Vous êtes sur la page 1sur 10

Proceedings of GT2007

ASME Turbo Expo 2007: Power for Land, Sea and Air
May 14-17, 2007, Montreal, Canada
Proceedings of GT2007
ASME Turbo Expo 2007: Power for Land, Sea and Air
May 14-17, 2007, Montreal, Canada

GT2007-27064
GT2007-27064

A NEW SLIP FACTOR MODEL FOR AXIAL AND RADIAL IMPELLERS


Xuwen Qiu, Chanaka Mallikarachchi, and Mark Anderson
Concepts NREC
217 Billings Farm Road
White River Jct., Vermont 05001, USA
ABSTRACT
a = exit throat width
b = passage width
c = chord
C = absolute velocity
C = absolute tangential velocity
Cm = meridional velocity (on Z-R plane)
Cslip = slip velocity (equation (3))
F = shape factor defined in equation (13)
m = meridional distance on the Z-R plane
n = exponent in Carters rule
s = pitch at the blade exit; s = 2R2 / Z
t = thickness at the blade trailing edge
U = blade rotating speed
W = the relative velocity
Z = number of blades

This paper proposes a unified slip model for axial, radial, and
mixed flow impellers. For many years, engineers designing
axial and radial turbomachines have applied completely
different deviation or slip factor models. For axial applications,
the most commonly used deviation model has been Carters
rule or its derivatives. For centrifugal impellers, Wiesners
correlation has been the most popular choice. Is there a
common thread linking these seemingly unrelated models?
This question becomes particularly important when designing a
mixed flow impeller where one has to choose between axial or
radial slip models.
The proposed model in this paper is based on blade loading,
i.e., the velocity difference between the pressure and suction
surfaces, near the discharge of the impeller. The loading
function includes the effect of blade rotation, blade turning, and
the passage area variation. This velocity difference is then
used to calculate the slip velocity using Stodolas assumption.
The final slip model can then be related to Carters rule for
axial impellers and Stodolas slip model for radial impellers.

Subscripts:
1: impeller inlet
2: impeller exit
p: pressure side of the blade
s: suction side of the blade

This new slip model suggests that the flow coefficient at the
impeller exit is an important variable for the slip factor when
there is blade turning at the impeller discharge. This may
explain the interesting slip factor trend observed from
experiments, such as the rise of the slip factor with flow
coefficient in Eckardt A impeller. Some validation results of this
new model are presented for a variety of applications, such as
radial compressors, axial compressors, pumps, and blowers.

INTRODUCTION
For an impeller with an infinite number of blades, the exit flow
angle should be the same as the exit blade metal angle. In the
real world, however, the exit flow angle deviates from the blade
guidance at the impeller exit due to the finite number of blades.
Correctly predicting flow deviation is a critical task in meanline
and throughflow modeling because the exit flow angle is
directly related to the work input and the pressure rise across
the impeller.

NOMENCLATURE
= flow angle, positive if same as rotation direction
b = blade metal angle
= blade camber angle
d/dm = blade turning rate
= deviation angle

Although the flow deviation phenomenon occurs in both axial


and radial impellers, researchers in their respective fields have
treated the subject quite differently. Axial impeller designers
use deviation angle, which is defined as the difference
between the exit flow angle and the blade metal angle (Figure
1).

= slip factor, American definition, equation (2)


= slip factor, European definition, equation (5)
2 = exit flow coefficient, 2 = C 2 m /U 2

= 2b 2

= angular velocity of the blade rotation


= meridional inclination angle
= density
= blade stagger angle

(1)

For radial impellers, engineers use a slip factor to model the


deviation of the flow. To further complicate matters, there are
two different definitions of this parameter.

Copyright 2007 by Concepts ETI , Inc.

Obviously, the Wiesner slip factor does not vary with any flow
parameters; however, according to Equation (6), is a
function of the exit flow coefficient even though is constant.
For a backswept blade with 2b < 0 , , it would decrease with

The one that is more commonly used in America is defined as:

= 1

C slip
U2

(2)

increasing 2 . The variation of the slip factor with the flow


parameter is one of the focuses of this study and will be further
discussed later in more detail.

With the slip velocity C slip being defined as:


C slip = C 2 C 2

(3)

A large number of researchers, such as Stodola [3], Eck [4],


Stanitz [5], Wiesner, Paeng and Chung [6], Pampreen and
Musgrave [7], Backstrom [8], etc., have come up with
correlations for the slip factor of centrifugal impellers. However,
although these correlations typically work well for one series of
impellers, they work less well, or fail badly, for other types. For
example, Wiesners model works fine for Eckardts O impeller
(Figure 3b) but is significantly off for Eckardts A impeller
(Figure 4b). A model that can be applied to a wide variety of
machines is still difficult to find. Most of the correlations relate
the slip factor to the blade geometry parameters, such as the
exit blade angle, number of blades, and sometimes, the inletto-exit radius ratio. Could it be that some key geometry
parameters are missing from the consideration of those
correlations?

where C 2 is the actual exit tangential velocity and C 2 is the


would-be tangential velocity if the flow exits at the same angle
as the blade metal angle, and therefore is calculated as:
C 2 = U 2 (1 + 2 tan 2 b )

(4)

where 2 = C m 2 /U 2 is the flow coefficient defined at the


impeller exit.

Furthermore, all existing slip models are only the function of


the blade geometry, not a function of any flow parameters. For
instance, Wiesners slip factor is a constant once the blade
geometry is defined. However, in many cases, tests have
shown the slip factor is a function of flow rate. How can these
contradictions be reconciled?
The axial impeller is in a better state comparatively, as far as
the deviation angle is concerned. Carters rule [9] and its
derivatives, such as the correlation from Lieblein [10], have
done a fairly good job in predicting the axial impeller deviation
angle. However, the method for estimating the deviation angle
for axial impellers is significantly different from the way the slip
factor is calculated for radial machines. A dilemma arises when
a mixed flow impeller is considered: should a deviation model
based on axial cascades be used or should a slip factor model
based on radial impellers be used? This calls for a unified slip
factor model that can be applied to axial, mixed-flow, and radial
impellers.

FIGURE 1. VELOCITY TRIANGLE AT IMPELLER EXIT


The other definition that is more frequently used in Europe, as
suggested by Japikse and Baines [1], is defined as:

C 2
C 2

(5)

For a no inlet swirl case, this value is the ratio of the actual
impeller work input to the ideal impeller work input with no slip.

This paper tries to address these issues by proposing a unified


model for axial, mixed-flow, and radial impellers. Besides the
traditional parameters involved, such as blade angle and the
number of blades, this model considers an additional important
geometrical variable the blade turning rate, and a flow
variable the exit flow coefficient. This model has been
validated with a variety of applications ranging from axial and
radial compressors and pumps to blowers.

It is worthwhile to spend a little more time with these two


definitions because they have quite often caused confusion
among engineers. The relationship between the two can be
easily derived as follows:

= 1

1
1 + 2 tan 2b

(6)

DERIVATION OF THE NEW SLIP FACTOR MODEL

For a radial exit impeller ( 2 b = 0 ), the two definitions give the


same results; however, for backswept blades ( 2b < 0 ) or

The relative eddy has often been considered the main


mechanism for the flow deviation in centrifugal impellers. An
observer traveling with the impeller can see a relative
circulation inside the flow passage, with the flow coming
radially inwards on the pressure side and outwards on the
suction side. Stodola argued that this relative motion is
responsible for the reduction of the absolute tangential velocity
at the impeller exit. He assumed that near the impeller
discharge, the relative eddy is circular in shape with a diameter
of a (same as the exit throat AC in Figure 8) and rotates as a

forward-swept blades ( 2 b > 0 ), the two definitions can be


significantly different. For example, Wiesners slip factor [2]
correlation follows the American convention. A simple form of
Wiesners correlation is as follows:

= 1

cos 2b
Z 0 .7

(7)

Copyright 2007 by Concepts ETI, Inc.

solid body with angular velocity ; therefore, the slip velocity


can be calculated as follows:
Cslip

a
s cos 2b
= = 2
2
2

C slip = F * {

(8)

(10)

Our next goal is to find a way to estimate the blade loading at


the impeller exit throat AC , so that the right hand side of
equation (10) can be evaluated. If we assume that the velocity
difference between D and A is small, the loading at the exit
throat AC can be approximately calculated through the
loading on the arc

= 1

t2
s2 cos 2 b

Z2

Z2

F cos 2 b sin 2

Z2

(15)

Equation (15) can be written in the following form:

= 1 radial turn passage

(16)

where

where the F-factor can be estimated with the following


equation:

= 1

DISSECTING THE NEW SLIP FACTOR MODEL

(12)

sin(

U2

(11)

DC can be related to the pitch length at the impeller exit ss


through a shape factor F,

C slip

Fs 2 2 d
F 2 s 2 sin 2 b d b

4 cos 2 b dm 2
4 2b2
dm 2

Ws W p

F = 1 sin

(14)

For an impeller with a finite rotation speed, a final form of slip


factor can be obtained when equation (14) is divided by the
rotational speed.

Cumpstys [12] work, the loading on the arc DC can be


expressed as follows:

DC = F * s2

W 2 s 2 d

4 dm 2

W 2 s 2 sin 2 2 b d b
}

8 2 b2
dm 2

DC . Following Johnston [11] and

= 2 sin cos
DC
d cos sin db
W (

)
dm
dm
b

A comment should be made here about applying equation (14)


to the stators before we reach the final form of the slip factor.
The first term in the parentheses is related to the blade
rotation, which is usually the dominant term for the radial
impeller. For stators where is zero, the first term drops out
and the second term becomes the dominant term that
determines the flow deviation. An important conclusion here is
that for a radial vaned diffuser, it is not correct to use any
existing slip models developed for radial impellers, such as
Stodolas model or Wiesners correlation, which essentially
models only the first term in equation (14). This is in contrast
with axial applications where Carters rule can be applied to
both rotors and stators. This is because in axial cases, the first
term in equation (14) is always zero because is zero,
regardless of whether it is a rotor or a stator. Therefore, both
axial rotors and stators follow the same deviation mechanism
that is dominated by the second term.

Therefore, the slip velocity can now be related to the blade


loading, i.e., the velocity difference on the blade suction and
pressure surfaces at the impeller discharge:
W Wp
a
= s
2
4

Strictly speaking, all the parameters inside the parentheses,


such as W,,, d/dm, should be evaluated at the arc DC
location. However, for practical purposes, they are
approximated with the values at the impeller exit.

The angular velocity in Stodolas model is the same as the


wheel speed, but in the reversed direction. Eck [4] further
extended Stodolas theory by arguing that the slip velocity is
the result of the velocity difference on the pressure and suction
surfaces. This unequal velocity distribution not only includes
the effect of the relative eddy, but also has the effect of the
blade turning. Eck further stated that the linear velocity
gradient is equivalent to a rotation of the total flow across the
passage and he recalculated the angular velocity of this
rotation as follows:
W Wp
(9)
= s
2a

C slip =

s 2 cos 2 b sin 2

radial =

F cos 2b sin 2
Z2

(16a)

is the decrement due to the radial rotation effect;

+ 2 b ) cos 2 b sin 2

(13)

turn =

Fs22 d

4 cos 2b dm 2

(16b)

is the decrement due to blade turning, and

The detailed derivation of the F-factor can be found in the


Appendix at the end of this paper.

passage =

Combining equations (10), (11), and (12), we can obtain a final


expression for the slip velocity:

F2 s2 sin 2b db

4 2b2
dm 2

(16c)

is the decrement due to the passage width variation.

Copyright 2007 by Concepts ETI, Inc.

a) The radial term and Stodolas model


b) The turning term and Carters rule

First, note the introduction of the inclination angle 2 in (16a).

The turning term is the result of extra loading from the


streamline curvature, or in other words, the blade turning. The
key geometry parameter in equation (16b) is the blade-turning
rate, d/dm. In axial cases, for any non-zero camber airfoils,
the turning rate is usually larger than zero. For radial impellers,
the turning rate could be negative, zero, or positive. Lots of
analytical work of the slip factor was based on impellers with
logarithmic spiral blades, as such was the case of the pioneer
slip factor work by Busemann [14]. The logarithmic spiral blade
assumption conveniently sets the blade-turning rate to zero,
and therefore, the turning effect never shows up in the slip
model. In hindsight, this assumption could be the root cause of
why blade-turning has never been factored into slip factor
modeling, since most slip factor correlations are compared
back to the original Busemann work. Readers should
distinguish between the backswept exit blade angle (i.e., the
negative blade angle) and the negative blade-turning rate. The
backswept blade angle could have a positive blade-turning rate
as long as the blade angle is increasing towards the impeller
discharge, which is, in fact, the case for a lot of pumps and
blowers.

Because of the presence of 2 , the new model can now be


applicable to the mixed-flow impeller, extended from the pure
radial application as most radial slip models do. It provides a
smooth transition from radial to axial impellers. It is also
interesting to notice that this term is also consistent with Lewis
conformal transformation work for mixed-flow machines [13].
His analysis has established that for a mixed-flow impeller, the
slip factor is given by that of an equivalent radial impeller with
M blades where

M =

Z2
sin 2

(17)

which is exactly the form that Z 2 and 2 are presented in


(16a).
In terms of loading, the radial term reflects the presence of the
Coriolis force. This term is the root cause for the completely
different approaches of the existing slip/deviation modeling
when dealing with axial and radial impellers. For axial
impellers, this term is zero, which is consistent with the fact
that the Coriolis force is zero because the flow direction is the
same as the rotation axis. For radial impellers, this term is
typically the dominant term. Most of the slip factor models for
radial machines account mainly for this effect. For example,
the Stodolas model can be written as follows:

= 1

cos 2b

(18)

Z2

which is consistent with the new slip model when 2 =

The slip factor decrement due to blade turning is significant if


the blade-turning rate d/dm is non-zero. In other words, if
there is a blade angle change near the impeller discharge, the
blade turning effect will affect the slip factor in spite of whether
it is an axial or radial impeller. According to equation (16b), if
the blade angle decreases towards the impeller exit,
d / dm < 0 , the turning term reduces the blade loading and,
therefore, causes the slip factor to increase with the exit flow
coefficient. Contrarily, if d / dm > 0 , the slip factor decreases
with the increasing flow rate.

and
2
the shape factor F is set to 1.0. Furthermore, if Z 2 in equation
(18) is replaced by M in equation (17), Stodolas model
becomes identical to 1 radial , provided that F = 1.0.

For an axial impeller without a large flow passage variation,


this is the only term that causes deviation because the radial
term is zero. From the velocity triangle at the impeller exit, we
can calculate the slip velocity based on the deviation angle,
assuming that the deviation angle is relatively small.

The introduction of the shape factor F in the new slip model is


also important. It is known that Stodolas model typically underpredicts the slip factor, especially for backswept impellers.
Since the shape factor F is always less than 1.0, it provides the
necessary adjustment to the original Stodola formula. Another
important application of the F -factor is that it can be used to
establish the applicable range of the current model. It is well
known in slip factor modeling that when the blade radius ratio
exceeds certain limits, the slip factor steeply decreases with an
increase of the radius ratio [2][8]. In the current model, the F factor approximately represents where the loading is
estimated, which apparently needs to be inside the blade
passage in order for the calculation to be meaningful. This
requires

R1
<F
R2

C slip = C m {tan( 2 b + ) tan( 2 b )}


C 2 m / cos 2 2 b

(20)

On the other side, the slip velocity caused by turning can also
be estimated from equation (14).

Cslip =

FC2 m s2 d

4 cos 2b dm 2

(21)

For most of the axial airfoil, we may assume a linear


distribution of the blade angle along the chord, therefore,
1b
d
. Equating (20) and (21) we can then obtain

2b
dm
c cos

(19)

Fs2 cos 2b d Fs2 cos 2b

4
4
c cos
dm
F cos 2b
=
( s2 / c)
4 cos

If the radius ratio is larger than F, the current method becomes


invalid. Adjustments similar to Wiesners model will be needed,
which is beyond the scope of this paper. However, it is
interesting to see how the F -factor becomes a natural limit for
the radius ratio.

(22)

Copyright 2007 by Concepts ETI, Inc.

the partial success that some slip models, such as the one by
Wiesner, have enjoyed with radial impellers. Once again we
should be careful with the slip factor definition convention.
Even though the slip factor based on the American convention
(equation [2]) is constant when the turning term is negligible,
the slip factor based on the European-convention (equation [5])
always varies with the flow rate when the exit blade angle is
not zero according to equation (6). Therefore, one should
never attempt to apply a European slip factor correlation that
does not include a flow parameter, such as the ones by
Stanitz, Balje [15], etc. for impellers with non-radial exits.

Comparing equation (22) with Carters rule, which is written in


the following form,

= m( s2 / c) n

(23)

we found that the current model is similar to Carters rule if we


set n=1 and the factor m as follows:
m=

F cos 2 b
F cos( + 2 b 0 )
=
4 cos
4 cos

(24)

Here, 2b 0 is the airfoil exit blade angle at zero stagger. Of


course, in Carters formula, he chose n=1/2. However, Carter
did mention that some difficulty was encountered in deciding
to what power s/c should be raised in this expression. In
general, theoretical results indicate a linear relationship with
s/c, while a root variation appears to fit test results better.
Once again, the analysis seems to be indicating that the linear
variation should be the preferred choice.

In some radial impellers, however, the turning term cannot be


ignored. In that case, the slip factor will show a significant
variation with the flow conditions. This may explain why a
robust correlation has been so difficult to achieve for radial
impellers when blade-turning rates and the flow conditions are
not taken into consideration.
Finally, the current slip model has identified a unique flow
parameter that correlates with the slip factor. This parameter is
the exit flow coefficient, 2 . As mentioned before, most of the
existing slip models are only functions of geometric parameters
and are independent of any flow variables. However, for a lot
of cases, test data have consistently indicated that the slip
factor is dependent on flow conditions. The new slip model
suggests that the slip factor is a linear function of exit flow
coefficient and the rate of the variation is decided by the bladeturning rate near the impeller discharge. If this holds true, the
slip factor should correlate well with 2 for different rotational
speeds and flow rates. That indeed is the case that will be
shown in the validation study later on in this paper.

For the m factor, Carter suggested that the factor is a function


of the stagger angle alone. Equation (24) is in agreement with
that statement, since 2b 0 is fixed for a particular airfoil, and F
is very close to 1 if the trailing edge thickness is only a small
fraction of the pitch (the second term in equation [13] is zero
for an axial machine). Setting 2b 0 = 10 for a typical airfoil,
equation (24) can be used to plot a relationship between the
stagger angle and the factor m, which is compared with
Carters original correlation in Figure 2. The scale and the
trend of the two lines agree surprisingly well. This analysis
confirms that the current model can be reduced to a form
similar to Carters rule for an axial impeller.

c) The passage variation term


0.4

The passage variation term is missing from any existing


deviation models. The term accounts for the flow acceleration
or diffusion near the discharge because of the passage
variation, db / dm . For instance, when the passage is
expanding, the loading increases, which causes the slip factor
to decrease. Typically this term is very small compared with
the radial term and the turning term, thus the contribution of
this term to the slip factor is minimal. Therefore, we are not
going to devote any more attention to this term in this paper
and it will always be assumed to be zero in all of our validation
studies.

Carter's m-factor

0.35
0.3
0.25
0.2
0.15
0.1

Current Model
Carter

0.05
0
-60

-40

-20

20

40

60

In summary, the radial term in the slip factor is consistent with


Stodolas model and Lewis conformal mapping work for the
mixed-flow impeller. The introduction of the impeller exit
inclination angle in the equation allows the model to be
applicable to mixed-flow impellers and have a smooth
transition to axial applications. The new model is found to be in
good agreement with Carters rule when applied to axial
impellers. The presence of the shape factor F is also important.
Not only does the F-factor provide the necessary adjustment to
Stodolas model, but also sets a natural blade radius ratio limit
beyond which the model becomes invalid. While the radial term
is only significant for the radial and mixed-flow impeller, the
turning term is important whenever significant blade turning
exists at the blade discharge section this is true for axial,
radial, and mixed-flow impellers as well as stators. Neglecting
the turning term and the blade turning parameter in the slip
factor modeling may be one of the main reasons why a robust
correlation has not been found for a wide range of impellers.
Finally, the new model suggests that the slip factor varies

Stagger angle

FIGURE 2. COMPARISON OF THE MODEL PREDICTION


OF CARTERS M FACTOR FOR CIRCULAR ARC AIRFOILS
Another interesting observation of equation (22) is that the
deviation angle is only a function of blade geometry, not any
flow parameters. This is not true for the slip factor, which varies
with the exit flow coefficient 2 linearly, as shown in equation
(16). This provides a justification for the researchers of axial
machines to model the deviation angle instead of the slip
factor.
For radial machines, things become more complicated. If the
turning term is very small compared with the radial term, the
turning term can be ignored and the slip factor remains a
constant independent of the flow conditions. This explains

Copyright 2007 by Concepts ETI, Inc.

linearly with the exit flow coefficient. The direction and the
slope of the variation are determined by the blade-turning rate
near the impeller discharge.

the data, the new model was slightly on the low side while the
Wiesner model was a little on the high side. Stodolas model
under-predicted the slip factor by a significant margin.

CASE STUDY AND VALIDATION

The test data showed a visible downward trend with the exit
flow coefficient. This may have been indicating a slight positive
flow turning near the discharge, which could have been a result
of the large wake region presented at the impeller exit. If the
blade-turning rate was set to be 5.0m-1, which corresponded to
0.29 of turning per mm of length, a better match could be
obtained, as represented as a solid line in Figure 3B (note that
to obtain this better match, the F factor had to be reduced
from its default value to 0.65).

To validate the new slip model, a variety of different impellers


were studied. The applications used in the study ranged from
radial, axial compressors and pumps to blowers. One of the
difficulties of this effort was to find cases that had a detailed
geometry definition at the impeller exit, so that the blade
turning rate could be properly set. Five different cases are
discussed here, which represents the typical results achieved
during our study using the new slip model.

The Eckardt rotor A had a backswept angle of 30 (Figure


4A). The blade first turned to a more radial direction and then
turned back to 30 at the impeller exit. In this case, the blade
turning rate was estimated at 9.0m-1, which corresponded to
(0.5) per mm of radius increment. The default F factor based
on equation (13) was 0.72. In order to compare it with
Eckardts original plot, the European slip factor based on
equation (5) was plotted against the exit flow coefficient in
Figure 4B.

a) Radial compressors: Eckardt Rotor O and Rotor A


The Eckardt rotors are arguably the most famous impellers in
the centrifugal compressor world. Eckardts [16] laser
velocimeter measurements still provide some of the most
accurate and detailed flow field data of a centrifugal impeller.
The Eckardt rotor O geometry information is shown in Figure
3A. The blade turning rate near the exit was close to zero. The
default F-factor calculated was 0.86. The slip factor variation
with the exit flow coefficient is presented in Figure 3B. The test
data was reproduced from Eckardts original plots in [16].

The rotor A slip factor test data showed a distinctive upward


trend with the increase of flow coefficient. Eckardt attributed
this unusual trend to the S-shaped blade design of impeller A,
namely, the turning back of 30 from the radial direction near
the impeller discharge. Sturge and Cumptsys [17] also
confirmed this trend in their two-dimensional numerical
simulation.

Parameters

Values

R2(m)

0.2

2b(degree)
2(degree)

0
90

Parameters Values

Z2
F
d/dm(m-1)

20
0.86
0

2b(degree) -30
2(degree) 90

R2(m)

0.2

Z2
20
F
0.72
d/dm(m-1) -9

FIGURE 3A. ECKARDT ROTOR O GEOMETRY

FIGURE 4A. ECKARDT ROTOR A GEOMETRY

0.95

Slip Factor (European) '

0.95

Slip Factor

0.9

0.85

RPM=10000
RPM=12000
RPM=14000
RPM=16000
RPM=18000
Current Model (no turning)
Current Model (with turning)
Wiesner
Stodola

0.8

0.75

0.9

0.85

R PM =10000
R PM =12000
R PM =14000
R PM =16000
C urrentM odel
W iesner
Stodola

0.8

0.75

0.7
0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8
0.7

Exit Flow Coefficient 2

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

Exit Flow Coefficient 2

FIGURE 3B. ECKARDT ROTOR O SLIP FACTOR

FIGURE 4B. ECKARDT ROTOR A SLIP FACTOR

The new model did a good job matching the test data, and the
same can be said about Wiesners correlation. Compared with

Copyright 2007 by Concepts ETI, Inc.

The current slip model matched the data very well in this case.
The negative blade-turning rate was obviously responsible for
the upward trend of the slip factor. Wiesner and Stodolas
models both failed badly for this case they both predicted an
opposite
downward
trend.
This
case
emphatically
demonstrated that the blade-turning rate could be an important
factor in the slip factor calculation. This was also the first time
that a slip factor model could be used to explain the unusual
upward trend of the slip factor for this famous impeller.

10
9
8

deviation angle

7
6
5
4

100%

Finally, it was impressive to notice how all the data points on


four different speed lines closely followed a single curve. This
indicated that the slip factor correlated very well with a single
flow parameter the exit flow coefficient, which was an
important conclusion from the current slip model.

90%
70%

Current Model

Carter

0
0.2

0.25

0.3

0.35

0.4

0.45

0.5

0.55

0.6

0.65

0.7

Exit Flow Coefficient 2

b) An axial compressor: Rotor 37

FIGURE 5B. ROTOR 37 DEVIATION ANGLE

The NASA Rotor 37 (Figure 5A) was chosen for the study
because of its detailed measurement of the flow field at the
impeller exit [18]. In a way, it was not an ideal case for the
deviation angle study because the presence of the shock
structure may have significantly altered the flow direction.
Nevertheless, the study demonstrated the main points that the
authors wanted to make.

0.95

slip factor

0.9

For this case, the blade-turning rate near the discharge was
estimated to be 17.0m-1, corresponding to 0.97 per mm of
meridional length. The default F-factor was 0.998. The radial
term contribution to the slip factor, in this case, was zero.

0.75

0.7
0.2

-38.9
0
36
0.998
17

0.3

0.35

0.4

0.45

0.5

0.55

0.6

0.65

0.7

FIGURE 5C. ROTOR 37 SLIP FACTOR


In this study, we had demonstrated that when it was applied to
an axial impeller, the new slip model could produce
comparable results to Carters rule. The current model was
capable of dealing with an axial impeller in a consistent
manner as for a radial impeller. A unified slip factor model for
radial and axial impellers is important because it bridges the
gap between two completely different approaches taken by
axial and radial deviation modeling. It also provides a potential
model for the cases with a mixed-flow exit, which used to fall in
a gray area when the slip factor or deviation was considered.
c) A pump case
This pump (Figure 6A) had test data for power input measured
at different flow rates using a torque meter. Adjustments were
made to the measured power values for bearing losses as well
as for leakage, windage, and recirculation losses using the
PUMPAL1 program from Concepts NREC. The adjusted
power value was now considered as the aerodynamic work
input based on the Euler turbomachinery equation, which could
be used to calculate the flow turning and the slip factor.

Parameters Values
2b(degree)
2(degree)
Z2
F
d/dm(m-1)

0.25

Exit Flow Coefficient 2

Figure 5C shows the slip factor vs. the exit flow coefficient. The
test data showed an obvious downward trend in this plot as the
flow coefficient increased. The current model predicted a linear
decrease of the slip factor because of the positive bladeturning rate. If you take out the 100% supersonic speedline,
which potentially has the shock wave altering the flow
direction, the rest of data points follow the predicted trend quite
well. Carters rule also predicted the slip factor would decrease
with the flow coefficient, but the scale was on the high side.

0.218

100%
90%
70%
Current Model
Carter

0.8

Figure 5B plots the deviation angle vs. the exit flow coefficient.
The data are scattered without any apparent upward or
downward trend. The scattering of the data is somewhat
disappointing, which may have been caused by the presence
of the shock structure in the flow passage. As stated earlier,
the deviation model based on the current slip model was not a
function of flow conditions and it appears on the plot as a flat
line. In the calculation based on Carters rule, the m-factor was
estimated as 0.33. Carters deviation angle did not vary with
the flow coefficient either. Although scattered, most of the data
points laid in the range of 1.5 of the current model prediction,
which was close to the angle measurement error margin of 1
in the test.

R2(m)

0.85

Applying the current model to pump applications had been a


more difficult task compared to the compressor and fan/blower
cases. The test data in Figure 6B showed an apparent
downward trend, which indicated a positive blade-turning rate

FIGURE 5A. ROTOR 37 GEOMETRY


1

PUMPAL is a registered trademark of Concepts ETI, Inc.


Copyright 2007 by Concepts ETI, Inc.

according to the current model. The estimated turning rate in


-1
this case from the actual geometry was about 4.92 m . To
match the slope with the test data presented, a value of 16.4
-1
m was needed. The F-factor also needed to be slightly
adjusted from its default 0.52 to 0.60 to obtain the nice match
shown in Figure 6B.

because it missed the F-factor, which was as small as 0.52,


and thus, had a significant effect on the slip factor.
d) A blower case
The blower case (Figure 7A) had Brake Horsepower (BHP)
measurements for a range of flow rates. Similar to the pump
case, adjustments were made to the BHP values for the
windage loss and recirculation loss using the FANPAL2
program from Concepts NREC. The adjusted power value was
then used to derive the slip factor.

This case was typical for most of the pump cases we had
studied, which usually required an increase of the blade turning
rate value to obtain a good match. One of the differences
between a compressor and a pump is the working fluid. For a
pump, it is usually water and the viscosity of water is much
larger than that of air. It could be argued that the friction force
in the volute was responsible for reducing the impeller exit
tangential velocity, thus increasing the slip velocity and causing
the slip factor to drop as the flow increased. Although the
mechanism for the extra deviation was uncertain, it was
interesting to note how we could artificially adjust the bladeturning rate to match the test data, which was important in
establishing a meanline model for an impeller.

Parameters Values

Parameters Values
R2(m)

R2(m)

0.232

2b(degree)
2(degree)
Z2
F
d/dm(m-1)

-58.5
90
9
0.63
13.12

FIGURE 7A. A BLOWER CASE GEOMETRY

0.073

2b(degree) -60
2(degree) 90

0.9

Z2
7
F
0.52
d/dm(m-1) 4.92

Slip Factor

0.8

0.7

Data
Current Model

0.6

Weisner

FIGURE 6A. A PUMP CASE GEOMETRY

Stodola

0.5
1

0.4
0
0.95

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

Exit Flow Coefficient 2

Current Model
Weisner

FIGURE 7B. BLOWER CASE SLIP FACTOR

Stodola
0.9

Slip Factor

0.05

Data

The blade-turning rate for this blower was estimated to be


13.12m-1. The default F-factor was 0.63. In Figure 7B, the new
model matched the data well except for the points at low flow
coefficients, which may have been partially attributable to the
recirculation loss model used during the data processing.
Neither the Wiesner nor Stodola model predicted the
downward trend for this case. Their predictions became worse
when the flow coefficient increased.

0.85

0.8

0.75

CONCLUSIONS
0.7
0

0.05

0.1

0.15

0.2

A new slip factor model is proposed that can be applied to


axial, radial, and mixed flow impellers. This model is based on
Stodola and Ecks slip factor modeling work as well as blade
loading analysis. According to this model, the slip factor is
mainly affected by the impeller rotation on the radial plane and
the blade turning when the blade-turning rate at the impeller
discharge is significant. For an axial impeller, there is no radial
effect and the slip factor is proved to be comparable to popular
deviation models such as Carters rule. For radial impellers, if
the blade turning effect can be ignored, the new slip factor

0.25

Exit Flow Coefficient 2

FIGURE 6B. PUMP CASE SLIP FACTOR


The Wiesner model did an excellent job for this case. Although
it did not predict the downward trend with the flow coefficient,
the Wiesner slip factor was very closely compared with the test
data. This showed why this amazing correlation of Wiesners is
still the most popular slip model for radial impeller applications.
The Stodola model under-predicted the slip factor, as usual,

FANPAL is a trademark of Concepts ETI, Inc.


Copyright 2007 by Concepts ETI, Inc.

[2] Wiesner, F. J., 1967, A Review of Slip Factors for


Centrifugal Impellers, ASME Journal of Engineering for
Power, 89, pp. 558-572
[3] Stodola, A., 1945, Steam and Gas Turbines, McGraw-Hill,
New York.
[4] Eck, B., 1973, Fans, Pergamon, Germany
[5]Stanitz, J. D., 1952, Some Theoretical Aerodynamic
Investigations of Impeller in Radial and Mixed-Flow Centrifugal
Compressors, Trans ASME, 74, pp.437-476.
[6] Paeng, K. S. and Chung, M. K., 2001, A New Slip Factor
For Centrifugal Impellers, Proc. Inst. Mech. Eng., Vol 215 Part
A, pp.645-649
[7] Pampreen, R.C. and Musgrave D. S., 1978, A Method of
Calculating the Slip Factor of Centrifugal Compressors from
Deviation Angle, Journal of Engineering for Power, Vol. 100,
pp.121-128.
[8] Backstrom T W, 2006, A Unified Correlation for Slip Factor
in Centrifugal Impellers,, Journal of Turbomachinery, Vol 128
[9] Carter A. D. S. and Hughes H. P., 1946, A Theoretical
Investigation Into the Effect of Profile Shape on the
Performance of Airfoils in Cascade, Report and Memo. No.
2384, British ARC.
[10] Lieblein S., 1960, Incidence and Deviation Angle
Correlations for Compressor Cascades, Trans. ASME, Journal
of Basic Engineering, 82, 575-587
[11] Johnson J. P., 1986, Radial Flow Turbomachinery,
Lecture in series Fluid Dynamics of Turbomachinery. ASME
Turbomachinery Institute, Ames, Iowa.
[12] Cumpsty N. A., 1989, Compressor Aerodynamics, Addison
Wesley Longman, England.
[13] Lewis R. I., 1996, Turbomachinery Performance Analysis,
John Wiley & Sons, New York.
[14] Busemann A., 1928, Das Forderhohenverhaltniss radialer
Kreiselpumen mit logarithmischspiraligen Schaufeln, Z.
Angew. Math. Mech. 8, pp. 372-384.
[15] Balje O. E., 1981, Turbomachines. A guide to design,
selection and theory. John Wiley and Sons, New York
[16] Eckardt D., 1980, Flow Field Analysis of Radial and
Backswept Centrifugal Compressor Impellers, Part 1: Flow
Measurement Using a Laser Velocimeter, 25th ASME Gas
nd
Annual Fluids Engineering
Turbine Conference and 22
Conference, New Orleans. Aymposium Performance
Prediction of Centrifugal Pumps and Compressors.
Proceedings published by ASME.
[17] Sturge D. P. and Cumptsy N. A., Two-Dimensional
Method for Calculating Separated Flow in a Centrifugal
Impeller, Journal of Fluids Engineering, Trans. ASME, Series
I, Vol 97, 1975, pp. 581-597.
[18] Moore R. D. and Reid L., 1980, Performance of SingleStage Axial-Flow Transonic Compressor with Rotor and Stator
Aspect Ratio of 1.19 and 1.26, Respectively, and with Design
Pressure Ratio of 2.05, NASA Technical Paper 1659.

model is similar to Stodolas model. For many radial impellers,


however, the blade turning term is significant and it is the key
factor that controls the trend of slip factor variation.
The method was validated with a variety of cases, ranging from
radial and axial compressors to pumps and blowers. The
reported cases have shown that the new model can be applied
successfully to both radial and axial machines. It also provided
sound explanations for the trends of slip factor variation with
flow conditions. Although no mixed-flow case was presented in
our validation study due to lack of test data for such machines,
it should be a safe interpolation to state that the proposed
method can also be applied to mixed flow impellers. The
agreement between the radial term (Equation 16a) and Lewis
conformal transformation work [13] also provided an additional
assurance for the successful application of the proposed
method to mixed-flow impellers.
One of the difficulties during the validation effort was that the
blade-turning rate at the impeller discharge was not commonly
available in the old databases. The requirement of knowing this
parameter can also limit this models effectiveness as a
meanline preliminary design tool when the blade geometry is
not fully specified. Nevertheless, a reasonable starting guess
with follow-up refinement of this parameter should still
constitute a good design practice.
The significance of this new slip model is threefold. First, the
new model provides a consistent basis for the calculation of
slip factor for both axial and radial impellers. This not only fills
a theoretical gap between the two in the field of meanline
modeling, it also allows direct application of the current model
to impellers with mixed-flow exits. Second, the model has
identified an important geometry parameter, the blade-turning
rate at the impeller discharge, as a key factor in the slip factor
calculation. This parameter was completely missing from any
existing slip factor models for radial types of impellers, while it
was hidden inside the axial deviation model through the
camber angle and chord. The omission of this variable may be
one of the main reasons why a robust slip factor model has
never been achieved. Third, the new model has made it clear
that a single flow parameter, exit flow coefficient, is correlated
with the slip factor. The slope of the correlation is determined
by the blade-turning rate. The identification of this flow
parameter may go a long way in future slip factor data
processing and meanline modeling.
Finally, since the model is based on blade loading, which is a
common factor for any rotor and stator in a turbomachinery
application, it is also possible that the current model could be
applied to other areas, such as turbine rotors, radial diffusers,
or inlet guide vanes.

ACKNOWLEDGEMENTS
APPENDIX:
FACTOR- F

The authors want to thank Dr. David Japikse of Concepts


NREC for the insightful technical comments and devoted
discussions he offered during the preparation of this paper.
Thanks also go to Mrs. Ellen Reney for her help in editing this
paper.

Introduction

OF

THE

SHAPE

Referring to Figure 8, a line AC is drawn from point A that is


perpendicular to the adjacent blade. Line AC is referred to as
exit throat in this paper. The goal in the following calculation is
to determine the length of arc DC .

REFERENCES
[1] Japikse D., Baines N. C., 1994,
Turbomachinery, Concepts ETI, Vermont.

CALCULATION

to

Copyright 2007 by Concepts ETI, Inc.

The angle between the two adjacent blades is:

2
Z2

The chord AB can then be calculated as:


AB = 2 R2 sin

2
Since OBA =

, therefore

ABC = OBA 2b =

2b
2

FIGURE 8. SKETCH FOR F-FACTOR CALCULATION

Now we can calculate the length of line BC .

+ 2b
BC = AB cos(ABC ) = AB sin
2

The radius reduction from the impeller tip (point B) to point C


can now be approximately calculated as:
BE = BC cos( 2b )sin


= 2 R2 sin
+ 2b cos( 2b )sin 2
sin

2 2

The presence of 2b is because the whole calculation is


projected to the radial plane to obtain the reduction in radius.
Therefore, the radius of the arc DC is
OC OB BE

= R2 1 2 sin sin
+ 2b cos( 2b )sin ( 2 )
Z
Z
2 2

For a blade with finite thickness, the length of arc DC is then

DC = OC *

t2
cos 2b

The F-factor is the ratio between the length of DC and AB :

t2
+ 2b cos( 2b )sin 2
F = 1 2 sin sin
Z
Z
s
cos
2b
2
2 2

10

Copyright 2007 by Concepts ETI, Inc.

Vous aimerez peut-être aussi