JOURNAL OF PROPULSION AND POWER

Vol. 33, No. 5, September–October 2017

Potential of Variable-Geometry Method for Compressor Range

Extension for Turbocharged Engines

Qiangqiang Huang∗ and Xinqian Zheng†

State Key Laboratory of Automotive Safety and Energy, Tsinghua University, 100084 Beijing,
People’s Republic of China
DOI: 10.2514/1.B36004
Increasingly stringent requirements on fuel economy and emissions are propelling turbocharging technology to
improve the power density of engines. In the future, turbocharged engines with ultrahigh-power density must be
equipped with high-pressure compressors. However, the narrow stable operating range of a compressor at a high-
pressure ratio is always a restriction. The variable-geometry method, which refers to the combination of a variable-
inlet prewhirl and variable diffuser vanes in this paper, will be a preferred choice for the range extension of
compressors, and so estimating its potential for range extension is of long-term value. This paper investigated the
Downloaded by TSINGHUA UNIVERSITY on October 11, 2017 | http://arc.aiaa.org | DOI: 10.2514/1.B36004

performance of a centrifugal compressor adopting the variable-geometry method via a steady three-dimensional
Reynolds-averaged Navier–Stokes simulation. The combination of variable diffuser vanes, ranging from −10 to
10 deg, and a variable-inlet prewhirl, ranging from −20 to 60 deg, has the potential to improve the stable operating
range from 23.5 to 63.0% at a pressure ratio of 4.8. The corresponding increase in the low-end engine torque is
estimated to be 53%. The combination shows advantages in terms of operating range and efficiency performance over
only adjusting the diffuser vanes or simply changing the inlet prewhirl. The contributions from the variable-inlet
prewhirl and variable diffuser vanes to the shifts of the surge line and choke line are discussed as well.

Nomenclature Subscripts
A = throat area c = choke
Bd;x = blockage factor at the diffuser throat, in which x denotes d = diffuser
the throat and d denotes the diffuser i = impeller
c = absolute velocity s = surge or engine shaft
cp = specific heat capacity at constant pressure u = circumferential component
D = diameter w = working
D2 = diameter at the impeller outlet x = position of throat
Hu = fuel low heating value z = axial component
i = incidence p












0 = total parameter
Mu2 = tip-speed Mach number, U2 ∕ γRT 01 1 = impeller inlet
m_ = mass-flow rate of air 2 = rotor/stator interface
m_ d;c = diffuser choking mass flow
m_ i;c = impeller choking mass flow
N = rotational speed I. Introduction
Pet = engine effective power output
p
R
=
=
pressure
stable operating range
T HE variable-geometry method, which aims at helping a
compressor operate at an operating point with the desired
performance by adjusting the geometric settings, has been widely
T = temperature
adopted by many compression applications. For gas-turbine
T tq = torque
compressors or industrial compressors, which usually work
u = impeller blade-tip velocity
continuously for long times, the variable-geometry method can
W = relative velocity
ensure a relatively high-efficiency operation, even under off-design
w = angular velocity
conditions or inlet fluctuations. For aeroengine compressors, which
α = prewhirl angle or air–fuel ratio
often operate in complex environments or under transient conditions,
γ = specific-heat ratio
such as startup and shutdown, the method is often used to suppress
ηet = effective engine efficiency
flow instabilities, such as surge or rotating stall, and to enhance the
ηi = impeller adiabatic efficiency, which is calculated from
stable operating range. In other words, the variable-geometry method
the inlet to the rotor/stator interface
makes a compressor operate efficiently over a wide range.
ρ = density
In axial compressors, the variable-geometry method generally
ϕ = _
flow coefficient, m∕ρ 1 ND2 
3
involves variable-inlet guide vanes (VIGVs) and variable stator vanes
ψ = pressure-rise coefficient, ΔP∕ρ1 N 2 D22 
[1]. In centrifugal compressors with vaned diffusers, the variable-
geometry method is usually implemented using VIGVs and variable
Received 13 September 2015; revision received 19 November 2016; diffuser vanes [2]. Depending on the operating conditions, the
accepted for publication 6 May 2017; published online 27 July 2017. stagger angles of these vanes are adjusted by mechanical linkages
Copyright © 2017 by the American Institute of Aeronautics and Astronautics, outside the pressure-containing body. The gaps between the impeller
Inc. All rights reserved. All requests for copying and permission to reprint and diffuser vanes can also be adjusted [3,4].
should be submitted to CCC at www.copyright.com; employ the ISSN 0748- Turbocharger compressors, which are usually centrifugal, play a
4658 (print) or 1533-3876 (online) to initiate your request. See also AIAA significant role in developing more powerful, cleaner, and more
Rights and Permissions www.aiaa.org/randp.
*Master Candidate, Department of Automotive Engineering, Turbomachinery economical internal combustion engines. For an engine working at
Laboratory; [email protected]. high altitudes, the compressor delivering a high-pressure ratio (PR) is
†
Associate Professor, Department of Automotive Engineering, able to maintain constant inlet manifold pressure and ensure efficient
Turbomachinery Laboratory; [email protected] (Corresponding combustion in the engine [5]. For diesel engines, which are heavily
Author). restricted by increasingly stringent emission regulations, the high-
1197
 1198 HUANG AND ZHENG

pressure intake that accompanies high exhaust-gas recirculation can important when evaluating the necessity of employing the variable-
reduce NOx and soot emissions simultaneously [6]. For the gasoline geometry method. The differences between the variable-inlet
engine, by adopting new technology, such as the Miller cycle to prewhirl and variable diffuser vanes, and their combination for the
improve fuel economy, high-PR compressors are employed to purpose of range extension have not been reported comprehensively.
compensate the volumetric efficiency [7]. However, the operating A comparison over a wide speed range and vane-angle range will be
range of a compressor is limited by the surge line, especially at high valuable to a decision maker when constructing a technology
PRs. With a left shift of the surge line or an extension of the operating road map.
range, the low-end torque of an engine can also be improved. With the The potential for range extension through a variable-inlet prewhirl
development of turbocharging technology, an important issue that has been studied in [19]. If an inlet prewhirl manages to greatly
remains to be solved is the extension of the compressor change the performance map, what is the mechanism for the change?
operating range. Previous research answered this question and explained how the
The variable-geometry method can be an effective way to widen positive inlet prewhirl shifted the surge line and choke line. However,
the operating range of a turbocharger compressor. Especially for a a variable-inlet prewhirl may not always influence the performance
compressor with a vaned diffuser, which usually has an efficiency map significantly. After combining a variable-inlet prewhirl with
advantage over that with a vaneless diffuser, the variable-geometry variable diffuser vanes, whether the potential will be enhanced
method can make up for the shortcoming of a reduced operating greatly or only slightly is still a question. We can look at this issue in
range. In the future of high boosting technology, the compressor that another way by considering the effectiveness of a variable-inlet
uses the variable-geometry method will be a preferred choice for prewhirl for compressors with different diffuser vane angles.
automotive industries. As mentioned earlier, VIGVs and variable Actually, it turns out that, if a variable-inlet prewhirl is applied after
Downloaded by TSINGHUA UNIVERSITY on October 11, 2017 | http://arc.aiaa.org | DOI: 10.2514/1.B36004

diffuser vanes are the representatives of the variable-geometry closing the diffuser, the contribution from the variable-inlet prewhirl
method for turbocharger compressors. to the range extension is relatively limited. Thus, in this paper, some
Mohtar et al. [8] and Herbst et al. [9] validated the range-extension discussion about a variable-inlet prewhirl on when the effects of the
effect of VIGVs on a turbocharger compressor through experiments variable-inlet prewhirl will be significant is given.
and simulations, respectively. In Herbst et al.’s work, a vane angle of In this paper, we focus on the variable-geometry method, which is
20 deg leads to a reduction in surge mass flow of up to 21%. the combination of a variable-inlet prewhirl and variable diffuser
Coppinger [10] and Mohseni et al. [11] investigated the influence of vanes. The main body consists of three parts. First, the numerical
the configurations of the inlet guide vanes (IGVs) and flow passage. It methods are stated and validated. In the second part, the
was found that the performances with different configurations were performances of a centrifugal compressor with different degrees of
quite different. Actually, the objective of using VIGVs is just to prewhirl and diffuser vanes are shown, and then the contributions
implement a variable-inlet prewhirl, which means that the from the variable-inlet prewhirl and variable diffuser vanes to the
circumferential component of the velocity of the inlet flow is shifts of the surge line and choke line are discussed. Third, the
nonzero and can be adjusted. Inlet radial vanes [12,13], reinjecting corresponding improvement in the low-end torque of the engine from
rotating gas [14], and inlet volute [15] can also be used to achieve a applying the variable-geometry method is estimated.
variable-inlet prewhirl. To avoid the influence of the specific type and
configuration of the swirl generator, this paper exploits the inlet II. Flow Modeling and Simulation Procedure
boundary condition of a velocity direction to implement a variable-
A. Investigated Compressor and Meshing
inlet prewhirl.
Oatway and Harp [16] presented a comprehensive review on the The compressor-geometry model is shown in Fig. 1a. It consists of
flow-control method for the range extension of turbocharger a vaned diffuser and an impeller with splitters. Each vane can be
compressors, and their experimental results indicated that, between rotated around the middle point of the centerline by 10 or −10 deg.
the 110 and 45% diffuser-throat areas, a flow range (the operating Table 1 shows the detailed configuration of the compressor. Because
range to the surge flow) of approximately 3.5:1 was achieved without the geometry is axisymmetric, only one impeller passage and one
excessive efficiency penalty at a PR of 2.4, and so they conclude that diffuser passage were meshed, as shown in Fig. 1b. The side surfaces
using variable diffuser vanes is the most practical method for of the mesh were periodically matched. The total number of nodes is
extending flow-range control. Variable diffuser vanes directly approximately 1.05 million, and the mesh of the impeller uses 0.88
influence the component performance and the matching between the million nodes. The grid independence has been validated in our
diffuser and the impeller, which will definitely change the surge flow, previous work [20]. The width of the cells nearest the wall is
choking flow, and performance. A numerical research also showed 0.002 mm. The shroud gap of the impeller is 0.5 mm. The clearance in
the effectiveness of variable diffuser vanes on widening the operating the diffuser was not meshed.
range [17]. An experimental research done by Ubben and Niehuis [4]
showed that the clearance between the vanes and the wall of the B. Numerical Methods
diffuser influences the operating range; however, there is no The commercially available code NUMECA FINE/Turbo was
recommendation or standard for the width of the clearance at present. employed to solve steady Reynolds-averaged Navier–Stokes
Additionally, because the clearance is between solid bodies, it can be equations. The central scheme and the fourth-order Runge–Kutta
quite small. Therefore, in this paper, the clearance is always assumed scheme were selected for spatial discretization and temporal
to be zero. discretization, respectively. The Spalart–Allmaras single-equation
Much research has been done on a variable-inlet prewhirl and model was used as the turbulence model. The width of the first layer
variable diffuser vanes separately, but there has been little of cells, 0.002 mm, ensured that y was below 10 in all simulations of
investigation on their combination. Simon et al. [18] adjusted an inlet this paper, which must be satisfied for modeling the viscous sublayer
prewhirl and diffuser vanes simultaneously for centrifugal in the turbulence model [21]. The rotor/stator interface was modeled
compressors. The combination proves to offer efficiency improve- using the full nonmatching mixing-plane method. An open test case
ment over the entire operating range compared with using only a of the compressor, Radiver [22], was used to validate the numerical
variable-inlet prewhirl or variable diffuser vanes. Furthermore, the method. The measurements on Radiver were carried out at the
authors also suggested how to determine the optimal combinations of Institute of Jet Propulsion and Turbomachinery at RWTH Aachen
diffuser and prewhirl angles for the purpose of efficiency University, Germany. Part of the investigation was funded by the
improvement. Their work emphasizes the efficiency benefit brought Deutsche Forschungsgemeinschaft. Radiver is a single-stage
by the combination of different variable-geometry methods. centrifugal compressor with a vaned diffuser. The steady and
However, the potential of variable-geometry methods for extending unsteady experimental results of the test case were published. The
the operating ranges of compressors has not been determined compressor geometry, figures, animations, and supplementary
quantitatively, and the corresponding improvements in low-end information have been compiled for the test case. The diffuser on the
engine torque are still unclear as well. These predictions are very test rig is adjustable, and data for different setups of diffuser vanes
 HUANG AND ZHENG 1199

c1u c1
α
c1z

Rotor/Stator Open 10°

interface
Prototype
Close 10°

a)
Fig. 2 Adiabatic efficiency performances of the compressor Radiver
Downloaded by TSINGHUA UNIVERSITY on October 11, 2017 | http://arc.aiaa.org | DOI: 10.2514/1.B36004

Diffuser passage Impeller passage from tests and simulations.

Inlet

Outlet
b)
Fig. 1 a) Compressor model; b) mesh.

have been released. The numerical validation was conducted with the
setup, in which the diffuser-vane angle is equal to 16.5 deg, and the
ratio between the diffuser inlet radius and the impeller outlet radius is
1.14. The grid size of the single-passage mesh of the compressor is 2
million. The simulation results show that y near the wall is lower
than 10 in the whole field, which satisfies the requirement for
turbulence modeling. Fig. 3 Total pressure at the diffuser outlet of the compressor Radiver.
Performances and flowfields from experimental tests and
simulations are compared at 80% design speed, as most of the
published experimental data are at this speed. The computed choking
mass flow at 80% speed is 2.278 kg∕s, which differs from the
experimental choking mass flow of 2.191 kg∕s. This implies that the
effective throat area was slightly overestimated by the simulation.
The overestimation of choking flow is ascribed to the turbulence
model, which greatly influences the development of the boundary
layer and the tip clearance, which may change with rotational speed.
After the mass-flow rates are normalized using the choking mass
flows from tests and simulations, it turns out that the simulation
results have a good agreement with the test results on most of the
parameters, as shown in Figs. 2 and 3. The compressor efficiency and
PR characteristics are well captured by the simulation, which proves
the fidelity of the simulation for predicting compressor performance.
The efficiency is calculated from the impeller inlet to the diffuser
outlet. To examine the ability of the simulation to capture flow

Table 1 Compressor parameters

Fig. 4 Absolute velocity and swirl angle at the impeller outlet of the
Parameters Value and unit compressor Radiver.
Design rotational speed N 111,700 rpm
Impeller outlet diameter D2 1.00 × 102 mm
Tip-speed Mach number Mu2 1.69 features, at the operating point where the dimensionless mass-flow
Total PR at the design point 5.50 rate is equal to 0.83, spanwise plots of quantities at the impeller outlet
Flow coefficient at the design point ϕ 0.004 are shown in Fig. 4. Figure 5 shows the flowfields at the diffuser inlet,
Pressure-rise coefficient at the design point ψ 0.003
Impeller throat area Ai 1.74 × 103 mm2
which validates the agreement between the tests and the simulations.
Diffuser-throat area Ad 5.67 × 102 mm2 The magnitudes of the velocity and the swirl angle are successfully
predicted. Additionally, via simulation, quantity distributions at the
 1200 HUANG AND ZHENG

Absolute velocity [m/s]

Hub

Shroud Test
Hub

Shroud Simulation

Position of the cut

Incidence [°]
Hub

Shroud Test
Hub

a)
Shroud Simulation
Downloaded by TSINGHUA UNIVERSITY on October 11, 2017 | http://arc.aiaa.org | DOI: 10.2514/1.B36004

Fig. 5 Absolute velocity and incidence near the diffuser inlet of the
compressor Radiver.

impeller outlet and diffuser inlet are captured. To sum up, the
numerical methods used in this paper have been validated for
investigating the performance and flow features.

C. Simulation Procedure
For the inlet boundary condition, the inlet prewhirl angle was set
by the velocity direction, and the absolute total pressure and total
temperature were 101.325 kPa and 298.15 K, respectively. No-slip
and impermeability conditions were imposed on all solid surfaces.
Simulations were conducted at 100, 90, 80, 60, and 40% maximum
speed. Because the peak points of the pressure characteristics
provide a convenient engineering approximation for the surge limit,
the peak PR point of each speed line was taken as the surge point. b)
The authors are not saying that flow instabilities, such as surge
and rotating stall, occur at the maximum PR points based on
steady computations. Compressor-performance maps with different
geometric setups were simulated, producing a large amount of data.
To compare the stable operating ranges for all the cases, the use of
the maximum PR point as the left end of each compressor
characteristic is a reasonable compromise.

III. Extending the Stable Operating Range

of the Compressor
A. Performance and Range Extension
The computed mass-flow rate at the design point is 0.48 kg∕s. As
shown in Fig. 6a, both closing of the diffuser and positive prewhirl
can shift the surge line left, at the expense of a PR drop. When the PR
is above 2.2, the shift becomes quite obvious. Compared with the case
“close 10 deg,” the combination of closing the diffuser and positive c)
prewhirl shifts the surge line more to the left at PRs above 4.0. Fig. 6 Compressor-performance maps and range extension.
Figure 6b shows the right shifts of the choke line that are induced by
the variable-geometry method. At 90% speed and below, opening the
diffuser significantly increases the choking flow. At 80% speed and  
m_
above, the effect of a negative prewhirl on the choking flow is very R 1− s × 100% (1)
_ c PRconst
m
obvious. For the case of combining negative prewhirl and opening of
the diffuser, the choking flow is increased at all speeds. Especially at
80 and 90% speed, the increase resulting from the combination is
much greater than that resulting from only opening the diffuser or Actually, the benefits of the combined variable-geometry method
only imposing negative prewhirl. The stable operating range R is lie not only in widening the flow range, but also in improving
defined by Eq. (1). Figure 6c shows the extended range resulting from efficiency. As shown in Fig. 7, the efficiency performance resulting
the combined effect of the variable-inlet prewhirl from −20 to 60 deg, from the combination is higher than that resulting from only closing
and variable diffuser vanes from −10 to 10 deg. At the PR of 4.8, the diffuser. For instance, for a given operating point at which the PR
the combined variable-geometry method increases R from 23.5% is 4.0 and the mass flow is 0.23 kg∕s, the combined variable-
(i.e., 1–0.39∕0.51) to 63.0%. geometry method and close 10 deg both can achieve high efficiency at
 HUANG AND ZHENG 1201

Relative velocity (m/s)

250 close 10°

200
Entropy(J/K)

150 400
350
60° prewhirl+close 10°
100 300
250
50 200
150
0 100
50
0
-50
-100
-150
Downloaded by TSINGHUA UNIVERSITY on October 11, 2017 | http://arc.aiaa.org | DOI: 10.2514/1.B36004

Fig. 7 Efficiency improvement by using variable-geometry method.

The entropy field of “close 10°” minuses

90% speed, but the combined variable-geometry method has an the entropy field of “60° prewhirl+close 10°”
advantage of 6.0 points in efficiency.
Closing the diffuser forces the impeller to operate at the left branch Fig. 9 Entropy contour and relative-velocity contours.
of the efficiency characteristic, which heavily exacerbates impeller
performance. As shown in Fig. 8, the impeller efficiency of close
10 deg is quite low. After exerting a positive inlet prewhirl, the
performance of the diffuser almost does not change, but the impeller
efficiency increases significantly. Thus, the positive inlet prewhirl Wno prewhirl
compensates for the poor impeller performance caused by closing Winlet prewhirl α c no prewhirl
the diffuser. As shown in Fig. 9, the separation vortex of 60 deg cinlet prewhirl
prewhirl  close 10 deg at the inducer inlet is much smaller than that
ino prewhirl u1
of close 10 deg. At a given mass-flow rate, a positive inlet prewhirl
can reduce the Mach number and incidence at the impeller inlet,
which has been proved through analyzing the velocity triangle in Impeller
Fig. 10. Thus, a positive inlet prewhirl relieves the heavy flow iinlet prewhirl
separation at the inducer inlet, which directly reduces the entropy
increase there and contributes to improving the impeller efficiency.
Furthermore, an inlet prewhirl was shown to be able to adjust the
matching relation between the impeller and the diffuser [23,24],
which greatly contributes to the shift of the impeller characteristic and
the efficiency gains. Fig. 10 Velocity triangle at the impeller inlet.
An experimental work done by Simon et al. [18] also proved the
efficiency benefits arising from combining these two variable-
geometry methods, in which the IGV was installed, the diffuser could
be replaced, and there was no volute. The authors concluded that, in potential of the combined variable-geometry method for improving
their case, the efficiency gains with a combined adjustment amount to efficiency. On the other hand, the authors think that flow loss of inlet
approximately 2% and up to 6%; these gains are similar in magnitude guide vans, deviation of the prewhirl angle, leakage flow in the
to the efficiency gains in this paper. This paper may estimate the clearance between the adjustable vanes and the diffuser wall, and
the interaction between the diffuser and the volute may affect the
efficiency performance. These negative aspects could result in an
efficiency improvement that is not as high as our estimate for the
combined geometry method. In the future, the penalty estimations
from these effects should be discussed in the case of combining both
variable-geometry methods. It is worthwhile investigating how to
reduce their impacts on performance.

B. Discussion on the Shift of Surge Lines

Whether the impeller or the diffuser triggers flow instability in the
prototype will be determined, which is the basis for analyzing the
influences of the variable-geometry method on the surge line. It is
inferred that flow instability is triggered by the impeller when the PR
is below 2.2 in the prototype, that is, the rotational speed is lower than
60% maximum speed. At 60% speed, closing the diffuser, which
makes the diffuser operate more stably, does not result in a significant
displacement of the surge line toward lower flow, and so it is not the
diffuser that controls the surge flow at 60% speed. However, at 80%
speed, closing the diffuser causes the surge line to move considerably
to the left, which indicates that the diffuser triggers flow instability at
Fig. 8 Impeller adiabatic efficiency and diffuser loss coefficient. 80% speed. From the point of view of the matching between the
 1202 HUANG AND ZHENG

impeller and the diffuser, the diffuser will become larger for matching Zheng et al. [23] showed that the positive inlet prewhirl influences the
as the rotational speed increases [25], which means the diffuser choking flows of the impeller and the diffuser, as in Eqs. (2) and (3),
characteristic would tend to lie in the higher mass-flow region respectively, in which m _ i;c is the choking flow of the impeller, m
_ d;c is
compared with the impeller characteristic. Thus, it is reasonable to the choking flow of the diffuser, and c1u is the circumferential
infer that the diffuser triggers flow instability at 80% speed and component of the inlet velocity. Therefore, if the choking flow of the
above, which leads to the left shift of the surge line at PRs above 2.2 impeller or the diffuser is very small, the change in the choking flow
through the closing of the diffuser. by the inlet prewhirl is also very small. For the prototype at PRs below
To provide more insight for the flowfield at maximum PR points, 2.2 (i.e., at 40% speed and 60% speed) and for the case of close 10 deg
streamlines and pressure on blade surfaces are shown in Fig. 11. At 100 at PRs below 4.0 (i.e., at 90% speed and lower speeds), the influence
and 80% speeds of datum case, circulation is found between the of the inlet prewhirl on the shifts of characteristics is negligible, and
impeller outlet and the diffuser inlet, although its area in the meridional so the shifts of the surge lines are also not significant. This is the so-
views is relatively small. These circulations, which are due to curvature called ineffectiveness of the inlet prewhirl caused by the low mass
effect, considerably affect flow stabilities at the diffuser [26]. Because flow at the surge point.
of the change of matching relation between the impeller and the
diffuser at 80% speed, the impeller tends to operate at a more unstable ∂m
_ i;c −3m_ i;c u1
 (2)
region, which is indicated by a larger separation vortex near the inducer ∂c1u cp T 01 − c1u u1  u2x ∕2
tip. After closing the diffuser by 10 deg, the separation vortex in the
inducer grows larger and extends out of the impeller tip. However,
60 deg prewhirl relieves the extent of the separation vortex in the ∂m_ d;c −3m _ d;c u1
 (3)
Downloaded by TSINGHUA UNIVERSITY on October 11, 2017 | http://arc.aiaa.org | DOI: 10.2514/1.B36004

inducer, which benefits flow instabilities in the impeller and drives the ∂c1u cp T 01 − c1u u1  u22
impeller to work at a lower surge mass flow.
The contribution from variable diffuser vanes on the shift of the The inducer stall plays an important role in the instability of the
surge line has been discussed earlier, and the effect of the variable- impeller performance. The complex interaction between the shock
inlet prewhirl will be discussed in the following section. wave and the boundary layer, which may cause the increase of
As shown in Fig. 6a, the positive inlet prewhirl fails to shift the blockage, can aggravate the inducer stall [28]. As indicated by
surge line of the prototype to the left at PRs below 2.2, and it also fails Fig. 10, the positive inlet prewhirl is able to relieve the relative
to shift the surge line of close 10 deg at PRs below 4.0. In the Mach number at the inducer inlet. Especially when the relative Mach
experimental research done by Rodgers [27], the IGVangle of 27 deg number approaches unity, the relief resulting from the positive inlet
fails to make the surge line move to the left at PRs below 3.7, which prewhirl will contribute significantly to stabilizing the performance
shows a similar trend with our results. The low mass flow and low of the impeller and the left shift of the surge line. Figure 12 shows the
inlet relative Mach number at the surge point are the two reasons for relative Mach-number fields at the surge points where the surge mass
the ineffectiveness of the positive inlet prewhirl. flows will reduce obviously after the positive inlet prewhirl is exerted
The inlet prewhirl influences the characteristics of components and (see Fig. 6a). In both of the fields, there are regions where the relative
their matching. However, if the choking flow of the compressor is too Mach number exceeds unity. After exerting the positive inlet
small, the effects of the positive inlet prewhirl are also small. With prewhirl, the stability of the impeller will be improved. As shown in
some assumptions (one is that the efficiency of the impeller is 100%), Fig. 13, the slope of the impeller PR characteristic at the surge point of

Fig. 11 Streamlines in the meridional views and pressure distributions at the peak PR points.
 HUANG AND ZHENG 1203

Fig. 12 Relative Mach-number contours with supersonic area in the inducer.

Downloaded by TSINGHUA UNIVERSITY on October 11, 2017 | http://arc.aiaa.org | DOI: 10.2514/1.B36004

5.7 the Euler equation [26]. Because the inlet prewhirl can adjust the
work input into the fluid, the total temperature and the total pressure
100% speed
5.6 at the throttle would usually drop with the positive inlet prewhirl, and
rise with the negative inlet prewhirl, which cause changes in the
choking mass flow. Zheng et al. [23] investigated the influence of the
Impeller pressure ratio

5.5 Positive slope inlet prewhirl on the matching between the impeller and the diffuser.
If the impeller and diffuser are supposed to match well in the case of
5.4 Negative slope zero inlet prewhirl, they will become mismatched in the case of
nonzero inlet prewhirl. With the positive inlet prewhirl, the throat area
5.3 of the diffuser is too large for the matching impeller. An intuitive
corollary is that the throat area of the diffuser is too small for the
5.2 matching impeller in the case of the negative inlet prewhirl. Thus,
after imposing negative prewhirl, the diffuser is too small for a good
60° prewhirl + close 10° matching and the compressor must still be choked at the diffuser at
5.1
Close 10° 100% speed. As shown in Fig. 15, the compressor with −20 deg
prewhirl is choked at the diffuser, as the static pressure on the rotor/
5 stator interface and near the diffuser inlet does not vary with the static
0.2 0.22 0.24 0.26 0.28 0.3 0.32
pressure imposed at the outlet. Opening the diffuser greatly increases
Corrected Mass flow rate/kg·s-1 the stage choking flow at 90% speed and below, because the diffuser
Fig. 13 Impeller PR characteristics. is always controlling the choking flow at these speeds. As shown in
Fig. 16, if negative prewhirl is imposed, the compressor with an open
diffuser is still choked at the diffuser at 90% speed.
Although the inlet prewhirl has been proven to influence choking
close 10 deg is positive, which indicates the high instability of the
mass-flow rates in many cases in Fig. 6, the choking mass-flow rates
impeller. However, the slope at the surge point with 60 deg inlet
barely change with the inlet prewhirl in some cases, such as close
prewhirl and the closing of the diffuser vanes by 10 deg becomes
10 deg at 90% speed and prototype at 60% speed. Table 2 shows the
negative, which proves that the relief of the excessive relative Mach
choking mass-flow rates for which the inlet prewhirl has almost no
number by the positive inlet prewhirl does stabilize the impeller
impact on the choking mass-flow rate. To reveal the mechanism,
performance. Figure 14 shows the relative Mach-number fields at the
first, this problem should be analyzed from the one-dimensional
surge points where the surge mass flows are barely reduced after the
perspective, and then plots of corresponding parameters should be
positive inlet prewhirl is exerted (see Fig. 6a). In both of the fields,
shown in support of the explanation. As described in the last
there are no supersonic regions. It seems that the positive inlet
paragraph, the compressor is choked at the diffuser in these cases, and
prewhirl is ineffective for reducing surge flow if the relative Mach
so a basic choking mass-flow equation of the diffuser is presented as
number at the inducer inlet does not exceed unity. This is the so-called
Eq. (4). By applying the ideal gas law and the sonic-speed equation,
ineffectiveness of the inlet prewhirl, caused by the low inlet relative
Eq. (4) can be rearranged into Eq. (5). Taking the impeller-efficiency
Mach number at the surge point. equation, Eq. (6), into account, the choking mass-flow-rate equation,
Eq. (7), is obtained. T 02 is the total temperature at the rotor/stator
C. Discussion on the Shift of Choke Lines interface, which is used to measure work input. Bd;x and ηi are the
Opening the diffuser vanes by 10 deg enlarges the throat area of blockage factor at the diffuser throat and the impeller efficiency,
each diffuser from 5.67 × 102 to 7.68 × 102 mm2 , but only increases respectively. These three aerodynamic factors considerably influence
the choking flow at 100% speed from 0.515 to 0.517 kg∕s, which the choking mass-flow rates. Figure 17 shows that the inlet prewhirl
implies that the diffuser controls the choking flow at 100% speed for does change the work input. For instance, in the case of close 10 deg
the prototype and that the choking flow of the impeller is just slightly at 90% speed, 60 deg prewhirl reduces T 02 by 27 K at the mass-flow
larger than that of the diffuser. The choked component turns out to be rate of 0.25 kg∕s, which should greatly affect the choking mass-flow
the impeller after opening the diffuser, and so the increase of the rate. However, because 60 deg prewhirl also helps to improve
choking flow is not proportional to and is much less than the increase impeller efficiency by 5.8 points, as shown in Fig. 8, and the diffuser
of the throat area of the diffuser. performance almost never changes, 60 deg prewhirl does not change
As shown in Fig. 10, inlet prewhirl means that the velocity of fluids the compressor choking mass-flow rate. The same situation occurs in
at the compressor inlet has a non-zero circumferential component. the case of the prototype at 60% speed. Although −20 deg prewhirl
Therefore, work input from the impeller would decrease according to increases the work input, the impeller performance is worsened and
 1204 HUANG AND ZHENG

Fig. 14 Relative Mach-number contours without supersonic flow in the inducer.

Downloaded by TSINGHUA UNIVERSITY on October 11, 2017 | http://arc.aiaa.org | DOI: 10.2514/1.B36004

1200 340 the diffuser performance does not change, as shown in Figs. 18
100% maximum speed -20° prewhirl, outlet and 19, which keeps Bd;x the same and causes ηi to decrease. The
Static pressure at the rotor/stator interface(kPa)

-20° prewhirl, R/S interface 335 influences of these three factors counteract one another, and so
1000 the compressor choking mass-flow rate remains 0.24 kg∕s.
Static pressure at the outlet(kPa)

Pressure(kPa) 330  γ1∕2γ−2

Shroud 1100 2
800 m_ d;choke  ρ02 a02 Bd;x Ad;x (4)
900
325 γ1
A B 700
 −1∕2  
600 Hub 320 T 02 p02 2 γ1∕2γ−2
Near the inducer 500 _ d;choke  ρ01 a01 Bd;x Ad;x
m (5)
T 01 p01 γ 1
Pressure(kPa) 315
400 350
 
p02 η T − T 01  γ∕γ−1
Shroud
330
A
310  1  i 02 (6)
p01 T 01
200 Hub A B 310 B
305
Near the diffuser inlet 290  −1∕2  γ∕γ−1
T 02 T
0 300 _ d;choke  ρ01 a01 Bd;x Ad;x
m ηi 02 − ηi  1
T 01 T 01
0.48 0.49 0.5 0.51 0.52 0.53 0.54  γ1∕2γ−2
2
Corrected Mass flow rate(kg·s-1) × (7)
Fig. 15 Static-pressure distributions of the choking points at the γ1
maximum speed.

IV. Improving the Low-End Torque of an Engine

Here, a rough estimation of the ability of the variable-geometry
method to improve low-end engine torque is obtained (in a way
similar to [19]). Figure 20 shows two compressor working lines at
270
90% maximum speed
-20° prewhirl+open 10°, outlet low engine-shaft speeds. One is for the prototype and the other is for
800
Static pressure at the rotor/stator interface(kPa)

-20° prewhirl+open 10°, R/S interface

the compressor employing the variable-geometry method. The surge
250
margin (SM) is defined by Eq. (8) [26], and it is controlled above 12%
in this study. The corner point of a working line has the greatest risk of
Static pressure at the outlet(kPa)

Pressure(kPa)
800
Shroud flow instability, and so it also represents the highest boost level that a
600 700
230 compressor delivers to an engine at low speeds. W and W 0 are the
600 corner points of the working lines in the original range and extended
C D
Hub
500 range, respectively. S is the surge point at the same rotational speed of
Near the inducer 400 210 W. S 0 is the corresponding surge point of W 0 . The SM of W is 13.1%
400
and that of W 0 is 16.0%. Both of the working lines almost hit the
Pressure(kPa)
230 190
Shroud 220
C
200 210
Hub C D
D Table 2 Choking mass-flow rates in selected cases
200 170
190 Choking mass-flow
Near the diffuser inlet
Case description Inlet prewhirl, deg rate, kg/s
0 150 Close 10 deg at 90% speed 0 0.26
0.46 0.48 0.5 0.52 0.54 60 0.25
Corrected Mass flow rate(kg·s-1)
Prototype at 60% speed 0 0.24
Fig. 16 Static-pressure distributions of the choking points at 90% −20 0.24
speed.
 HUANG AND ZHENG 1205

405 540 6
Total temperature at rotor/stator interface [K], 60% speed

Total temperature at rotor/stator interface [K], 90% speed

W'(0.25, 5.0)
S'(0.21, 5.0)
5
400 530
Extended range

Total pressure ratio

4
395 520 S(0.23, 2.8)
3

390 510 Original range

2

W(0.26, 2.75)
-20° prewhirl+prototype, 60% speed 1
385 500 Working line in the original range
prototype, 60% speed
Working line in the extended range
60° prewhirl+close 10°, 90% speed
close 10°, 90% speed
0
380 490 0 0.1 0.2 0.3 0.4 0.5 0.6
0.05 0.1 0.15 0.2 0.25 0.3 Corrected mass flow rate/kg·s-1
Corrected mass flow rate [kg/s] Fig. 20 Estimated working lines in the original range and the extended
Downloaded by TSINGHUA UNIVERSITY on October 11, 2017 | http://arc.aiaa.org | DOI: 10.2514/1.B36004

Fig. 17 Total temperatures at rotor/stator interface in selected cases. range.

critical SM. The PRs of W and W 0 can be used to estimate the

improvement in the low-end torque.
90% 0.8  
-20° prewhirl+prototype, impeller PR ∕m _
prototype, impeller SM  1 − w w × 100% (8)
-20° prewhirl+prototype, diffuser PRs ∕m _ s Nconst
prototype, diffuser
Impeller adiabatic efficiency

85% 0.6 Assuming that the mixture of air and fuel maintains its stoichiometry,
Diffuser loss coefficient

60% rotational speed the engine-shaft torque should be directly related to the amount of
compressor intake. The ratio of pressures at W 0 and W is 1.82. If the
compression process is assumed to be reversible and adiabatic, and the
80% 0.4 intercooling is neglected, the ratio of the air densities at W 0 and W is
approximately 1.53 (1.821∕γ , in which γ is 1.4). Assuming that the
effective efficiency of the engine remains the same at W 0 and W, the
ratio of the brake torques at W 0 and W should be the same as the ratio of
75% 0.2 the mixture amounts or the ratio of the intake amounts, as indicated by
Eq. (9). Thus, with the variable-geometry method, the low-end torque
of an engine is estimated to increase by 53%.
70% 0 Pe ηet mH
_ u
0.1 0.15 0.2 0.25 T tq   ∝ PR1∕γ (9)
ws αws
Corrected mass flow rate [kg/s]
Fig. 18 Impeller adiabatic efficiency and diffuser loss coefficient at 90%
speed. V. Conclusions
To develop ultrahigh-power-density turbocharged engines,
compressors are required to provide high-pressure air over a wide
mass-flow range. The combination of the variable-inlet prewhirl and
variable diffuser vanes was employed to extend the compressor
operating range. Numerical simulations were used to investigate the
performance of a centrifugal compressor with a diffuser-vane angle
varying from −10 to 10 deg, and inlet prewhirl varying from −20 to
60 deg. The potentials of the combined method for extending the
compressor operating range and improving low-end engine torque
were estimated. The following conclusions are drawn:
1) Combining variable diffuser vanes, with angles between −10
and 10 deg, and variable-inlet prewhirl, between −20 and 60 deg, has
the potential to improve the stable compressor operating range from
23.5 to 63.0% at a PR of 4.8. (Such a high PR can be a preferred
choice for future turbocharged engines.) The shifts of the choke and
surge lines contribute to this extension. The combination of positive
inlet prewhirl and a closed diffuser has the advantage of moving the
surge line left compared to only closing the diffuser or only using the
positive inlet prewhirl. In addition to extending the flow range,
another advantage of the variable-geometry method is its ability to
achieve a higher-efficiency performance by adjusting the diffuser
vane angle and prewhirl angle together.
2) For the prototype, at 60 and 40% speed, the influences of the inlet
prewhirl on the surge flow and choking flow are negligible. For the case
of closing the diffuser, even at 80 or 90% speed, the inlet prewhirl still
Fig. 19 Difference in entropy fields at the mass-flow rate of 0.24 kg∕s. has little effect. As indicated by the equations, if the original choking
 1206 HUANG AND ZHENG

flow of a compressor is too small, the effect of the inlet prewhirl on [12] Galindo, J., Serrano, J. R., Margot, X., Tiseira, A., Schorn, N., and
shifting the characteristic will also be small. This is one of the reasons Kindl, H., “Potential of Flow Pre-Whirl at the Compressor Inlet of
why the inlet prewhirl has little influence at low speeds. Automotive Engine Turbochargers to Enlarge Surge Margin and
3) When the relative Mach number in front of the inducer Overcome Packaging Limitations,” International Journal of Heat and
Fluid Flow, Vol. 28, No. 3, 2007, pp. 374–387.
approaches unity, the relief of the relative Mach number by the doi:10.1016/j.ijheatfluidflow.2006.06.002
positive inlet prewhirl will contribute significantly to stabilizing [13] Sezal, I., Lang, M., Aalburg, C., Chen, N., Erhard, W., Greco, A. S. D.,
the performance of the impeller and moving the surge line to the left. Tapinassi, L., and Gadamsetty, R. K., “Introduction of Circumferentially
This is another reason for the great displacement of the surge line at Non-Uniform Variable Guide Vanes in the Inlet Plenum of a Centrifugal
high speeds after employing the positive inlet prewhirl. Compressor for Minimum Losses and Flow Distortion,” Journal of
4) The inlet prewhirl does not always change the compressor Turbomachinery, Vol. 138, No. 9, 2016, Paper 091008.
choking mass-flow rates, which are influenced by impeller efficiency [14] Kyrtatos, N., and Watson, N., “Application of Aerodynamically Induced
and work input. The positive inlet prewhirl reduces the work input Prewhirl to a Small Turbocharger Compressor,” Journal of Engineering
from the impeller, but it may also contribute to improving impeller for Power, Vol. 102, No. 4, 1980, pp. 943–950.
doi:10.1115/1.3230365
efficiency, which counteracts the drop in work input. The negative [15] Whitfield, A., and Abdullah, A. H., “The Performance of a Centrifugal
inlet prewhirl increases the work input, but it may worsen the Compressor with High Inlet Prewhirl,” Journal of Turbomachinery,
performance improvements. The combined effects lead to no change Vol. 120, No. 3, 1998, pp. 487–493.
in the compressor choking mass-flow rate. doi:10.1115/1.2841744
5) Benefitting from the range extension of the compressor, the [16] Oatway, T. P., and Harp, J. L., “Investigations of a Variable Geometry
maximum pressure of intake for a turbocharged engine can be Compressor for a Diesel Engine Turbocharger,” U.S. Army Tank-
Downloaded by TSINGHUA UNIVERSITY on October 11, 2017 | http://arc.aiaa.org | DOI: 10.2514/1.B36004

improved by approximately 82%, and the low-end torque is estimated Automotive Command Rept. SR-21, Warren, MI, 1973.
to increase by 53%. [17] Jiao, K., Sun, H., Li, X., Wu, H., Krivitzky, E., Schram, T., and
Larosiliere, L. M., “Numerical Investigation of the Influence of
Variable Diffuser Vane Angles on the Performance of a Centrifugal
Compressor,” Proceedings of the Institution of Mechanical Engineers,
Acknowledgments Part D: Journal of Automobile Engineering, Vol. 223, No. 8, 2009,
This research was supported by the National Natural Science pp. 1061–1070.
Foundation of China (grant number 51176087). It was also [18] Simon, H., Wallmann, T., and Monk, T., “Improvements in Performance
sponsored by the State Key Laboratory of Automotive Safety and Characteristics of Single-Stage and Multistage Centrifugal Compres-
Energy under project number KF16102. The authors would like to sors by Simultaneous Adjustments of Inlet Guide Vanes and Diffuser
thank the Institute of Jet Propulsion and Turbomachinery of the Vanes,” Journal of Turbomachinery, Vol. 109, No. 1, 1987, pp. 41–47.
doi:10.1115/1.3262068
RWTH Aachen University for providing the test case, Radiver, a
[19] Zheng, X., and Huang, Q., “Potential of the Range Extension of
centrifugal compressor with a wedge diffuser. Compressors with a Variable Inlet Prewhirl for Automotive
Turbocharged Engines with an Ultra-High-Power Density,” Proceed-
ings of the Institution of Mechanical Engineers, Part D: Journal of
References Automobile Engineering, Vol. 229, No. 14, 2015, pp. 1959–1968.
[1] Smith, L. H., Jr., “NASA/GE Fan and Compressor Research [20] Zheng, X., Huenteler, J., Yang, M., Zhang, Y., and Bamba, T., “Influence
Accomplishments,” Journal of Turbomachinery, Vol. 116, No. 4, of the Volute on the Flow in a Centrifugal Compressor of a High-
1994, pp. 555–569. Pressure Ratio Turbocharger,” Proceedings of the Institution of
[2] Whitfield, A., “Review of Variable Geometry Techniques Applied to Mechanical Engineers, Part A: Journal of Power and Energy, Vol. 224,
Enhance the Performance of Centrifugal Compressors,” International No. 7, 2010, pp. 1157–1159.
Compressor Engineering Conference, July 2000, Paper 2367. doi:10.1177/095440541022400702
[3] Ziegler, K. U., Justen, F., Rothstein, M., Gallus, H. E., and Niehuis, R., [21] Anon., User Manual, FINE/Turbo v8.7, NUMECA International,
“Research on a Centrifugal Compressor of Variable Geometry,” Brussels, Belgium, 2010, pp. 9–11, Chap. 4.
International Compressor Engineering Conference, July 2000, [22] Ziegler, K. U., Gallus, H. E., and Niehuis, R., “A Study on Impeller–
Paper 1371. Diffuser Interaction—Part I: Influence on the Performance,” Journal of
[4] Ubben, S., and Niehuis, R., “Experimental Investigation of the Diffuser Turbomachinery, Vol. 125, No. 1, 2003, pp. 173–182.
Vane Clearance Effect in a Centrifugal Compressor Stage with doi:10.1115/1.1516814
Adjustable Diffuser Geometry—Part I: Compressor Performance [23] Zheng, X., Huang, Q., and Liu, A., “Loss Mechanisms and Flow Control
Analysis,” Journal of Turbomachinery, Vol. 137, No. 3, 2014, for Improved Efficiency of a Centrifugal Compressor at High Inlet
Paper 031003. Prewhirl,” Journal of Turbomachinery, Vol. 138, No. 10, 2016,
[5] Rodgers, C., “Turbocharging a High Altitude UAV C.I. Engine,” AIAA pp. 101011–101011.
Paper 2001-3970, July 2001. doi:10.1115/1.4033216
[6] Aoyagi, Y., Kunishima, E., Asaumi, Y., Aihara, Y., Odaka, M., and [24] Huang, Q., Zheng, X., and Wang, A., “Mechanism and Flow Control on
Goto, Y., “Diesel Combustion and Emission Using High Boost and High the Mismatching of Impeller and Vaned Diffuser Caused by Inlet
Injection Pressure in a Single Cylinder Engine,” JSME International Prewhirl for Centrifugal Compressors,” Turbo Expo, American Soc. of
Journal, Series B: Fluids and Thermal Engineering, Vol. 48, No. 4, Mechanical Engineers Paper GT2016-56244, Seoul, South Korea,
2005, pp. 648–655. 2016.
doi:10.1299/jsmeb.48.648 [25] Casey, M., and Rusch, D., “The Matching of a Vaned Diffuser with a
[7] Blair, J., and Bower, G., “Development of a Miller Cycle Powersports Radial Compressor Impeller and Its Effect on the Stage Performance,”
Engine,” Soc. of Automotive Engineers TP 2014-32-0090, 2014. Journal of Turbomachinery, Vol. 136, No. 12, 2014, pp. 121004–
[8] Mohtar, H., Chesse, P., Yammine, A., and Hetet, J. F., “Variable Inlet 121004.
Guide Vanes in a Turbocharger Centrifugal Compressor: Local and doi:10.1115/1.4028218
Global Study,” Soc. of Automotive Engineers TP 2008-01-0301, 2008. [26] Cumpsty, N. A., Compressor Aerodynamics, Kreiger, Malabar, FL,
[9] Herbst, F., Stöber-Schmidt, C.-P., Eilts, P., Sextro, T., Kammeyer, J., 2004, pp. 4, 107, 367.
Natkaniec, C., Seume, J., Porzig, D., and Schwarze, H., “The Potential [27] Rodgers, C., “Centrifugal Compressor Inlet Guide Vanes for Increased
of Variable Compressor Geometry for Highly Boosted Gasoline Surge Margin,” Journal of Turbomachinery, Vol. 113, No. 4, 1991,
Engines,” Soc. of Automotive Engineers TP 2011-01-0376, 2011. pp. 696–702.
[10] Coppinger, M., “Aerodynamic Performance of an Industrial Centrifugal doi:10.1115/1.2929136
Compressor Variable Inlet Guide Vane System,” Ph.D. Dissertation, [28] Senoo, Y., Hayami, H., Kinoshita, Y., and Yamasaki, H., “Experimental
Loughborough Univ., Loughborough, England, U.K., 1999. Study on Flow in a Supersonic Centrifugal Impeller,” Journal of
[11] Mohseni, A., Goldhahn, E., Van den Braembussche, R. A., and Seume, Engineering for Power, Vol. 101, No. 1, 1979, pp. 32–39.
J. R., “Novel IGV Designs for Centrifugal Compressors and Their doi:10.1115/1.3446453
Interaction with the Impeller,” Journal of Turbomachinery, Vol. 134,
No. 2, 2012, Paper 021006. F. Liu
doi:10.1115/1.4003235 Associate Editor