Noise of Turbocharger Compressors

Hua Chen

To cite this version:

Hua Chen. Noise of Turbocharger Compressors. 17th International Symposium on Transport Phe-
nomena and Dynamics of Rotating Machinery (ISROMAC2017), Dec 2017, Maui, United States.
�hal-02376818�

HAL Id: hal-02376818

https://hal.archives-ouvertes.fr/hal-02376818
Submitted on 22 Nov 2019

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.
 Noise of Turbocharger Compressors

Hua Chen
Marine Engineering College, Dalian Maritime University
1 Linghai Road, Dalian, China 116026
[email protected]

ABSTRACT  Density (kg/m3)

 Torque (N-m)
Noise produced by turbocharger compressors is becoming a Ω Angular speed of rotor (rad./s), rotor rps (1/s)
major concern of turbocharging industry. But confusion ex-  Flow coefficient based on impeller outlet blade
ists regarding the sources of the noise. This paper discusses speed
this noise from an aerodynamicist's point of view. Blade
passing frequency noise and flow instability related noise are Subscript
in particularly examined. Focus is on how these two noises 0 Stagnation or total state
are generated and how they may be controlled at source 1 Impeller inlet
level. The aerodynamic causes of the noises are described 2 Impeller exit
and identified, and measures to reduce them at compressor R Rotation
design stage are proposed. r Radial direction
RI Rotational instability
KEYWORDS u Tangential direction
z Axial direction
noise, turbocharger, compressor, instability
Superscripts
NOMENCLATURE ~ Root square mean

A Flow passage area (m2) 1. INTRODUCTION

B Blade number
BPF Blade passing frequency (Hz) Noise produced by turbochargers is a concern for their users.
F Force (N) With higher charging pressures for downsized internal com-
f Frequency (Hz) bustion engines, the noise is becoming louder and there is a
f* dimensionless frequency = f /rps of rotor growing demand to control it. Compared with other aspects
h Specific enthalpy (J/kg) of turbochargers and with aero engines, the noise of turbo-
L A characteristic length of impeller, eq. (1) chargers draws little attention until recently. Feld et al. (1)
LE Leading edge described the design and set up of an acoustic test facility for
m Circumferential wave number large turbocharger compressors for marine applications. In-
• fluences of inlet distortion and vaned diffuser to blade pass-
m Mass flow rate (kg/s)
ing frequency (BPF) noise were studied. Clay and Moch (2)
P Pressure (N/m2) reported the development of an acoustic test chamber for
P.S. Pressure side small turbochargers. Rämmal and Åbom (3) discussed sound
RI Rotating instability generating mechanisms of turbocharger compressors and
r Radius (m) described a model for evaluating the passive acoustic effect
rps Revolution per second (1/s) of turbocharger turbines to engine exhaust pulses. Brand et
SPL Sound pressure level (dB) al. (4) presented their work of acoustic decouplings to
S.S. Suction side against turbocharger noise being transferred to engine ex-
T Time period of vortex shedding (s) haust system. Schweizer and Sievert (5) examined the noise
TLV (LE) Tip leakage vortex generated by rotor-bearing-shaft system. Tanna et al (6)
U Impeller blade peripheral (tip) speed (m/s) portrayed their work on reducing the BPF noise of turbo-
URANS Unsteady Reynolds averaged Navier-Stoke eqs. charger compressors employing a ported shroud. Lee et al (7)
V Absolute velocity (m/s) conducted an experimental study of the noise radiated by a
turbocharger. A turbocharger test rig was mounted on an
Greek Symbols anechoic chamber, and noise was measured with a free field
 Angle of relative velocity, measured from meridi- microphone placed at 0.8 m from the inlet of the compressor.
onal direction (rad.) A narrow band noise was found at a frequency about 3 times
0 Half of the shear layer thickness, see Fig. 9(a) of rotor rev. and was the dominant frequency in most config-

1
urations. This was attributed to the experimental setup. characteristics by modern mufflers and after-treatment de-
Sheng (8) gave a presentation on turbocharger noises identi- vices, turbine noise is less of a concern. The compressor on
fying a number of them. Kabral et al. (9) proposed an acous- the other hand is a flow diffusion device and subject to flow
tic method for investigating turbocharger flow instability in instability. The instability generates noise. For these reasons,
which acoustic features of the sound generation and scatting this paper discusses compressor noise from aerodynamic
by the compressor near surge were investigated. The results point of view. The focus will be the underline mechanism of
show that broadband noise at near surge condition is consid- the noise, that is, how various noises are generated by the
erably higher than at other conditions, and a notable hump compressor. The measures to control these noises will also
centred at 45% shaft speed was found and considered to be be addressed.
rotating stall related. Karim et al. (10) presented their study
of 'hiss/whoosh' or broad noise of a turbocharger compressor 2. BLADE PASSING FREQUENCY (BPF) NOISE
at off-design conditions, and identified the best configuration
of inlet guide vanes and grooves/steps in front of the impel- This noise is mainly caused by blade loading. The noise
ler. Dehner R et al. (11) tested a turbocharger compressor caused by blade thickness should be small and will not be
for instability at the low-flow range using acoustic signals. discussed here. The pressure difference (loading) on the two
They recognized the link between surges and incidence angle surfaces of a blade generates a pressure disturbance when
at impeller inlet and others. Surge noise identification and the blade rotates. This disturbance propagates as sound
control was discussed by Kuang X. et al. (12). Åbom and wave, its frequency is equal to blade number x blade rotating
Rämmal (13) reported their development of a compact si- frequency, and its amplitude increases with the pressure
lencer for noise at KHz range. Broatch et al. (14) carried out difference. Figures 1 and 2 show the measured compressor
experimental and computational studies of the broad band operating condition and casing blade loading of a 400mm,
noise of a turbocharger compressor at a peak pressure ratio 30o backsweep centrifugal impeller operated at a subsonic tip
of 2.24:1. Biet and Baar (15) described an acoustic climate Mach number of 0.6. The blade loading (and the BPF noise)
camber for acoustic measurement of turbochargers at cold reduces and shifts toward impeller LE with compressor mass
environmental conditions. Gupta et al. (16) investigated the flow, from operating points (A) to (D). When flow instability
methods to reduce turbocharger whistle noise transmitted occurs, operating point (G), the loading, particularly the front
into passenger cabin. Identification of the noise sources, aero part, greatly reduces. The change of the loading with mass
or mechanical, was mentioned. Galindo et al. (17) studied flow is related to the change of incidence angle to the blade.
the effects of real running shroud clearance of a small turbo-
charger compressor on its acoustic behavior in near surge
condition, taking into account the effects of rotation, thermal
expansion and shaft motion. No significant changes in the
behavior were found. It was suggested that in this working
condition, the tip clearance is immersed in a region of
strongly swirling flow, therefore cannot establish any coher-
ent noise source mechanism.

CR: Clearance Ratio

Majority of the noise generated by turbochargers are flow
= Clearance / diffuser gap
related. This asks for a good understanding of the fluid dy-
namics side of this type of noise, which is currently unsatis-
factory. Often the source of a noise is misidentified, or simply
described with general term of 'instability' or 'rotating stall'.
Of the turbine and compressor in a turbocharger, more nois-
es are generated and transmitted by the compressor. Be-
cause of the flow acceleration nature of the turbine, thicker Fig. 1 Head coefficients of a centrifugal compressor
and heavier turbine casing, and the relatively good damping with two shroud clearances. (18)

2
 Fig. 2 Casing measurement deducted blade loading at operating points (A), (B), (D) and (G) of Fig. 1
Clearance ratio CR = 12.7%, solid and broken lines for full and splitter blades respectively. (18)

Figures 3 gives the calculated blade loading of a small

turbocharger compressor working at supersonic flow
condition. A striking feature of it is the sharp pressure
LE spike by unsmooth blade
spike at the LE. This is due to LE bow shock and the
curvature & LE shock
discontinuity of the blade surface curvature in the LE
region. The LE shock is responsible for the 'buzz-saw'
type noise commonly detected in front of transonic
compressors and fans (19). The shock strength may
be reduced by using a sharp and thin LE. The LE spike
can also be reduced by using curvature smooth blades
(20). Both LE spike and passage shocks can cause flow
to separate, the link between flow separation and noise
will be discussed later.
Pressure rise due to
Figure 4 presents calculated static pressure distribu-
passage shocks
tion at inlet duct of a small turbocharger compressor.
The impeller has six main and six splitter blades.
Twelve circumferential waves are visible, and the influ-
Fig. 3 Shroud pressure loading of a turbocharger
ence of the sixth harmonic of rotating frequency is cut-
compressor at 547m/s tip speed & peak efficien-
off rapidly.
cy point by CFD

3
 man ears, changes in U2 may have different effects on
human perception to the BPF noise. Vr2/U2 > 0 is im-
peller exit flow coefficient. 2 is the impeller exit flow
angle and is approximately equal to impeller exit blade
angle, commonly referred to as backsweep angle since
2 <0 in turbocharger compressors. An increase of the
backsweep angle (2 becomes more negative) reduces
blade loading, particularly at high flows, see Figure 6,
hence decreases BPF noise. Large backsweep angle
also increases the stagger of the blades, resulting in
Fig. 4 Instantaneous pressure wave pattern by higher blocking by the blades to sound propagation
URANS at inlet duct of a 6+6 bladed turbocharger downstream.
compressor operating at inducer tip speed of
401.5m/s and near choke

Figure 5 gives calculated SPL at the compressor inlet

and outlet. The inlet is dominated by high order har-
monics of full blade passing (6 x rotating speed), par-
ticularly even number modes 2, 4 and 6. The outlet on
the other hand is dominated by the fundamental mode
of full blade passing. According to the flow field meas- Inlet
urement by Roduner et al. (21), the high order har-
monics are damped by vaneless diffuser.

URANS is not the most accurate CFD method for sound

source calculation. More sophisticated methods such as
Large Eddy Simulation (LES) and Detached Eddy Simu-
lations (DES) are more accurate but require more
computational power. CFD is usually used only in
source computing, sound transmission can be calculat-
ed by simpler methods.

The overall blade force is connected directly with com-

pressor power. Using the Euler's turbomachinery eq., Harmonics of 6 x rotating speed
the power of a compressor is
• • •
  m(h 02 h01 )  mU 2Vu 2  U 1Vu1   mU 2Vu 2
 Vr 2  2
  1 A1Vz1 1  tan  2 U 2
 U2 
Outlet
The total blade force F and the force on single blade
F/B, where B is blade number, are

Vz1  Vr 2  2
F   / L   /( L)  1 A1 1  tan  2 U 2
U2  U2 
Vz1  Vr 2  2
F / B  1 A1 1  tan  2 U 2 / B (1)
U2  U2 
Where L is a characteristic length, and may be approx-
imated by the impeller tip radius. So, the force creat-
ing the BPF noise increases with compressor size A1,
compressor inlet flow coefficient Vz1/U2 and blade tip
Harmonics of 6 x rotating speed
speed U2. U2 also links to the BPF, so depending on
where the BPF is on the sensitivity curves of the hu-

4
 Fig. 5 Calculated pressure wave amplitude and strong tonal and broadband noises are reported when
SPL dB (in blue) at inlet and outlet of the com- a compressor is closed to surge. At this condition, the
pressor in Fig. 4 compressor as a flow diffusion device is working at its
maximum capacity. They are several noise generating
mechanisms by flow instability inside compressors, and
they are discussed here. We restrict ourselves to the
open shroud impellers with shroud clearance and
vaneless diffusers. Figure 8 shows different compres-
sor instabilities on the map of such a compressor and
the associated casing pressure signals taken close to
impeller LE.

Fig. 6 Effects of impeller blade exit angle and exit

flow coefficient on stage loading coefficient

Blade number is very important to the efficiency and

BPF noise of small turbocharger compressors. Viscous
friction loss is high in such compressors, and reducing
the blade number often brings efficiency benefit. This,
however, needs to be balanced with BPF noise consid-
eration. F/B, the force per blade, increases with reduc-
tion of blade number B. Use of full bladed impellers is
an effective way to reduce BPF noise and solve this
conflict. Not only the single inducer blade loading is
reduced and BPF moves up to the less sensitive range
of human ears, but also the damping to the BPF noise
is increased, because the damping is proportional to e-
Bx
below sound wave cut-off. Figure 7 demonstrates
these effects. In compressor design stage, more blades
may be achieved with shorter blades to maintain blade
'picth-chord' ratio so that aero performance would not
be affected. Mechanical properties and manufacturing
capability of the impeller may also be kept.

3. NOISE DUE TO FLOW INSTABILITY

A turbocharger compressor usually composes of three

components: a centrifugal impeller, a diffuser and a
volute housing. The impeller and the diffuser deceler-
ate flow and are subject to certain diffusion limits be-
fore the flows separate. The volute housing can both Figure 7 Measured on engine SPL level of BPF
decelerate and accelerate flow depending on its mass noise at exit of two turbocharger compressors.
flow rate: if the rate is higher than the one it was de- The large inducer blade number led to a signifi-
signed for, the flow will accelerate in it; if the rate is cant reduction of BPF noise and a up-shift of its
lower than the design value, the flow will diffuse in it frequency
and may eventually stall. Flow separations are un-
steady and companied with vortices generation. Often

5
Fig. 8 Performance map of a turbocharger compressor with a vaneless diffuser, various instability regions
are identified with associated casing static pressure signals taken in front of impeller LE. (22)

Deep surge (the red line and beyond in the figure)

under which the compressor mass flow rate can be- Starting from the lowest speed-line in Figure 8, a
come negative will not be discussed because this con- noise of 2 x rotor rotating frequency was observed.
dition is to be avoided in compressor applications. To This noise occurs before the more familiar rotating stall,
the right of the pre-stall/surge points, the compressor and maybe caused by a Kelvin-Helmholtz instability
is stable as a system. These points are also the peak recently reported by Bousquet et al (23). Kelvin-
pressure ratio points: to their right, the speed-lines or Helmholtz instability refers to an instability of shear
pressure rise curves have negative slopes and are sta- layers. The large velocity gradient across a shear layer
ble; to their left, the slope of the speed-lines becomes makes it unstable. In centrifugal impellers, increased
positive and the speed-lines become unstable, because incidence and loading near LE when compressor mass
if a small disturbance reduces compressor mass flow, flow rate is reduced generate a LE tip leakage flow
the compressor pressure ratio will drop as well, and from blade p.s. to the s.s. This leakage flow is of high
this drop diminishes the ability of the compressor to vorticity and moves with a different speed from those
deliver mass downstream, and compressor mass flow of surrounding fluids, forming a unstable shear layer,
will further decreases, resulting a positive feedback Figure 9a). Vortices are then produced by the shear
cycle of mass flow reduction. Mild surge is defined as layer, migrate downstream and breakup under in-
the surge without global mass flow reversal. It is a creased back pressure, Figure 9b). Like vortices
resonance of compression system under instability. In shedding, this process produces noise. In this com-
compressor applications it is also better avoided but its pressor, CFD predicted a two-lobe (circumferential
mechanism is well understood, and the noise it created wave number m = 2) wave with a dominating frequen-
is of low frequencies as shown in Figure 8. So, we cy about 6 times of rotor rotating frequency, Figure
focus on other types of instability. 9c).

3.1Kelvin-Helmholtz instability related

6
 Vorticity

T = 4.50/a, 0 is shown in (a)

a) Time averaged magnitude of vorticity at 90% pan b) Vortices (circled) formation, migration and breakup

Due to K-H
instability

c) Pressure amplitude and phase at 8 numerical probes in 8 inducer channels. fR is impeller rotating frequency
Fig. 9 URANS calculation of a centrifugal compressor operating near stall (NS) showing vortices generated
by Kelvin-Helmholtz instability. The compressor has a 21 vanes diffuser (23).

When the TLV eventually reaches or is near the LE of

3.2 Subsonic inducer rotating stall related the adjacent blade, as sketched in Figure 11, the in-
ducer loading of this blade is destroyed, compare Fig-
When the mass flow of the compressor in Figure 8 ure 2(d) and Figure 2(g), causing this blade to stall
further reduces, the TLV becomes stronger, larger and and block the flow to its passage, and relieving the
moves in more tangential direction due to increased LE previously stalled blade. In travelling to the next blade,
blade loading and higher back pressure. Figure 10 breakup of the TLV slows it down. And because of im-
displays high speed camera images of TLV (LE tip peller rotation, the stalled blade channels or stall cells
leakage vortex) at two flow coefficients of a subsonic will rotate in the same direction of impeller in the ab-
axial compressor. The smaller flow corresponds to pre- solute frame of reference, but in slower speeds. This
stall or near rotating stall condition. The vortex core generates a sub-synchronous noise in the rotating fre-
enlargement, which signals vortex breakdown, and the quency of the stall cells which is a fraction of impeller
vortex core movement towards tangential direction are rotating frequency. If number of stall cells is m, then a
visible under the reduced mass flow.

7
noise with a frequency of m x rotating frequency of unstable component of the compression system. Its
stall cells is also observed. stall often triggers the stage stall and surge (25-27,
see Figure 15 as well). Because its inflow contains a
large circumferential velocity component, the diffuser
stall often occurs in the form of circumferential waves
whose frequencies are linked to shaft speed. Vaneless
diffuser can also have radial wave modes. Different
number of stall cells may be formed at different oper-
ating conditions and with dissimilar geometries, result-
ing various noise frequencies being observed. Figures
12 and 13 give examples of one- and two-cells rotat-
ing stalls of vaneless diffusers in two industrial com-
pressors respectively. Note the strong disturbance
caused by lobe passage in Figure 13. Three or four
lobes vaneless diffuser rotating stalls were also often
observed. Figures 14 and 15 show an example.

Fig. 10 High speed images of the cavitating tip

leakage vortex of a subsonic axial compressor at
two flow coefficients  = 0.35 and 0.25 and three
time instances. Compressor is near stall at the
smaller  value. (24)

(a) Pressure at diffuser inlet

Fig. 11 Sketch of trajectories of TLV (broken red

line) and overtip leakage flow (solid red line from
5 to 3) in rotating stall of a helicopter compres-
sor, based on casing pressure signals from trans-
ducers 1 - 9. (25)

3.3 Diffuser stall related

While impeller inducer may stall, but it rarely triggers (b) FFT results of pressure signal
compressor surge directly. With the centrifugal force Fig. 12 Single lobed (m =1) stall of vaneless dif-
contribution to the overall pressure rise, centrifugal fuser in an industrial centrifugal compressor.
impeller's speed-lines do not have large positive slopes Blade Tip Mach no. = 0.85. (28)
even in rotating stall conditions. A vaneless diffuser on
the other hand always decelerates flow and is the most

8
 Time averaged wall pressure
vs. flow coefficient

Blade Tip Mach No. = 1.1

A: measured at 1.05 x Impeller tip radius

D: measured at 1.64 x Impeller tip radius

Pressure Spectra at A at
(a) Pressure at diffuser inlet (1) to (5) operating points

(b) FFT results of pressure signal

Fig. 13 Two lobed (m =2) stall of vaneless diffus-
er in an industrial centrifugal compressor. Blade
Tip Mach no. = 0.5. (28) Numbers in the brackets indicate stall cell number and
rotating speed of the cells divided by impeller speed
Fig. 15 Measured wall pressure spectra of a tur-
bocharger compressor at five operating points at
vaneless diffuser inlet (location A). (29)

Rotating speed of stall cells is typically < 50% of im-

peller rotating speed, but the lobe passage speed can
be higher if lobe number is >1. If the stall pattern is
nearly symmetric, only the lobe passage may be
picked up by FFT and real stall cells rotation may be-
come invisible. Figure 16 presents such a case.

Fig. 14 Measured mid span flow velocities in Chen and Sheng et al (30-31) proposed a rotating
vaneless diffuser at operating condition 5 (see stall model for vaneless diffusers that allows estimation
Fig. 15) of a turbocharger compressor, showing of the rotating speed and the number of the stall cells.
three stalled regions or stall cells. (29) They show that working with backsweep impellers
vaneless diffusers can have two stalls at different flows,
that is, in between the two stalls there is a stable re-
gion. This may explain why in Figure 8 and in many

9
turbocharger compressor maps there exists a negative sage between blades 1 and 2 is stalled, the main pas-
slope region to the left of the first positive slope region sage shock on blade 1 s.s. causes TLV and overtip vor-
on the speed-lines. tices to breakdown in shroud region, blocking the in-
flow to the passage. The blockage to the flow has
reached the LE of blade 2 p.s., causing higher inci-
dence to the blade 2. The larger incidence and the ad-
ditional mass flow trying to get into the passage
formed by blades 2 and 3 increase the Mach number
on the s.s. of blade 2 thereby the shock strength in
this passage, leading to severer flow breakdown and
blockage in it (bottom picture of the figure) and re-
lieving the previously stalled passage.

(a) Pressure at diffuser inlet

Fig. 17 URANS calculated relative Mach number

at 92% span of a centrifugal compressor (9+9
bladed) at onset of inducer rotating stall at two
time instances. Bottom picture was taken 1/8
rotor rev. after the top one. (32)

The pressure signal of Figure 17 was analysed by FFT

and shows distinct tonal feature peaked at about 5
times of rotor frequency, Figure 18. The rotating stall
frequency itself is about 87% of the rotor frequency.
The five times noise is likely caused by interaction be-
tween the blade passing and rotating instability, see
next section.

(b) FFT results of pressure signal Am/Am(f*=0)

Fig. 16 nearly symmetric bi-lobed stall of
vaneless diffuser in an industrial compressor.
Blade Tip Mach no. = 0.65. (28)

3.4 Supersonic inducer stall related

At high compressor speeds, the stall mechanism of

impeller inducers is slightly different from that at low
compressor speeds, and shocks play an important role
f*
in inducer rotating stall. Near the stall, the LE shock is
Fig. 18 Frequency Spectra of pressure signal by
detached from the blade under increased back pres-
numerical probes located in vaneless diffuser at
sure, and the flow at the LE is subsonic, making the
r/r2 = 1.2 and 98% span on relative frame of ref-
propagation of flow information in tangential direction
erence with three different downstream throttle
and rotating stall possible. Figure 17 depicts the in-
valve cases. Case 2 corresponds to Fig. 17. (32)
ducer rotating stall inception of a compressor with
vaneless diffuser. In the top picture of the figure, pas-

10
In addition to the noise generated by this rotating stall, properly, can be and have been used to reduce the
the flow separation introduced by shocks is highly un- severity of the mixing and broad band noise.
steady and causes the shocks foot at the boundary
layer to oscillate. The oscillation generate a narrow
band noise whose energy peaks at Strouhal number
Str = li x f/V∞ = 0.01~0.03 (33), where f is the noise
frequency, li is the distance shock foot oscillates and
V∞ the free stream velocity before the shock. For tur-
bocharger compressors, the frequency of this noise is
estimated in KHz to 10 KHz order. The strength of the
noise is to be studied.

3.5 Scatter by rotor blades

The noise signals produced by compressor instability

may get scattered by the rotation of the impeller
blades when they propagate. The circumferential mode
m at frequency fm is scattered into modes mR&RI at fre-
quencies fR&RI,

mR&RI = i*B – j*m;

fR&RI = i*B*R – j*fm ; (2)
i, j = …, -2, -1, 0, 1, 2, …

where B is rotor blade no. and R the rotating frequen-

cy of the rotor. This scattering adds to the frequency
spectra of compressors in stall and mild surge (during
mild surge cycles, stalls happens intermittently, see for
example (22)). Figure 19 shows measured sound
filed near the LE of an axial subsonic fan at the onset
Fig. 19 Frequency spectrum (top) and decom-
of rotating instability (RI). The fan rotor has 24 blades.
posed circumferential mode amplitudes (bottom)
The dominating modes of RI is 21 and 22. i = j = 1 in
eq. (2) corresponds to BPF - RI in Figure 19. of a subsonic axial fan showing scattering effect
of rotor blades. (34)
3.6 Broad band noise

Broad band noise is associated with turbulence. vorti-

ces are generated when compressors work in stall and
mild surge conditions. These large flow structures will
breakdown under the adverse pressure gradient into
smaller flow structures and eventually into turbulences
as previously shown in Figures 9 and 10. Figure 20
further demonstrates the mechanism of vortices
breakdown where a simulated breaking down of a large
vortex in compressor discharge section into smaller
vortices and turbulences at instabilities of a turbo-
charger compressor is shown. Another source of broad
band noise is the mixing of the incoming flow and the
reversed flow at compressor inlet in stall and mild
(a) Peak pressure ratio, single large vortex
surge conditions (see for example (35-36)). Casing
grooves, inlet vanes and inlet silencers, designed

11
 ter than asymmetric ones in noise aspect. How to link
this observation to compressor design is to be explored.

(b) Just in stall, two smaller vortices

(c) Deep into stall, more small vortices

Fig. 20 Calculated flow structure at discharge
section of a turbocharger compressor under vari-
ous operating conditions. (26)

3.7 Reduce noise due to instability

The best method to reduce the noise produced by

compressor instability is to make compressor stable. Fig. 21 Measured 500-4500Hz SPL (dB) maps
Delay diffuser stall or make the stall less severe can of a turbocharger compressor with elliptical
reduce the noise. One technique is to increase the tur- and blunt TEs.
bulence of impeller exit flow so that the diffuser
boundary layer is less likely to separate. One way of Increase the flow area under the volute tongue can
doing it is to use blunt impeller TE, instead of sharp make flow more axisymmetric and improve compres-
one. The blunt TE sheds vortices downstream energis- sor stability. Ported shroud is a very effective way to
ing the diffuser flow. Figure 21 shows the results. This improve the stability of centrifugal compressors, it also
method will slightly reduce compressor efficiency and reduces the strength of inducer passage shock (37).
as the figure indicates, the noise level at the choke end These two functions combined, ported shroud can
is slightly increased. Another technique to make the greatly reduce noises due to flow instability and shocks.
diffuser more stable is to reduce the diffusion in the See Figure 22. Ported shroud may increases BPF
diffuser, such as using a small diffuser exit-to-inlet noise in inlet pipe due to the enlarged casing front and
diameter ratio. The loss of diffusion in the diffuser interaction of port supporting ribs with BPF noise. It
maybe compensated by using a larger volute, at least may also increase broad band noise with the flow sep-
for low pressure ratio applications. Diffuser width and arations inside the port. A silencer and other measures
meridional contour can also be employed to control the (37) can solve the problem.
diffuser instability. An examination of Figures 12, 13
and 16 suggests that nearly symmetric stalls are bet-

12
 Ported
Reduced tonal noise
dB

Engine rpm
Frequency (KHz)
Fig. 22 Measured exhaust noise of a ported
shroud compressor during engine transient. The
compressor has 11 full blades. Non ported case is
given in Fig. 7.

Impeller design has significant effects on compressor

stability and associated noise. The aim is to have
negative slopes for impeller speed-lines. High back-
sweep angles and small shroud clearance should be
used where possible. For high speed machines, the
Mach number on blade s.s. should be restricted and
the diffusion through the impeller controlled. 10~20o
blade tip extension at shroud (mixed flow type tip)
reduces the diffusion along the shroud and improves
stability. Figure 23 shows noise maps of three turbo-
charger compressors. Design 1 is a radial type design (a) Kreischen noise (2000-2800Hz)
and Designs 2 and 3 are of mixed flow tip. Both show
improved noise characters to Design 1. Design 3 uses
forward lean and full blades to further enhance stability,
and displays the best noise character.

13
 5. ACKNOWLEGEMENTS
The author is grateful to his employer NLETT for allow-
ing the publication of this paper.

6. REFERENCES

1. Feld H., Aschenbrenner A. and Girsberger R.,

(2001), Investigation of acoustic phenomena at
the inlet and the outlet of a centrifugal compressor
for pressure ratio 4.5. ASME Turbo Expo, 2001-
GT-0314.
2. Clay D. and Moch S. W., (2002), Development of a
new test facility for evaluation of turbocharger
noise emissions. Proc. IMechE 7th Int. Conf. Turbo-
chargers & Turbocharging, C602/032/2002. Lon-
don.
3. Rämmal H. and Åbom M., (2007), Acoustic of tur-
bochargers. SAE paper 2007-01-2205.
4. Brand J., Fallen M. and Kammer H., (2008), Fu-
ture Technologies against turbocharger noise
transferred to exhaust systems. SAE paper 2008-
01-0891.
5. Schweizer B. and Sievert M., (2009), Oscillations
of Automotive Turbocharger Turbine. J. Sound and
Vibration, Vol. 321, pp. 955-975.
6. Tanna R. P. et al., (2010), Development of ported
shroud compressor housing with reduced com-
pressor blade pass source acoustic level. IMechE
Ninth International Conference on Turbochargers
and Turbocharging, C001/008/2010, pp115-124,
London.
7. Lee Y., Lee D., So Y. and Chung D., (2011), Con-
(b) Hiss (800-1800Hz) noise trol of Airflow Noise From Diesel Engine Turbo-
Fig. 23 Measured noise (dB) maps of three small charger. SAE paper 2011-01-0933.
turbocharger compressors with different design 8. Sheng X., (2012), 涡轮增压器噪声和振动. 贯彻落实
philosophies ‘国四’排放标准技术交流会专题报告. (Turbocharger
noises and vibrations), 17-18，May 2012，山东潍坊.
4. CONCLUSION 9. Kabral R., Rammal H. and Abom M., (2013),
Acoustic methods for investigating turbocharger
Factors affecting the BPF noise of compressors are flow instability. SAE paper 2013-01-1879.
discussed. Size, flow coefficient, speed, backsweep 10. Karim A., Miazgowicz K., Lizotte B. and Zouani A.,
angle and blade number etc. are shown to be involved. (2013), Computational aero-acoustics simulation
Use of full bladed impellers is demonstrated to be an of compressor whoosh noise in automotive turbo-
effective measure to reduce the BPF noise. chargers. SAE paper 2013-01-1880.
11. Dehner R. et al., (2013), Instability at the low-flow
Both tonal and broad band noises by flow instabilities range of a turbocharger compressor. SAE paper
are examined for their aerodynamic sources. These 2013-01-1886.
include Kelvin-Helmholtz instability, vortex breaking 12. Kuang X. et al., (2014), Surge noise identification
down, subsonic and supersonic inducer rotating stalls, and control of automotive turbochargers. SAE pa-
vaneless diffuser rotating stall, stall in volute discharge per 2014-01-2053.
section, shocks boundary layer interaction and blade 13. Åbom M. and Rämmal H., (2014), Turbocharger
scattering effect. To reduce the noises, compressors noise - Generation and Control. SAE paper 2014-
should avoid positive slopes on their speed-lines. Vari- 36-0802.
ous measures to improve compressor acoustic charac- 14. Broatch A., Galindo J., Navarro R. and Garcí a-
ter, such as energising the diffuser flow, ported shroud, Tíscar J., (2014), Methodology for experimental
and impeller design philosophy such as mixed flow tip validation of a CFD model for predicting noise
and forward lean etc, are illustrated.

14
 generation in centrifugal compressors. Intl. J. Heat 29. Ohuchida S. and Tamaki H., (2013), Experimental
and Fluid Flow, 50(2014) 134-144. study of rotating stall in vaneless diffuser of a cen-
15. Biet C. and Baar R., (2015), Turbocharger Test trifugal compressor. ASME Turbo Expo, GT2013-
Bench Extension for Acoustic Measurements at 95468.
Cold Environment Conditions. SAE paper 2015-01- 30. Chen H. et al, (2012), A 3-D compressible flow
1672. model for weak rotating waves in vaneless diffus-
16. Gupta K, Vikram M. and Manta E., (2015), Study ers, Parts I & II. ASME J. Turbo. Vol. 134: 89–97
of Turbocharger Whistle Noise and Its Reduction & 98-105.
Into Passenger Cabin. SAE paper 2015-26-0129. 31. Sheng F., Chen H., Zhu X. and Du Z. (2012), A
17. Galindo J., Tiseira A., Navarro R. and Lprz M., three-dimensional compressible flow model for ro-
(2015), Influence of tip clearance on flow behavior tating waves in vaneless diffusers with unparallel
and noise generation of centrifugal compressors in walls. IMechE J. Mech. Eng. Sci., 226(9) 2230–
near-surge conditions. Intl. J. Heat and Fluid Flow, 2249.
52(2015) 129-139. 32. Vagnoli, S. and Verstraete T., (2014), Numerical
18. Schleer M., Song S. and Abhari R., (2008), Clear- investigation of inlet distortion on the stall incep-
ance Effects on the Onset of Instability in a Cen- tion of a radial compressor. ASME Turbo Expo,
trifugal Compressor. ASME J. Turbo., July 2008, GT2014-25516.
Vol. 130 / 031002-1. 33. Clemens N. and Narayanaswamy V., (2014), Low-
19. Cumpsty N., (1989), Compressor Aerodynamics, frequency unsteadiness of shock wave/turbulent
Longman Group Ltd., Pearson Education Ltd, Har- boundary layer interaction. Annu. Rev. Fluid Mech.
low, Essex, U.K. ISBN 0-582-01364-X. (There is 2014, 46: 469-92.
also a 2004 version with new Preface) 34. Pardowitz B., Tapken U., Neuhaus L. and Enghardt
20. Bi Q., Chen H., Tong D. and Lu Y., (2016), A de- L., (2014), Experiments on an axial fan stage:
sign method for globally curvature-smooth centrif- time resolved analysis of rotating instability modes.
ugal compressor blades. ASME Turbo expo, ASME Turbo Expo, GT2014-26323.
GT2016-56416. 35. Tamaki H., (2015), Effect of Piping System on
21. Roduner C., Kupferschmied P., Köppel P. and Testing of Centrifugal Compressors for Turbo-
Gyarmathy G., (2000), On the Development and chargers. Turbocharging Seminar 2015, Sept.
Application of the Fast-response Aerodynamic 2015, Tianjin, China.
Probe System in Turbomachines—Part 2: Flow, 36. Gancedo M., Guillou E. and Gutmark E., (2013),
Surge, and Stall in a Centrifugal Compressor. Experimental investigation of flow instability in a
ASME J. Turbo., July 2000, Vol. 122, pp 517-526. turbocharger ported shroud compressor. ASME
22. Zheng X. and Liu A., (2015), Phenomenon and Turbo Expo, GT2013-95134.
mechanism of two-regime-surge in a centrifugal 37. Chen H. and Lee V., (2013), Casing Treatment and
compressor, ASME J. Turbo., Vol. 137, 081007-1. Inlet Swirl of Centrifugal Compressors. ASME J.
23. Bousquet Y., Carbonneau X. and Trebinjac I., Turbo. July 2013, Vol. 135 / 041010.
(2015), Numerical investigation of Kelvin- 38. Greitzer E., Tan C. and Graf M., (2004), Internal
Helmholtz instability in a centrifugal compressor flow. Cambridge University Press. ISBN 0 521
operating near stall. ASME Turbo Expo, GT2015- 34393 3.
42495. 39. LandauL. and Lifshitz E., (1987), Fluid Mechanics,
24. Tan D. et al, (2014), Visualization and time re- 2nd Ed., Pergamon Press, Oxford, England.
solved PIV measurements of the flow in the tip re-
gion of a subsonic compressor rotor. ASME TE,
GT2014-27195.
25. Buffaz N. and Trébinjac I., (2012), Impact of tip
clearance size and rotating speed on the surge on-
set in a high pressure centrifugal compressor.
ASME Turbo Expo, GT2012-68427.
26. Chen H. et al. (2008), Numerical simulations of
onset of volute stall inside a centrifugal compres-
sor. ASME Turbo Expo, GT2008-50036.
27. Pullan G. et al, (2012), Origin and structure of
spike-type rotating stall. ASME Turbo Expo,
GT2012-68707.
28. Bianchini A. et al, (2014), Some guidelines for
experimental characterisation of vaneless diffuser
rotating stall in stages of industrial centrifugal
compressors. ASME Turbo Expo, GT2014-26401.

15