View PDF - Conference Proceedings

Three Park Avenue, New York, N.Y. 100164990
The Society shall not be responsible for statements or opinions advanced In papers or discussion at meetings of the Society or of its Division:for
Sections, or printed in Its publications. Discussion is printed only if the paper is published In an ASME Journal. Authorization to photocopy
for internal or personal use is granted to libraries and other users registered with the Copyright Clearance Center (CCC) provided
S3/article is paid to CCC, 222 Rosewood Dr., Danvers, MA 01923. Requests for special permission or bulk reproduction should be addressed to the ASME Technical Publishing Department.
Copyright 0 1999 by ASME
Rights Reserved
Printed in (ISA
N. Karamanis, R.F. Martinez-Botas and C.C. Su
Department of Mechanical Engineering
Imperial College of Science, Technology & Medicine
A detailed flow investigation downstream of two mixed-flow
turbocharger turbines has been carried out at 50% and 70% design
speeds, equivalent to 29,400 and 41,300 rpm respectively. The
measurement technique used was laser Doppler velocimetly (LDV).
The measurements were performed at a plane 9.5 mm behind the
rotor trailing edge, they were resolved in a blade-to-blade sense to
fully examine the nature of the flow. The results confirmed the
performance tests and indicated the improved performance of the
rotor with a constant inlet blade angle relative to the rotor with a
nominally constant incidence angle.
There is a wide application of turbochargers in the automotive
industry in order to achieve improved engine power output and
efficiency, and reduction of exhaust emissions. The approach in the
present work is to improve the efficiency characteristics of the
turbocharger turbine as a means to make better use of the engine
exhaust gas energy. The geometry chosen for this study is of a mixedflow nature. This type of rotor design has been investigated for many
years both experimentally and theoretically, most of these studies
have focused on the design, manufacture, and performance evaluation
under steady and pulsating flow conditions (Abidat et al., 1992;
Arcoumanis et al., 1995; Baines et al., 1979; Chen et al..1996; Chou
and Gibbs, 1989; Wallace and Pasha, 1972), whereas the information
of the velocity distribution at this type of rotor is rare. Hence the aim
of the present investigation is to explore the exit flow characteristics
of mixed flow turbines using a laser Doppler velocimeter system.
Arcoumanis et al. (1997) presented circumferentially averaged
exit LDV measurements; this approach does not reveal the detailed
nature of the exit flow, hence it is difficult to point out specific areas
of design improvement. The work presented in this paper resolves
the flow in a blade-to-blade sense, thus revealing the detailed flow
structure. The flow conditions at the rotor exit are strongly influenced
by the passing blade. The designer's aim is to ensure that the flow
leaves the rotor with the lowest possible level of mixing loss and
swirl; the diffuser performance could otherwise be compromised. The
flow in the vicinity close to the trailing edge of the blade is threedimensional and complex due to the periodical nature correlated to
the passing blade. It is therefore, particularly difficult to obtain
measurements in this region of the flow.
There is little experimental data at the exit of radial-inflow
turbines. Yea and Baines (1990) and Baines and Yea (1991)
measured the flow field characteristics at the rotor exit of a twinentry vaneless radial - inflow turbine using a L2F velocimeter. The
measurements were conducted at an axial distance of 30 nun
downstream of the blade trailing edge. The results showed that the
flow velocity and flow angle profiles were symmetric about the
absolute flow velocity
Saunter mean diameter
mass flow rate
plexiglass tube inner radius
radial distance
blade speed
relative flow velocity
axial distance from the blade trailing edge
absolute flow angle
relative flow angle
blade angle
deviation angle
0 total value
turbine inlet
e turbine exit
m meridional direction perpendicular to the trailing edge
0 tangential direction
is isentropic
Presented at the International Gas Turbine & Aeroengine Congress & Exhibition
Indianapolis, Indiana — June 7-June 10, 1999
Downloaded From: on 12/22/2014 Terms of Use:
The experimental results suggested a correlation between the exit
flow angle and the turbine performance; a reduction in exit flow
angle resulted in an increase in the overall turbine efficiency.
Given the lack of experimental data on the flow behaviour of
mixed-flow turbines, the aim of the present study is to observe the
periodicity of the exit flow characteristics of mixed-flow turbines
with the rotor revolution, using a LDV system. This study will allow
an improved understanding of the complicated rotor exit flow field,
and its correlation with swirl and exit kinetic energy loss. Besides,
the measured results can provide accurate validation for CFD models
of the turbine flow.
turbine axis. At design point the flow velocity was almost constant
and the flow angle was minimum indicating small exit swirl. While
away from the design condition, the flow velocity became higher near
the rotor hub and the flow angle increased significantly towards the
sidewall of the exit duct, indicating large swirl occurred in the wall
Zaidi and Elder (1993) reported the flow measurements of a
radial-inflow turbine using a L2F velocimeter. The measurements
were conducted at two different axial locations of 50 mm and 100
trim from the rotor trailing edge. The experimental results showed a
similar swirling flow pattern at both locations. This exit flow was
characterised by a highly turbulent flow region behind the hub end
where no velocity was obtained, a centre annulus region of uniform
flow velocity and flow direction and an outer flow region with a
similar flow direction to that in the centre annulus region, but the
flow velocity which increased rapidly towards the outer wall. At both
measured positions, it was found that the turbulent flow region
appeared to be a little larger at 50 mm position than that at 100 mm
position because the wake produced by the rotor hub decayed further
Benisek (1994) measured the flow velocity with a L2F
velocimeter and multi-port cobra probes at the rotor exit regions of a
radial-inflow turbine. The measurements were made at the rotor exit
of 5 ram and 42 mm downstream of the rotor blade. The results of
both measured planes showed that the agreement between the laser
and probe measured flow velocity and angle was poor. This was
because that the pneumatic probe was not capable of obtaining
accurate measurements in the highly turbulent flow region. The
investigations had demonstrated that conventional pneumatic probes
were not suitable for measuring the flow at the rotor exit which was
strongly swirling and highly unsteady.
Murugan et al. (1996) performed a three-dimensional flow field
investigation in the exit region of a radial-inflow turbine using a
LDV system at the well off design speed of 1,000 rpm. The flow
velocities were measured at three cross-sectional planes (A, B and C)
of 2.54, 7.62 and 15.24 nun downstream of the rotor exit. The
tangential velocities at the three cross sections showed that the
degree of swirl was higher near the tip region and the levels of the
tangential velocities reduced in the downstream direction. The radial
velocities showed that there was a general radially inward movement
of the flow due to the loss of centrifugal force as the flow left the
rotor. At the downstream cross-section C, it was noted that there
were some slightly outward radial velocities along the mid-passage
near the rotor hub due to the wake behind the hub-end. The axial
velocities were higher along the suction surface than those along the
pressure surface at the first cross-section. Then they became mostly
uniform at downstream cross-sections.
Arcoumanis et al. (1997) first employed a WV system to
explore the flow field characteristics in the exit region of two mixedflow turbine rotors (B with a nominally constant incidence angle and
C with a constant inlet blade angle). The measurements of the axial
and tangential velocity components were conducted at three different
axial planes of 9.5, 31.5 and $0 mm from the trailing edge of the
blades. The circumferentially averaged results showed that the rotor
B generated higher tangential velocity than rotor C at all locations
and for both rotational speeds. It was also noted that the exit flow
angles were reduced with increasing rotational speeds. In addition,
the flow angles of rotor C were considerably lower than those of rotor
B which may explain the improved steady performance of rotor C.
The turbocharger facility (Arcoumanis et al., 1997) consists of
the research turbine, an air supply system, a power absorber in the
form of a centrifugal compressor and a data acquisition system. The
turbine rotors tested here are of the mixed-flow type; a photo and a
schematic diagram illustrating this type of rotor geometry can be seen
in Fig. 1. Table 1 gives the geometric characteristics of the two
rotors, one shorter in length with a nominally constant incidence
angle (rotor B) and the other with a constant inlet blade angle (rotor
A). The volute feeding the rotor is single-entry, non-symmetric and
nozfteless. The steady-state performance was evaluated by means of
the energy balance method (Arcoumanis et al., 1997), where the
turbine actual output power is estimated by measuring the power
absorbed by the loading device (centrifugal compressor) and the heat
discharged to the bearing lubricating oil.
Table 1: Mixed-flow rotors A and B geometry
Rotor type
Tip mean diameter (mm)
Inlet blade height (rnm)
Number of blades
Exducer tip diameter (mm)
Rotor length (mm)
Exit mean blade angle
Inlet blade angle
The laser Doppler velocimetry system comprised an Argon-ion
laser (Spectra Physics) operating at a wavelength of 0.514 pm and
power of up to I W, an optical unit dividing the laser beam into two
of equal intensity and bringing the two beams to an intersection
volume, a photomultiplier, and a frequency counter (TSI model 1995)
interfaced to a microcomputer. The intersection volume was
approximately 1,160 pm in length and 64.0 gm in diameter, with a
fringe spacing of 4.65 inn.
It should be emphasised that the measured parameters are
correlated to the rotor blade rotation, to observe any periodic nature
of the flow. Therefore, the results presented here are blade-to-blade
gated measured quantities with a resolution of 1 ° over an angular
displacement of 30° within the blades. The digital output of the
frequency counter was fed into a microcomputer via a DOSTEK
DMA interface card and sample of 40,000 data was recorded over
many turbine rotations and then resolved to nearly 1,000 data per
Downloaded From: on 12/22/2014 Terms of Use:
For measuring the flow at the rotor exit, a plexiglass tube was
added and the radial distribution of the axial and tangential velocities
was measured at a distance of 9.5 mm from the blade trailing edge,
as shown in Fig. 1, with a near-forward scattering angle of 60 0 to the
optical axis.
A possible source of uncertainty in LDV measurements is the
size of the seeding droplets and their ability to follow the flow
fluctuations. The silicone oil droplets were generated by an air-blast
atomiser and added to the flow in the divergent section upstream of
the volute. The atomiser has been shown to produce droplets with
Sauter mean diameter, dp of up to 2 tun which corresponds to a
maximum effective Stokes number of 0.1. This implies that the
droplets, on average, follow the flow fluctuations although occasional
larger droplets may occur, giving rise to an uncertainty due to their
size which can be considered to be negligible.
Other uncertainties such as those due to velocity gradients were
minimised by aligning the smaller measuring volume across the
velocity gradient and collecting the light off the optical axis. The
statistical uncertainty was estimated to be less than 1.8% and 4.4%
with a 95% confidence level, in the mean and rms value of the
velocity, respectively. The maximum counting ambiguity of the TSI
counter having a clock frequency of 1000 MHz was 1.1% of the
Doppler frequency shift. The whole optical system was mounted on a
3-dimensional traversing mechanism with a positional uncertainty of
0.025 mm. The axial velocity component turbulence intensity as a
function of the rotor position (time) and the radius (space), exhibits
values, between 0.14 - 0.2 dose to the mid-span at the exit of rotor A
that remain the same with rotational speed. Rotor B demonstrates
values between 0.12-0.23, and 0.12 - 0.2 for 50% and 70% of the
equivalent design speed respectively.
( la )
49.5 rum
l--- Plexiglas Tube
_ -111011■1•■101EW
R = 39.8 mm
Steady flow performance
The steady-state performance of rotors A and B was investigated
at five different rotational speeds covering the range between 50%
and 90% equivalent design speeds. The total-to-static efficiency
characteristics and the swallowing capacity of the two mixed-flow
turbine rotors, are presented in Figs. 2 -3.
Rotor A exhibits a pattern of increasing efficiency with
rotational speed up to the 80% equivalent design speed after which
the efficiency remains nearly constant. Similarly rotor B follows the
same trend with an increase in efficiency up to 70% equivalent
design speed. The efficiency curves of rotor B become flatter than
rotor A at speeds between 80% and 90% design speed.
The swallowing capacity of rotor B is higher than that of rotor A
(Fig. 3), however, the latter has the best overall performance across
the whole range of speeds as shown in Fig. 2. The peak efficiencies
for rotor A were higher than rotor B by 2.76 percentage points at the
50% design speed and by 2.96 points at the 90% design speed.
Rotor A, at high velocity ratios, demonstrates more compact
efficiency curves than rotor B, a feature that implies a reduced
sensitivity to the rotational speed with rotor A. This is expected to
improve the transient performance of the turbine since a change in
the rotational speed will produce a small change in efficiency levels.
Arcounumis et al. (1995) indicated that the mixed-flow turbine tested
had an improved unsteady cycle-averaged efficiency when compared
to a radial inflow turbine.
Tota l-to-static efficiency
Fig. I (a) Photograph of the mixed flow turbine rotor B, and (b)
Schematic of the laser measurement location (dashed line) at the
exit of the mixed-flow rotor.
o 50% Design speed
a 60% Design speed
+ 70°4 Design speed
X 80% Design speed
o 90% Design speed
Velocity ratio ( / C,)
( 2a )
Downloaded From: on 12/22/2014 Terms of Use:
Tota l-to-s ta t ic e ffic ie ncy
Velocity measurements
Measurements of the axial and tangential mean and rms velocity
components at the exit of rotors A and B were obtained at the peak
efficiency condition of the 50% and 70% equivalent design speeds
corresponding to 29,400 and 41,300 rpm, respectively (see Table 2).
0 50% Designc4.cserre
a 60% Design speed
+ 70% Design speed
x 80% Design speed
*90% Design speed
Rotor Design
Speed (%)
Table 2: LDV Test Conditions
Velocity ratio ( U / Cs
50% Design speed
A 609'. Design speed
+ 70°,'. Design speed
x 80% Design Speed
o 90% Design speed
Pressure ratio ( P o , / P.
o 50% Design speed
a 60% Design speed
+ 70% Design speed
x 80% Design speed
o 90% Design speed
Poi / Pe
m (T01)°3/ Fel
Given that there was no appreciable efficiency increase beyond
the 70% design speed point, it was decided to limit the LDV data to
this maximum speed. These are taken at an axial plane of x = 9.5
mm from the trailing edge of the blades. The radial distance is
normalised with the exit pipe inner radius, R. The ensemble values
of the mean axial, as well as, the absolute and relative tangential
velocity components are shown in Figs. 4 - 6.
At the exit of both rotors A and B the velocity distribution
shows a significant variation with radius, as well as, angular position.
Further more, at the location where the axial component is relatively
high the tangential component remains low and the vice versa; this
pattern extends throughout the whole angular range within the blade
passage. The tangential velocities (Fig. 5) exiting close to the
pressure surface are higher compared to those near the suction
surface, both rotors tend to follow a similar flow pattern. This may be
attributed to the pressure field acting between the pressure and
suction surfaces, which drives the flow to come into equilibrium
behind the trailing edge of the blade (Murugan et al., 1996). On the
other hand, the axial velocity contour plots (Fig. 4), exhibit higher
values along the suction rather than the pressure surface.
In more detail, the tangential component magnitude is high close
to the pressure surface, with a non-uniform pattern showing a double
peak in magnitude one peak at a radial location near the tip and the
other close to the mid-span. This phenomenon suggests the presence
of a trailing edge vortex, mainly attributed to the effect of the passing
blade. Further more, low values of axial velocity near the shroud,
suggest a swirling flow pattern, near the pressure surface. At the
mid-passage, the tangential development follows a forced vortex
assumption (as radius increases, there is an increase in the tangential
velocity and a decrease in the axial velocity, assuming negligible
radial flow); the double peak is no longer present hence, the tip
vortex shows a negligible effect at this angular position. Finally, the
flow near the suction surface becomes more axial, hence less
swirling, exhibiting a local peak of magnitude of the axial velocity
component near the mid-span. Both rotors follow the same flow
pattern, rotor B however demonstrates higher values of tangential
component magnitude at the local peak regions suggesting higher
levels of swirl near the pressure surface.
The relative to the blade tangential velocity component We
development (Fig. 6), shows a decrease in value from shroud to hub,
with rotor A exhibiting a smoother gradient than that of rotor B. In
Fig.2 Efficiency characteristics of mixed flow turbine
(a) rotor A and (b) rotor B.
Pressure ratio ( Po, / P. /
( 3b )
Fig.) Swallowing capacity of mixed flow turbine
(a) rotor A and (b) rotor B.
Downloaded From: on 12/22/2014 Terms of Use:
1 95
Fig. 4 Exit axial velocity C. (m/s) contours at x = 9.5 mm of the mixed-flow rotor A for the (a) 50% and (b) 70%,
and of the mixed flow rotor B for the (c) 50% and (d) 70% of the equivalent design speed.
close to zero values imply a more axial flow. The level of swirling
flow near the pressure surface is higher than that near the suction
surface, where the flow exhibits a local peak of positive absolute
angle. This phenomenon is more pronounced for rotor B, confirming
previous comments.
The deviation flow angle 8 = p—o, , is an indication of the level
of flow guidance in the blades, where the relative flow angle is
defined as 13 = tan .' [(U-00)/C.) and 0*, is the design blade camber
line angle. The closer to zero the value of the deviation angle, the
lower the departure of flow from the blade camber line is. Hence,
while positive values of deviation angle imply an under-turned flow,
negative values imply an over-turned flow.
general the contour is rather uniform with angular position, except in
the mid-span between the pressure surface and the mid-passage,
where rotor B shows a non-uniformity with low values of velocity, a
phenomenon that may be related to the tip vortex. On the contrary, a
local peak of magnitude is exhibited at the shroud region close to the
suction surface. An increase in speed from 50% to 70% extinguishes
this regional non-uniformity, as the gradient becomes smoother with
speed, for both of the rotors.
Flow angles are key velocity-diagram parameters because they
link the axial and swirl (tangential) velocity components. The
computed flow angle contours are shown in Figs. 7 - 8. The absolute
flow angle a = tan -I (C0/C,„), is an indication of the level of swirl in
the flow. Positive values of absolute angle imply a more swirling,
Downloaded From: on 12/22/2014 Terms of Use:
. 10
r- -
r 60
1'7 55
I-7 5°
Fig. 5 Exit tangential velocity Co (m/s) at x 9.5 mm of mixed-flow rotor A for the (a) SO% and (b) 70%
and of the mixed flow rotor B for the (c) 50% and (d) 70% of the equivalent design speed.
Rotor B exhibits a non-uniform distribution of deviation angle
with a concentration of high negative value between the pressure
surface and the mid-passage, close to the mid-span. Moreover, the
swirl of the flow near the pressure surface is higher than that near the
suction surface, where the flow exhibits a local peak of positive
deviation angle. The lack of flow guidance becomes less obvious with
an increase in rotation speed from 50% to 70%, as the local peaks of
deviation angle are reduced.
Rotor A, on the other hand, demonstrates a smoother deviation
angle contour plot than that of rotor B, with almost zero deviation
values. The flow angle development seems to be smoother as the
rotational speed increases from 50% to 70%. The results also
indicate, that the flow is better guided at the exit of rotor A with
almost zero deviation, in comparison to the negative deviation at the
exit of rotor B.
A clear difference between swirling flows obtained with the two
rotors is that, the tangential mean velocity is consistently lower for
rotor A, especially at the forced vortex region near the mid-passage.
Since both rotors have equal blade counts and equal exit dimensions,
the increase in chord length of the latter translates to an increase in
the solidity of the rotor. Therefore, the reason for the lower swirl
velocities with almost zero deviation of rotor A, should be attributed
to its increased solidity.
It is well known, that the performance of a turbine depends on
the exit flow angle. As the exit swirl velocities become smaller for a
given rotational speed, the exit absolute flow angle will decrease and
Downloaded From: on 12/22/2014 Terms of Use:
i 160
rl 150
Rotor hub
• We
— 16
E l 160
au 150
ri 130
I ' ll
Fig. 6 Exit relative tangential velocity. We (this) at x = 9.5 mm of mixed-flow rotor A for the (a) 50%
and (b) 70% and of the mixed flow rotor B for the (c) 50% and (d) 70% of the equivalent design speed.
the mid-passage, as radius increases, there is an increase in the
tangential and a decrease in the axial velocity components, so that the
flow follows a forced vortex assumption. The flow becomes less
swirling in the proximity of the suction surface of the rotor blade.
The tip vortex exhibited in the vicinity of the exit duct wall, as
an interaction of the tip clearance flow and the boundary layer
development on the duct wall, occurs at the 50% of the design speed,
it becomes less significant however at the 70% of the design
operating conditions.
The results revealed that the shorter rotor B generated higher
swirl velocities rather than the longer rotor A. The reason for the
lower swirl velocities with almost zero deviation of the former,
should be attributed to its increased solidity.
the flow will become more axial. The reduced loss of kinetic energy
can thus be expected to lead to an increase in efficiency, as long as
there is no significant increase in the internal loss generation which
is controlled by the relative exit velocity. The results also indicate,
that the flow at the exit of rotor A is less swirling than that of rotor
B, which may also explain the less efficient performance of the latter.
The LDV gated measurements reveal a complex flow pattern at
the exit of both rotors (A of constant inlet blade angle and B of
constant incidence), especially at a radial location between the
shroud and the mid-span, attributed to the passing blade effect At
Downloaded From: on 12/22/2014 Terms of Use:
• 40
- 35
Fig. 7 Exit absolute flow angle a (degrees) contours at x = 9.5 mm of mixed-flow rotor A for the (a) 50% and (b) 70%,
and of rotor B for the (c) 50% and (d) 70% of the equivalent design speed.
Turbine and Aeroengine Congress and Exposition, Houston, Texas,
June 5-8, Paper 95-01-210.
Arcounaanis, C., Martinez-Botas, R. F., Noun, J. M., and Su, C.
C., 1997, "Performance and exit flow characteristics of mixed-flow
turbines", International Journal of Rotating Machinery, Vol. 3, No. 4,
The total-to-static efficiency parameter is specifically designed
to penalise a turbine for exhaust kinetic energy, and thus it is rather
obvious that for similar mass flow through the turbine a machine
with exit swirl will suffer. Therefore, the relatively high levels of exit
swirl of rotor B contribute to its less efficient performance.
pp. 277-293.
Baines, N. C., Wallace, F. J., and Whitfield, A., 1979,
Abidat, M., Chen, H., Baines, N. C., and Firth, M. R., 1992,
"Computer aided design of mixed flow turbines for turbocharger",
Trans. of the ASME, Journal of Engineering for Power, Vol. 101,
pp. 440-449.
Babies, N. C. and Yeo, J. H., 1991, "Flow in a radial turbine
under equal and partial admission", IMechE Conference on
Turbomachinety, Paper C4231002.
'Design of a highly loaded mixed flow turbine", Proc. lnstn. Mech.
Engrs., Part A: Journal of Power and Energy, Vol. 206, pp. 95-107.
Arcoumanis, C., Hakean, I., Khezzar, L., Martinez-Botas, R F.,
and Baines, N. C., 1995, "Performance of a mixed flow turbocharger
turbine under pulsating flow conditions", ASME International Gas
Downloaded From: on 12/22/2014 Terms of Use:
Fig. 8 Exit deviation flow angle 8 (degrees) contour at x = 9.5 mm of mixed-flow rotor A for the (a) 50% and (b) 70%,
and of rotor B for the (c) 50% and (d) 70% of the equivalent design speed.
Murugan, D. M., Tabalcoff, W., and Hamad, A., 1996, "Threedimensional flow field measurements using WV in the exit region of
a radial inflow turbine", Experiments in Fluids, Vol. 21, pp. 1-10.
Wallace, F. J. and Pasha, S. G. A., 1972, "Design, construction
and testing of a mixed-flow turbine", The 2nd International JSME
Symposium on Fluid Machinery and Fluids, Tokyo, pp. 213-224.
Yeo, J. H. and Baines, N. C., 1990, "Pulsating flow behaviour in
a twin-entry vaneless radial-inflow turbine", IMechE Conference on
Turbocharging and Turbochargers, Paper C4051004.
Zaidi, S. H. and Elder, R. L., 1993, "Investigation of flow in a
radial turbine using laser anemometry, ASME, Ins, Gas Turbine and
Aeroengine Congress and Exhibition, Cincinnati, Ohio, May 24-27,
ASME paper 93-01-55.
Benisek, E., 1994, "Comparisons of laser measurements and
pneumatic measurements in a turbocharger turbine", 1MechE
Conference on Turbocharging and Turbochargers, Paper
Chen, H., Hakeem, I., and Martinez-Botas, R. F., 1996,
"Modelling of a turbocharger turbine under pulsating inlet
conditions", Proc. Instn. Mech. Engrs., Pan A: Journal of Power and
Energy. Vol. 210, pp. 397-408.
Chou, C. and Gibbs, C. A., 1989, "The design and testing of a
mixed-flow turbine for turbochargers", SAE Paper 890644.
Dale, A. and Watson, N., 1986, "Vaneless radial turbocharger
turbine performance", IMechE Conference on Turbocharging and
Turbochargers, Paper CI10/86.
Downloaded From: on 12/22/2014 Terms of Use: