George A.

Christou1
Department of Aeronautics and Astronautics,
Massachusetts Institute of Technology,
Cambridge, MA 02139
e-mail: gchristo@mit.edu
Choon S. Tan
Department of Aeronautics and Astronautics,
Massachusetts Institute of Technology,
Cambridge, MA 02139
e-mail: choon@mit.edu
Borislav T. Sirakov
Honeywell Turbo Technologies,
Torrance, CA 90504
e-mail: Bobby.Sirakov@Honeywell.com
Vai-Man Lei
Honeywell Turbo Technologies,
Torrance, CA 90504
e-mail: VaiMan.Lei@Honeywell.com
Giuseppe Alescio
Honeywell Turbo Technologies,
Torrance, CA 90504
e-mail: Giuseppe.Alescio@Honeywell.com
Characterizing Flow Effects of
Ported Shroud Casing Treatment
on Centrifugal Compressor
Performance
This paper presents an investigation of the effects of ported shroud (PS) self-
recirculating casing treatment used in turbocharger centrifugal compressors for increas-
ing the operable range. The investigation consists of computing three-dimensional flow in
a representative centrifugal compressor with and without PS at various levels of approxi-
mations in flow physics and geometrical configuration; this provides an enabler for
establishing the causal link between PS flow effects and compressor performance
changes. It is shown that the main flow path perceives the PS flow as a combination of
flow actuations that include injection and removal of mass flow and injection of axial
momentum and tangential momentum. A computational model in which the presence of
the PS is replaced by imposed boundary conditions (BCs) that reflect the individual flow
actuations has thus been formulated and implemented. The removal of a fraction of the
inducer mass flow has been determined to be the dominant flow actuation in setting the
performance of PS compressors. Mass flow removal reduces the flow blockage associated
with the impeller tip leakage flow and increases the diffusion in the main flow path. Add-
ing swirl to the injected flow in the direction opposite to the wheel rotation results in an
increase of the stagnation pressure ratio and a decrease of the efficiency. The loss gener-
ation in the flow path has been defined to rationalize efficiency changes associated with
PS compressor operation. [DOI: 10.1115/1.4035664]

Introduction and Problem Statement static pressure experienced larger flow recirculation through the
PS cavity. As a consequence, the use of a PS casing treatment
The ported shroud (PS) self-recirculating casing treatment has
reduced the level of the individual passage mass flow nonuniform-
proven its effectiveness in centrifugal compressors for vehicular
ity. Zheng et al. [11] experimented with a nonaxisymmetric PS
turbochargers in increasing the compressor operable range [1].
casing treatment that took into account the volute-tongue induced
Fisher was one of the first researchers who measured the compres-
pressure nonuniformity and reported a decrease of up to 10% in
sor operability range extension with the use of a PS on a number
surge mass flow compared to a symmetric PS slot.
of impellers [1]. Uchida et al. [2] and Iwakiri and Uchida [3]
It can thus be inferred from the published results that the effect of
showed that the introduction of a PS decreased the extent of
the PS on the main compressor flow path may be perceived as a com-
reversed flow near the inducer shroud region, thus decreasing the
bination of flow actuations that consist of mass flow removal down-
surge mass flow compared to a conventional compressor [2]. Hun-
stream of the impeller leading edge (LE), mass flow, and momentum
ziker et al. [4] postulated that the preswirl contained in the recir-
(axial and tangential) injection upstream of the wheel inlet. While
culating flow decreased the blade incidence angle, assisting in
the research to-date on PS compressor flow has been useful, there is
avoiding flow separation. Qiu et al. [5] and Fisher [1] hypothe-
a lack of technical clarity on which aspects of PS flow actuations are
sized that the flow recirculating through the PS cavity blocks part
conducive for improving compressor operation. Thus, the overall
of the passage flow, reducing the inlet flow area and leading to a
goal of the research is to first clarify and analyze the quantitative role
decrease of the incidence angle of the main core flow [5]. Yama-
of each of the PS flow actuations (mass flow removal/addition and
guchi et al. [6] and Tamaki et al. [7,8] added vanes inside the PS
axial and tangential momentum injection) on the compressor per-
cavity that introduced counterswirl to the recirculating flow which
formance. This is followed by formulating a framework that synthe-
increased the recirculating PS flow compared to a PS compressor
sizes the individual PS flow actuations to: (i) determine the overall
with no vanes and further increased the surge margin. Similar
impact of PS flow on the compressor performance and (ii) explain
results were found by Sivagnanasundaram et al. [9] where vanes
why the PS compressor operates the way it does.
were inserted into the PS cavity. The vanes decreased the swirl
This paper is organized as follows: To set the tone of this paper,
angle of the exiting PS flow to 30 deg in the direction of the wheel
we first put forward a rationalization of the operation of PS com-
rotation, thus increasing the compressor pressure ratio and surge
pressor in terms of the various PS flow actuations working in uni-
margin. Hunziker et al. [4] measured a decrease in the pressure
son. The approach is described next. Results are then presented to
nonuniformity in the inducer of a PS compressor compared to the
support the role of the flow actuations in setting the PS compres-
nonported variant. Thus, it was hypothesized that the PS
sor operation; these include the effects of mass flow removal/
decoupled the inducer from the circumferentially nonuniform
injection, angular momentum injection, and integrated mass flow
flow in the rest of the stage [4]. Yang et al. investigated how the
removal and angular momentum injection. This paper ends with a
PS influenced the volute-tongue induced pressure nonuniformity
summary of the key findings.
[10]. Individual flow passages that coincided with higher levels of

1
Corresponding author. Operation of PS Compressor
Contributed by the International Gas Turbine Institute (IGTI) of ASME for
publication in the JOURNAL OF TURBOMACHINERY. Manuscript received November 13,
The operation of PS compressor is rationalized through synthe-
2016; final manuscript received December 20, 2016; published online March 21, sizing the effects of the various PS flow actuations as follows.
2017. Editor: Kenneth Hall. Flow recirculation through the PS cavity implies that the inducer

Journal of Turbomachinery Copyright V

C 2017 by ASME AUGUST 2017, Vol. 139 / 081005-1

Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jotuei/936123/ on 04/02/2017 Terms of Use: http://www.asme.org/ab

 works at a higher mass flow and closer to peak g compared to the
remaining compressor. This results in the decrease of the inducer
recirculation zone and subsequent decrease of flow blockage and
entropy generation. The effect of removing a fraction of the inducer
mass flow, a PS flow actuation involving mass flow removal, is
shown to further decrease both flow blockage and entropy genera-
tion compared to nonported operating points (OPs) of same inducer
mass flow. This is due to the removal of the inducer tip leakage flow
and is shown to increase the blade loading. Additionally, mass flow
removal results in the deceleration of the compressor main stream,
increasing the static pressure rise at the impeller outlet, thus enhanc-
ing the stability margin potential. Injection of the flow at the PS out-
let with counterswirl, a PS flow actuation involving angular
momentum injection, further increases the compressor pressure ratio
and inducer static pressure rise compared to coswirl mass flow injec-
tion. The higher inducer static pressure gives rise to higher mass
flow through the PS cavity, thus amplifying the effect of mass flow
removal in increasing the stagnation pressure ratio. The combined
effect of mass flow removal and counterswirl injection results in the
inducer working at an OP further away from surge compared to
coswirl mass flow injection for equal compressor outlet mass flow.
Because a larger fraction of mass flow is recirculated through the PS
cavity and the inducer is working at a higher mass flow, the flow
blockage upstream of the blade LE is lower compared to coswirl
injection OPs. Thus, the effectiveness of counterswirl on increasing
the inducer static pressure rise and therefore PS mass flow is
extended to OPs closer to surge compared to coswirl injection. It is
therefore inferred that the synergistic effect of counterswirl injection
and mass flow removal is responsible for the larger range extension
and stagnation pressure ratio compared to a PS compressor with pre-
swirl in the PS flow.
Approach
The centrifugal compressor used in this study is a Honeywell
Turbo Technologies (HTT) ten-bladed commercial vehicle centrif-
ugal compressor. The compressor has a backsweep of 25 deg at the
impeller trailing edge with respect to the radial direction, exducer
tip diameter of Dexducer,tip ¼ 88 mm, exit/inlet wheel area ratio
0.475, ratio of volute throat area to radius 0.57 in. (14.48 mm), and
a vaneless diffuser of area ratio 1.3. The PS cavity inside the
inducer shroud includes four ribs positioned circumferentially,
with a PS slot of 2.5 mm width [12]. The height of the tip gap is 2%
of the span at the LE and 6% at the trailing edge of the impeller.
ANSYS CFX has been employed to numerically model the per-
formance of the compressor. The computational model of the full
wheel PS compressor (Fig. 1) contains three domains: the wheel
diffuser-volute and the port-inlet. The entire compressor is discre-
tized using a structured grid of 4.5 million hexahedral elements.
The tip gap is discretized using 12 elements along the span.
Steady-state Reynolds-averaged Navier–Stokes computations
have been carried out using the shear stress transport (SST) turbu-
lence model [13]. Stagnation pressure, stagnation temperature,
and that flow be axial are prescribed at the compressor inlet as
boundary conditions. Mass flow is specified at the volute
Fig. 1 Full wheel PS compressor computational domain;
frozen rotor type interface between wheel and port-inlet domain
(A and B) and between wheel and diffuser-volute domain (C)
081005-2 / Vol. 139, AUGUST 2017
Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jotuei/936123/ on 04/02/2017 Terms of Use: http://www.asme.org/ab
discharge as the outlet boundary condition. Frozen rotor type
interfaces have been used between wheel and port-inlet domain
(interfaces A and B) and between the wheel and diffuser-volute
(interface C) shown in Fig. 1. The upper bound of reduced fre-
quency encountered is approximately 0.4; quasi-steady calcula-
tions at similar values of reduced frequency are in agreement with
gas stand measurements at Honeywell Turbo.
The computed compressor stagnation pressure ratio pc and effi-
ciency g, at 0:84 Nref , 1:16 Nref , and 1:48 Nref shown in Fig. 2,
are compared against the gas stand data. Measurement uncertain-
ties in the gas stand data are assessed based on techniques
described in Ref. [14]. The results based on computational fluid
dynamics (CFD) overestimate pc by up to 3.5% at 1:48 Nref ,
while the discrepancy between experimental and computed g is
less than two points for OPs between peak efficiency and surge.
The largest discrepancies are present for high mass flow OPs
where the steep curve suggests that the compressor is close to
choke. The observed accord between CFD and experimental
measurements implies that the CFD model and techniques
employed can be used to assess the PS compressor performance.
To clarify the effect of each individual PS flow actuation, a sin-
gle passage CFD computational model of the compressor that
does not include the PS cavity or the volute is first formulated.
Specifically, the PS cavity is removed from the computational
domain of the PS compressor, and the inlet/outlet regions of the
PS cavity are modeled through the application of boundary condi-
tions (BCs) at the PS outlet and slot. This approach allows for
direct and independent control of the injected/removed flow at the
two ends of the PS cavity. Variant A of the PS actuation model,
left sketch of Fig. 3, is used when modeling flow both injected at
the PS outlet and removed at the slot. However, it is also
employed for assessing the sole actuation of flow injection at the
PS outlet by placing a solid boundary at the PS slot. Variant B is
used for assessing the effect of mass flow removal at the PS slot
(no flow is injected at the PS outlet). Furthermore, variant B is
also used to model the performance of the nonported compressor
by placing a solid boundary at the PS slot. Variants A and B of
the PS actuation model are discretized using 2.4 and 1.7 million
hexahedral elements, respectively. The tip gap is discretized using
16 elements along the span for both variants.
The following approach is taken to assess whether the perform-
ance of the geometrically coupled compressor can be determined
through substituting the PS flow with boundary conditions that
reflect the PS flow actuations:
(1) An axisymmetric PS variant without the volute and ribs
inside the PS cavity is used as a baseline compressor. This
is done to remove the geometric asymmetry of the
compressor.
(2) OPs corresponding to operation near peak g and surge for
the axisymmetric PS compressor at 0:84 Nref and 1:16 Nref
are computed.
Fig. 2 Comparison between numerical and gas stand meas-
urements pc (left) and g (right) for ported compressor
Transactions of the ASME
 Fig. 3 Single passage PS actuation model used for modeling
flow at both PS outlet and slot or only at PS outlet (a); when
flow is removed at PS slot or for the nonported compressor var-
iant (b) is used
(3) Flow quantities at the PS outlet and slot of the axisymmet-
ric PS compressor, listed in Table 1, are then averaged and
used as BCs in the geometrically uncoupled PS actuation
model.
(4) The performance of the PS actuation model (variant A in
Fig. 3) is then assessed against that of the baseline axisym-
metric PS compressor.
Figure 4 shows the computed compressor pc and g for the axi-
symmetric PS compressor and the single passage PS actuation
model at 0:84 Nref and 1:16 Nref . Comparison between the two
shows that the compressor performance characteristics are cap-
tured through the use of the single passage PS actuation model.
The discrepancy between the two CFD models is approximately
1.5 points in g and less than 2% in terms of pc . The agreement
between geometrically coupled and uncoupled models implies
that the PS actuation model can be used as a platform to assess the
effect of individual flow actuations on the compressor
performance.
Results
The PS actuation model is used to assess the effect of the PS
flow actuations listed in Table 2. We will first elucidate the domi-
nant role of mass flow removal in reducing flow blockage,
increasing blade loading and enhancing pressure rise capability,
reducing loss generation, and increasing stability margin potential.
The attendant effect associated with only introducing upstream
counterswirl through angular momentum injection is shown to
increase the pressure ratio but decreases efficiency. However,
there is a beneficial amplifying effect of counterswirl injection
combined with mass flow removal in enhancing the operable
range and stagnation pressure ratio relative to a PS compressor
with preswirl—and this is demonstrated in the computations with
integrated flow actuations involving angular momentum injection
with mass flow removal. Finally, a back-to-back assessment of
discarding the PS flow at the PS slot versus recirculating it
through the PS cavity is shown. The results indicate that removing
the flow at the PS slot increases the compressor pressure ratio
while succumbing to negligible efficiency penalty compared to a
recirculating axisymmetric PS compressor.
Mass Flow Removal at PS Slot. The actuation of mass flow
removal at the PS slot is isolated from the flow actuations at the
PS outlet by using variant B of the PS actuation model of Fig. 3.
The approach taken is to select a baseline inlet mass flow m_ in at
Table 1 Flow quantities used as BCs at PS outlet and PS slot
in PS actuation model
Location Boundary condition
M
PS outlet _ Tt , aM
m,
PS slot m_
Journal of Turbomachinery
Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jotuei/936123/ on 04/02/2017 Terms of Use: http://www.asme.org/ab
Fig. 4 Compressor pc (left) and g (right) for PS actuation model
and axisymmetric PS compressor at 0:84 Nref and 1:16 Nref
Table 2 Examined PS flow actuations at their respective
locations
Flow actuation Location
Mass flow removal PS slot
Mass flow injection PS outlet
Angular momentum injection PS outlet
Angular momentum injection PS outlet
Mass flow removal PS slot
the compressor inlet and increase the fraction of mass flow
removed at the PS slot, uPS slot ¼ m_ PS slot =m_ in . The m_ in of the
compressor stays constant, while the m_ out decreases with increas-
ing values of uPS slot . The baseline OPs selected at 1:16 Nref are
listed in Table 3. The first OP corresponds to the peak g OP for
the nonported compressor, and the second OP corresponds to
operation closer to surge. The region upstream of the PS slot
(inducer) works at a higher mass flow compared to the exducer
region. To assess the effect of mass flow removal consistently, the
inducer and exducer regions of the impeller are treated separately
when compared against nonported cases. For instance, a case of
mass flow removal with an inlet mass flow of m_ in =m_ ref ¼ 0:85
and a fraction of removed mass flow of uPS slot ¼ 0:5 will be
compared against a nonported OP with m_ in =m_ ref ¼ 0:85 at the
inducer and a nonported OP with m_ in =m_ ref ¼ 0:43 at the exducer.
The effect of increasing the fraction of mass flow removed at
the PS slot on the compressor pressure ratio is shown for the OPs
listed in Table 3 in Fig. 5 on a m_ out basis. The solid line data
points correspond to single passage nonported OPs. The dashed
line data points are OPs with mass flow removed at the PS slot for
which the baseline is m_ in =m_ ref ¼ 1:21 (red circle in Fig. 5). The
dotted line data points correspond to m_ in =m_ ref ¼ 0:85 (green
circle in Fig. 5). Up to 50% of the m_ in is removed at increments of
10% (i.e., Dðm_ PS slot =m_ in Þ ¼ 0.1). The results indicate that the
removal of mass flow through the PS slot increases the compres-
sor pc by up to 13% when uPS slot ¼ 0:5.
Two scenarios are considered for determining the compressor
efficiency when removing mass flow at the PS slot. The first sce-
nario involves reusing the removed PS flow. This approach gives
an upper bound of efficiency. The second approach involves
Table 3 Baseline OPs used for assessment of mass flow
removal at PS slot on compressor performance at 1:16 Nref
Description m_ in =m_ ref
Peak g 1:21
Near surge 0:85
AUGUST 2017, Vol. 139 / 081005-3

Fig. 5 Effect of mass flow removal at PS slot on compressor
pc at 1:16 Nref
discarding the removed PS flow and all the work that the compres-
sor has done on the flow, thus giving a lower bound. The approach
used to calculate the efficiency for both scenarios is shown in
Appendix A. The effect of mass flow removal on the compressor
efficiency is shown in Fig. 6. The solid line corresponds to
nonported OPs. The dashed line with square points (reuse PS
stream) and dotted line with square points correspond to mass
flow removal OPs corresponding to inlet mass flow of
m_ in =m_ ref ¼ 1:21. The dashed line with triangular points (reuse PS
stream) and dotted line with triangular points correspond to mass
flow removal OPs with m_ in =m_ ref ¼ 0:85.
When the PS stream is fully utilized, the effect of removing
mass flow increases the compressor efficiency compared to non-
ported OPs working at same m_ out by more than ten points for
mass flow removal fractions of uPS slot ¼ 0:5. However, when the
PS stream is discarded, the efficiency is found to decrease by up
to eight points compared to nonported OPs of equal m_ out .
With the impact of mass flow removal on the compressor per-
formance characteristics (pressure ratio and efficiency) elucidated,
we next proceed to establish the traceability of the compressor
performance changes to the underlying flow processes. These flow
processes include the reduction of flow blockage, increase of
impeller blade loading and pressure rise capability, decrease of
loss generation, and enhancement of the stability margin
potential.
Mass flow removal leads to flow blockage reduction both
upstream and downstream of the PS slot. This is traced to two
mechanisms: (i) removing a fraction of the inlet flow results in the
inducer seeing a larger mass flow compared to the remaining
impeller (and therefore further away from surge) and (ii) to the
(partial) removal of the inducer tip leakage flow. The streamwise
Fig. 6 Effect of mass flow removal at PS slot on compressor g
at 1:16 Nref
081005-4 / Vol. 139, AUGUST 2017
Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jotuei/936123/ on 04/02/2017 Terms of Use: http://www.asme.org/ab
evolution of the impeller flow blockage, expressed as a fraction of
the geometric area, is quantified in Fig. 7. Flow blockage at cross-
flow planes along the impeller is quantified using the blockage
calculation method of Khalid, based on the concept of two-dimen-
sional boundary layer thickness [15]. The OP at which 50% of the
flow is removed at the PS slot (dotted line) has an inducer mass
flow of m_ in =m_ ref ¼ 0:85. The OP corresponding to the dotted line
has the same m_ in with the solid curve and the same m_ out with the
dashed curve.
The implications of the results in Fig. 7 are as follows:
First, upon removing a portion of the inducer flow through the
PS slot, the inducer in effect operates at a higher mass flow than
the rest of the impeller and with lower flow blockage. By compar-
ing the dashed to the solid line, both nonported OPs, the impact of
increasing the inducer mass flow in reducing the inducer flow
blockage can be quantified to be approximately 8% [16]. How-
ever, comparing the dotted line (OP corresponding to mass flow
removal) to the solid line (nonported OP corresponding to
m_ in =m_ ref ¼ 0:85 of equal inducer mass flow) shows that operat-
ing the inducer at higher mass flow is not the only source of flow
blockage reduction as there is an additional decrease of flow
blockage by approximately 10%. Therefore, the total reduction in
blockage between nonported OP (dashed red) and mass flow
removal OP (dotted green) is approximately 18%.
Second, the reduction in inducer flow blockage between solid
and dotted lines is traced to the removal of the recirculation zone
in the inducer shroud region driven by the inducer tip leakage
flow. Furthermore, by removing a portion of the inducer tip leak-
age vortex, the downstream exducer flow blockage is also reduced
by up to 10%.
To summarize, operating the inducer at higher mass flow condi-
tions than the rest of the stage only accounts for less than half of
the estimated changes in blockage between nonported and ported
compressors. Thus, allowing the inducer to operate at OP with
higher mass flow than the rest of the stage cannot by itself explain
the performance changes observed in a ported compressor. The
additional benefits of removing a portion of the inducer tip leak-
age flow are more significant (as will be seen in the results to be
presented below). All the effects associated with mass flow
removal occur simultaneously and cannot be separated in practice.
We next assess the effect of mass flow removal on the inducer
static pressure rise. The impeller static pressure rise coefficient
Cp ¼ ððp  pwheel inlet Þ=ð1=2qinlet U 2 tip ÞÞ is shown in Fig. 8.
Increasing the inducer mass flow from m_ in =m_ ref ¼ 0:43 (dashed)
to m_ in =m_ ref ¼ 0:85 (solid) has no influence on the inducer static
pressure rise at these OPs near surge. In the context of PS com-
pressor operation, it can be inferred that at near surge OPs, the
fact that the inducer works at a higher mass flow than the exducer
Fig. 7 Effect of increasing fraction of mass flow removed at PS
slot on impeller blockage with respect to baseline nonported
m _ ref 5 0:85 OP at 1:16 Nref
_ in =m
Transactions of the ASME
 does not contribute to the increase of the impeller static pressure
rise. Comparing the dotted to solid curve shows a more than
100% increase in the static pressure rise at the inducer exit upon
removing 50% of inducer mass flow. In contrast, the Cp rise in the
exducer between mass flow removal OP (dotted) and nonported
compressor of the same m_ out (dashed) is approximately 10%. The
computed changes in pressure rise coefficient are rationalized
below based on the use of a control volume analysis and method
of influence coefficients detailed in Appendix B.
We use the control volume analysis and method of influence
coefficients to quantify the following individual contributing
effects to the impeller static pressure rise [16]:
(i) diffusion of the main core stream, associated with changes
of main flow stream area between locations upstream and
downstream of the PS slot location, due to removing a
fraction of the inducer mass flow dm_
(ii) blade loading
(iii) compressibility
(iv) diffusion associated with changes of effective flow area,
Aeff ¼ A  Ab , due to changes in flow blockage
(v) changes in radius through the centrifugal force
The difference of Cp rise inside the inducer, between mass flow
removal OPs and nonported compressor OPs, accounts for more
than 75% of the static pressure difference at the impeller outlet.
Therefore, the region of primary interest is the inducer. The break-
down of the inducer DCp into its contributing mechanisms
between the mass flow removal OP at m_ in =m_ ref ¼ 0:85 with
uPS slot ¼ 0:5 and the baseline nonported compressor OP at
m_ in =m_ ref ¼ 0:85 is shown in Fig. 9. When removing 50% of the
inducer flow, the diffusion of the core flow due to mass flow
removal accounts for approximately 35% of the inducer static
pressure rise, the increase of the blade loading accounts for 50%,
and the increase of the Aeff (due to the decrease of the flow block-
age) for the remaining 10%.
The increase of blade loading for the mass flow removal OPs is
due to the decrease of flow blockage near the inducer shroud
which results in a larger turning of the flow. The inducer blade
loading DCp ¼ ððPpressure side  Psuction side Þ=ð0:5qinlet U 2 tip ÞÞ at
span values of 50% and 80% is shown in Fig. 10. The increase of
the blade loading due to mass flow removal at 80% span is
approximately three times higher than that estimated at 50% span.
This is in accordance with the computed flow blockage, shown at
a crossflow plane upstream of the PS slot in Fig. 11, which shows
that the reduction of flow blockage is primarily located at span
values greater than 70%.
In conclusion, the decrease of the flow blockage associated
with the removal of the tip leakage flow is responsible directly for
approximately 10% of the increase in the inducer Cp rise between
Fig. 8 Effect of mass flow removal at PS outlet on impeller Cp
rise with respect to baseline nonported m _ ref 5 0:85 OP at
_ in =m
1:16 Nref
Journal of Turbomachinery
Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jotuei/936123/ on 04/02/2017 Terms of Use: http://www.asme.org/ab
Fig. 9 Breakdown of inducer DCp rise between mass flow
removal OP at m _ ref 5 0:85 with uPS Slot 5 0:5 and baseline
_ in =m
nonported compressor OP m _ in =m
_ ref 5 0:85 at 1:16 Nref
mass flow removal and nonported compressor OPs. Increase of
the flow turning, due to the increase of blade loading, is shown to
be a consequence of the decrease of the flow blockage and
accounts for up to 50% of the static pressure rise for OPs near
surge. Therefore, up to 60% of the computed inducer Cp increase
can be attributed to the decrease in flow blockage.
The impact of mass flow removal on loss generation is exam-
ined next. Removing a portion of the inducer mass flow leads to a
decrease of the entropy generated both upstream and downstream
of the PS slot. This can be inferred from the results in Fig. 12
where the entropy generation upstream of the impeller inlet is on
the left chart and that inside the impeller on the right chart. The
entropy generation is calculated as the net sflux across a control
volume and normalized by the total enthalpy flux according to the
below equation
ð ð ð 
Tt;out sout dm_ out þ sPS out dm_ PS out  sin dm_ in
ð ð ð (1)
ht;out dm_ out þ ht;PS out dm_ PS out  ht;in dm_ in
Comparison between nonported compressor m_ in =m_ ref ¼ 0:85
(solid) and m_ in =m_ ref ¼ 0:43 (dashed) OPs shows that increasing
the inducer mass flow results in the decrease of the entropy gener-
ated upstream of the PS slot. Specifically, the decrease of the
entropy generated upstream of the wheel inlet and inside the
inducer corresponds to approximately five and four points of effi-
ciency, respectively. The smaller extent of the recirculation zone
near the compressor shroud for the nonported compressor
m_ in =m_ ref ¼ 0:85 OP is responsible for the lower entropy genera-
tion upstream of the PS slot location. These results imply that for
Fig. 10 Effect of mass flow removal on inducer blade loading
for mass flow removal OP at m _ ref 5 0:85 with uPS Slot 5 0:5
_ in =m
and baseline nonported compressor OP m _ ref 5 0:85 at
_ in =m
1:16 Nref
AUGUST 2017, Vol. 139 / 081005-5

Fig. 11 Flow blockage at crossflow plane upstream of PS slot;
_ ref 5 0:85 (left) and mass flow removal OP
_ in =m
nonported OP m
m _ ref 5 0:85 with uPS Slot 5 0:5 (right) at 1:16 Nref
_ in =m
a recirculating PS compressor, the inducer operating at a higher
mass flow would improve the compressor efficiency relative to
that of a nonported compressor OP of equal m_ out . The entropy
generated inside the impeller upstream of the PS slot for the solid
curve is approximately 20% higher than that of the mass flow
removal OP (dashed curve), which corresponds to approximately
one point lower overall efficiency. The decrease in loss generation
is due to the partial removal of the recirculation zone associated
with the inducer tip leakage flow. In the exducer region, Fig. 12
indicates that the entropy generated downstream of the PS slot for
the mass flow removal case is approximately 35% less than that
for the nonported compressor OP at m_ in =m_ ref ¼ 0:43. The
decrease of entropy generation (for the OP with mass flow
removal relative to the nonported compressor OP with equal m_ out )
occurs at spanwise locations greater than 50%. These computed
trends can be traced to the partial removal of the inducer tip leak-
age flow, which result in a lower loss generation (from viscous
mixing with the main core flow) in the downstream exducer
region.
The final item associated with mass flow removal considered is
its impact on compressor stability. The slope of the compressor
total-to-static pressure rise characteristic @Dpts =@ m, _ where
Dpts ¼ pout  pt;in , has been used as an indicator for the likeli-
hood of compressor instability [10,17,18]. When the slope
becomes positive, the compressor system can amplify disturban-
ces leading to compressor surge. The slope of the compressor
total-to-static pressure rise, @D^p ts =@ðm_ out =m_ ref Þ with
D^p ts ¼ Dpts =ð1=2qinlet U 2 tip Þ, is used to assess the effect of
mass flow removal on the compressor stability. Figure 13 shows
@D^ p ts =@ðm_ out =m_ ref Þ for nonported compressor OPs (solid)
and OPs with mass flow removal at a baseline inlet flow of
Fig. 12 Effect of mass flow removal on entropy flux upstream
of impeller inlet (left) and inside wheel (right) with respect to
baseline nonported m _ ref 5 0:85 OP at 1:16 Nref
_ in =m
081005-6 / Vol. 139, AUGUST 2017
Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jotuei/936123/ on 04/02/2017 Terms of Use: http://www.asme.org/ab
m_ in =m_ ref ¼ 1:21 (dashed) and m_ in =m_ ref ¼ 0:85 (dotted). The slope
of the total-to-static pressure rise for OPs with mass flow removal
is negative and has a lower value compared to nonported com-
pressor OPs of same m_ out . Even though a positive slope of the
total-to-static pressure rise may not necessarily indicate surge in
the case of centrifugal compressors, nevertheless a more negative
slope can be viewed to be stabilizing. Thus, higher static pressure
at the impeller outlet for OPs with mass flow removal (relative to
that for nonported compressor OPs) due to the diffusion of the pri-
mary stream and the decrease of flow blockage can be considered
as positive attributes for improving the compressor stability
margin.
Mass Flow Injection at PS Outlet. The assessment of mass
flow injection at PS outlet on the compressor operation is based
on variant A of PS actuation model (Fig. 3). The approach fol-
lowed is to select a baseline inlet mass flow m_ in at the compressor
inlet and increase the fraction of mass flow injected at the PS out-
let m_ PS outlet =m_ in . The m_ in of the compressor remains constant,
while the m_ out increases with increasing values of uPS Out . The
flow at the PS outlet is injected upstream along the axial direction
at the same Tt as the flow at the upstream main flow inlet. This is
considered justified since the focus of this investigation is to
assess only the actuation of mass flow injection.
The solid curve data points in Fig. 14 correspond to single pas-
sage nonported compressor OPs. The dashed curve data points
correspond to OPs where the mass flow is injected at the PS outlet
with respect to a baseline case of m_ in =m_ ref ¼ 0:63 (indicated with
the circle) at 1:16 Nref . The fraction of flow injected is m_ PS outlet =
m_ in ¼ 0:15; 0:35, and 0:5. The difference in the compressor per-
formance between OPs where mass flow is injected normal to the
PS outlet and that of nonported compressor OPs of the same m_ out
is found to be negligible. It is therefore inferred that the actuation
of mass flow injection results in the inducer working at a higher
mass flow than the rest of the compressor, in the framework of a
recirculating PS compressor, and therefore further away from
surge. However, there is no additional benefit from mass flow
injection, compared to the performance of the inducer of a non-
ported OP of equal m_ ind .
Angular Momentum Injection at PS Outlet. Injection of the
flow at the PS outlet in a direction opposite to the wheel rotation
increases the compressor stagnation pressure ratio and the impel-
ler static pressure rise while succumbing to a penalty in efficiency
compared to OPs where flow is injected with zero or coswirl. The
effect of angular momentum injection on the compressor perform-
ance is shown to decrease at OPs with high flow blockage
upstream of the blade LE.
Variant A of the PS actuation model (Fig. 3) is also used to
assess the effect of angular momentum injection at the PS outlet
on the compressor performance. For each investigated OP, the
compressor exit mass flow m_ out and mass flow fraction of the
Fig. 13 Effect of mass flow removal at PS slot on compressor
stability at 1:16 Nref
Transactions of the ASME
 Fig. 14 Effect of mass flow injection at PS outlet on compressor pc (left) and g (right) at
1:16 Nref

injected flow at the PS outlet uPS Out ¼ m_ PS outlet =m_ out are fixed.
Tt of the injected flow is equal to that of the inlet main flow. The
swirl angle of the injected stream a is varied from 50 deg  a
 50 deg. The investigated swirl angle range was selected as rep-
resentative from values calculated at the PS outlet of the axisym-
metric PS compressor shown in Fig. 4 for OPs from peak g to
surge. Positive values of a correspond to flow injected in the
direction of the wheel rotation (coswirl) and negative values to
counterswirl. The effect of a at the PS outlet on the compressor
performance is assessed for the OPs between peak g and near
surge at 1:16 Nref listed in Table 4. The mass flow fraction
injected for all the cases is set to uPS out ¼ 0.4.
The compressor pc and g are plotted as a function of the swirl
angle a of the injected flow in Fig. 15. Introduction of
counterswirl to the injected PS stream tends to increase pc and
decrease g. This effect is amplified at OPs working at higher m_ out .
For peak g OPs, a change of a corresponding to 100 deg results in
an increase of pc by approximately 12% and a decrease of effi-
ciency g by five points. However, for the near surge OP, the same
change in a results in less than 0.5% increase of pc and a 2.5 point
drop in efficiency g.
The compressor performance at OP near surge is less sensitive
to changes of the injected flow angle a (for an equal fraction of
injected mass flow at the PS outlet uPSout ) compared to the OP at
peak efficiency. The different behavior can be understood in terms
of the relative flow angle at the impeller blade LE, b1 . The differ-
ence of b1 between counterswirl (a ¼ 50 deg) and coswirl
(a ¼ 50 deg) is found to be up to 12 deg in the shroud region
(>80% span) for the peak g OP. In contrast for the near surge OP
m_ out =m_ ref ¼ 0:63, the largest difference between counter- and
coswirl is found to be less than 5 deg at midspan, implying that
the incidence angle is almost the same for both coswirl and
counterswirl.
The circumferential velocity uh for the counterswirl injection
OPs at m_ out =m_ ref ¼ 1:21 and m_ out =m_ ref ¼ 0:63 is shown in Fig.
16. Positive values of uh (preswirl) indicate fluid to which the
compressor has already imparted work to and is therefore used as
an indicator to define the extent of the inducer recirculation zone.
For OPs near surge, there is a recirculation zone near the inducer
shroud region.
To assess how the recirculation zone near the inducer shroud
affects the impeller relative flow angle, the control volume (CV)
shown in Fig. 17 is used. Station i corresponds to a location
upstream of the blade LE, where the PS and main stream flow are
assumed to have mixed out entirely. Station e corresponds to a loca-
tion between Station i and the blade LE; the recirculation zone in
the region between Station i and the blade LE blocks a portion of
the geometric area, so that the flow passes through a smaller effec-
tive area Ae;eff . Mixing out of the PS flow, which contains angular
momentum, with the main stream flow results in a flow at Station i
corresponding to solid body rotation of angular velocity Xi and uni-
form uaxial;i . The angular velocity Xi is given by
rPS
Xi ¼ 2uuaxial;PS tan aPS (2)
r2 i
where
m_ PS
u¼
m_ PS þ m_ inlet
The swirl angle at Station i is given by

Table 4 Baseline OPs used to assess influence of angular

momentum injection on compressor performance at 1:16 Nref

Description m_ out =m_ ref uPS out

Peak g 1:21 0:4

Intermediate 0:83 0:4
Near surge 0.63 0:4 Fig. 15 Effect of angular momentum injection at PS outlet on
compressor pc (left) and g (right) at 1:16 Nref

Journal of Turbomachinery AUGUST 2017, Vol. 139 / 081005-7

Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jotuei/936123/ on 04/02/2017 Terms of Use: http://www.asme.org/ab

 Ai rPS
tan ai ¼ 2u2 tan aPS 2 r (3)
APS ri
Equation (3) shows that ai is a function of the domain geome-
try, u and aPS . Therefore, ai is the same for all the OPs provided
that the mass flow fraction u and swirl angle a of the injected
stream are equal. The flow from Station i to Station e can be
viewed as a nozzle flow. Assuming incompressible and
inviscid flow, based on the analysis shown in Ref. [19], ai is
related to ae by pffiffiffiffiffiffiffiffiffiffiffiffi
ae =ai  Ae =Ai (4)
It can thus be inferred from Eq. (4) that for OPs near surge, the
decrease of the effective flow area upstream of the blade LE, due
to the recirculation zone, leads to the decrease of a in the absolute
frame. Reducing a in the absolute frame results in a subsequent
reduction of the absolute and relative frame flow angle difference
between coswirl and counterswirl angular momentum injection
OPs. The difference in the inlet flow angle between counter and
coswirl angular momentum injection situations decreases from the
peak g OP to OPs near surge along the speedline. This is because
the extent of the inducer recirculation zone, in both upstream and
spanwise direction, increases continuously when operating closer
to OP near surge as indicated in Fig. 16.
The efficiency penalty at the peak g OP m_ out =m_ ref ¼ 1:21
between a ¼ 50 deg and a ¼ 50 deg injection is found to be five
points as shown in Fig. 15. The entropy generation through the
impeller passage is shown in Fig. 18. The solid lines correspond
to the entropy generated across the entire passage span, while the
dashed lines correspond to the entropy generated in the top 20%
of the span. The data show that the counterswirl injection case
increases the impeller entropy generation by approximately 50%,
which corresponds to a decrease of five points in efficiency. More
than 60% of the higher loss in the impeller passage between
counterswirl and coswirl is accounted for by the higher entropy
generated in the shroud region.
The higher entropy generated in the impeller shroud region for
the a ¼ 50 deg OP is due to the higher mixing loss between tip
leakage and main flow. To assess this, a mixed out analysis of the
tip leakage and main flow is carried out. Since the tip leakage
flow rate is small compared to the main passage flow (<3%), the
following expression for the entropy generated due to mixing of
the two streams is used [20]:
Fig. 16 Contour of uh =Utip indicating that the extent of the
inducer recirculation zone increases when moving closer to
OPs near surge at 1:16 Nref
081005-8 / Vol. 139, AUGUST 2017
Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jotuei/936123/ on 04/02/2017 Terms of Use: http://www.asme.org/ab
Fig. 17 Control volume upstream of impeller inlet
Fig. 18 Effect of angular momentum injection on impeller
entropy generation at 1:16 Nref
( )

Ds m_ tip c1 2 Tt;tip  Tt;main
¼ 1þ M main
Cp m_ main 2 Tt;main
  
m_ tip Vtip cos d
þ ðc  1ÞM2 main 1  (5)
m_ main Vmain
where d is the relative angle between the tip leakage flow (sub-
script “tip”) and main flow (subscript “main”).
The entropy generated in the top 20% of the impeller span
based on CFD results and that computed based on the mixed out
analysis for the CV is given in Table 5. The results based on the
mixed out analysis are in accord with those from CFD. We thus
infer that the additional entropy generated in the top 20% span for
the counterswirl OP is due to the higher mixing losses.
Combined Effect of Mass Flow Removal and Angular
Momentum Injection. The results presented thus far provide a
framework for quantifying the combined impact of the PS flow
actuations and hence the operation of the PS compressor. We
summarize here the key results on compressor performance with
removal of mass flow in combination with injection of angular
momentum at equivalent OPs as delineated in Table 6.
The computed compressor pc and g for the OPs listed in Table
6 are shown in Fig. 19. The OP with combined mass flow removal
and counterswirl angular momentum injection (diamond) is found
to have the highest compressor pc , followed by the OP with only
mass flow removal (square). For OPs with flow injection at the PS
outlet, the mixing loss between the injected and inlet stream
upstream of the wheel inlet decreases the Pt leading to a penalty
in terms of the compressor pc . This additional loss, which scales
with ðuaxial;PS out  uaxial;in Þ2 , and decreases the overall compressor
efficiency by 1.5 point, is absent in the situation with mass flow
removal only. Relative to the nonported compressor OP
Table 5 Comparison of CFD and mixed out analysis values of
entropy generated Dsflux Tt; out =Dht; f lux in top 20% of span
OP CFD Mixed out
m_ out =m_ ref ¼ 1:21 and a ¼ 50 deg 0.042 0.044
m_ out =m_ ref ¼ 1:21 and a ¼ 50 deg 0.057 0.061
Transactions of the ASME
 Table 6 Table of computed OPs for assessing effect of combi- the OP with mass flow removal only (square) is approximately
nation of mass removal and angular momentum injection at equal to that for the OP with combined mass flow removal and
1:16 Nref coswirl injection (circle), while the efficiency at OP with equiva-
^_ ¼ 0:63 ^_ ¼ 0:63 ^_ ¼ 0:63 ^_ ¼ 0:63 m
^_ ¼ 0:89
lent mass flow removal and counterswirl injection (diamond) is
m m m m
approximately 3.5 points lower than that corresponding to coswirl
a ¼ 65 deg a ¼ 65 deg Mass removal Nonported Nonported
injection. The effect of mass flow removal only is responsible for
m_ out
0.63 0.63 0.63 0.63 0.89 the decrease in impeller entropy generation by approximately
m_ ref
25% relative to the OP for nonported compressor at m^_ ¼ 0:63
m_ in
m_ ref 0.63 0.63 0.89 0.63 0.89 with same m_ out . The decrease in entropy generation is due to the
m_ ind
0.89 0.89 0.89 0.63 0.89 reduction of the backflow region near the inducer shroud, yielding
m_ ref
an efficiency increase of approximately four points. The entropy
m_ PS outlet
m_ out 0.4 0.4 — — — generation for the OP with counterswirl injection (diamond)
m_ PS slot
0.4 0.4 0.4 — — results in a penalty of 1.5 points in efficiency relative to that of
m_ out
the OP with mass flow removal only. The higher efficiency pen-
a 65 deg 65 deg — — — alty is due to the higher entropy generation in the top 20% of the
impeller span. The higher entropy generation for the counterswirl
OP can be traced to higher tip leakage flow and therefore higher
corresponding to the same outlet mass flow (up-facing triangle), mixing loss with the main flow. This is in accordance with the
the pc for the OP with mass flow removal (square) is approxi- entropy generation mechanisms associated with counterswirl
mately 14% higher, while for the OP with combined mass flow injection described above.
removal and counterswirl injection (diamond) it is 15% higher.
The difference in Cp increase between OPs at the inducer exit
is shown to be approximately equal to that at the impeller exit, Discarding the Flow at the PS Slot Versus Recirculating.
implying that the static pressure increase of the impeller can be The results thus far have shown that the dominant flow actuations
traced to the increase in the inducer. For the OP with only mass are that of mass flow removal at the PS slot and counterswirl
flow removal, the impeller Cp increases by approximately 30% momentum injection at the PS outlet. In this section, the effect of
relative to that of the nonported compressor OP with same m_ out , discarding the flow removed at the PS slot is compared against a
while the Cp for the OP with combined mass flow removal and recirculating geometrically coupled PS compressor.
counterswirl is approximately 10% higher than that of the OP An axisymmetric PS compressor variant without the volute and
with mass flow removal only. It is therefore inferred that the ribs inside the PS cavity is used as the baseline recirculating com-
removal of mass flow is the dominant actuation in setting the PS pressor. This is done to remove the geometric asymmetry of the
compressor pc . compressor. A number of OPs for the axisymmetric PS compressor
If the PS cavity were included in the computational model, the are computed after which the mass flow recirculating through the PS
higher static pressure rise at the inducer due to counterswirl would cavity m_ PS =m_ ind , shown in Fig. 20, is used as a boundary condition
result in a higher pressure difference between the PS slot and out- on the PS slot of variant B of the PS actuation model. The m_ out
let. The higher pressure difference would increase between the two models is specified to be the same, while the com-
the recirculating mass flow ratio from m_ PS outlet =m_ out ¼ 0:4 to pressor inlet mass flow for the PS actuation model is equal to the
approximately m_ PS outlet =m_ out ¼ 0:43; likewise, the inducer inducer mass flow m_ ind of the axisymmetric PS compressor.
mass flow would then increase from m_ inducer =m_ out ¼ 0:89 to The computed compressor pressure ratio and efficiency at
m_ inducer =m_ out ¼ 0:92. Consequently, the inducer would be work- 1:16 Nref for both axisymmetric PS compressor and mass flow
ing at operating conditions further away from surge with increas- removal OPs are shown in Fig. 21. The calculated efficiency val-
ing fraction of recirculating flow through the PS cavity. Both of ues for the mass flow removal OPs are calculated taking into
these effects, resulting from the higher inducer static pressure rise account that the PS flow removed is discarded and all the work
associated with counterswirl angular momentum injection done on the PS stream is lost (see Appendix A). Comparison of
superimposed on mass flow removal, should serve to enhance the the compressor performance between the axisymmetric PS and
compressor operable range significantly. mass flow removal OPs on an m_ out basis shows that discarding
As noted above, the mixing of the flow injected at the PS outlet and not recirculating the PS stream increases the pc by up to 10%
with that of the main stream for OPs with mass flow removal and while succumbing to less than one point of efficiency penalty.
with co- or counterswirl injection results in the decrease of the The increase of the impeller pressure ratio between the mass
overall efficiency by approximately 1.5 points. The efficiency for flow removal OPs and recirculating PS is approximately 7%. The
remaining 3% is accounted for by the lower DPt upstream of the
wheel inlet for the mass flow removal case since there is no flow

Fig. 19 Combined effect of mass flow removal at PS slot and

angular momentum injection at PS outlet on compressor pc _ PS =m
Fig. 20 Fraction of recirculating mass flow m _ ind for axi-
(left) and g (right) at near surge OP at 1:16 Nref symmetric PS compressor at 1:16 Nref

Journal of Turbomachinery AUGUST 2017, Vol. 139 / 081005-9

Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jotuei/936123/ on 04/02/2017 Terms of Use: http://www.asme.org/ab


Fig. 21 Compressor pc (left) and g (right) for axisymmetric PS
compressor and single passage PS actuation model with flow
removal at PS slot at 1:16 Nref
exiting from the PS outlet to mix with the main inlet flow. The
flow exiting the PS outlet for the axisymmetric PS OPs contains
preswirl due to the work already added by the inducer. The
impeller relative inlet flow angle for the near surge OP
m_ out =m_ ref ¼ 0:59 is found to be approximately 10 deg smaller for
the axisymmetric PS compressor compared to that of the mass
flow removal OP. The higher relative inlet flow angle for the mass
flow removal OPs leads to higher blade loading which results in
higher compressor pc .
The entropy generation inside the impeller for the mass flow
removal OP is approximately 30% higher than that of the axisym-
metric ported OP. The higher entropy generation for the entire
flow passage is traced to the top 20% of the span in accordance
with the findings for counterswirl angular momentum injection at
the PS outlet. In addition to the higher entropy generated inside
the impeller, the mass flow removal OPs incur a penalty in effi-
ciency due to the discarding of the useful work of the PS stream.
For the near surge OP m_ out =m_ ref ¼ 0:59, this accounts for approxi-
mately 13 points in efficiency. For the axisymmetric PS compres-
sor, the entropy generated inside the PS cavity and upstream of
the wheel inlet accounts for approximately 17 points of efficiency.
The net result is a similar efficiency between recirculating and
mass flow removal OPs as shown in Fig. 21.
In conclusion, these results indicate that discarding rather than
recirculating the PS flow combines the effect of both mass flow
removal and removing the preswirl from the injected flow at the
PS outlet. The combined result of these actuations is the increase
of the compressor pressure ratio, while the difference in the effi-
ciency is shown to be negligible.
Summary and Conclusions
To quantify the effect of the PS flow actuations on the compressor
performance, a single passage CFD model of the compressor, which
does not include the PS cavity, is used as a computational platform.
The PS flow actuations are modeled through appropriate boundary
conditions at the PS outlet and slot locations. This approach allows
for a flow actuation to be isolated and its effect on the performance
individually assessed. The effect of three actuations on the compres-
sor performance is assessed; mass flow removal at the PS slot, as
well as mass flow, and angular momentum injection at the PS outlet.
The key findings of this assessment are as follows:
(1) Mass flow removal is the dominant PS flow actuation in set-
ting the PS compressor performance. Increasing the fraction
of mass flow removed results in the increase of the compres-
sor pressure ratio. The removal of mass flow results in the
deceleration of the compressor main stream and reduction of
the impeller flow blockage. The decrease in blockage is due
to the reduction of the recirculation zone extent near the
081005-10 / Vol. 139, AUGUST 2017
Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jotuei/936123/ on 04/02/2017 Terms of Use: http://www.asme.org/ab
inducer shroud through the removal of the inducer tip leakage
flow and is shown to increase the blade loading. The partial
removal of the recirculation zone decreases the entropy gen-
erated inside the impeller compared to a nonported compres-
sor OP of equal outlet mass flow.
(2) The introduction of counterswirl to the PS flow at the PS
outlet increases the compressor pressure ratio and static
pressure rise and decreases the efficiency. The higher pres-
sure ratio is a result of the impeller doing more work on the
flow due to the higher inlet flow angle compared to coswirl
injection. The decrease in efficiency is due to the higher
mixing loss between tip leakage and main flow. The effect
of angular momentum injection on the compressor per-
formance is shown to decrease at OPs with high flow block-
age upstream of the blade LE.
(3) No additional benefit from mass flow injection is found in
the inducer when compared to a nonported compressor of
same mass flow.
(4) Representing the PS flow in terms of its associated flow actua-
tions on compressor provides a framework for rationalizing the
operation of PS compressor and optimizing the attributes of PS
flow conducive to maximizing compressor performance.
Finally, we can infer from the findings the following design
implications:
(1) By removing the PS flow at the PS slot rather than recircu-
lating through the PS cavity, the compressor pressure ratio
increases by approximately 10% with a negligible penalty
in efficiency. These findings suggest that the development
of a casing treatment that allows for only mass flow
removal increases the compressor performance compared
to a PS compressor without the need to include a recircula-
tion channel in the compressor shroud.
(2) The framework allows optimizing the performance of com-
pressor in terms of a combination of PS flow actuations.
Once an optimal combination of flow actuations has been
determined, one can then determine the ported shroud
geometry to yield the desired combination of flow actua-
tions on the compressor flow.
Acknowledgment
The authors of this paper would like to acknowledge both the
financial support and technical expertise of Honeywell Turbo
Technologies. Specifically, we would like to thank Dr. Anish
Gupta, Sujatha Guntu, and Mithun Venkatesh for their assistance
in generating a number of computational meshes used in this
paper. Furthermore, we would like to thank Professor Nicholas
Cumpsty and Professor Edward Greitzer for their input and assis-
tance. We also thank the reviewers for their useful comments that
served to improve the clarity and technical content of this paper.
Nomenclature
Dimensional and Nondimensional Quantities
a¼ swirl angle
A¼ geometric area
Cp ¼ specific heat capacity at constant pressure
h¼ specific enthalpy
m_ ¼ mass flow rate
M¼ Mach number
N¼ angular frequency
p¼ pressure
P¼ power
q¼ dynamic head
r¼ radius
s¼ specific entropy
T¼ static temperature
u¼ velocity in stationary frame
Transactions of the ASME
 ð ð 
U¼ rotational velocity, U ¼ Xr
w¼ velocity in rotating frame Ploss; PS stream ¼ Tt;PS slot sPS _ PS slot
slot d m  sin dm_ PS slot
b¼ angle in relative frame
c¼ ratio of specific heats (A4)
d¼ angle between tip leakage and main stream
q¼ density In the case where the PS stream is discarded, all the work done by
sij ¼ shear stress tensor the compressor on the PS stream is lost, giving thus a lower bound
x¼ vorticity of efficiency. The lost power of the PS stream is then
ð ð
X¼ angular velocity
Ploss; PS stream ¼ ht;PS slot dm_ PS slot  ht;in dm_ PS slot (A5)
Subscripts and Superscripts
axial ¼ axial direction Appendix B: Separation of Impeller Static Pressure
b¼ blocked
Rise
blade ¼ blade
eff ¼ effective A method which allows for the separation of the static pressure
flux ¼ flux rise into individual contributing effects is described here [16]. The
in ¼ compressor inlet control volume shown in Fig. 22 consists of the flow passage
ind ¼ inducer between blade pressure side and suction side along an incremental
M¼ mass flow average meridional distance of dx. The radial direction coincides with the
main ¼ main flow z axis. An effective velocity in the meridional direction expressed
out ¼ compressor outlet through Eq. (B1) is defined, which takes into account that the flow
PS slot ¼ PS slot location does not occupy the geometric passage area A but rather the effec-
PS outlet ¼ PS outlet location tive area Aeff
m_
r¼ radial direction ueff ¼ (B1)
ref ¼ reference value qAeff
span ¼ spanwise direction
t¼ stagnation quantity Flow enters on the left side of the CV with m; _ ueff ; q; and p,
str ¼ streamwise direction while the flow and geometric quantities associated with the CV
tip ¼ tip leakage flow outlet are m_ þ dm, _ ueff þ dueff , q þ dq, p þ dp, and A þ dA,
wheel inlet ¼ wheel inlet Aeff þ dAeff . A portion of the incoming flow dm_ is removed
h¼ circumferential direction through an opening located on the shroud of the CV. Finally, the
1¼ leading edge location effect of the blade loading is taken into account by assigning PPS
and PSS on the sides of the CV. The PPressure side and PSuction side
Abbreviations values are taken to be constant along the meridional distance dx
CV ¼ control volume of the CV.
LE ¼ leading edge Conservation of mass is applied on the CV taking into account
OP ¼ operating point that the flow does not occupy the geometric passage area A but
PS ¼ ported shroud rather the effective flow area Aeff ¼ A  Ab
SST ¼ shear stress transport
TE ¼ trailing edge queff Aeff ¼ dm_ þ ðq þ dqÞðueff þ dueff ÞðAeff þ dAeff Þ (B2)

Conservation of momentum along the meridional direction,

assuming inviscid flow and that the pressure acts on the geometric
area A, is written as
Appendix A: Efficiency Calculation for Mass Flow
Removal at PS Slot ueff m_ þ ðueff þ dueff Þðm_ þ dmÞ
_ þ uinj dm_
The compressor efficiency is calculated through the use of ¼ pA  ðp þ dpÞðA þ dAÞ þ Dpblade Ablade sin f (B3)
entropy and stagnation enthalpy flux taking into account the com-
pression process of both PS and main flow streams in accordance where uinj is the velocity component of the flow removed in the
with the following equation: meridional direction, and Dpblade ¼ PPressure side  PSuction side .
Ploss; þ Ploss;
Main stream PS stream
g¼1 (A1)
Pcompressor

where Pcompressor is the compressor power input based on

ð ð ð
Pcompressor ¼ ht;out d m_ out  ht;PS slot d m_ PS slot  ht;in dm_ in (A2)

The lost power of the main stream is calculated through the fol-
lowing equation:
ð ð 
Ploss; Main stream ¼ Tt;out sout dm_ out  sin dm_ out (A3)

In the case where the PS stream is reused, the lost power of the PS
stream is estimated through Eq. (A4), giving an upper bound of Fig. 22 Control volume used to separate static pressure rise;
efficiency radially inward looking view

Journal of Turbomachinery AUGUST 2017, Vol. 139 / 081005-11

Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jotuei/936123/ on 04/02/2017 Terms of Use: http://www.asme.org/ab

 Equations (B2) and (B3) are combined while keeping only the [2] Uchida, H., Kashimoto, A., and Iwakiri, Y., 2006, “Development of Wide Flow
first-order terms. The effect of the centrifugal force in increasing Range Compressor With Variable Inlet Guide Vane,” R&D Rev. Toyota
CRDL, 41(3), pp. 9–14.
the pressure along the meridional direction is also included, where [3] Iwakiri, Y., and Uchida, H., 2006, “Numerical Fluid Analysis of a Variable
n is the angle between the radial and meridional directions. Geometry Compressor for Use in a Turbocharger,” R&D Rev. Toyota CRDL,
Finally, after nondimensionalizing using qref ¼ 1=2qinlet U 2 tip , the 41(3), pp. 15–21.
following expression for the impeller Cp is obtained: [4] Hunziker, R., Dickmann, H., and Emmrich, R., 2001, “Numerical and Experi-
mental Investigation of a Centrifugal Compressor With an Inducer Casing
Bleed System,” Proc. Inst. Mech. Eng., Part A, 215(6), pp. 783–791.
[5] Qiu, X., Krivitzky, E., and Bollweg, P., 2012, “Meanline Modeling of Ported
@Cp @Cp @Cp @Cp Shroud Centrifugal Compressors,” ASME Paper No. GT2012-68915.
Cp ¼ dm_ þ dq þ dAeff þ dr
@ m_ @q @Aeff @r [6] Yamaguchi, S., Yamaguchi, H., Goto, S., Nakao, H., and Nakamura, F., 2002,
“The Development of Effective Casing Treatment of Turbocharger
Dpblade Compressors,” Seventh International Conference on Turbochargers and Turbo-
þ Ablade sin f (B4) charging, London, IMechE Paper No. C602/008/2002.
Aqref
[7] Tamaki, H., 2012, “Effect of Recirculation Device With Counter Swirl Vane of
Performance of High Pressure Ratio Centrifugal Compressor,” ASME J. Tur-
where bomach., 134(5), p. 051036.
[8] Tamaki, H., Unno, M., and Kawakubo, T., 2009, “Aerodynamic Design to
  Increase Pressure Ratio of Centrifugal Compressors for Turbochargers,” ASME
@Cp ueff uinj
¼ 2þ (B5) Paper No. GT2009-59160.
@ m_ Aqref ueff [9] Sivagnanasundaram, S., Spence, S., Early, J., and Nikpour, B., 2013, “An
@Cp u2 eff Aeff Impact of Various Shroud Bleed Slot Configurations and Cavity Vanes on Com-
¼ (B6) pressor Map Width and the Inducer Flow Field,” ASME J. Turbomach., 135(4),
@q Aqref p. 041003.
[10] Yang, M., Martinez-Botas, R., Bamba, T., Tamaki, H., Zhang, Y., Zheng, X.,
and Li, Z., 2011, “Investigation of Self-Recycling-Casing-Treatment (SRCT)
@Cp qu2 eff
¼ (B7) Influence on Stability of High Pressure Ratio Centrifugal Compressor With a
@Aeff Aqref Volute,” ASME Paper No. GT2011-45065.
[11] Zheng, X., Zhang, Y., Yang, M., Bamba, T., and Tamaki, H., 2013, “Stability
Improvement of High Pressure Ratio Turbocharger Centrifugal Compressor by
@Cp qX2 r cos n Asymmetric Flow Control—Part II: Non-Axisymmetric Self Recirculation Cas-
¼ (B8)
@r qref ing Treatment,” ASME J. Turbomach., 135(2), p. 021007.
[12] Guillou, E., Gancedo, M., DiMicco, R. G., Gutmark, E., and Mohamed, A.,
2011, “Surge Investigation in a Centrifugal Compressor by Stereoscopic PIV,”
The first term in Eq. (B4) ð@Cp=@ mÞd _ m_ quantifies the effect of AIAA Paper No. 2011-742.
removing a fraction of the inlet flow through the PS slot on the [13] ANSYS Inc., 2013, “ANSYS CFX-Solver Theory Guide,” Canonsburg, PA.
static pressure. The additional static pressure rise from this mech- [14] Guillou, E., 2013, “Uncertainty and Measurement Sensitivity of Turbocharger
anism is due to the change in flow area of the main core stream Compressor Gas Stands,” SAE Technical Paper No. 2013-01-0925.
[15] Khalid, S. A., Khalsa, A. S., Waitz, I. A., Tan, C. S., Greitzer, E. M., Cumpsty,
from upstream to downstream of the PS slot location, thereby N. A., Adamczyk, J. J., and Marble, F. E., 1999, “Endwall Blockage in Axial
experiencing a larger level of diffusion. Changes in density are Compressors,” ASME J. Turbomach., 121(3), pp. 499–509.
taken into account through ð@Cp=@qÞ. The third term in Eq. (B4) [16] Christou, G. A., 2015, “Fluid Mechanics of Ported Shroud Centrifugal Com-
ð@Cp=@Aeff ÞdAeff quantifies how the static pressure is affected by pressor for Vehicular Turbocharger Applications,” Ph.D. thesis, Massachusetts
Institute of Technology, Cambridge, MA.
changes in the effective area of the compressor flow path due to [17] Hunziker, R., and Gyarmathy, G., 1992, “The Operational Stability of a Centrif-
blockage, while ðDpblade =Aqref ÞAblade sin f represents the influence ugal Compressor and Its Dependence on the Characteristics of the Sub-
of the blade loading. components,” ASME Paper No. 92-GT-284.
[18] Cumpsty, N. A., 2004, Compressor Aerodynamics, Krieger Publishing, Mala-
bar, FL.
[19] Greitzer, E. M., Tan, C. S., and Graf, M. B., 2004, Internal Flow Concepts and
References Applications, Cambridge University Press, New York.
[1] Fisher, F. B., 1988, “Application of Map Width Enhancement Devices to Tur- [20] Denton, J. D., 1993, “Loss Mechanisms in Turbomachines,” ASME J. Turbom-
bocharger Compressor Stages,” SAE Paper No. 880794. ach., 115(4), pp. 621–656.

081005-12 / Vol. 139, AUGUST 2017 Transactions of the ASME

Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jotuei/936123/ on 04/02/2017 Terms of Use: http://www.asme.org/ab