A Method to Estimate

the Performance Map

of a Centrifugal Compressor
Michael Casey1
e-mail: [email protected] Stage
Chris Robinson A novel approach to calculate the performance map of a centrifugal compressor stage is
e-mail: [email protected]
presented. At the design point four nondimensional parameters (the flow coefficient /,
the work coefficient k, the tip-speed Mach number M, and the efficiency g) characterize
PCA Engineers Limited,
the performance. In the new method the performance of the whole map is also based on
Studio 2, Deepdale Enterprise Park,
these four parameters through physically based algebraic equations which require little
Deepdale Lane,
prior knowledge of the detailed geometry. The variable empirical coefficients in the para-
Nettleham, Lincoln LN2 2LL, UK
meterized equations can be calibrated to match the performance maps of a wide range of
stage types, including turbocharger and process compressor impellers with vaned and
vaneless diffusers. The examples provided show that the efficiency and the pressure ratio
performance maps of turbochargers with vaneless diffusers can be predicted to within
62% in this way. More uncertainty is present in the prediction of the surge line, as this is
very variable from stage to stage. During the preliminary design the method provides a
useful reference performance map based on earlier experience for comparison with
objectives at different speeds and flows. [DOI: 10.1115/1.4006590]

Introduction than those tested, and does not provide a useful reference based
on the best experience derived from many different stages.
An important requirement during the preliminary design of a
The common alternative to these approaches is to use empirical
centrifugal compressor stage is the calculation of a reliable per-
correlation based methods for predicting performance maps. Such
formance map as a guide to the expected operating flow range and
methods require fairly detailed information of the stage geometry,
the sensitivity to speed variations when the design is completed.
at least on a one-dimensional (1D) basis. Examples of 1D methods
With this information the designer can assess if the design will be
are given by Swain [1], Aungier [2], Oh et al. [3], and Cumpsty
suitable for the application. For example, it is possible to check if
[4]. The authors’ experience with such methods, however, is that
a new design will provide adequate efficiency, pressure ratio, and
they often require fairly tedious tweaking of coefficients in the
surge margin on the low speed characteristics and sufficient choke
empirical models to generate a satisfactory performance map, and
margin at high speeds.
that the characteristic curves often have unreliable shapes.
This paper presents a new engineering approach to estimate the
The approach proposed here is based on the fact that well-
performance map for such preliminary design applications. This is
designed state-of-the-art compressor stages for a particular aero-
a difficult task as the final detailed geometry (such as throat areas,
dynamic duty tend to have fundamentally similar shapes of their
blade angles, blade number, blade thicknesses, etc.) and many
efficiency performance maps. This indicates that the duty itself is
aspects of the aerodynamic design (diffusion levels, flow angles,
an excellent guide to the form of the performance map. The
etc.) have not been finalized at this point. An additional difficulty
method proposed is entirely novel in that it uses four key nondi-
is that the performance map is needed very quickly, typically dur-
mensional parameters at the design point to determine the per-
ing engineering discussions on a proposed new development. The
formance map, rather than the geometry of the stage.
most accurate methods of determining the map for competing
The method has been applied to a wide range of stage types,
design solutions, by first completing the design and then making
which are categorized here as turbocharger or process style impel-
CFD simulations or even measurements, are not suitable in this
lers, with vaned and vaneless diffusers, giving four categories.
time frame.
Turbocharger impellers (which are similar to gas turbine impel-
In some cases, the new stage may be sufficiently similar to a
lers) are relatively long and un-shrouded, have an axial inlet
stage that has previously been tested such that the map of this
designed with an inducer, and are equipped with thin blades and
stage can be quickly scaled to give a good estimate of the
splitter vanes to make them suitable for transonic flow. Process
expected performance. The scaling approach is limited in that it
style impellers in multistage compressors are shrouded and
only works if such data is available, and will not work well if the
shorter, they usually have radial inlets, have thick blades most of-
design point of the new stage is sufficiently different from previ-
ten with no splitter vanes, and are generally designed for lower
ous experience, especially if the back-sweep and work coefficient
Mach numbers. The emphasis of the paper is on the application of
have changed as this changes the slope of the flow versus work
the method to predict performance maps for turbocharger style
characteristic. In addition, the scaling of maps provides no para-
impellers with vaneless diffusers, but examples are also given of
metric description, gives no speed lines at higher or lower speeds
some other categories of stage to demonstrate the validity of the
method over many applications.
The paper is organized as follows. First, some background to the
1
Also at Institute of Thermal Turbomachinery, University of Stuttgart, Pfaffen- key nondimensional parameters and their significance in this
waldring 6, D-70569 Stuttgart, Germany. method is given. This is followed by a description of the algebraic
Contributed by the International Gas Turbine Institute (IGTI) of ASME for publi-
cation in the JOURNAL OF TURBOMACHINERY. Manuscript received September 5, 2011;
equations used to parameterize the efficiency performance maps,
final manuscript received October 17, 2011; published online November 8, 2012. which explains how these equations allow the physical features of
Editor: David Wisler. the maps to be modeled for variations of flow and speed. The

Journal of Turbomachinery Copyright V

C 2013 by ASME MARCH 2013, Vol. 135 / 021034-1

Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/ on 08/10/2013 Terms of Use: http://asme.org/terms

 calibration of the coefficients in the efficiency equations using test objective of the preliminary design and detailed design but a good
data from several sources is then provided. A subsequent section estimate of the map is possible without this information.
describes the equations used for the variation of the work coefficient The design point efficiency g is defined as the polytropic effi-
with flow and speed, and the validation of these. A difficult aspect ciency; this defines the aerodynamic quality of the design and
is the prediction of surge and the pragmatic solution to this is allows a simple extension to multistage machines without so-
described in the penultimate section. Finally the paper demonstrates called reheat factors to account for the divergence of the isobars
the use of the equations to predict typical turbocharger performance in a T-s diagram. It is assumed that some system of correlations is
maps. A discussion of the results leads to the conclusions. available to characterize all the losses at the design point, and this
aspect of the preliminary performance prediction technique is not
described in any more detail here.
The nondimensional inlet flow coefficient / characterizes the
Background to the Method stage type, the flow channel width, and also provides a guide to
The objective is to calculate the values of two dependent varia- the efficiency level that may be expected, see Cumpsty [4]. Well-
bles (the polytropic efficiency and work coefficient) for specific designed radial compressors have a value of the flow coefficient
values of the independent variables (the flow coefficient and tip- close to / ¼ 0.09 6 0.01.
speed Mach number): The stage work coefficient k characterizes the work input of the
stage, including parasitic disk friction losses. Typical values are
g; k ¼ f ð/; MÞ (1) k ¼ 0.65 6 0.1. This parameter is determined essentially by the
outlet velocity triangle of the impeller, and so characterizes the
All other thermodynamic performance information can then be effect of blade number, impeller outlet width, and back-sweep
calculated from these variables, such as isentropic efficiency, angle on the work input.
pressure ratio, volume flow, and mass flow. Speed lines can then The stage tip-speed Mach number M characterizes the local
be generated for specific values of the tip-speed Mach number by Mach number levels in the stage and loading in terms of the stage
varying the flow coefficient and the full map comprises an array pressure and temperature ratio. Typical industrial process com-
of speed lines with different tip-speed Mach numbers. The equa- pressor stages have tip-speed Mach numbers between 0.5 and 1.0,
tions describing the functional dependency of these variables turbochargers for automotive applications between 0.5 and 1.5,
include several nondimensional parameters, many variable coeffi- turbochargers for larger diesel engines between 0.8 and 1.7, and
cients, and some fixed constants, and are described in the sections gas turbine compressors and refrigeration compressors from 1.0 to
below. 2.0. At the design condition the density variation through the stage
Before we consider these equations, it is worthwhile to consider defines the local meridional velocity levels, velocity triangles, and
the four key nondimensional parameters that are used: the work input of the stage. Operation at other speeds will cause a
•
change in the density variation and the change in velocity trian-
the expected design point polytropic efficiency g.
•
gles leads to a different matching of the components, probably
the global volume flow coefficient /.
•
with a lower performance, and a shift in volume flow coefficient.
the stage work coefficient k.
•
The method described here captures these effects of a change of
the stage tip-speed Mach number M.
speed away from the design point, at least on a global basis.
These values are the typical results of a preliminary design pro- For map prediction additional information is also needed to
cess based on 1D performance correlations. Together with the characterize the variation in performance away from the design
inlet flow conditions, gas properties, and impeller diameter, they point, which will differ for different stage types. For example,
define the design point duty by the following equations: stages with a vaneless diffuser tend to have a wider operating
range than stages with a vaned diffuser. Equations are needed for
_ 2 D2 ; k ¼ Dh=u2 ;
/ ¼ V=u M ¼ u2 =a1 the variation in the best point efficiency and flow on different
2 2
p2   gc speed lines and for the variation in performance with flow along
¼ 1 þ ðc  1ÞkM2 c1 (2) each speed line. In addition the change in work with flow and tip
p1
speed needs to be captured. The equations described below pro-
V_ ¼ /u2 D22 ¼ /Ma1 D22 vide a parameterized description of the performance map,
whereby different values of the empirical coefficients are used to
At first sight it may be surprising that these nondimensional pa- model the characteristics for different types of compressor stage
rameters also define the shape of the performance map with no but the structure of the equation system remains the same for all
detailed information about the stage geometry. However, the compressors.
design values of flow coefficient, work coefficient, efficiency, and
tip-speed Mach number determine the details of the stage geome-
try in the subsequent design. If this design is done well then the Variation of Efficiency With Flow and Speed
performance characteristic for this optimally designed stage can
be implied from these design values. Equations for the Variation of Efficiency With Flow. The
Consider a stage designed for a certain flow coefficient and tip- equations used to describe the variation of efficiency with the flow
speed Mach number with an optimally matched vaned diffuser. If go back to the system described in diagrams by Rodgers [5].
the diffuser vane setting angle were changed to give a lower throat Rodgers showed various diagrams, similar to Fig. 1, in which the
area, then the matching between the impeller and diffuser would variation of normalized efficiency with normalized flow is mod-
change and the value of the flow coefficient at peak efficiency eled with an equation of the form
would be reduced. Thus a change in the value of the design flow  
g /
coefficient would characterize this change in geometry. Alterna- ¼f ;M (3)
tively the smaller diffuser could be considered to be a diffuser that gp /c
is better matched at a higher tip speed for the same flow coeffi-
cient, so the effective design Mach number of the stage would be where the ratio of the efficiency relative to that at the peak effi-
increased by this geometrical change. In this way the values of the ciency point (g/gp) is a function of the flow coefficient relative to
flow coefficient and tip-speed Mach number at the design point the maximum flow coefficient at choke (///c) and the tip-speed
include information about the optimal geometry and internal stage Mach number. The lines plotted in Fig. 1 are an example of those
matching and can be used to characterize this. Exactly which geo- described by the equations given below and are not taken from
metrical features are needed to meet the performance target is the Rodgers. The key aspect is that at low tip-speed Mach numbers

021034-2 / Vol. 135, MARCH 2013 Transactions of the ASME

Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/ on 08/10/2013 Terms of Use: http://asme.org/terms

 This equation has the natural physical property that the efficiency
is zero at zero flow coefficient and the curves have a maximum at
the flow coefficient for peak efficiency (/p//c), and are horizontal
at this point. To match the test data the coefficient D takes differ-
ent values for different tip-speed Mach numbers and this provides
a small Mach number dependency on the shape of the characteris-
tics in the low flow region, see Fig. 2. For turbocharger stages
with vaneless diffusers the typical value of D varies from 2.1 on
the low speed characteristics to 1.7 on the high speed characteris-
tics, and this is clearly related to the physical increase in the
strength of the incidence losses at high Mach numbers. The equa-
tions do not include any specific details related to flow separation
or instability at low flows and retain a simple ellipselike shape
down to zero flow. This is a weakness in the prediction of the
onset of instability as in reality relatively small changes in the
slope of the pressure rise variation with flow can cause the onset
of instability.
Fig. 1 Variation of efficiency ratio with the ratio of the flow For flows above the peak efficiency these equations are further
coefficient relative to that at choke for a range of tip-speed
Mach numbers
modified to take account of the fact that when the maximum flow
is reached, the efficiency may differ from zero:

the operating range is largest and the flow at peak efficiency is 2 0 1H 31=H
/ /p
then furthest from the choke conditions. 6 
g 6 B/ c / c C 7
The same information is presented in Fig. 2, but plotted relative / > /p ; ¼ ð1  GÞ þ G61  B C 77 (6)
to the flow at peak efficiency (///p). The form shown in Fig. 2 is gp 4 @ /p A 5
1
needed in this method, as in the preliminary design phase the /c
design flow is known and the choke flow is not known as the
throat areas have still to be determined. This equation has a similar ellipselike shape to the equations
Rodgers [5] gave no equations for these curves but in several pub- below the peak efficiency point and this leads to a smooth transi-
lications (Swain [6], Casey [7], and Swain [1]) some analytic equa- tion between the two arcs around the peak efficiency. The curves
tions for such curves have been described. After extensive automatically have the maximum efficiency at the flow coefficient
experimentation with these published equations, and many other for peak efficiency and have the maximum flow with a vertical
forms, a new more physically based equation structure has been characteristic at the flow coefficient for choke. The coefficients H,
developed. Clearly the equations used are not actually a real physical G, and the ratio of the flow coefficient at peak efficiency to that at
model of the losses but a particular structure has been chosen in order choke (/p//c) are not constant but also vary with the tip-speed
to reproduce important physical effects in the stage characteristics. Mach number to give narrower characteristics of different shape
In the current work the variation of stage efficiency with flow as the Mach number increases, see Fig. 2.
along a speed line is a modified form of an elliptic curve: The exponent H has a similar function to the exponent D in Eqs.
  y2 1=2 (5). This exponent takes into account the fundamentally different
x2 y2 x shapes of efficiency characteristics for low Mach number and high
þ ¼ 1; ¼ 1 (4)
a2 b2 a b Mach number impellers. Low Mach number impellers tend to have
a smooth drop in efficiency ratio related to incidence losses as the
For flows below the peak efficiency point this equation is modi- flow increases above peak efficiency. A value of H near to 2, giving
fied to have a variable exponent to give the characteristic curves elliptical curves, is needed to match low-speed experimental data.
for efficiency ratio as a function of flow ratio as follows: High Mach number stages tend to have a small plateau of high effi-
" !D #1=D ciency close to the peak efficiency point and then drop much more
g /=/c sharply into choke. To match experimental data on the shape of
/ < /p ; ¼ 1 1 (5) high Mach number characteristics a value of H ¼ 3.5 is needed,
gp /p =/c
which is close to the exponent of 4 used by Swain [1].
Equations (6) includes an additional change to account for the
fact that the location of maximum flow at choke is not necessarily
at zero efficiency, but at an efficiency ratio of (1  G). High Mach
number impellers reach choke at relatively high efficiency (with
0 > G > 1), or even at peak efficiency if all sections operate at
unique incidence conditions, see Lohmberg et al. [8]. The vertical
choked characteristics start immediately below this point and not
when zero efficiency has been reached, as shown in the high
Mach number curves in Fig. 2. In contrast, low Mach number
impellers fall to an efficiency of zero well before the maximum
choke flow is reached (with G > 1), as high losses due to negative
incidence stall occur before choke, see Casey and Schlegel [9].
The flow continues to increase below the point of zero efficiency,
as also shown in the low Mach number curves of Fig. 2. In this
way G is also a function of the tip-speed Mach number.
The analysis of test data for typical turbocharger style stages indi-
cates that at high speeds a value of G ¼ 0.3 is needed—indicating
that choke occurs at an efficiency ratio of 0.7. At low speeds a value
Fig. 2 Variation of efficiency ratio with the ratio of the flow of G ¼ 2 is needed, which indicates that choke occurs well down the
coefficient relative to that at peak efficiency for a range of characteristic when the efficiency ratio reaches 1. The diagrams
tip-speed Mach numbers shown by Casey [7] include this effect but are simply elliptical at all

Journal of Turbomachinery MARCH 2013, Vol. 135 / 021034-3

Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/ on 08/10/2013 Terms of Use: http://asme.org/terms

 speeds with H ¼ 2. The equations of Swain [1] do not include these
choking effects and always become vertical at a maximum flow at
zero efficiency ratio, as they effectively have a value of G ¼ 1.
Neither system includes any variability of the strength of incidence
effects below the peak efficiency point.

Variation of Coefficients With Mach Number. Equations (5)

and (6) include many variable coefficients which depend only on
the tip-speed Mach number. These coefficients are constant along
each speed line but vary from one speed line to the next. At low-
tip-speed Mach numbers the flow is effectively incompressible
and there is only a small effect of Mach number, so the coeffi-
cients have a constant value at low speeds. Analysis of test data,
see below, has identified that the coefficients also tend to have a
constant value at high tip-speed Mach numbers, indicating that
the normalized shape of the high speed characteristics no longer
changes with Mach number. In the current method the coefficients
for intermediate Mach numbers are then determined from the two Fig. 3 Ratio of the flow coefficient at peak efficiency to that at
asymptotic values at low speed and at high speed with the use of a choke over a range of tip-speed Mach number
blending function.
This procedure is demonstrated with regard to the ratio of the trend of Swain appears unrealistic as at low Mach numbers constant
flow coefficient at peak efficiency to the flow coefficient at choke values are expected. It should be noted that at low speeds it is in
(/p//c), which appears in Eqs. (6). A low value of this coefficient any case difficult to be precise about choking as no test data ever
indicates a high operating range between peak efficiency and reaches this condition.
choke. Experience shows that the operating range of compressors The coefficient B in Eqs. (7) can be envisaged as a measure of
is largest at very low Mach numbers and remains constant at low the transonic capability of the stage, or as a sort of critical tip-
Mach numbers. The range then decreases with increasing Mach speed Mach number of the design. Analysis of test data has shown
number but again remains more or less constant at much higher that typical turbocharger stages tend to require a value of B of
Mach numbers. Equations have been selected which include this around 1.1 or 1.2 to match their characteristics. Process style
feature as a natural property: stages in multistage compressors tend to require a value of B of
    around 0.8 to 0.9. In all stages examined a good match with the
/p /p /p
¼ ð1  PÞ þP transition over the range of Mach numbers was given with values
/c /c LO /c HI of A between 0 and 1 and C between 4 and 5.
(7)
1 The coefficients D, G, and H in Eqs. (5) and (6) make use of a
P¼ ; t ¼ ðM  BÞðAM þ CÞ similar approach to determine their values at intermediate Mach
1 þ et
numbers from the values at high and low tip speeds:
The parameter P in this equation is the s-shaped logistic func-
tion, and varies from 0 to 1 with increasing tip-speed Mach num- D ¼ ð1  PÞDLO þ PDHI
ber. It blends the values of the coefficients between the low tip- G ¼ ð1  PÞGLO þ PGHI (8)
speed Mach number asymptote (subscript LO) and the high tip-
H ¼ ð1  PÞHLO þ PHHI
speed Mach number asymptote (subscript HI). At low Mach num-
ber the variation with Mach number is small as P remains close to
0, which is physically realistic as the flow is effectively incom- The blending function P is the same as in Eqs. (7)
pressible up to M ¼ 0.3. At high Mach number the equations also The equations given above describe the variation of efficiency
attain a constant asymptotic value with P close to 1, which with flow relative to that at the peak efficiency point along each
matches experience that the form of the characteristics remains speed characteristic. In addition to these equations some addi-
similar at high Mach numbers. The transition between these tional efficiency equations are needed for the peak efficiency and
extreme values reaches the halfway value between the upper and the flow coefficient at peak efficiency on each speed line. These
lower asymptotic values with a value of P ¼ 0.5 with t ¼ 0 at a equations are determined from test data on a more ad hoc basis, as
Mach number of M ¼ B, and the rate of transition around this outlined below.
value is determined by the constants A and C.
Different types of stages (process or turbocharger impeller, Validation of the Efficiency Equations With Test Data.
vaned and vaneless diffuser) have different coefficients for the ratio Having defined these equations it remains to demonstrate that these
of /p//c at high and low Mach numbers. Vaneless diffuser stages adequately describe the variation of efficiency with flow and speed
tend to have lower values of this ratio than vaned diffuser stages as in real compressor stages. An analysis of more than 45 different
vaned diffusers are usually matched such that they choke before the compressor stages covering a wide range of applications has been
impeller and this limits the range of the stage giving a larger value carried out to determine the most suitable values of the coefficients
of /p//c. Two lines demonstrating the value /p//c for different in the equations. The test data covers a variety of state-of-the-art
coefficients are shown in Fig. 3. The upper thin curve models pro- turbocharger and process compressor characteristics from measure-
cess stages with vaneless diffusers (with impellers having thick ments made by a variety of confidential sources and neither the
blades and no splitters) and the lower thick curve models modern sources nor the original data may be disclosed. Of the many per-
turbocharger stages with vaneless diffusers. Other curves are formance maps available to the authors only those for the best
needed for vaned diffusers. Swain has suggested two forms of vari- stages have been used for the correlations. No further information
ation of this ratio with the Mach number, but supplied no equations can be provided, but in all cases attention has been given to the con-
for this. These are shown as dashed lines in Fig. 3—the shorter sistency of the data and it is considered to be very sound and reli-
dashes are Swain’s original correlation and the longer dashes are able with typical efficiency errors in the range 62%.
his recent “loss-based” curve, see Swain [1]. Note that the curves With fine tuning of the available coefficients in the equations
of Swain show broadly similar trends, although they are derived for given above it is possible to match almost exactly the variation of
the impeller only, except at very low Mach numbers where the efficiency ratio with flow along any speed line for any specific

021034-4 / Vol. 135, MARCH 2013 Transactions of the ASME

Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/ on 08/10/2013 Terms of Use: http://asme.org/terms

 case, but for the purposes of a general method a fixed set of coeffi-
cient values is needed for particular stage types. The typical good
agreement using these fixed set of coefficients is shown in Fig. 4
giving variations of efficiency ratio with flow ratio for a typical
turbocharger stage with a vaneless diffuser. The measurement
data has been obtained along a number of constant speed lines and
the curves are the calculated variations at these speeds using the
equations of this paper. Taking into account inevitable experimen-
tal error (probably at least 62% on the characteristics of typical
turbocharger rigs) the agreement is exceptionally good.
The test data on this case show clearly that there is a need for
different characteristic shapes for the efficiency below the peak
efficiency point as the Mach number varies, using different values
of D in Eqs. (5). Note that the coefficients used here have not been
selected to match the performance of this particular stage but are the
mean values needed to match the performance of a range of similar Fig. 5 Ratio of the flow coefficient at peak efficiency to that at
turbocharger stages. This demonstrates that the method appears to choke over a range of tip-speed Mach number for turbocharger
be useful to adapt knowledge gained on the operating range of one compressor stages with a vaneless diffuser
stage to provide a reference for another of a similar type.
The equations for the variation of the ratio of the flow coeffi-
cient at peak efficiency to the maximum flow coefficient at choke stages with vaneless diffusers. The bottom axis is the impeller tip-
(/p//c) have also been determined from test data. The values at speed Mach number and the vertical axis shows the variation of
very low Mach number have been calibrated from data published flow coefficient at peak efficiency relative to that at a low tip-
by Gülich [10] for the operating range of centrifugal pumps (with speed Mach number. The process compressor stages are shown as
M ¼ 0), and this causes the trend at low Mach numbers to differ solid symbols and are modeled empirically by the full curve
from the curves of Swain, as already indicated in Fig. 3. At higher shown. The open symbols represent turbocharger style stages and
Mach numbers the validation process is slightly more difficult and are modeled by the dashed line.
requires some fine engineering judgment. First, the test data to- The flow coefficient at the peak efficiency point tends to
ward choke usually do not have sufficiently low pressure ratios increase as the tip-speed Mach number increases. This is caused
that the maximum flow at choke is achieved, especially on the partly by the effect of the change in the density on the velocity tri-
low Mach number speed lines. Secondly, the efficiency curves are angles at impeller inducer inlet with speed. It is also related to the
very flat near the peak efficiency point and the small number of mismatch with the diffuser as this becomes too small for the
test points along each characteristic combined with the experi- impeller at low speeds forcing the peak efficiency of the stage to
mental error does not uniquely define the flow coefficient at the move to a lower flow coefficient. At very high tip-speed Mach
peak efficiency point. These issues have been tackled by using the numbers there is a drop in the flow coefficient at peak efficiency
equations given here to model each characteristic individually and related to the onset of choking in the impeller inlet, which clearly
to determine the required flow ratio (/p//c) from the modeled happens earlier in process stages. This is believed to be related to
characteristics. Figure 5 shows the variation of the flow ratio the general use of thicker blades and a lack of splitter vanes in
determined by this process from the analysis of test data from sev- process compressors. Polynomial equations have been developed
eral turbocharger stages with a vaneless diffuser. In these cases for the curves shown here.
with turbocharger style impellers the choke flow is defined by the Figure 7 shows the variation of the peak efficiency, relative to
choke of the impeller inlet. The differences are related to the dif- that at the design point over a range of tip-speed Mach numbers.
ferent throat areas, blade thicknesses, and shroud diameter of the The horizontal axis is the ratio of the Mach number to the design
different impellers. Mach number. Here the scatter appears to be very large but this is
Figure 6 shows the additional characterization of the variation slightly deceptive. The test data includes many of turbocharger
of the flow coefficient at peak efficiency with Mach number for stages whose apparent efficiency drops rapidly toward low tip-

Fig. 6 Ratio of the flow coefficient at peak efficiency to that at

low speed over a range of tip-speed Mach number for vaneless
Fig. 4 Normalized efficiency ratio of a vaneless turbocharger stages. Turbocharger compressor stages shown as a dashed
stage using Eqs. (5), (6), (7), and (8) compared with test data at line and open symbols and process compressor stages shown
different tip-speed Mach numbers as a full line and full symbols.

Journal of Turbomachinery MARCH 2013, Vol. 135 / 021034-5

Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/ on 08/10/2013 Terms of Use: http://asme.org/terms

 slip factor the work input coefficient can be expected to decrease
linearly with the impeller outlet flow coefficient.
We convert Eq. (9) into a form which takes the density change
from the stage inlet to the impeller outlet into account; so that the
flow coefficient based on the inlet volume flow can be used, as
follows:

1 D2 qt1
/2 ¼ / (10)
p b2 q2

If we consider compression in the impeller as a polytropic pro-

cess with a kinematic degree of reaction of cd (see Casey and
Schlegel [9]) and a polytropic exponent of nd, then the density ra-
tio from inlet to outlet can be derived as
Fig. 7 Ratio of peak efficiency to that at design over a range of
tip-speed Mach number for stages with vaneless diffusers, with q2   1
¼ 1 þ ðc  1Þcd kM2 nd 1 (11)
and without the correction for heat transfer effect qt1

speed Mach number, as depicted by the dotted line. This is not an and the Euler work input coefficient can be expressed in terms of
aerodynamic effect but is an error in efficiency measurement due the inlet volume flow coefficient as
to ignoring the heat transfer effects from the turbine in the effi-
ciency determination on a turbocharger test rig; see Casey and cs /D2 tan b02
Fesich [11]. The strength of this effect depends on the amount of kEuler ¼ 1  þ (12)
u2 b2 p ½1 þ ðc  1Þc kM2 nd11
heat transfer, and for adiabatic stages there is only a small fall in d
efficiency as the Mach number is reduced below the design point.
The experimental error in the efficiency determination in these When operating at a flow coefficient near to the peak efficiency
tests is probably not better than 62%. point, backswept impellers have a fairly constant value of the
degree of reaction (around 0.6), so this and the slip factor are
Summary of the Calculation Procedure for the Efficiency. For taken as constant along the characteristic. The polytropic expo-
the calculation of efficiency on a given speed line, the first step is nent can be approximated from the stage efficiency.
to calculate the values of the variable coefficients (D, G, H, and The design value of the work coefficient includes not only the
/p//c) that correspond to this tip-speed Mach number, from Eqs. Euler work but also the disk friction work so that
(7) and (8). The peak efficiency and the flow coefficient at peak  
efficiency need to be determined from the equations for the curves kdf
kd ¼ kEuler d 1 þ (13)
shown in Figs. 6 and 7. For a given inlet flow coefficient this then /
allows the efficiency to be determined from Eqs. (5) for flows
below the peak efficiency or Eqs. (6) for higher flows. where the value of kdf is a disk friction coefficient and can typi-
At this point it is useful to summarize what has been achieved cally be taken as 0.003 or calculated from the design value of the
with regard to the prediction of efficiency. The objective was to disk friction power using correlations from Daly and Nece [12].
calculate the value of the polytropic efficiency for specific values The Euler work coefficient varies linearly with flow so that from
of the independent variables (the flow coefficient and tip-speed Eq. (12) at the design point we derive a linear expression at the
Mach number). A completely new set of parametric equations has design tip-speed Mach number and then have
been developed for this purpose, based on the key nondimensional
parameters which determine the performance at the design point. cs
kEuler d ¼ 1  þ /d kkd (14)
The equations describing this functional dependency are physi- u2
cally realistic. They also include many variable coefficients
(which are a function of the tip-speed Mach number) and some The value of the coefficient in this equation, which is the slope
fixed constants, and typical values of these have been given of the Euler work coefficient curve with a variation in the flow,
above. The excellent agreement shown in Fig. 4, which is typical can be determined at the design point as
of other cases, demonstrates that the objective has been achieved  
with these equations and coefficients for turbocharger style stages cs
to within an error band of roughly 62% in the efficiency, which kkd ¼ kEuler d  1 þ =/d (15)
u2
probably corresponds to the accuracy of these measurements.
The slip factor is assumed to stay constant and needs to be
Variation of Work Input Coefficient. The second part of the specified, but the coefficient for the slope changes with the Mach
functional relationship of Eq. (1) is the variation of the work input number, as can be seen in Eq. (12). An approximate relationship
coefficient with flow and tip-speed Mach number. This variation for this slope, which is related to the change in density at impeller
is based on the Euler equation following an approach described by outlet with speed follows from Eq. (12), and assumes that the
Casey and Schlegel [9]. Under the assumption that the flow has no impeller efficiency is similar to the stage efficiency so that the
swirl at inlet to the impeller, then the Euler work due to the adia- polytropic exponent can be calculated:
batic work input of the impeller can be approximately estimated
from the velocity triangles. The work is related directly with the   1
flow coefficient at impeller outlet /2 ¼ cm2 =u2 , the assumed slip kki 1 þ ðc  1Þcd kd Md2 nd 1
¼  1 (16)
velocity cs =u2 , and the impeller outlet blade angle b02 , as follows: kkd 1 þ ðc  1Þcd kd Mi2 nd 1
cu2 cs
kEuler ¼ ¼ 1  þ /2 tan b02 (9) Equations (15) and (16) provide the slope of the work coeffi-
u2 u2
cient curve at each speed and the complete equation for the varia-
Note that the impeller outlet angle is negative for a backswept tion of the work coefficient with flow and tip-speed Mach number
impeller with the notation used here. For a constant value of the is then given as

021034-6 / Vol. 135, MARCH 2013 Transactions of the ASME

Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/ on 08/10/2013 Terms of Use: http://asme.org/terms

   
kdf cs bated by the fact that different components (diffuser or impeller)
ki ¼ 1þ 1  þ /i kki (17) may be responsible for the instabilities, different phenomena may
/i u2
occur (rotating stall, and surge), test procedures to identify insta-
bilities may differ, and that the compression system itself may
In this expression index i represents the different speed lines.
play a role.
The novel aspect of these equations is that the variation of the
With such a large number of effects it is not surprising that any
work coefficient with flow is calculated with no prior knowledge
systematic comparison of different stages shows the wide variabil-
of the impeller outlet velocity triangle or geometry. The value of
ity of the operating range of compressor stages. As an example of
the work coefficient at the design flow coefficient together with
the typical large amount of scatter, Baines [13] shows that the ra-
the slip factor determines the slope of the work input characteris-
tio of the flow at the onset of instability relative to that at choke
tic at the design speed. The slope is then adjusted to take into
may vary by at least a factor of 2 for a range of different designs.
account the effect of the speed on the work input, via the density
It is clearly not possible in a preliminary design method with no
change at impeller outlet. The determination of this slope in this
detailed information on the internal aerodynamics, geometry, and
way eliminates the need for any specific knowledge of the blade
loading of the stage to predict the stability point exactly. In fact,
geometry and ensures that the flow-work characteristic is in ac-
even with more detailed knowledge of the geometry and internal
cordance with the Euler equation.
flow, and even with the help of computational fluid dynamics, it is
still exceptionally difficult to predict the onset of instability with
Validation of Work Coefficient Curves With Test Data. The any certainty.
available test data has been used to examine the variation of work The best that can be done in the framework of this method is to
coefficient with flow and Mach number. Note that it is necessary derive a logically consistent set of equations for the variation of
to define a value of the slip factor (typically in the range 0.1 to the flow at the stability point, and to suggest a range of typical
0.2), a value of the disk friction coefficient (typically between coefficients that can be used in these equations to express an opti-
0.002 and 0.004), and the degree of reaction (typically between mistic, a realistic, and a pessimistic stability line. A summary of
0.50 and 0.65) for these curves, so that some limited additional in- the instability points of the vaneless turbocharger stages that have
formation about the stage geometry is useful for this. As this in- been analyzed is presented in Fig. 9. This shows the ratio of the
formation is usually known early in a preliminary design process measured flow coefficient at the onset of instability relative to that
this is not difficult. at the choke point (/s//c) on each speed line, as denoted by the
Figure 8 shows an example of a turbocharger impeller tested tip-speed Mach number. The large scatter is immediately apparent
with a vaneless diffuser in a dedicated compressor test stand, and this would be larger if poor quality stages had been included.
which avoids the heat transfer effects on the work coefficient The dark symbols are conventional stages without the use of a
found in turbocharger gas stands. The large white dot in the mid- recirculating bleed system for enhancing the map width. The open
dle of the diagram is the design point of the stage at a tip-speed symbols are stages using some form of recirculating bleed, which
Mach number of 1.25 and the variation of the work coefficient clearly improves the operating range on the high speed character-
with flow and with Mach number on different characteristics istics. This diagram also includes the upper and lower limits of the
(from 0.6 to 1.45) away from this point is well captured by the correlation of Baines [13] as dotted lines. Note that the original
equations above. The symbols are the measured work coefficient curves of Baines are expressed in terms of a pressure ratio and
values roughly situated on the lines determined from the equa- here these have been converted approximately to a tip-speed
tions. Note that the equations reproduce the increase in work coef- Mach number using sensible engineering assumptions.
ficient as the flow reduces and the change in work with tip-speed The lowest operating range of the well-designed stages exam-
Mach number. This determines the slope of the pressure rise char- ined is slightly better than the correlation of Baines and the dot-
acteristic, which is different for different stages, and would not be dashed line shows the equations developed for this pessimistic
obtained by a simple scaling of measured characteristics. The variation of the instability point. The other dashed line expresses
method, as described here, does not capture the additional fall in the correlation for stages with a wide range and this agrees partly
work coefficient as the impeller chokes toward high flow on the
high speed characteristics, but a method to include this has already
been described by Casey and Schlegel [9], and is currently being
included in the method.

Prediction of Stability Limit. One of the most difficult

aspects of compressor flows is the prediction of the stability limit
due to the onset of rotating stall or surge. The difficulty is exacer-

Fig. 9 Variation of the ratio of the flow coefficient at instability

Fig. 8 Variation of work coefficient with flow and tip-speed with that at choke for vaneless turbocharger stages (/s//c). The
Mach number for a turbocharger stage with a vaneless diffuser open symbols are stages with inlet bleed recirculation.

Journal of Turbomachinery MARCH 2013, Vol. 135 / 021034-7

Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/ on 08/10/2013 Terms of Use: http://asme.org/terms

 with the upper optimistic limit quoted by Baines. The full line in
this diagram is the pragmatic equation used for a mean range,
which is considered realistic for a well designed stage. This is
based on the equations already described for the peak efficiency to
choke flow ratio (/p//c) in Eq. (8), as modified in Eqs. (18):
   
/s /s /
¼ ð1  PÞ þP s
/c /c LO /c HI
(18)
1
P¼ ; t ¼ ðM  B s ÞðAs M þ C s Þ
1 þ et

The coefficients used for the realistic surge line shown in Fig. 9
are as follows: (/s//c)LO ¼ 0.225, (/s//c)HI ¼ 0.835, As ¼ 0.0,
Bs ¼ 1.25, and Cs ¼ 4.75.

Prediction of Overall Operating Range. During the develop-

ment of the correlations presented here different ways of compar- Fig. 11 Flow coefficient—Mach number diagram for turbocharger
ing the performance of different stage types has been developed. compressors with vaned diffusers
One of these is shown in Figs. 10 and 11. Figure 10 shows the
variation of the choke flow coefficient, the peak efficiency flow ler at high speeds and then becomes much too small at low
coefficient, and the surge flow coefficient as a function of the tip- speeds. This causes the impeller to operate at high incidence and
speed Mach number for the typical vaneless diffuser turbocharger forces the stage characteristic to move to a lower flow coefficient.
stage, and Fig. 11 shows this variation for a turbocharger stage This effect is much less strong in the vaneless diffuser. A vaneless
with a vaned diffuser. In both cases the flow coefficients are nor- diffuser with no blades can accept higher flows at lower speeds as
malized relative to the flow coefficient that would occur at peak it does not choke and the impeller is not forced to operate at such
efficiency at low Mach number /p0, which is lower than the a low flow coefficient.
design flow coefficient. The lines given in these diagrams are
the curves given by the equations in this paper, and the points are
the measured surge, peak efficiency, and choke points of the Prediction of Performance Map. As a demonstration of the
stages used for validation at different tip speeds. capability of the method described here, Figs. 12 and 13 compare
The variation with tip-speed Mach number shown provides a the predicted and the measured characteristics of a typical turbo-
useful presentation of the operating envelope and resembles a typ- charger compressor stage with a vaneless diffuser over a range of
ical “a-Mach” diagram used to visualize the available operating speeds. The measured characteristics were obtained on a turbo-
range of compressor cascades at varying incidence and flow Mach charger gas stand using a small compressor impeller with a back-
number levels in the preliminary design of axial compressors, see sweep of approximately 30 deg over a range of speeds from
the examples given by Casey [7]. The variation with Mach num- 40,000 to 230,000 rpm. The predictions are based on the equations
ber of the different lines in these diagrams agrees well with the given in this paper, whereby the coefficients used were derived
measurement data and the lines are consistent with each other. from a whole range of turbocharger stages. The measured points
This provides further justification for the coefficients used to on all speed lines agree very well (to within 62%) with the pre-
determine these individual curves. dictions, and the measured surge line lies typically between the
The difference between these two diagrams also shows up limits of the predicted pessimistic and optimistic lines.
some interesting aspects related to the effect of matching on the The design point of this stage is denoted as a large open circle
characteristics at different speeds. First, the much wider range of on the third fastest speed line. The nondimensional values of the
stages with vaneless diffusers can be seen from the spacing key parameters (efficiency, flow coefficient, work coefficient, and
between the surge and choke lines in these diagrams. Secondly, tip-speed Mach number) and the impeller diameter have been
the shift in the flow coefficient at the peak efficiency point with
Mach number is very different for the two cases. In the vaned dif-
fuser case (Fig. 11) the diffuser is generally matched to the impel-

Fig. 12 Predicted and measured pressure ratio performance

map for a turbocharger with a vaneless diffuser over a wide
Fig. 10 Flow coefficient—Mach number diagram for turbocharger range of rotational speeds. The coefficients used are derived
compressors with a vaneless diffuser from an analysis of many stages, including this one.

021034-8 / Vol. 135, MARCH 2013 Transactions of the ASME

Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/ on 08/10/2013 Terms of Use: http://asme.org/terms

 coefficient at the design point of only 0.5, compared to a value of
0.6 to 0.75 for the other stages. The consequence of this is that the
steeper flow coefficient versus work coefficient relationship from
Eq. (9) leads to steeper pressure rise characteristics. This effect
cannot be predicted simply by scaling characteristics of other
stages and demonstrates the predictive nature of the method
adopted here.

Discussion and Outlook

The work presented here allows an estimate of the performance
map of a centrifugal compressor stage to be obtained from limited
design data during the preliminary design phase, without prior
knowledge of the stage geometry. Most of the examples presented
are for turbocharger compressor stages with vaneless diffusers,
but in work not described in detail here the method has also been
adapted to predict other categories of compressor stages by the
use of different empirical coefficients in the equations.
Fig. 13 Predicted and measured efficiency performance map The equations provide a fully parameterized system of alge-
for a turbocharger with a vaneless diffuser over a wide range of braic functions for the shape of the performance map, which can
rotational speeds. The coefficients used are derived from an also be used in different modes, in which the coefficients are
analysis of many stages, including this one, and a correction adjusted to exactly match the measured performance of a particu-
for the effect of heat transfer at low speeds is made. Symbols lar stage. This then provides an algebraic set of equations for a
as in Fig. 12. performance map that could be used in whole engine performance
calculations, allowing the performance for any flow and speed to
be calculated extremely rapidly with very little data required and
no detailed knowledge of the geometry. Alternatively a single
measured speed line can be approximated and other speed lines
then generated by the equations.
The method has been developed on the basis of the inherent non-
dimensional performance coefficients of the stages, and because of
this can be adapted to other gases and other applications relatively
easily. It has already been adapted to generate a stage stacking tool
for multistage process compressors allowing the behavior of several
stages of a multistage compressor to be analyzed.
An obvious limitation of the method is that it only considers the
global performance data and not the component losses (impeller,
diffuser, and volute) which are the cause of this behavior. The
method can naturally be developed further to include the individ-
ual components so that separate sets of performance curves for
each component can be developed and stacked to produce a more
accurate stage performance, which is actually the procedure sug-
gested by Rodgers [5] and used by Swain [6]. The extension to
Fig. 14 Predicted and measured pressure ratio performance include the separate diffuser and impeller characteristics would
map for a turbocharger with high back-sweep over a wide range also allow the method to identify in more detail the choking and
of rotational speeds. This stage was not used to derive the instability limits of each component and to study changes in com-
correlations used here. ponent matching with speed, and is currently being pursued.
The fact that such a simple technique can predict the shape of
characteristics of compressor stages is of great relevance to the use
specified at this point and all other curves are then calculated from of performance correlations. There are many competing correlation
this point using the correlations and equations given in this paper. systems in the literature, see for example the extensive but nonex-
In addition to these nondimensional coefficients, the slip factor, haustive list provided by Oh et al. [3]. All of these seem to do
the degree of reaction and the disk friction coefficient are speci- equally good jobs of predicting performance maps when the empiri-
fied, whereby standard values from other impellers may be used if cal coefficients have been tweaked appropriately. The important
these are not known, but no other information on the geometry is finding from this paper is that, provided the methods predict the
needed. As this compressor stage has been tested in a turbocharger peak efficiency point reasonably well, the variation away from this
with heat transfer from the turbine, the efficiency shown here point is then fairly similar for all well-designed stages of a similar
includes a correction for the effect of heat transfer according to type and duty and can be predicted without the need for correlations
the approach of Casey and Fesich [11], whereby the empirical of individual effects or information about the geometry.
coefficient needed for this has been adjusted to match the test
data. Without this correction the apparent peak efficiency would
not drop sharply as the speed decreases as shown in Fig. 13. The Conclusions
curves also include an extrapolation to a higher rotational speed, The theoretical analysis and correlations in this paper provide a
which was not measured, to demonstrate that the system still pro- novel way of predicting centrifugal compressor performance
duces sensible characteristics at higher speeds. maps during the preliminary design phase. The basis of the
A further case is shown in Fig. 14. This stage was not included method is the scientific analysis of measured stage characteristics
in the stages used to generate the coefficients in the correlation using classical nondimensional parameters and the approximation
equations, and is extremely well-predicted with the equations and of these with suitably designed parametric equations. The nondi-
coefficients derived from the other stages. Of great interest is the mensional parameters predict a reasonably accurate performance
fact that this stage has a very high back-sweep leading to a work map for a particular duty, based on experience from other stages.

Journal of Turbomachinery MARCH 2013, Vol. 135 / 021034-9

Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/ on 08/10/2013 Terms of Use: http://asme.org/terms

 This map can be used as a reference to quickly assess the operat- b2 ¼ impeller outlet width (m)
ing range of a compressor against objectives for a new design. c¼ absolute velocity (m/s)
This method also produces an important new insight into the off- cs ¼ slip velocity (m/s)
design performance—that for a well-designed stage this is also D2 ¼ impeller tip diameter (m)
defined by the design duty. The weakest aspect of the method is h¼ specific enthalpy (J/kg)
the considerable uncertainty that remains with the prediction of kkd ¼ coefficient in Eq. (13)
the surge line, but an optimistic, a realistic, and a pessimistic esti- kdf ¼ disk friction coefficient
mate can be made. It is hoped that the same equations when M¼ tip-speed Mach number
adapted to predict component performance will be more success- n¼ polytropic exponent
ful in respect of surge prediction. p¼ static pressure (N/m2)
P, t ¼ coefficients in logistic function
The novel features of the method are: u2 ¼ impeller blade tip speed (m/s)
• A new set of parameterized algebraic equations has been V_ ¼ volume flow rate (m3/s)
developed that describes the variation of efficiency with
changes in flow and speed away from the design point. Greek Symbols
• The method requires the values of only four key nondimen-
b2 ¼ impeller blade back-sweep angle (deg)
sional parameters at the design point (the efficiency, flow
c¼ isentropic exponent
coefficient, work coefficient, and tip-speed Mach number) to
cd ¼ degree of reaction
define the whole performance map and requires very little
g¼ polytropic efficiency
knowledge of the geometry.
•
k¼ work input coefficient, k ¼ Dh=u22
The equations are designed to automatically produce physi-
w¼ pressure rise coefficient
cally realistic behavior of the characteristic curves and _ 2 D2
/¼ global flow coefficient, / ¼ V=u 2
include empirical coefficients that adjust the shape of the effi-
/2 ¼ impeller outlet flow coefficient
ciency characteristics to match experimental data on different
x¼ angular velocity (rad/s)
speed lines of compressor stages.
• The variation of work input with changes in flow and speed
away from the design point is determined from the Euler Subscripts
equation, using empirical data for the disk friction, the degree 0¼ at M ¼ 0
of reaction and the slip factor, also with no detailed knowl- 1¼ inlet total conditions
edge of the geometry. 2¼ outlet total conditions
• The coefficients in all these equations have been calibrated c¼ choke point
from extensive test data on a range of compressor stages. d¼ design point
Typical values of the necessary coefficients for well-designed Euler ¼ based on Euler equation
turbocharger stages with vaneless diffusers are given in the HI ¼ at high speed
paper. i¼ value on the ith speed line
• A new and useful form of presentation of the operating enve- LO ¼ at low speed
lope of a compressor stage called a “flow coefficient Mach,” m¼ meridional component
or a /-Mach diagram, has been developed. This diagram is p¼ peak efficiency point
different for different stage types and highlights interesting s¼ isentropic, stall
aspects of the stage behavior related to mismatch at off- t¼ total
design speeds of stages with vaned and vaneless diffusers. u¼ circumferential component
• The method has been described mainly for turbocharger com-
pressor stages with vaneless diffusers but with different val-
References
[1] Swain, E., 2005, “Improving a One-Dimensional Centrifugal Compressor Per-
ues of the empirical coefficients it is possible to adapt the formance Prediction Method,” Proc. IMechE Part A, 219, pp. 653–659.
same equations for use with stages with vaned diffusers and [2] Aungier, R. H., 2000, Centrifugal Compressors—A Strategy for Aerodynamic
with industrial process impellers. Design and Analysis, ASME, New York.
[3] Oh, H. W., Yoon, E. S., and Chung, M. K., 1997, “An Optimum Set of Loss
Models for Performance Prediction of Centrifugal Compressors,” Proc. IMechE
Part A, 211, pp. 331–338.
Acknowledgment [4] Cumpsty, N. A., 1989, Compressor Aerodynamics, Longman Group, Pearson
The first author would like to thank the Rector of the University Education Ltd, Harlow, Essex, UK.
[5] Rodgers, C., 1964, “Typical Performance Characteristics of Gas Turbine Radial
of Stuttgart for his permission to take a sabbatical term in 2009, Compressors,” ASME J. Eng. Power, 86, pp. 161–175.
which was material to the development of the method described [6] Swain, E., 1990, “A Simple Method for Predicting Centrifugal Compressor Per-
here. The authors are indebted to the session organizer, Greg Hol- formance Characteristics,” Proceedings of IMechE Conference, C405/040.
brook, and the reviewers of the paper whose constructive com- [7] Casey, M. V., 1994, “Computational Methods for Preliminary Design and Ge-
ometry Definition in Turbomachinery,” AGARD-LS-195, AGARD Lecture Se-
ments on the first draft allowed many improvements to be made. ries on Turbomachinery Design using CFD, NASA-Lewis.
Many companies involved in confidential development projects [8] Lohmberg, A., Casey, M., and Ammann, S., 2003, “Transonic Radial Compres-
with PCA Engineers and with the University of Stuttgart are sor Inlet Design,” Proc. IMechE Part A, 217, pp. 367–374.
acknowledged for the provision of confidential measured com- [9] Casey, M. V., and Schlegel, M., 2010, “Estimation of the Performance of Tur-
bocharger Compressors at Extremely Low Pressure Ratios,” Proc. IMechE Part
pressor performance data used to carry out this analysis. The as- A, 224, pp. 239–250.
sistance of Ed Swain, Peter Came, and Jeremy Tapper is [10] Gülich, J. F., (2007), Centrifugal Pumps, Springer, Berlin.
gratefully acknowledged. [11] Casey, M. V., and Fesich, T. M., 2010, “The Efficiency of Turbocharger
Compressors With Diabatic Flows,” ASME J. Eng. Gas Turbines Power, 132,
p. 072302.
Nomenclature [12] Daily, J. W., and Nece, R. E., 1960, “Chamber Dimension Effects on Induced
a1 ¼ inlet speed of sound (m/s) Flow and Frictional Resistance of Enclosed Rotating Disks,” ASME J. Basic
Eng., 82, pp. 217–232.
a, b, A, B, [13] Baines, N. C., 2005, Fundamentals of Turbocharging, Concepts/NREC, West
C, D, E, F, G, and H ¼ coefficients in equations River Junction, VT.

021034-10 / Vol. 135, MARCH 2013 Transactions of the ASME

Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/ on 08/10/2013 Terms of Use: http://asme.org/terms