Slip Factor Prediction for Impellers with

Straight, Back-swept Blades

T.W. von Backströma
Received 26 September 2019, in revised form 4 November 2019 and accepted 2 December 2019

Abstract: Slip factor accounts for the deviation of the flow Φ flow coefficient = 𝑉𝑉𝑟𝑟 /𝑈𝑈𝑒𝑒
angle from the trailing edge blade angle at the exit of radial σ slip factor = 1 − ∆𝑊𝑊/𝑈𝑈𝑒𝑒
impellers. Accurate values are required to predict impeller Ω rotor angular velocity
torque and energy input. The slip factor prediction method
for back-swept radial bladed impellers is based on the so- Subscripts
called single relative eddy (SRE) method, which is an 1-5 version of F in SRE method
approximation of the classical, two-dimensional analytical c critical (beyond which slip factor depends on RR)
solution to the inviscid flow problem. The relatively cor corrected for radius ratio beyond the critical
accurate prediction of the slip factors of 19 impellers found e exit
in four data sets published since its formulation, shows the eq equivalent
reliability of the SRE method. The characteristics of i inlet
straight-bladed impeller layouts are explored and r radial
incorporated into the SRE method. It turns out that the ϑ circumferential
SRE method as developed for logarithmic spiral blades Superscript
predicts the inviscid flow slip factor of 42 different straight ˊ ideal (non-slip) value
bladed impeller geometries accurately, when a correction
for the critical radius ratio at low blade numbers is
1 Introduction
introduced. Predicted slip factor values are also compared
Designers of centrifugal impellers need methods to
to new experimental data for five different impellers with
accurately predict the impeller drive shaft torque and power
straight, back-swept blades, over a range of three blade
input into the flow. This is especially important in the design
angles and three blade numbers. Agreement is excellent
of simple centrifugal fans where the cost of the electric motor
near the flow coefficient corresponding to the volute design
may exceed the cost of the fan itself. No method developed
angle.
specifically for back-swept, straight-bladed centrifugal
impellers could be found in the literature. The question is
Additional keywords: Slip factor, Relative eddy, Rotor then whether classical, inviscid, two dimensional analytical
blades, Straight blades, Back-swept blades, SRE methods for logarithmic spiral blades apply to these blades,
and whether generally used, semi-empirical slip factor
Nomenclature prediction methods are adequate.
Lewis [1], Visser et al. [2] and Hassenpflug [3]
Roman
recalculated the classical, two-dimensional analytical
c blade chord length
solutions to the inviscid flow problem solved by Busemann
BH Busemann [4] data from Hassenpflug [3]
[4], Theodorsen [5, 6], and others. Visser et al. [2]
F solidity influence coefficient in SRE method
summarized the previous work, and applied conformal
F effective area coefficient in method of Qiu [16]
mapping to find exact solutions in terms of Fourier
f length (figure 1)
coefficients given by the Gauss hypergeometric series and the
RMS root mean squared (equation 8)
Beta function, for flow through two-dimensional radial
RR radius ratio; 𝑟𝑟𝑖𝑖 /𝑟𝑟𝑒𝑒
impellers with logarithmic spiral blades. These authors did
RRCF radius ratio correction factor
consider unswept, straight (radial) blades insofar as they
r radial position from axis
represent a limiting case of the logarithmic spiral blade.
SRE single relative eddy
Hassenpflug [3], however, was the only one, after
s blade spacing (pitch)
Theodorsen [5], to extend his approach from logarithmic
TH Theodorsen [5,6] data from Hassenpflug [3]
spiral blade shapes to straight, back-swept blades.
TSF theoretical slip factor, from Memardezfouli [21]
The rigorous, two-dimensional, inviscid, analytical
U rotor tangential velocity
solutions of Busemann [4], Visser et al. [2] and Hassenpflug
V absolute velocity
[3] depend on the number of blades, Z, the blade exit angle,
x length (figure 1)
βe and the impeller radius ratio, RR. RR is the ratio of the
Z number of blades
radial location of the rotor blade leading edge to that of the
Greek trailing edge. These methods apply the Kutta condition (local
β blade angle (from radial direction) flow parallel to the blade at the blade exit) and a similar, so
called shock-free condition, at the blade inlet. To ensure zero-
a. Department of Mechanical and Mechatronic incidence flow at the rotor blade leading edges at all through-
Engineering, Stellenbosch University, South Africa. flow conditions, the use of adjustable inlet guide vanes would
E-mail: [email protected] be implied in practice. To avoid this complication, many
R & D Journal of the South African Institution of Mechanical Engineering 2019, 35, 55-63 55
http://dx.doi.org/10.17159/2309-8988/2019/v35a7
http://www.saimeche.org.za (open access) © SAIMechE All rights reserved.
 Slip Factor Prediction for Impellers with Straight, Back-swept Blades

practical radial turbomachines have impellers with inducers the single relative eddy (SRE) method of Von Backström [12,
that guide the flow into the impellers, but sometimes the high 13, 14], while some recent methods, e.g. Ji et al. [15] and Qiu
solidity near the rotor entrance fulfills this function et al. [16] add the effects of blade curvature.
automatically. Like all the semi-empirical slip factor
prediction methods mentioned in this paper, the SRE neglects 3 The SRE method
the flow conditions at the impeller inlet. Von Backström [12] showed that five of the common
approximate methods (Stodola [10], Stanitz [11], Wiesner
2 Slip factor [9], Eck [17] and Csanady [18] are actually simplified
Slip factor accounts for the flow deviation with respect to the versions of the SRE method. It emerged during its derivation
blade trailing edge angle at the impeller exit. In the that the eddy-induced slip factor, equation 4 above), depends
fundamental case of two-dimensional, inviscid flow, slip on the blade solidity, c/se at the impeller exit multiplied by a
factor is determined by the vorticity of the internal flow universal coefficient, F, which is a function of the impeller
relative to the rotating impeller, resulting in a so-called blade exit angle and the number of blades:
relative eddy or eddies. These present as a reduction in the 1
absolute circumferential velocity component at the rotor exit σ = 1 − ∆𝑊𝑊𝑒𝑒 ⁄Ω𝑟𝑟𝑒𝑒 = 1 − (5)
1 + 𝐹𝐹 × 𝑐𝑐/𝑠𝑠𝑒𝑒
compared to the ideal case where the exit relative flow
follows the local blade direction. Engineering and teaching The coefficient, F, is the ratio of the sum of the average
texts like Aungier [7], Lewis [1], Whitfield and Baines [8] eddy-induced velocities on the suction and pressure surfaces,
typically non-dimensionalise the slip velocity through to that along the impeller exit rim. This coefficient depends
division by the ideal absolute circumferential velocity on βe and Z, and has been developed further in successive
component at the blade trailing edges: papers by Von Backström [12, 13, 14] to improve the
′
agreement between the SRE method and the numerical values
σ = 𝑉𝑉θ𝑒𝑒 ⁄𝑉𝑉𝛳𝛳𝛳𝛳 (1) from the two-dimensional analytical solution of Hassenpflug
The ideal circumferential component is determined from [3], henceforth called the BH data. The best empirical
the ideal, no-slip rotor exit velocity triangles as: approximation was [13]:
′
𝑉𝑉𝛳𝛳𝛳𝛳 = 𝑈𝑈𝑒𝑒 − 𝑉𝑉𝑟𝑟𝑟𝑟′ 𝑡𝑡𝑡𝑡𝑡𝑡𝛽𝛽𝑒𝑒′ (2) 3
𝐹𝐹 = 𝐹𝐹4 = 2 + �2.7 + � � 𝑐𝑐𝑐𝑐𝑐𝑐𝛽𝛽𝑒𝑒 � 𝑐𝑐𝑐𝑐𝑐𝑐𝛽𝛽𝑒𝑒 (6)
Assuming no pre-rotation at the rotor inlet, the non- 𝑍𝑍
dimensionalised slip velocity then equals the fractional As explained by Von Backström [14], the digit 2, outside
reduction in rotor torque and in power demand due to slip. the brackets in equation 6 ensures that F4 = 2 when 𝛽𝛽𝑒𝑒 =
The slip factor equals the ratio of torque or power, with slip 90° , resulting in excellent agreement with the BH data at
included, to torque or power, without slip. high values of βe. A remark of Hassenpflug [3] that, for an
In this paper, however, the slip factor is written in terms impeller with 2 straight radial blades (𝑐𝑐𝑐𝑐𝑐𝑐𝛽𝛽𝑒𝑒 = 1.0), the
of the eddy-induced slip velocity at the rotor blade exit, given inviscid flow slip factor should be exactly 1/2, led to further
by: investigation. For RR = 0.5, and the conditions mentioned
′
∆𝑊𝑊𝑒𝑒 = ∆𝑉𝑉𝛳𝛳𝛳𝛳 = 𝑉𝑉𝛳𝛳𝛳𝛳 − 𝑉𝑉𝛳𝛳𝛳𝛳 (3) above, c/se = 1/ (2π), and so F must be equal to 2π to make F
× c/se equal to the integer, 1. This is achieved by replacing
The slip velocity is then normalised through division by
the (3/ Z) in equation 6 by the more meaningful (π /Z), the
the circumferential speed, Ue of the rotor blade trailing edges,
half angle between adjacent blades, and by replacing the 2.7
to give the slip factor as:
by (1.5 π-2). These substitutions allow us to rewrite F in
σ = 1 − ∆𝑊𝑊𝑒𝑒 / 𝑈𝑈𝑒𝑒 = 1 − ∆𝑊𝑊𝑒𝑒 ⁄Ω𝑟𝑟𝑒𝑒 (4) terms of π as:
This definition is chosen because, unlike equation 2, it is π
𝐹𝐹 = 𝐹𝐹5 = 2 + �(1.5π − 2) + � � 𝑐𝑐𝑐𝑐𝑐𝑐𝛽𝛽𝑒𝑒 � 𝑐𝑐𝑐𝑐𝑐𝑐𝛽𝛽𝑒𝑒 (7)
independent of the through-flow, as it does not contain the 𝑍𝑍
radial velocity component. This makes it useful for This empirical correction ensures perfect agreement with
comparing rotors that do not operate at exactly the same flow the BH data for 2 straight radial blades (βe = 0), and an error
coefficients, as does Wiesner [9]. Also, Whitfield and Baines of 0.3 % for Z > 2. When Z = 1, the error is < 2.5 %. The
[8] state that the definition in equation 4 is more commonly comparative curve fit of Paeng and Chung [19] requires five
adopted than equation 1, and support the notion that it is the equations for similar accuracy, and is not defined when Z =
slip velocity, ∆We that is important. Visser et al. [2] state 1.
that for inviscid methods the use of equation 4 results in lower The current paper applies the SRE method to impellers
sensitivity to through-flow. Such solutions disregard effects with straight, back-swept blades, by making the initial
such as boundary layer growth and separation as well as wake assumption that the slip factor depends primarily on the blade
formation that contribute to flow deviation at blade trailing solidity, as defined by the sum of the blade lengths divided
edges. Since these effects are dependent on flow rate, they by the impeller circumference. The initial objective is then
may contribute to some extent to slip in practice. to write the solidity for straight blades in terms of the number
Simple, approximate slip factor prediction methods like of blades, their exit angle and the radius ratio for straight,
those of Stodola [10] and Stanitz [11] consider the slip factor back-swept blades. As a first approximation the solidity
to be dependent on the number of blades and blade trailing influence coefficient, F5, as developed for logarithmic spiral
edge angle only, but to be independent of blade shape. Some blades is applied, and the resulting prediction compared to the
other methods include corrections for the effect of the experimental data.
impeller inner to outer radius ratio (RR), e.g. Wiesner [9] and
R & D Journal of the South African Institution of Mechanical Engineering 2019, 35, 55-63 56
http://dx.doi.org/10.17159/2309-8988/2019/v35a7
http://www.saimeche.org.za (open access) © SAIMechE All rights reserved.
 Slip Factor Prediction for Impellers with Straight, Back-swept Blades

The validity of the solidity as a primary variable that Table 1 Comparison of slip factor prediction methods:
determines slip, has however been questioned by Huang et al. Huang [20]
[20]. This really relates to the effect of radius ratio, RR, as it Case 1 2 3 4 5 % RMS
is the only relevant variable that appears in the solidity but difference
not in its influence coefficient, F. This concern had been Z 13 15 15 15 17 -
addressed in the SRE method by assuming RR = 0.5 𝛽𝛽𝑒𝑒 (°) 30.0 30.0 41.8 47.0 41.8 -
Huang et 0.868 0.866 0.883 0.895 0.895 0.0
whenever RR < 0 [12]. It implies that slip factor does not al. data
depend on solidity when RR < 0.5. It also assumes that the Wiesner 0.845 0.860 0.870 0.876 0.881 1.6
drop-off in slip factor at larger radius ratios starts at RR = 0.5 SRE-F5 0.844 0.861 0.869 0.874 0.882 1.7
and that the drop-off rate is correctly predicted by the basic
SRE method. This assumption needs reconsideration, (RMS) value of the difference, to well below 2 %, as does
however, as discussed later. To check the validity of the SRE Wiesner [9] (table 1).
method, its performance will be compared to that of other Memardezfouli and Nourbakhsh [21] compare various
more recent data sets and prediction methods. slip factor prediction methods to their measured data for five
pump impellers. They compare the methods of Wiesner [9],
4 Objectives Stodola [10] and Stanitz [11], as well as an early version of
The general objective of this study is to investigate the slip the SRE [12] and their own theoretical slip factor (TSF),
factor of impellers with straight, back-swept blades, and to which is a superposition of an axi-symmetrical meridional
develop a method to predict their performance. Specific plane analysis and a blade-to-blade singularity method.
objectives of the work are to: Unfortunately they did not, in their application of the SRE
1. Investigate the geometric peculiarities of straight- method, correct the blade reference angle from radial to
bladed impellers tangential (i.e. change cosines in the SRE equations to sines).
2. Derive the SRE slip factor equations for the blade The SRE equation they list under their table also has 1 – RR2
solidity of straight bladed impellers. in it instead of 1 – RR. Table 2 shows the correct version of
3. Compare predicted and published slip factors. the SRE method, and compares, in the last column, the RMS
4. Select, design and manufacture a family of straight- differences. At their relatively large exit blade angles (60 to
bladed radial impellers. 72 °.) the SRE method agrees slightly better with their data
5. Measure the slip factor of the family of straight- than their own TSF method (2.1 % RMS difference compared
bladed back-swept impellers to 2.4), with both showing deviations of about half of that of
6. Compare predicted and measured slip factors Wiesner [9].
7. Make recommendations regarding the use of the Table 2 Comparison of slip factor prediction methods:
SRE and other methods for straight, back-swept Memardezfouli [21]
blades.
Case 1 2 3 4 5 % RMS
difference
5 Review of recent literature Z 6 8 6 6 4 -
Slip factors predicted by the SRE method will be compared 𝛽𝛽𝑒𝑒 (°) 66.0 66.0 60.0 72.0 66.0 -
Experiment 0.77 0.80 0.74 0.82 0.72 0.0
in terms of RMS differences, to those from four data sets
TSF 0.79 0.83 0.78 0.82 0.71 2.4
published since the SRE papers. The RMS value of the Wiesner 0.82 0.85 0.80 0.84 0.76 4.6
difference between xi and yi for i = 1 to N is defined as: SRE-F5 0.789 0.832 0.769 0.817 0.717 2.1

1 𝑁𝑁
𝑅𝑅𝑅𝑅𝑅𝑅 = � � (𝑥𝑥𝑖𝑖 − 𝑦𝑦𝑖𝑖 )2 (8) Ji et al. [15] developed a slip factor correlation for radial
𝑁𝑁 1 and mixed-flow impellers, based on the Euler turbo-
machinery equation. It incorporates the dependence of slip
The reference row in each of the tables below is the one
factor on the flow coefficient to better represent experimental
showing the zero RMS difference. The objective is to gauge
data. The relative eddy effect is introduced through a Stodola
how well the SRE predicts slip factor compared to these data
[10] type approach, where a circular eddy is fitted into each
sets. Due to space constraints the prediction methods from
blade passage near its exit. The Stodola model is inaccurate
literature are not discussed in detail.
Huang et al. [20] used commercial centrifugal compressor for low blade numbers (Z < 16 and high blade angle (βe > 50°)
software to design six impellers with 13, 15 and 17 blades (Von Backström, [14]). To this they then add the contribution
with exit angles of 30, 41.8 and 47 °, and modelled their the of blade curvature. Table 3 shows that the SRE with F5
performance at mass flows from 80 to 120 % of the design does about as well as their method (2.7 % RMS difference
flow, at zero tip clearance, using CFD. They investigated the compared to 2.8) with the Wiesner RMS difference even
effect of blade curvature, referring to a change in blade angle lower at 2.0 %.
relative to the local radial direction along the blade length. Qiu et al. [16] presented the development and validation
They found that it could affect the variation of slip factor with of a slip factor model for design and off-design, similar to that
flow rate, although, in their examples, the presented of Ji et al. [15]. They incorporate the flow coefficient, φe at
maximum variation in slip factor over their flow range is the rotor exit to cater for off-design variation in slip factor,
about 2 %. For the examples we could reproduce, the SRE and present four test cases that can be used as benchmarks.
predicted the slip factors, in terms of the root mean squared The Eckardt [22] O and A rotors, as well as a pump and
blower test case, all have relatively high sensitivity to flow
R & D Journal of the South African Institution of Mechanical Engineering 2019, 35, 55-63 57
http://dx.doi.org/10.17159/2309-8988/2019/v35a7
http://www.saimeche.org.za (open access) © SAIMechE All rights reserved.
 Slip Factor Prediction for Impellers with Straight, Back-swept Blades

coefficient. Their method, which is sensitive to flow 𝑟𝑟𝑖𝑖

coefficient, is very accurate, especially at mid flow range. 𝑅𝑅𝑅𝑅 = (10)
𝑟𝑟𝑒𝑒
The SRE method, does however capture some flow sensitivity
From the above, it is clear that βi is not defined if
when the slip factor is converted from our definition to that
𝑠𝑠𝑠𝑠𝑠𝑠𝛽𝛽𝑒𝑒 /𝑅𝑅𝑅𝑅 > 1 , since the straight blade will then not
of Eckardt [22], as proposed by Ji et al. [15]. The predicted
intersect the inner radius circle. This limits the possible blade
slip factor then varies from 0.90 for φ = 0.2, to 0.92 forφ =
exit angles to low values if RR is low, e.g. to 30° when RR =
0.6. For the blower case, RR was taken from the relevant
0.5, as in this study. Conversely, the use of large back-sweep
figure as 0.67. For the four test cases in table 4 the RMS
angles requires large values of RR. This implies that the
difference of 1.9 %, of the SRE method is 70% larger than the
critical radius ratio RRc where the slip factor starts dropping
1.1 % of Qiu et al. [16], but it is still relatively small and
off, may be exceeded in many designs.
about one third of that of Wiesner [9].
Table 3 Comparison of slip factor prediction methods: Ji
[15]
Case 1 2 3 4 5 % RMS
difference
Z 4 5 6 16 20 -
𝛽𝛽𝑒𝑒 (°) 64.5 64.5 64.5 20.7 0.0 -
Experiment 0.721 0.768 0.829 0.832 0.880 0.0
Ji et al. 0.753 0.805 0.836 0.871 0.872 2.8
Wiesner 0.743 0.786 0.812 0.861 0.877 2.0
SRE-F5 0.710 0.752 0.784 0.865 0.886 2.7

Table 4 Comparison of slip factor prediction methods: Qiu

[16]
Case 1 2 3 4 % RMS
difference
Z 20 20 7 9 -
𝛽𝛽𝑒𝑒 (°) 0.0 30.0 60.0 58.5 - Figure 1 The geometry of impellers with straight, back-
Experiment 0.875 0.940 0.815 0.750 0.0 swept blades
φ 0.45 0.60 0.13 0.20 -
Qiu et al. 0.875 0.940 0.825 0.770 1.1 In figure 1 the blade chord length is then found by noting
Wiesner 0.877 0.886 0.819 0.843 5.4 that the blade length, c = x – f, with:
SRE-F5 0.886 0.919 0.794 0.730 1.9 𝑥𝑥 = 𝑟𝑟𝑒𝑒 𝑐𝑐𝑐𝑐𝑐𝑐𝛽𝛽𝑒𝑒 (11)
In spite of being insensitive to flow coefficient, the SRE 𝑓𝑓 = 𝑟𝑟𝑖𝑖 𝑐𝑐𝑐𝑐𝑐𝑐𝛽𝛽𝑖𝑖 (12)
method performs consistently well (RMS difference of So:
typically 2 %) compared to the recently published data and 𝑐𝑐 = 𝑟𝑟𝑒𝑒 𝑐𝑐𝑐𝑐𝑐𝑐𝛽𝛽𝑒𝑒 − 𝑟𝑟𝑖𝑖 𝑐𝑐𝑐𝑐𝑐𝑐𝛽𝛽𝑖𝑖 (13)
methods in the 19 test cases listed in tables 1 to 4. This
establishes the use of an accurate, inviscid relative eddy The blade solidity at the impeller outer radius is the sum
model as a basis for a relatively accurate practical slip factor of the blade lengths divided by the rotor circumference. For
prediction method. It is therefore reasonable to consider it for straight blades it is:
impellers with back-swept straight blades too. As the method 𝑐𝑐𝑐𝑐 𝑍𝑍
� �= [𝑐𝑐𝑐𝑐𝑐𝑐𝛽𝛽𝑒𝑒 − 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝛽𝛽𝑖𝑖 ] (14)
of Qiu et al. [16] is sensitive to flow coefficient, but 2π𝑟𝑟𝑒𝑒 2𝜋𝜋
insensitive to radius ratio, and is based on an extensive data It has been shown before (Von Backström, [12]) that for
set, it will also be applied. The equations used to calculate the logarithmic spiral blades the solidity is:
Qiu et al. [16] slip factors are given in the appendix of their
𝑐𝑐𝑐𝑐 𝑍𝑍(1 − 𝑅𝑅𝑅𝑅)
paper. For application here, they were simplified to remove � �= (15)
the effects of disc coning (for mixed-flow impellers), blade 2π𝑟𝑟𝑒𝑒 2𝜋𝜋 𝑐𝑐𝑐𝑐𝑐𝑐𝛽𝛽𝑒𝑒
thickness and fluid density ratio. The well-tried empirical Note that for radial blades, when cosβe = cosβi = 1, both
method of Wiesner [9] on the other hand, is insensitive to equations above result in a solidity equal to Z (1-RR)/(2π).
flow coefficient but can account for radius ratio, so it will also From the solidity equations (13) and (14) above it follows
be considered. that, for a given straight-bladed impeller geometry, the
equivalent radius ratio of the logarithmic spiral bladed
6 SRE slip factor for straight-bladed impeller with the same solidity is:
impellers 𝑅𝑅𝑅𝑅𝑒𝑒𝑒𝑒 = 1 − 𝑐𝑐𝑐𝑐𝑐𝑐𝛽𝛽𝑒𝑒 [𝑐𝑐𝑐𝑐𝑐𝑐𝛽𝛽𝑒𝑒 − 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝛽𝛽𝑖𝑖 ] (16)
Figure 1 shows the geometry of a two-dimensional straight- If this radius ratio, RReq is used, the logarithmic spiral SRE
bladed impeller. The blade angle at the outer radius re, is βe. theory will render the same slip factor as the straight bladed
The blade angle, βi at the inner radius can be determined with SRE theory for a straight-bladed impeller with a particular
the sine rule: number of blades and blade exit angle.
𝑠𝑠𝑠𝑠𝑠𝑠𝛽𝛽𝑒𝑒 From equations 5 and 14 it follows that the SRE slip factor
𝛽𝛽𝑖𝑖 = 𝑠𝑠𝑠𝑠𝑠𝑠−1 � � (9) for straight bladed impellers is:
𝑅𝑅𝑅𝑅
where the radius ratio, RR is defined as:
R & D Journal of the South African Institution of Mechanical Engineering 2019, 35, 55-63 58
http://dx.doi.org/10.17159/2309-8988/2019/v35a7
http://www.saimeche.org.za (open access) © SAIMechE All rights reserved.
 Slip Factor Prediction for Impellers with Straight, Back-swept Blades

1
𝜎𝜎 = 1 − (17)
𝐹𝐹𝐹𝐹
1+ [𝑐𝑐𝑐𝑐𝑐𝑐𝛽𝛽𝑒𝑒 − 𝑅𝑅𝑅𝑅 𝑐𝑐𝑐𝑐𝑐𝑐𝛽𝛽𝑖𝑖 ]
2𝜋𝜋
The intention is to use the coefficient, F = F5, equation 7
listed above.

7 Literature for straight bladed impellers

Straight, radial blades (𝛽𝛽 e = 0) can be classified as members
of the logarithmic spiral family as well as of the straight blade
family. Consequently they have already been incorporated
into the development of the SRE and other theories in the
literature. Apart from data on impellers with radial blades
with no back sweep, no literature specifically addressing
impellers with back-swept, straight blades could be found.
Figure 4 Slip factor vs Solidity for TH data, Example 3
An available data set of analytically determined, back-swept,
straight bladed impellers, however, is that of Theodorsen [5, adjusted radius ratios, agree reasonably well with the TH slip
6] as reported by Hassenpflug [3], henceforth called the TH factors for solidities above about 0.6 (representing 8 or more
data (geometry in tables 5 to 7). He presents graphs of the blades), but that the accuracy deteriorates rapidly at lower
slip factor for two-dimensional inviscid flow against values. This indicates that a more accurate model for back-
equivalent sparsity (the inverse of solidity) for three straight- swept straight bladed impellers with low numbers of blades
bladed backswept impellers with 1 to 12, 1 to 14 or 1 to16 is required. When figures 2 to 4 are overlaid (not shown) it
blades, representing 42 test cases in total. In figures 2 to 4 also turns out that at the higher solidities the three sets of data
we compare the SRE slip factors, calculated with F5 against coincide well, indicating that for high solidities the slip factor
the real solidity, given by equation 14, extracted from the TH is indeed completely determined by the solidity and the
data. unique influence coefficient, F.
The top lines in the figures, designated by SlipLn, show The reason that the SRE does not agree well with the
that the SRE slip factors for straight blades using the un- inviscid solutions for less than 8 blades (solidity = 0.7), is that
according to the BH data, the drop-off in slip factor with
radius ratio already starts at RR < 0.5 when Z < 8. It is
however possible to correct for this effect. We propose the
following simple, empirical procedure. First calculate the
SRE slip factor with F5, for RR = 0.5. Then calculate the
critical radius ratio RRc = 2σ -1, where the slip factor starts
dropping with increasing RR. This follows from the
observation that on the Busemann [4] graphs of slip factor vs.
RR, that σ ≈ 0.5 + RRc/2. The radius ratio correction
coefficient is RRCF = (RR - RRc)/(1 - RRc). For RR > RRc,
the coefficient, F5 is then multiplied by 1 - RRCF1.5. The
exponent, 1.5 was chosen to blend the constant slip factor
region smoothly with the drop-off region. The result is so
good that SRE prediction and the TH data practically
coincide, as seen in figures 2 to 4 and tables 5 to 7 (RMS
differences of 2.0 %, 0.7% and 0.9 %). What is especially
gratifying is that the standard SRE with this simple
Figure 2 Slip factor vs Solidity for TH data, Example 1 modification could almost perfectly represent all of the 42
straight-bladed impellers of the TH data set. The Wiesner
RMS differences vary between 8 and 12 %.
Table 5 Comparison of slip factor prediction methods: TH
data [3] Ex.1
Case 1 2 3 4 5 % RMS
difference
Z 1 2 4 8 12 -
𝛽𝛽𝑒𝑒 (°) 20.0 20.0 20.0 20.0 20.0 -
RR 0.532 0.532 0.532 0.532 0.532 -
Theodorsen, 0.208 0.361 0.569 0.753 0.827 0.0
1
Wiesner 0.031 0.403 0.633 0.774 0.830 8.7
SRE-F5 0.383 0.501 0.640 0.786 0.829 10.5
RRc 0.0 0.003 0.279 0.537 0.659 -
SRE-F5Cor 0.241 0.381 0.585 0.768 0.829 2.0
Figure 3 Slip factor vs Solidity for TH data, Example 2
R & D Journal of the South African Institution of Mechanical Engineering 2019, 35, 55-63 59
http://dx.doi.org/10.17159/2309-8988/2019/v35a7
http://www.saimeche.org.za (open access) © SAIMechE All rights reserved.
 Slip Factor Prediction for Impellers with Straight, Back-swept Blades

Table 6 Comparison of slip factor prediction methods: TH

data [3] Ex.2
Case 1 2 3 4 5 % RMS
difference
Z 1 2 4 8 12 -
𝛽𝛽𝑒𝑒 (°) 13.08 13.08 13.08 13.08 13.08 -
RR 0.352 0.352 0.352 0.352 0.352 -
Theodorsen, 0.283 0.449 0.629 0.769 0.829 0.0
2
Wiesner 0.013 0.392 0.626 0.770 0.827 12.3
SRE-F5 0.385 0.500 0.638 0.766 0.827 5.1
RRc 0.0 0.001 0.275 0.532 0.655 -
SRE-F5Cor 0.295 0.442 0.630 0.766 0.827 0.7

Table 7 Comparison of slip factor prediction methods: TH

data [3] Ex.3
Case 1 2 3 4 5 % RMS
difference Figure 5 Open impeller and volute
Z 1 2 4 8 12 -
𝛽𝛽𝑒𝑒 (°) 35.0 35.0 35.0 35.0 35.0 - 0, 22.52 and 28.02 °. Each impeller has 4, 8 or 16 blades. The
RR 0.610 0.610 0.610 0.610 0.610 - impellers were driven at 1500 rpm by a 1 kW electrical motor
Theodorsen, 0.193 0.343 0.563 0.771 0.838 0.0 through a variable speed drive, resulting in a rotor rim speed
3
Wiesner 0.095 0.443 0.657 0.789 0.841 7.6
of 23.6 m/s. Speed was measured at 1 second intervals with
SRE-F5 0.381 0.506 0.649 0.777 0.837 11.8 an electronic tachometer with an accuracy of ± 0.05%.
RRc 0.0 0.012 0.297 0.554 0.674 - The flow through the fan was measured with a calibrated
SRE-F5Cor 0.210 0.352 0.565 0.769 0.837 0.9 150 mm diameter bell mouth nozzle with a discharge
coefficient of 0.980. The fan exit velocity was measured at a
An important finding is that the standard SRE method, radius of 158.5 mm exactly above the fan axis. The volume
with an RRc correction accurately models inviscid flow flow through the fan was adjusted by a plate throttle fitted to
through two-dimensional impellers with any given number of the flange at the end of the volute.
straight back-swept blades. This is an important
consideration for designers of series of impellers with
varying numbers of blades.

8 Fan design, manufacture and testing

A series of simple test fans was constructed to gather
experimental data on the slip factor of centrifugal impellers
with straight back-swept blades. The rotor, volute and set-up
are shown in figures 5 and 6. The fan inlet tube intrudes into
the fan housing up to the impeller eye, leaving an axial gap
of 1.5 mm (1 % of eye diameter). The fan casing and
logarithmic spiral volute are 95 mm wide in the axial
direction, and the volute width increases linearly in the radial
Figure 6 Layout of experimental rig
direction from 10 mm at its start to 185 at its end, giving a
volute angle of 72 ° (from radial). The start and end of the A wedge probe (figure 7) was used to measure the flow
volute are exactly above the rotor axis when the fan is velocity magnitude and angle at ten points across the fan
oriented such that the flow leaves the volute horizontally. impeller width (figure 8). The probe was calibrated in a small
Each impeller consists of a 300 mm diameter circular, flat wind tunnel against a standard Pitot tube, and had a velocity
back-plate attached to the fan shaft, and a 300 mm front plate coefficient of 0.950. The pressure transducers had an
with a 150 mm diameter inlet eye. The blade leading edges accuracy of ±1% and a range of 1 000 Pa, and were calibrated
are at the edge of the inlet eye, at 75 mm from the axis, and against a Betz water manometer with an accuracy of 1 Pa.
the trailing edges at the rotor rim at 150 mm radius, thereby A null method was used to align the probe to the flow
fixing RR at 0.5. This is acceptable, as it is expected to have direction and determine the flow angle to an estimated
an effect only for the 4-bladed impeller, and the comparison accuracy of 1°. This measurement approach corresponds to
with the TH data has already shown that the SRE method can that of Memardezfouli and Nourbahksh [21]. They found that
handle variations in RR. The rear disc of the fan was made the radial velocity components measured with a wedge probe
of 6 mm thick Aluminium plate and the front of 4 mm. The like ours, at three different circumferential positions, agreed
blade thickness is 2 mm. The laser cut blades have tabs that perfectly only at the nominal design flow coefficient (= 0.14)
fit into laser cut slots in the front and back plates, and were of the volute. The flow weighted average radial and
welded to the plates on the outside. This ensures smooth, circumferential velocity components were determined by
unobstructed flow channels. The internal width of the numerical integration as presented by Memardezfouli and
impeller between front and back discs is 40 mm. The blade Nourbahksh [21]. Our integration width was 42.2 mm to
inlet angles are 0, 50 or 70 ° with corresponding exit angles allow for a 7.5 ° radial expansion angle per side of the jet
R & D Journal of the South African Institution of Mechanical Engineering 2019, 35, 55-63 60
http://dx.doi.org/10.17159/2309-8988/2019/v35a7
http://www.saimeche.org.za (open access) © SAIMechE All rights reserved.
 Slip Factor Prediction for Impellers with Straight, Back-swept Blades

leaving the impeller. The velocity components were then coefficient of 0.14, the measured slip factors are expected to
corrected to account for the difference between the radial be the most reliable near that value. The SRE method for the
position of the measuring station and the exit radius of the
impeller, by considering the flow areas and conservation of
angular momentum. It was found that the average mass flow
balance between the bell mouth and the exit probe
measurement was within 0.1 %, but the standard deviation of
the difference was a relatively high 8.8 %. Since the
circumferential velocity component was about four times as
large as the radial component, a small error in measured angle
has a large effect on the radial component, and virtually no
effect on the circumferential component.

Figure 9 Measured and predicted slip factors versus exit

flow coefficient: 𝑍𝑍 = 4, 𝛽𝛽𝑒𝑒 = 22.52°

Figure 7 Wedge probe

Figure 10 Measured and predicted slip factors versus exit

flow coefficient: 𝑍𝑍 = 8, 𝛽𝛽𝑒𝑒 = 0°
Figure 8 Diagram of measuring positions across impeller
width
The measured slip velocity is calculated as the difference
between the measured and ideal circumferential velocity
components at the rotor rim:
′
∆𝑊𝑊𝑒𝑒 = ∆𝑉𝑉𝛳𝛳𝛳𝛳 = 𝑉𝑉𝛳𝛳𝛳𝛳 − 𝑉𝑉𝛳𝛳𝛳𝛳 (18)
The radial velocity component at the impeller exit is
calculated from the volume flow as determined by the inlet
nozzle.

9 Comparison of predicted and

measured slip factors
Figures 9, 11 and 13 present measured and predicted slip
factors for the impellers with 4, 8 and 16 blades, and a blade
exit angle of 22.52 °. Figures 10, 11 and 12 relate to 8-bladed
impellers with exit angles of 0.0, 22.52 and 28.02 °. Since the
volute design corresponded more or less with an exit flow Figure 11 Measured and predicted slip factors versus exit
flow coefficient: 𝑍𝑍 = 8, 𝛽𝛽𝑒𝑒 = 22.52°

R & D Journal of the South African Institution of Mechanical Engineering 2019, 35, 55-63 61
http://dx.doi.org/10.17159/2309-8988/2019/v35a7
http://www.saimeche.org.za (open access) © SAIMechE All rights reserved.
 Slip Factor Prediction for Impellers with Straight, Back-swept Blades

eddy-induced slip factor is not sensitive to flow coefficient, Table 8 Comparison of slip factor prediction methods:
but the Qiu et al. [16] method predicts only a small effect for Present
these cases, and the measured data show no clear trends for Case 1 2 3 4 5 % RMS
our data. The equations used to calculate the Qiu et al. [16] difference
slip factors are given in the appendix of their paper. For Z 4 8 8 8 16 -
application here, their equations were simplified to remove 𝛽𝛽𝑒𝑒 (°) 22.52 0.0 22.52 28.02 22.52 -
RR 0.50 0.50 0.50 0.50 0.50 -
the effects of disc coning, blade thickness and fluid density
Experimental 0.617 0.734 0.734 0.741 0.854 0.0
ratio. φ 0.169 0.186 0.186 0.178 0.190 -
𝛽𝛽𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 (°) 73 76 71 73 78 -
F (Qiu et al.) 0.396 0.854 0.750 0.739 0.900 -
𝑑𝑑𝑑𝑑⁄𝑑𝑑𝑑𝑑 (m-1) 5.308 0.0 5.308 6.862 5.308 -
Qiu et al. 0.727 0.665 0.752 0.774 0.851 4.7
Wiesner 0.591 0.767 0.776 0.781 0.862 2.4
SRE-F5 0.641 0.765 0.769 0.772 0.866 2.1
RRc 0.281 - - - - -
SRE-F5Cor 0.597 0.765 0.769 0.772 0.866 2.0

10 Conclusions
The equations for the SRE method applied to impellers with
straight, back-swept blades have been derived. It was found
that in straight-bladed impellers the possible exit blade angles
are constrained by the impeller radius ratio. This means that
when radius ratios are moderate to low, say below 0.5, the
exit blade angle is then limited to a maximum of 30 °.
The previously published universal solidity influence
Figure 12 Measured and predicted slip factors versus exit coefficient in the SRE method has been slightly adjusted to
flow coefficient: 𝑍𝑍 = 8, 𝛽𝛽𝑒𝑒 = 28.02° achieve better numerical agreement with exact analytical data
for two-dimensional inviscid flow through logarithmic spiral
blades. Comparison with methods and measurements
published since the previous SRE papers show that, in general
the SRE method predicts slip factors with a typical RMS
difference of about 2 %. This establishes the use of an
accurate, inviscid relative eddy model as a basis for a
relatively accurate practical slip factor prediction method,
and questions in principle the use of the inaccurate Stodola
relative eddy model as basis, before adding addition terms to
account for flow variations and blade curvature.
A new radius ratio correction for cases with low numbers
of blades ensures excellent agreement to the analytical
prediction of slip factor for impellers with straight, back-
swept blades. The slip factor, as defined and used in these
cases is independent of flow coefficient. A set of impellers
with straight, back-swept blades was designed, manufactured
Figure 13 Measured and predicted slip factors versus exit and tested. It was found that their slip factors vary very little
flow coefficient: 𝑍𝑍 = 16, 𝛽𝛽𝑒𝑒 = 22.52° when the flow coefficient is close to the volute design flow
coefficient. The SRE method predicts these slip factors to
Slip factors and absolute flow angles at rotor exit for each
within 2 % RMS.
fan are presented at the same four throttle settings. The flow
In general it is recommended that standard slip factor
angles agree more or less with the volute angle of 72 °. For
prediction methods be used for impellers with straight, back-
each impeller the middle two flow coefficients resulted in
swept blades. For the range for radius ratio, blade number and
more or less the same slip factor, with the one at the higher
blade exit angle of straight-bladed impellers, it turns out that
flow corresponding better with the design flow coefficient.
accounting for the critical radius ratio is more important than
For these cases (table 8) the RMS difference between
to model the sensitivity to flow coefficient.
experiment and prediction was 2.0 % for the SRE compared
to 4.7 % for Qiu et al. [16] and 2.4 % for Wiesner [9]. Since
the method of Qiu et al. is blind to the effect of radius ratio, Acknowledgement
it over predicts the slip factor for the low blade number of 4 Stephen Ryall designed and installed the volute, and did the
in figure 9, where the critical radius ratio is less than 0.5. experimental work.
Also, since radial blades (βe =βi = 0) have no effective
curvature, unlike back-swept, ”straight” blades, the
curvature term is zero, and so it under predicts in figure
10.
R & D Journal of the South African Institution of Mechanical Engineering 2019, 35, 55-63 62
http://dx.doi.org/10.17159/2309-8988/2019/v35a7
http://www.saimeche.org.za (open access) © SAIMechE All rights reserved.
 Slip Factor Prediction for Impellers with Straight, Back-swept Blades

References [14] T. W. von Backström. Relative-eddy induced slip in

[1] R. I. Lewis. Turbomachinery Performance Analysis. centrifugal impellers for engineering students. R&D
Wiley, New York, 1996. Journal, 21(1):21–27, 2007.
[2] F. C. Visser, J. J. H. Brouwers and R. Badie. Theoretical [15] C. Ji, X. D. Ruan, P. Darlo. and X. Fu. A new correlation
analysis of inertially irrotational and solenoidal flow in for slip factor in radial and mixed-flow impellers. In
two-dimensional radial-flow pump and turbine impellers Proceedings: Institution of Mechanical Engineers, Part
with equiangular blades. Journal of Fluid Mechanics, A: Journal of Power and Energy, 225(1):114–119, 2011.
269:107–141, 1994. [16] X. Qiu, D. Japikse, J. Zhao and M. R. Anderson.
[3] W. C. Hassenpflug. The incompressible two- Analysis and validation of a unified slip factor model for
dimensional potential flow through blades of a rotating impeller design and off-design conditions. ASME
radial impeller. Mathematical and Computer modelling, Journal of Turbomachinery, 133(4), 2011.
52(9-10):1299–1389, 2010. [17] B. Eck. Fans. Pergamon Press, Germany, p. 37, 1973.
[4] A. Busemann. Das förderhöhenverhältniss radialer [18] G. T. Csanady. Head correction factors for radial
kreiselpumpen mit logarithisch-spiraligen schaufeln. impellers. Engineering (London), 190:195, 1960.
Zeitschrift fur Angewandte Mathematik und Mechanik, [19] K. S. Paeng and M. K. Chung. A new slip factor for
8(5):372–84, 1928. centrifugal impellers. In Proceedings: Institution of
[5] T. Theodorsen. Theory of wing sections of arbitrary Mechanical Engineers, Part A: Journal of Power and
shape. NACA Technical report No 311, 1931. Energy, 215(5):645–649, 2001.
[6] T. Theodorsen and I.E. Garrick. General potential theory [20] J. M. Huang, K. W. Luo, C. F. Chen, C. P. Chiang, T. Y.
of arbitrary wing sections. NACA Report No 452, 1933. Wu and C. H. Chen. Numerical investigations of slip
[7] R. H. Aungier. Centrifugal Compressors: A Strategy for phenomena in centrifugal compressor impellers.
Aerodynamic Design and Analysis. The American International Journal of Turbo & Jet-Engines,
Society of Mechanical Engineers, New York, 2000. 30(1):123–132, 2013.
[8] A. Whitfield and N.C. Baines. Design of Radial [21] M. Memardezfouli and A. Nourbaksh. Experimental
Turbomachines. Longman Scientific and Technical, UK, investigation of slip factors in centrifugal pumps.
1990. Experimental Thermal and Fluid Science, 33(5):938–
[9] F. J. Wiesner. A review of slip factors for centrifugal 945, 2009.
impellers. Journal of Engineering for Gas Turbines and [22] D. Eckardt. Flow field analysis of radial and backswept
Power, 89(4):558–566, 1967. centrifugal impellers, Part 1: Flow measurement using
[10] A. Stodola. Steam and Gas Turbines. McGraw-Hill, laser velocimeter. In 25th ASME Gas Turbine Conference
Reprinted by Peter Smith, New York, 1927. and 22nd Annual Fluid Engineering Conference, New
[11] J. D. Stanitz. Some theoretical aerodynamic Orleans, 1980.
investigations of impellers in radial- and mixed-flow
centrifugal compressors. Transactions of the ASME,
74:473–476, 1952.
[12] T. W. von Backström. A unified correlation for slip
factor in centrifugal impellers, ASME Journal of
Turbomachinery, 128(1):1–10, 2006.
[13] T. W. von Backström. A compact equation for the
prediction of eddy-induced slip in centrifugal impellers,
In Proceedings: Institution of Mechanical Engineers,
Part A: Journal of Power and Energy, 220(8): 911–915,
2006.

R & D Journal of the South African Institution of Mechanical Engineering 2019, 35, 55-63 63
http://dx.doi.org/10.17159/2309-8988/2019/v35a7
http://www.saimeche.org.za (open access) © SAIMechE All rights reserved.