage flow is an additional contributor to the unsteady stator surface pressures measured at 80% span, providing an additional pressure peak per blade passing and thus, more high frequency content (twice the blade passing frequency). Nomenclature < > ensemble-averaged bpf blade pass frequency c x axial chord CL clocking configuration C P coefficient of pressure N EA number of samples in ensemble P static pressure PS pressure side ρ density R rotor S stator SS suction side U t rotor tip speed Subscripts i instantaneous in compressor inlet Introduction As the increase in computational resources allows for multistage calculations to become more common, data acquired in a multistage environment are needed to validate the models and verify that the important flow physics, such as blade row interactions, are captured. Not only are the levels of unsteadiness larger in an embedded compressor stage, but the frequency content of the unsteadiness is markedly different from that of single stage and repeating stage machines. The flow field in compressors is dominated by unsteady blade row interactions, where the viscous wakes shed from upstream blade rows is one of the most important aerodynamic forcing functions, especially in the rear stages of a high pressure compressor. The convection and decay of wakes as they propagate and interact with downstream rows have been studied by many authors. Early studies 1-4 examined how the downstream blade rows chop a wake into segments. Others 5-7 have examined the broadening of wakes through viscous mixing and the benefits of inviscid stretching known as wake recovery. Furthermore, the action of chopping the low momentum fluid of an upstream wake creates a negative jet and affects the surface pressures of the downstream row. The suction side of the chopping row experiences a local increase in pressure while the pressure side has a reduction in surface pressure from the rotor wake 2,8 . The wake’s role as a forcing function for vibrations on the downstream vane row is often considered using an average, representative wake. Variations in compressor hardware due to machining tolerances or wear, in addition to blade row interactions, introduce blade-to-blade differences in wake shedding, and this is commonly referred to as wake variability. It is uncommon for these effects to be considered in CFD calculations or quantified in

Upload: others

Post on 03-Dec-2021

1 views

Category:

Documents


0 download

TRANSCRIPT

1

1

ISABE-2015-20104

Effects of Blade Row Interactions on Unsteady Stator Surface Pressures in an Embedded Compressor Stage

Natalie R. Smith and Nicole L. Key

Purdue University

West Lafayette, IN 47907 USA

Abstract

Blade row interactions drive unsteady blade

forces in compressors. This paper presents a

perspective on understanding how the pitchwise

variations in the flow at the exit of the rotor affect the

surface pressures on the downstream vane. The rotor

wakes and tip leakage flows are the primary unsteady

flow features that drive unsteady lift on the vane.

However, those rotor flow features are affected by

their interaction with the wakes from upstream vanes.

Thus, the interaction with Stator 1 and Rotor 2 must

be understood to adequately characterize the

interaction between Rotor 2 and Stator 2. This paper

utilizes vane clocking, or the circumferential shift in

successive vane rows of similar counts, to illuminate

these blade row interaction effects on the downstream

vane surface pressure distribution. To accomplish

this, experiments were performed in a three-stage

axial compressor where high-frequency response

pressure transducers were flush-mounted in the Stator

2 pressure and suction surfaces at 50% and 80%span.

Results show that Rotor 1 – Rotor 2 interactions

contribute significantly to the changes in unsteady

stator surface pressure over the course of a rotor

revolution. Also, the Rotor – Rotor 2 interaction

levels change in the pitchwise direction downstream

of the rotor, and thus, the changes in surface pressure

are affected by vane clocking. Furthermore, the rotor

tip leakage flow is an additional contributor to the

unsteady stator surface pressures measured at 80%

span, providing an additional pressure peak per blade

passing and thus, more high frequency content (twice

the blade passing frequency).

Nomenclature

< > ensemble-averaged

bpf blade pass frequency

cx axial chord

CL clocking configuration

CP coefficient of pressure

NEA number of samples in ensemble

P static pressure

PS pressure side

ρ density

R rotor

S stator

SS suction side

Ut rotor tip speed

Subscripts

i instantaneous

in compressor inlet

Introduction

As the increase in computational resources

allows for multistage calculations to become more

common, data acquired in a multistage environment

are needed to validate the models and verify that the

important flow physics, such as blade row

interactions, are captured. Not only are the levels of

unsteadiness larger in an embedded compressor

stage, but the frequency content of the unsteadiness is

markedly different from that of single stage and

repeating stage machines. The flow field in

compressors is dominated by unsteady blade row

interactions, where the viscous wakes shed from

upstream blade rows is one of the most important

aerodynamic forcing functions, especially in the rear

stages of a high pressure compressor.

The convection and decay of wakes as they

propagate and interact with downstream rows have

been studied by many authors. Early studies1-4

examined how the downstream blade rows chop a

wake into segments. Others5-7

have examined the

broadening of wakes through viscous mixing and the

benefits of inviscid stretching known as wake

recovery. Furthermore, the action of chopping the

low momentum fluid of an upstream wake creates a

negative jet and affects the surface pressures of the

downstream row. The suction side of the chopping

row experiences a local increase in pressure while the

pressure side has a reduction in surface pressure from

the rotor wake2,8

.

The wake’s role as a forcing function for

vibrations on the downstream vane row is often

considered using an average, representative wake.

Variations in compressor hardware due to machining

tolerances or wear, in addition to blade row

interactions, introduce blade-to-blade differences in

wake shedding, and this is commonly referred to as

wake variability. It is uncommon for these effects to

be considered in CFD calculations or quantified in

2

2

experiments. However, a few authors have assessed

wake variability both experimentally and

numerically. Sherman et al.9 used a mean wake

analysis to demonstrate the need to consider

individual blade variability, as the mean wake was

only representative of about half the rotors in their

single-stage compressor. Similarly, Boyd and

Fleeter10

found that less than 50% of the rotor wakes

in their facility fell within the confidence level of the

grand mean average wake. Furthermore, this

variability resulted in differences in aerodynamics

response by as much as 100%. Key et al.11

used

different averaging techniques to isolate variability

effects locked to the rotor revolution versus

variations associated with the pitchwise proximity to

the downstream stator for all three rotors in a three-

stage compressor. Sanders and Fleeter12

evaluated the

blade-to-blade variability in a 1.5 stage high-speed

compressor and showed that clocking the inlet guide

vane (IGV) had a profound effect on how the

chopped IGV wakes would combine with the rotor

wakes and change the forcing on the downstream

stator. At off-design conditions, they observed that

even small rotor wake variability could result in large

differences in unsteady lift on the downstream stator.

The rotor tip leakage flow is another feature of

the unsteady rotor exit flow field that drives unsteady

lift on the downstream vane. The tip leakage flow

develops as the flow from the pressure side of the

rotor blade releases to the suction side through the

rotor tip gap. The effects of tip leakage flow on

compressor performance and stall margin have been

studied by several researchers13-14

. Additionally,

recent efforts have shown that the characteristics of

the tip leakage flow (size, unsteadiness, penetration

into the passage) can change in the pitchwise

direction depending on the rotor’s interaction with

the upstream vane’s wake15-17

.

Since the rotor exit flow field is not only

unsteady but also varies in the pitchwise direction

due to interactions with the upstream vane row, the

unsteady pressures on the downstream vane row will

be affected by the pitchwise location of the upstream

vane row. The relative setting or phasing of

successive vane rows with similar vane counts is

called vane clocking, and it affects stage

performance, unsteady blade forces, and acoustics.

Vane clocking is a useful tool for a researcher

because it allows for a better understanding of how

blade row interactions, including the pitchwise

variations in the rotor exit flow field, affect unsteady

lift. Several authors18,19

have explored how the vane

surface pressure unsteadiness changes with vane

clocking. A computational study of a 1.5 stage

turbine by Griffin et al.20

found lower surface

pressure unsteadiness was associated with the lower

pressure loss clocking configuration. However, a

study by Saren et al.18

reported the maximum

efficiency clocking configuration had both more

unsteadiness and higher frequency content. Thus,

vane clocking affects unsteady surface pressure, but

the trend is not clearly tied to vane performance.

The focus of this research is to utilize vane

clocking to provide insight to the drivers of unsteady

lift on the downstream vane row in a multistage

compressor environment. This information will

provide guidance in development of appropriate

models for predicting forced response and blade

vibration phenomena.

Experimental Approach

The experiments were performed in the Purdue

3-Stage Axial Compressor Research Facility. The

compressor models the rear stages of a high-pressure

compressor with engine representative Reynolds and

Mach numbers. The design corrected speed is 5,000

rpm. The compressor is scaled up (24-in tip diameter)

to allow measurements with high spatial resolution.

The flow path consists of an inlet guide vane (IGV)

followed by three stages. The IGV and rotor airfoils

are double circular arc designs, and the stators are

NACA 65-series airfoils. The rotor blade counts

decrease by three through the machine: 36, 33, and

30, respectively. The IGV, Stator 1, and Stator 2 rows

each have 44 vanes, and Stator 3 has 50. All four

vane rows are shrouded, feature circular leading

edges, and are individually indexable to enable vane

clocking. For this study, the IGV and Stator 1 are

moved in unison while the relative position of Stator

2 and Stator 3 are fixed, thus isolating clocking

effects to the Stator 1 wake interaction with Stator 2.

Stator 2 turns the flow 30 degrees, has an aspect

ratio of 0.8333, an inlet Reynolds number based on

chord of 4.6x105, and an inlet Mach number of 0.34

at peak efficiency. Additional facility details are

available in Ref. 21.

This study addresses the implications of blade

row interactions and wake variability on the unsteady

surface pressures of the second stator row. The inlet

conditions to the vane row were characterized using a

TSI miniature platinum cross-film sensor with a TSI

Inc. IFA 100 anemometer. The film was calibrated in

a jet over the full range of anticipated angles,

velocities, and densities. Data were acquired at 50%

and 80% span for 20 circumferential traverse

positions at a rate of 300kHz for 500 rotor

revolutions. Also at the Stator 2 inlet, time-resolved

total pressure data were acquired with a Kulite LQ-

062 sensor embedded in a miniature Kiel-head

(0.083-in. diameter) probe. The same 20

circumferential positions were traversed, but a more

detailed radial traverse was completed that

3

3

incorporated a higher resolution in the tip region:

increments of 2% span from 78% to 100% span,

increments of 4% span from 4% to 20% span and

70% to 78% span, and increments of 5% span from

20% to 70% span.

To acquire unsteady surface pressures, high-

frequency response pressure transducers were

embedded into two Stator 2 vanes. Each vane was

instrumented with eight Kulite LQ-062 pressure

transducers. The sensors were positioned at 50% and

80% span, at 10%, 20%, 30%, and 40% axial chord

(cx) on the pressure side of one vane and the suction

side of another, such that they measure the same vane

passage, shown in Fig. 1. They are reverse embedded

so the measured passage remains unaffected by the

potting material. The sensor at 20%cx at 50%span on

the pressure side was not functional, and therefore,

no data will be presented at that location. The

transducers did not have screens, which resulted in a

high frequency response, as much as 100-150 kHz

per manufacturer’s specifications. Details regarding

the pressure calibration can be found in Ref. 22.

In this study, the surface pressure distribution

fluctuations and unsteadiness will be discussed. A

once-per-revolution signal from an optical

tachometer is used to phase-lock the pressure data to

the rotor and calculate an ensemble average

revolution of static pressure, ⟨ ⟩,

⟨ ⟩

, (1)

where NEA is the number of rotor revolution used in

the ensemble average (500 revolutions), and Pi is the

instantaneous static pressure. The unsteady static

pressure measurements are presented in terms of a

pressure coefficient, CP, normalized by the dynamic

pressure at the inlet. The pressure unsteadiness that is

not locked to the rotor is evaluated using the root

mean square (RMS) of CP based on the ensemble

average,

⟨ ⟩ √

∑ ⟨ ⟩

, (2)

where ρ is the inlet density and Ut is the rotor tip

speed.

This study will address the effects of blade row

interactions and rotor wake variability on Stator 2

surface pressure at part speed. Detailed discussion of

the Rotor 2 wake variability has been presented in

Ref. 17, where data were acquired at 74%Nc near the

Rotor 2 first torsion resonance and some of these data

are also presented in this paper to introduce the inlet

flow to Stator 2. Two loading conditions are

presented: a nominal loading condition which is

along the operating line for peak efficiency at

100%Nc and a high loading condition. Stator 2

surface pressure data have been acquired for six

clocking configurations at these two loading

conditions at 74%Nc.

Vane clocking is used to place the downstream

vane at different pitchwise locations to explore the

variations in unsteady lift associated with the

fluctuating stator inlet flow. The vane clocking

configurations are defined by a clocking offset (CL)

which is expressed as the difference in

circumferential location between the upstream (IGV

and Stator 1) and downstream (Stator 2 and Stator 3)

vane rows in terms of percent vane passage (vp).

Results at Mid-Span

The results at mid-span are explored where the

main flow feature driving unsteady vane lift is the

rotor wake. The rotor wakes are measured with a

cross-film anemometer, and they are presented in

terms of absolute flow angle. Measurements were

acquired at twenty circumferential positions in the

pitchwise direction across one vane passage. This

section presents the Rotor 2 wake characteristics at

different pitchwise positions.

Figure 1: Stator 2 vanes with flush-mounted

pressure sensors at 50% and 80%span.

4

4

Figure 2a shows the ensemble-averaged absolute

flow angle downstream of Rotor 2 for 20 positions

across the Stator 2 pitch. The relative frame velocity

deficit associated with the rotor wake is manifested

as an increase in flow angle in the absolute reference

frame, which is directly measured by the sensor.

Each Rotor 2 wake is observed as an increase in

absolute flow angle on the order of 10-20°. The shape

of the Rotor 2 wake depends both on location within

the rotor revolution and pitchwise location with

respect to the vanes. The blade-to-blade Rotor 2 wake

variability (across a rotor revolution) is driven by the

Rotor 1 – Rotor 2 interactions (as discussed in Ref.

17), resulting in a 3-beat behavior based on the blade

count difference of Rotor 1 and Rotor 2. There are

36 Rotor 1 blades and 33 Rotor 2 blades. Thus, there

are three portions of the rotor revolution where the

Rotor 1 wakes pass through the Rotor 2 passage and

are visible as small increases in flow angle in

between the Rotor 2 wakes. There are also three

portions of the revolution where the Rotor 1 wakes

interact with the Rotor 2 blades, and thus, the wakes

combine and are indistinguishable at the Stator 2

inlet.

The strength of this 3/rev modulation associated

with rotor-rotor interactions changes across the vane

passage. This is highlighted by focusing on the

circumferential positions with the strongest and

weakest 3/rev amplitude modulation. These positions

are at 30%vp and 80%vp. The 30%vp position

consists of rotor wakes, which are consistently

thicker. Fig. 2b shows the largest (max) and smallest

(min) wakes shed from Rotor 2 for these two

circumferential locations. For 80%vp, these are the

wakes shed from rotor blade 19 (max) and 3 (min),

and at 30%vp, these are rotor blade number 19 (max)

and 23 (min). At 30%vp, the Rotor 2 wakes are

somewhat similar in shape, whereas at 80%vp, the

maximum and minimum wakes feature significantly

different shapes. These differences in wake shapes

between pitchwise measurement positions can be

attributed to a couple interactions. First, there is an

effect from the downstream stator’s potential field.

This can be noted by the higher mean flow angles at

80%vp compared to 30%vp in Fig. 2b. Secondly,

there is a change in the average wake width and

depth with pitchwise position, which is associated

with the Stator 1 wake interaction with Rotor 2. This

was presented in detail by Key et al.11

whose research

in the same facility showed that the downstream

stator’s potential field contributes to changes in the

mean levels of the flow angle in the wakes, but the

differences in wake shape are driven by blade row

interactions from the upstream rows. Specifically, at

different times within the Rotor 2 blade pass period,

the rotor will have differing levels of interaction with

the Rotor 1 wake, which will affect the unsteady lift

on the rotor and thus, the shedding of the boundary

layers into the rotor wake.

Table 1 summarizes the difference in wake width

and depth for the largest and smallest wakes at 30%

and 80%vp at high loading to the average (Avg)

wake. The maximum and minimum wakes from

blades 19 and 23, respectively, at 30%vp are larger

(both wider and deeper) than the overall average

wake. At 80%vp, the maximum wake is not as large

as 30%vp, and the minimum wake is significantly

smaller than the Avg wake.

Finally, the amount of Rotor 2 wake variability,

or the amount of variation in Rotor 2 wakes

compared to the mean wake, changes with pitchwise

position. While this can be observed in Fig. 2, it is

also noted in Table 1 as a percent difference between

the maximum and minimum wakes (ΔMM) at the

Figure 2: Stator 2 inlet absolute flow angles, α, for 20 pitchwise locations (a), with maximum and

minimum Rotor 2 wakes (b), and frequency content (c) at 50%span for high loading.

(a) (b) (c)

5

5

two circumferential positions. The maximum and

minimum wakes at 30%vp are 34.3% different in

width and 39.1% different in depth, and at 80%vp,

they are 58.2% different in width and 58.5% different

in depth. This indicates more wake variability at

80%vp, and this is a result of stronger Rotor 1 –

Rotor 2 interactions.

These differences between pitchwise locations

are important because they indicate that a different

stator clocking position could result in the

downstream stator experiencing a different amount of

inlet flow variability. Differences in inlet flow can

have effects on both aerodynamic performance and

unsteady aeromechanic forcing. Figure 2c shows the

magnitude of the Fourier decomposition for the

ensemble-averaged revolution of absolute flow angle

at 30% and 80%vp. The strength of the Fourier

transform magnitudes at Rotor 2 and Rotor 1 blade

pass frequencies differ at the two pitchwise positions.

The 33/rev (Rotor 2 blade pass frequency) is stronger

at 30%vp than at 80%vp. At 80%vp, there is a larger

spectral magnitude at 36/rev (Rotor 1 blade pass

frequency), and the 3/rev frequency associated with

the Rotor 1 and Rotor 2 interaction is present, and

this is also where the rotor wake shapes were most

different.

As discussed, Stator 2 will experience blade-to-

blade variations in the inlet flow angle due to Rotor 1

– Rotor 2 interactions. The flow angle data at the

Stator 2 inlet showed that the Rotor 2 wakes can vary

by as much as 58% in wake width and depth due to

rotor-rotor interactions at a particular pitchwise

position. This difference in unsteady aerodynamics

forcing will affect the unsteady stator surface

pressure.

To better quantify the changes in the Stator 2

surface static pressure resulting from these inlet

variations, the signals from the Stator 2 sensors

nearest the leading edge (10%xc) on the suction side

and pressure side at 50%span for high loading are

considered. Figure 3 shows a section of the

ensemble-averaged rotor revolution from blade pass

period 1 to 4 of the suction side signal. The three

blade pass periods are segmented, shown by the

vertical dashed lines. For each blade pass period, the

range of the pressure coefficient is calculated based

on the difference between the maximum and

minimum value in the period, ΔCP. Also for each

blade pass period, the average coefficient of pressure,

Avg CP, is calculated and depicted as the horizontal

dashed lines in Fig. 3. These values will be used to

evaluate the effects of Rotor 1 – Rotor 2 interactions,

and later, vane clocking.

Figure 4a shows the ensemble-average pressure

coefficient for the full rotor revolution at 50%span on

the suction side near the leading edge, with the three

blade pass periods from Fig. 3 noted. The 3/rev

beating is apparent. The influence of the Rotor 2

wake variability affects both the mean pressure

coefficient value and the variation in pressure

coefficient during the blade pass period. Both of

these effects are shown in Fig. 4 for the pressure side

and suction side, where Fig. 4b shows the ΔCP and

Fig. 4c shows the Avg CP for each blade pass period.

Figure 4b shows the range of pressure coefficient

for each Rotor 2 blade pass period as a percent based

on the mean coefficient (ΔCP). The blade pass

periods with the highest changes in pressure

coefficient are twice as large as those with the

smallest changes in pressure coefficient for the

suction side, and they are associated with the portions

of revolution where the Rotor 1 and Rotor 2 wakes

have combined to provide a large wake at the stator

inlet. The pressure side experiences changes of about

1.5 times between the lowest and highest pressure

responses.

The mean pressure coefficient (Avg CP) for each

blade pass also changes with the Rotor 1 – Rotor 2

interactions as shown in Fig. 4c; the pressure

coefficients have been normalized by the overall

average pressure coefficient. The suction side is more

strongly affected, with changes of mean pressure per

blade pass of about 4%. These results indicate that

the Rotor 2 wake variations from the mean wake,

which range from 34% to 58% depending on

Table 1: Rotor 2 wake variability at two pitchwise

positions at midspan for a high loading condition.

W D

Avg 10.9%bp ΔMM 15.22° ΔMM

30%vp

ΔMax

(%) 37.2

34.3

42.6

39.1 ΔMin

(%) 2.1 2.5

80%vp

ΔMax

(%) 32.1

58.2

28.1

58.5 ΔMin

(%) -16.5 -19.2

1 1.5 2 2.5 3 3.5 40.3

0.35

0.4

0.45

0.5

0.55

< C

>

P

Rotor Blade Pass Period

avg CP

CP

Figure 3: Definition of Avg CP and ΔCP

6

6

pitchwise location across the passage, can alter the

downstream stator’s unsteady surface pressure by

100% and the mean pressure by 4%.

The 3/rev pattern drives the blade-to-blade

variability, but the pattern is markedly repeatable.

The pressure coefficient trace shown in Fig. 4a can

be split in to three sections of eleven blade pass

periods and then overlaid on one another, as shown in

Fig. 5a. While the eleven Rotor 2 blade pass periods

each exhibit a unique pressure response due to the

interaction with the Rotor 1 wake, this pattern is

repeatable across one third of the rotor revolution.

The small differences that exist are attributed to

blade-to-blade variability (such as manufacturing

variations) as opposed to variations caused by Rotor

1 – Rotor 2 interactions. Similarly, the ensemble-

averaged flow angles at the Stator 2 inlet for a single

circumferential position may be similarly split and

overlaid, shown in Fig 5b. Again, the pattern formed

by the Rotor 1 wake interaction with Rotor 2 is

unique for each of the eleven blade pass periods, but

the pattern is repeatable. This indicates that a reduced

computational domain utilizing a third of the rotor

wheel could properly capture the important forcing

function and resulting unsteady vane surface

pressures.

There is a phase shift between the data shown in

Figs. 5a and 5b due to the circumferential and axial

position of the measurements. This is on the order of

two blade pass periods. Therefore, the blade pass

period which has Rotor 1 wakes between Rotor 2

wakes is blade pass period 2 in Fig. 5b. The

corresponding blade pass period in Fig. 5a is blade

pass period 4.

The Rotor 2 exit data showed changes in wake

shapes with respect to the pitchwise position.

Though these differences are noteworthy, the

downstream vane is positioned at one particular

pitchwise location. Therefore, to understand the

impact of the pitchwise variation of inlet conditions

on vane surface pressure response, Stator 2 was

clocked to six different circumferential positions.

Figure 6 shows how the Stator 2 suction surface

response near the leading edge (10%cx) changes for

two clocking configurations (CL2 and CL5).

Although results are only shown for the suction side,

the pressure side exhibited similar trends when

phase-shifted. The two clocking configurations

shown in Fig. 6 contain differences in the Rotor 1 –

Rotor 2 interaction effects just like the two pitchwise

positions highlighted in Fig. 2. The 3/rev modulation

is altered between the two configurations in both

magnitude and phase. The shift in phase of Rotor 1 –

Rotor 2 interactions is apparent in Fig. 6b, which

shows the change in CP for each rotor blade pass

period. Clocking configuration CL2 experiences the

minimum change in CP within a rotor wake passing

for blade numbers 6, 17, and 28 while CL5

experiences this minimum for blade numbers 1, 12,

and 23.

Figure 4: Mid-span Stator 2 surface pressure

at 10%cx (a) ,changes in pressure response (b),

and average pressure (c).

Figure 5: Comparison of sections of the 3/rev

modulation in the Stator 2 surface pressure

coefficient (a) and inlet flow angle (b).

7

7

Figure 6c reveals another difference in the Rotor

1 – Rotor 2 interactions for different clocking

configurations. The mean pressure per Rotor 2 blade

pass period follows the 3/rev pattern and the level

changes by about 4% for clocking configuration CL2.

In contrast, clocking configuration CL5 experiences

changes in mean CP within only 1% through a rotor

revolution. Based on the observations of the rotor

exit flow field, it is concluded that these clocking

effects on the Stator 2 unsteady surface pressures are

not necessarily associated with the Stator 1 wake

interacting with the Stator 2 surface directly, but

rather how the Stator 1 wake affects the Rotor 2 flow

field, which is then driving the unsteady surface

pressure on Stator 2.

Effects of Tip Leakage Flow

An additional unsteady flow feature affecting

Stator 2 surface pressures is the Rotor 2 tip leakage

flow. Similar to the flow angle data at different

positions across the stator pitch presented in Fig. 2 at

mid-span, the pitchwise variation in the Rotor 2 exit

flow field was evaluated in the tip region. To best

visualize the tip leakage flow, a detailed radial

traverse with a high-frequency response total

pressure probe was performed including 26 radial

positions at 20 circumferential positions across a

vane pitch. Figure 7 shows an ensemble-averaged

revolution of the RMS of the pressure coefficient at

the same two pitchwise locations as shown in Fig. 2,

30% and 80%vp. Regions of high RMS indicate

unsteadiness that is not phase-locked to the rotor

revolution and indicates the rotor wake and tip

leakage flow.

Figure 7a shows the Rotor 2 exit flow at 80%vp

where the high unsteadiness caused by the rotor tip

leakage flow extends down into the rotor passage to

approximately 70%span. Alternatively, the flow field

acquired at 30%vp shows that the tip leakage flows

only penetrate down to 85%span. Additionally, the

strength of the unsteadiness in the tip flows is

reduced at 30%vp. These differences in tip leakage

flow characteristic between pitchwise positions are

due to interactions with the upstream Stator 1 wake,

as discussed by previous studies17

. Therefore, the

surface pressures along Stator 2 at 80%span may or

may not be affected by the tip leakage flows, and this

depends on the pitchwise position of Stator 2 relative

to Stator 1, or the clocking configuration.

Figure 6: Mid-span Stator 2 suction surface

pressure response for two clocking

configurations (CL2 and CL5)

Figure 7: Contours of Stator 2 inlet unsteadiness, <CP,RMS> at 80%vp (a) and 30%vp (b)

8

8

Figure 8 shows the difference in Stator 2 surface

pressure near the leading edge of the suction side for

clocking configurations CL3 and CL6 at 80%span.

As before, the modulation of the pressure response is

different between the two clocking configurations,

but unlike the mid-span data, there are extra

disturbances in the pressure signal with each blade

pass period. This is more clear in Figs. 8b and 8c.

Clocking configuration CL3 has the familiar 3/rev

modulation, though it is weak for the change in CP

per blade pass period. However, clocking

configuration CL6 has a strong 6/rev pattern with

large changes in CP with each blade pass period. The

mean pressure for each blade pass period also has the

6/rev pattern for CL6, where there are smaller peaks

within the 3/rev pattern. These extra disturbances are

due to the presence of the rotor tip leakage flow in

the inlet conditions at 80%span for this clocking

configuration.

To clearly define how these fluctuations differ

between clocking configurations, the CP for an

average rotor blade pass period is calculated. The

ensemble-averaged revolution of data is broken into

33 segments for the 33 Rotor 2 blade pass periods

and then those are ensemble-averaged together to

create a mean Rotor 2 blade pass period. The average

pressure coefficient RMS, CP,<RMS>, for a mean Rotor

2 blade pass period is shown in Fig. 9 for the high

loading condition. The Rotor 2 wake is located at

20%bpp. The disturbance near 85%bpp for CL6 is

associated with the rotor tip leakage flows.

The amount by which the Stator 2 surface

pressure unsteadiness is elevated due to the Rotor 2

tip leakage flow changes with each clocking

configuration because as the position of Stator 1

changes, the interactions between the Stator 1 wake

and Rotor 2 tip leakage flow changes with respect to

the fixed Stator 2 position.

The extra disturbance within each rotor blade

pass period changes the frequency content of the

data, shifting it to higher frequencies. The Fourier

transforms of the full ensemble-averaged revolution

of pressure coefficient for these two clocking

configurations were calculated for each of the four

chordwise positions measured along the stator. The

dominant frequencies are blade pass frequencies

(bpf) and their harmonics, specifically 33/rev (R2

bpf), 66/rev (second harmonic of R2 bpf), and 30/rev

(R3 bpf). The magnitude of the Fourier

decomposition at each of these frequencies is shown

in Fig. 10 as a function of axial chord for two

clocking configurations. As expected, the Rotor 2 bpf

is dominant, but there are shifts in the frequency

content with vane clocking and axial position.

Clocking configuration CL6, which contained an

extra disturbance due to the presence of the rotor tip

leakage flow, has stronger magnitudes of frequency

content at both Rotor 2 bpf and its second harmonic.

This difference between the clocking configurations

is particularly strong at the sensor nearest to the

leading edge (10%cx) for the 66/rev frequency. The

second harmonic of Rotor 2 blade pass frequency is

significant because it is an indicator that the rotor tip

leakage flow is acting as a second disturbance within

each blade pass period. This difference in the 66/rev

magnitude between the clocking configurations

decreases along the chord. Also the Rotor 3 bpf

Figure 8: 80%span Stator 2 surface pressure

response near the leading edge for two

clocking configurations (CL3 and CL6)

Figure 9: Mean Stator 2 surface pressure

response near the leading edge for two

clocking configurations at 80%span

9

9

(30/rev) shows no difference between the two

clocking configurations, and the magnitude increases

further downstream where the sensors are in closer

proximity to Rotor 3.

Summary of Rotor-Rotor Interactions and Vane

Clocking Effects

Finally, the changes in average CP per blade

passing event due to Rotor 1 – Rotor 2 interactions

were considered for all six clocking configurations.

The difference between the maximum and minimum

average CP per blade pass period shown in Figs. 6c

and 8c was calculated for each clocking

configuration, and these results are shown in Fig. 11

for 10%cx at 50% and 80%span for nominal and high

loading conditions. These data are shown as a

function of clocking position, and they are shown

twice assuming periodicity. Only the data for the

sensor closest to the leading edge is presented, but

the others exhibit similar trends. A sinusoidal trend

exists in all cases, where all have a minimum at

clocking configuration CL5. This configuration

represents the clocking configuration with the

smallest variations in rotor wakes due to Rotor 1 –

Rotor 2 interactions and thus, the smallest variations

in the unsteady lift experienced by the stator between

different rotor blade pass periods.

Additionally, the effects of vane clocking on

stator surface pressure associated with the Rotor 1 –

Rotor 2 interactions are stronger at 50%span where

vane clocking changes the unsteady surface pressure

due to rotor-rotor interactions by as much as 2.25% at

high loading. The vane clocking effects at 50%span

are more significant at high loading, compared to

those at nominal loading, by about 1%.

Thus, although vane clocking has a significant

impact on the rotor exit flow field in the tip region,

the variations in the vane unsteady pressure between

different rotor blade pass periods is not as significant

as those at mid-span. Furthermore, the differences

between the nominal and high loading conditions are

significantly reduced at 80% span compared to those

shown at 50% span, and for some clocking

configurations, the differences in unsteady pressure

fluctuations are larger at nominal loading. Both of

these results were unexpected and show the

significant variations in unsteady pressure envelopes

along the vane span, thus, justifying the effort to

acquire data at more than one spanwise location.

Conclusions

Stator 2 unsteady surface pressure was

investigated in the Purdue 3-stage axial compressor

using high-frequency response pressure transducers

embedded at 50% and 80%span on the Stator 2

pressure and suction surfaces. Results from six

clocking configurations were presented at part speed

conditions with a focus on two loading conditions:

nominal and high loading. The unsteady Stator 2 inlet

flow was characterized including a discussion of

rotor wake variability due to Rotor 1 – Rotor 2

interactions.

The Rotor 2 wakes vary in width and depth due

to Rotor 1 – Rotor 2 interactions. Additionally,

significant differences in the Rotor 2 wakes were

measured at different pitchwise locations, and this

was associated with differing levels of interaction

between Rotor 2 and the upstream Stator 1 wakes

over different portions of the rotor blade pass period.

The variations in the Rotor 2 wakes with respect to

the mean wake were 34% to 58% depending on the

Figure 10: Stator 2 surface pressure frequency

content for two clocking configurations at

80%span along the vane chord

Figure 11: Effects of Rotor 1 – Rotor 2

interactions and vane clocking on Stator 2

average surface pressure per blade pass period

10

10

circumferential position at which the wakes were

measured.

Clocking was utilized to place the downstream

vane at different pitchwise locations to understand

how the different inlet flow conditions affected the

unsteady vane surface pressure. While the mean

pressure only changed by 4% for the different

clocking locations, the change in unsteady pressure

for one blade pass period was as high as 100% at

mid-span.

At 80% span, the rotor tip leakage flow added an

additional unsteadiness to the Stator 2 surface

pressures. This was not present at all of the clocking

configurations because the interaction of the Stator 1

wake with Rotor 2 causes very different

characteristics of the tip leakage flow for different

pitchwise locations (including radial penetration).

While the tip region had this additional excitation, the

variations in unsteady pressure between the different

rotor blade pass periods were significantly reduced

compared to the results at mid-span.

Blade row interactions are important factors

when considering unsteady flows in compressors.

This paper reveals that the effects from rotor-rotor

interactions can be significantly altered through vane

clocking, and these effects differ with spanwise

location and loading condition. While previous work

has looked at changes in vane surface pressure with

clocking, this research has carefully tied in the flow

physics at the stator inlet that is driving the changes

in vane surface pressure. Rather than the Stator 1

wake interaction with Stator 2 directly, the key driver

for the measured differences is the interaction of the

Stator 1 wake with Rotor 2, which provides

circumferentially varying flow field in the absolute

reference frame. These results can help focus the

direction of improved models for unsteady

aerodynamic excitations that can be used in forced

response and aeroelastic solvers.

Acknowledgements

The authors are grateful to Rolls-Royce for

granting permission to publish this work. The GUIde

IV Consortium funded portions of this research, and

the authors are grateful for this support. Efforts of

William L. Murray III throughout instrumentation

setup and data acquisition are also much appreciated.

References 1. Kemp, N.H. and Sears, W.R., 1955, “The Unsteady

Forces Due to Viscous Wakes in Turbomachines,” J.

Aeronaut. Sci., Vol. 22 no.7, pp. 478–483.

2. Meyer, R.X., 1958, “The Effect of Wakes on the

Transient Pressure and Velocity Distributions in

Turbomachines,” Trans. ASME, Vol. 80, pp. 1544–

1552.

3. Lefcort, M.D., 1965, “An Investigation Into Unsteady

Blade Forces in Turbomachines,” ASME J. Eng. Power,

Vol. 87, pp. 345–354.

4. Kerrebrock, J.L. and Mikolajczak, A.A., 1970, “Intra-

Stator Transport of Rotor Wakes and Its Effect on

Compressor Performance,” J. of Engineering for

Power, Vol. 92, pp. 359-368.

5. Smith, L.H., 1966, “Wake Dispersion in

Turbomachines,” J. of Basic Engineering, Vol. 88, pp.

688-690.

6. Deregel, P. and Tan, C.S., 1996, “Impact of Rotor

Wakes on Steady-State Axial Compressor

Performance,” ASME Paper 96-GT-253.

7. VanZante D.E., Strazisar A.J., Wood J.R., Hathaway

M.D., and Okiishi T.H., 2000, “Recommendations for

Achieving Accurate Numerical Simulation of Tip

Clearance Flows in Transonic Compressor Rotors,” J.

of Turbomachinery, Vol. 122, pp.733-742.

8. Sanders, A. J. and Fleeter, S., 2001, “Multi-Blade Row

Interactions in a Transonic Axial Compressor, Part II:

Rotor Wake Forcing Function & Stator Unsteady

Aerodynamic Response,” ASME Paper No. 2001-GT-

0269.

9. Sherman, P.J., Dudley, R., and Suarez, M., 1996, “The

Stochastic Structure of Downstream Pressure from an

Axial Compressor – II. An Investigation of Blade-to-

Blade Variability,” Mechanical Systems and Signal

Processing, Vol. 10 No. 4, pp. 423-437.

10. Boyd, D.M. and Fleeter, S., 2003, “Axial Compressor

Blade-to-Blade Unsteady Aerodynamic Variability,” J.

of Propulsion and Power, 19(2).

11. Key, N.L., Lawless, P.B., and Fleeter, S., 2010, “Rotor

Wake Variability in a Multistage Compressor,” J. of

Propulsion and Power, Vol. 26, No. 2, pp. 344-352.

12. Sanders, A.J. and Fleeter, S., 2002, “Rotor Blade-to-

Blade Wake Variability and Its Effect on Downstream

Vane Response,” J. of Propulsion and Power, Vol. 18,

pp. 456-464.

13. Wisler, D.C., 1985, “Loss Reduction in Axial-Flow

Compressors Through Low-Speed Model Testing,” J.

Engineering for Gas Turbines and Power, 107(2), pp.

354-363.

14. Day, I., 1993, “Stall Inspection in Axial Flow

Compressors,” J. of Turbomachinery, Vol. 115, pp. 1-9.

15. Mailach, R., Lehmann, I., and Vogeler, K., 2008,

“Periodic Unsteady Flow Within a Rotor Blade Row of

an Axial Compressor – Part II: Wake-Tip Clearance

Vortex Interaction,” J. of Turbomachinery, 130, pp.

041005-1 – 041005-10.

16. Krug, A., Busse, P., and Vogeler, K., 2014,

“Experimental Investigation of the Steady Wake-Tip

Clearance Vortex Interaction in a Compressor

Cascade,” J. of Turbomachinery, Accepted Manuscript,

TURBO-14-1244.

17. Smith, N.R., Murray III, W.L., and Key, N.L., 2015,

“Considerations for Measuring Compressor

Aerodynamics Excitations Including Rotor Wakes and

Tip Leakage Flows,” ASME Paper GT2015-43508,

Presented at the ASME Turbo Expo Montréal, Canada,

June 15-19, 2015.

18. Saren, V.E., Savin, N.M., Dorney, D.J., and Zacharias,

R.M., 1997, “Experimental and Numerical

11

11

Investigation of Unsteady Rotor-Stator Interaction on

Axial Compressor Stage (With IGV) Performance,”

Unsteady Aerodynamics and Aeroelasticity of

Turbomachines, Proceedings of the Eighth International

Symposium, Stockholm, Sweden, pp. 407-424.

19. Dorney, D.J., Sharma, O.P., and Gundy-Burlet, K.L.,

1998, “Physics of Airfoil Clocking in a High-Speed

Axial Compressor,” ASME Paper 98-GT-82.

20. Griffin, L.W., Huber, F.W., and Sharma, O.P., 1996,

“Performance Improvement Through Indexing of

Turbine Airfoils: Part 2 – Numerical Simulation,”

ASME J. of Turbomachinery, Vol. 118, pp. 636-642.

21. Brossman, J.R., Ball, P.R., Smith, N.R., Methel, J.C.,

and Key, N.L., 2014, "Sensitivity of Multistage

Compressor Performance to Inlet Boundary

Conditions," J. of Propulsion and Power, Vol. No. 2,

pp.407-415.

22. Murray III, W.L., 2014, “Experimental Investigation of

a Forced Response Condition in a Multistage

Compressor,” MS Thesis, Aeronautics and

Astronautics, Purdue University, West Lafayette.