Mechanism and Machine Theory 87 (2015) 70–81

Contents lists available at ScienceDirect

Mechanism and Machine Theory

journal homepage: www.elsevier.com/locate/mechmt

A new curve for application to the rotor profile of rotary

lobe pumps
Chiu-Fan Hsieh
Department of Mechanical and Computer-Aided Engineering, National Formosa University, 64 Wunhua Road, Huwei, Yunlin, Taiwan, ROC

a r t i c l e i n f o a b s t r a c t

Article history: This paper proposes a new type of curve, an elliptical roulette, which it initially applies to rotary
Received 12 April 2014 lobe pumps. Once the new rotor profile has been mathematically modeled using the principle
Received in revised form 21 December 2014 of gearing, equations can be derived to assess the presence or absence of undercutting in the
Accepted 30 December 2014
tooth profile. The effect of the new profile properties on pumping performance is evaluated
Available online xxxx
using a specially developed three-dimensional fluid analysis model. The proposed curve is tested
using six novel rotor profiles based on the elliptical axial ratio parameter λ (a shorter axis divided
Keywords: by a larger axis). These six new cases and a traditional case are analyzed under the same volume
Rotary lobe pump
and clearance conditions and their differences compared. The results show that a smaller elliptical
Gear
axial ratio design produces better flow characteristics. In particular, an elliptical axial ratio smaller
Flow characteristics
Undercutting than 0.6 not only achieves high discharge efficiency but allows vibration and noise to be controlled
by the flow rate fluctuation coefficient, which approximates the traditional indicator. The
proposed curve can thus serve as a useful reference for pump or rotary fluid machine design.
© 2015 Elsevier Ltd. All rights reserved.

1. Introduction

Rotary lobe pumps, such as blower pumps, vacuum pumps, and hydraulic pumps, are fluid-extracting apparatuses used widely in
industry. A type of positive displacement pump, they use two or more rotating lobes, in a specially shaped cylinder, that intermesh
with each other using timing gears to draw fluid in from one side to deliver to the other. These two rotors rotate in opposite directions
with a constant gear ratio and maintain only tiny clearances during operation.
The majority of studies relevant to this paper focus on the geometric design and rotor analysis of the Roots type rotary pump. For
example, Litvin and Feng [1] developed a geometry for screw Roots blowers in which the addendum tooth profile is a single circular
arc. They then studied the conjugation of surfaces or rotors, the synthesis of rotor surfaces with two contact lines, and the avoidance of
singularities. Shung and Pennock [2] proposed a unified compact equation describing the geometry and geometric properties of the
different types of trochoid and the geometric properties of a conjugate envelope. Demenego et al. [3] developed a tooth contact
analysis (TCA) computer program and discussed avoidance of tooth interference and rapid wearing through modification of the
rotor profile geometry of a cycloidal pump whose one pair of teeth is in mesh at every instant. Wang et al. [4] proposed certain design
constraints for five-arc Roots vacuum pumps, including the conditions for avoiding tooth undercutting and carryover. Mimmi and
Pennacchi [5–8] examined rack generated rotors with three lobes and considered profiles comprising an arc, an involute, and an
epitrochoidal arc with a constant trochoid ratio. Using these profiles, they analyzed admissible compositions that would avoid unrea-
sonable rotor profiles and undercutting. They also investigated the load dynamics on rotors of a three-lobe Roots blower and found
that increasing the reservoir and using helical shaped rotors improves dynamic load regularity and decreases system vibrations.

E-mail address: cfhsieh@nfu.edu.tw.

http://dx.doi.org/10.1016/j.mechmachtheory.2014.12.018
0094-114X/© 2015 Elsevier Ltd. All rights reserved.
 C.-F. Hsieh / Mechanism and Machine Theory 87 (2015) 70–81 71

Nomenclature

a elliptical larger axis

b elliptical shorter axis
rp radius of the base circle (radius of the pitch circle)
λ elliptical axial ratio parameter (a shorter axis divided by a larger axis)
n number of lobes
Si coordinate system i where i = 1,2,f
Ri position vector of rotor i represented in coordinate system Si, i = 1,2
Mij coordinate transformation matrix form system j to system i
Oi origin of the coordinate system Si, i = 1,2,f
ϕ rotation angle
ψ parameter of the elliptical profile
θ parameter of the base circle (or pitch circle)
ζ normal angle at a point on the rolling ellipse
N1 normal vector represented in coordinate system S1
V(12)
1 sliding velocity represented in coordinate system S1
η area efficiency
QI increment of average flow rate
I coefficient of flow rate fluctuation

Liu et al. [9] and Tong and Yang [10–12] proposed a novel design method, the derivation function method, used mainly to design
rotor profiles for high-sealing lobe pumps and synthesize lobe pump profiles with given flow rate functions. In particular, this method
can be used to reshape original pitch pairs to obtain the desired profiles of the generated pairs. Subsequently, Hwang and Hsieh
[13,14] proposed a new rotor profile with a variable trochoid ratio and investigated how to achieve high volumetric efficiency and
high sealing. Valdès et al. [15] used a semi-empirical Knudsen–Dong law to calculate internal leaks in dry Roots vacuum pumps.
Their method predicts the conductance of each clearance within a few percent in a wide pressure range covering the transient flow
regime. Burmistrov et al. [16] developed an angular coefficients method for calculating channel conductance in Roots pumps based
on the similarity between laws of radiation and diffusive reflection of molecules. This method enables accurate calculation of the
conductance of Roots pump channels for different rotor positions.
For flow simulation, Voorde et al. [17] applied the fictitious domain method to study a three-blade lobe pump and tooth compres-
sor. Houzeaux et al. [18] presented a finite element method for the simulation of rotary positive displacement pumps. This method,
however, although it solves leakage problems in the gap and gear intersection, has two drawbacks: First, generating configurations
is tedious if the geometry is complicated. Second, one Navier–Stokes solver must solve each configuration flow, meaning that simu-
lations using many configurations require a huge amount of memory. Strasser [19] used a commercial computational fluid dynamics
(CFD) solver, FLUENT, to simulate the gear pump. Similarly, Huang and Liu [20] used the renormalization group k-ε model, PISO algo-
rithm, and second-order upwind difference scheme to solve the governing equations of an involute-type three-lobe positive discharge
blower. All these studies [18–20], however, focused on the involute external gear pump, and most used primarily a two-dimensional
numerical approach by means of computational fluid dynamics or CFD tools.
What the above literatures does make clear, however, is that volume efficiency and leakage level are the key points in pump
design, meaning that topics related to the method of rotor geometric design and flow analysis are important. This present paper
therefore develops a new geometric design method, one based on an elliptical roulette (the path traced by a point attached to an
ellipse as it rolls without slipping along a fixed circle), which it then applies to the rotor profile of rotary lobe pumps.
To assess the performance of this new curve, a three-dimensional fluid analysis model of rotary lobe pumps is constructed that
enables analysis of pump flow characteristics. Here, the effects of this new design on fluid are exemplified using a two-lobed rotor.
The results thus provide a useful reference for the design of rotary lobe pumps. The next sections outline and discuss the mathematical
model of the new rotor profile, the undercutting analysis, and the fluid analysis models.

2. Mathematical models of rotors and undercutting

2.1. Mathematical model of the new curve

As Fig. 1(a) shows, the equation of ellipse can be expressed as

R ¼ ½xðψÞ; yðψÞ ¼ ½−a cos ψ; b sin ψ: ð1Þ

When the rolling ellipse rotates counter-clockwise around the circumference of the base circle in a pure rolling motion, the point p
on the rolling ellipse will trace an elliptical path. As Fig. 1(b) illustrates, because the rolling motion is pure, the length of curve 12 on the
72 C.-F. Hsieh / Mechanism and Machine Theory 87 (2015) 70–81

(a) Elliptical rolling

(b) Geometric relation

Fig. 1. Principle of elliptical path traced.

ellipse should be equal to the length of curve 120 on the base circle. If the base circle radius is rp, the relation between the lengths of the
two curves can be given as
Z ψ 1
2 2
rp θ ¼ ∂ψ xðψÞ þ ∂ψ yðψÞ 2 ¼ sðψÞ: ð2Þ
0

In which, the domain of θ is from 0 to π/n. n is the lobed number. If the lobed number is two (n = 2), then the circumference of a
rolling ellipse equals to a quarter (1/4) of the circumference of the base circle.
As Fig. 2 shows, the two mating rotors rotate in opposite directions around parallel axes. Coordinate systems S1, S2, and Sf are
rigidly attached to the rotational axes of rotor 1, rotor 2, and the frame, respectively. The addendum curve is an elliptic trajectory
curve whose equation can be derived and represented in coordinate system S1 as follows based on the geometric relation:

2 3 2 3
x1 a cosðζ þ θÞð cos ψ−1Þ þ b sinðζ þ θÞ sin ψ þ r p cos θ
r1 ¼ 4 y1 5 ¼ 4 a sinðζ þ θÞð cos ψ−1Þ−b cosðζ þ θÞ sin ψ þ rp sin θ 5 ð3Þ
1 1

where a is the ellipse's larger axis and b is its shorter axis. The normal angle ζ at the contact point 2 on the rolling ellipse is

−1 ∂ψ xðψÞ
ζ ¼ tan : ð4Þ
∂ψ yðψÞ
 C.-F. Hsieh / Mechanism and Machine Theory 87 (2015) 70–81 73

Fig. 2. Coordinate system.

2.2. Equation of meshing

The conjugate segment must satisfy the meshing equation [24]:

ð12Þ
N1  V1 ¼0 ð5Þ

where N1 is the normal vector expressed as

∂r1  0 0 
N1 ¼  k ¼ x1 i þ y1 j  k ¼ Nx1 i þ Ny1 j: ð6Þ
∂θ

The relative velocity V1(12) is then determined by

   
ð12Þ ð12Þ ð12Þ
V1 ¼ V x1 i þ V y1 j ¼ −2r p sin ϕ þ 2y1 i þ −2x1 þ 2r p cos ϕ j: ð7Þ

Substituting Eqs. (6) and (7) into Eq. (5) yields the following equation of meshing:
   
f ¼ −2r p sin ϕ þ 2y1 Nx1 þ −2x1 þ 2rp cos ϕ Ny1 ¼ 0 ð8Þ

where
 0 0  0 0   0 0 
Nx1 ¼ rp cos θ−a 1 þ ζ ψ cosðζ þ θÞ þ a þ ψ aζ −b cos ψ cosðζ þ θÞ þ b þ ψ bζ −a sin ψ sinðζ þ θÞ ð9Þ

 0 0  0 0   0 0 
Ny1 ¼ r p sin θ−a 1 þ ζ ψ sinðζ þ θÞ þ a þ ψ aζ −b cos ψ sinðζ þ θÞ− b þ ψ bζ −a sin ψ cosðζ þ θÞ ð10Þ

0 s0 ðψÞ
ψ ¼ ð11Þ
rp

0 2
0 abψ sec ψ
ζ ¼ : ð12Þ
b2 þ a2 tan2 ψ

2.3. Mathematical model of the conjugate curve

The equation of conjugate curve r2 is determined through coordinate transformation:

r2 ¼ M21 r1 ð13Þ
74 C.-F. Hsieh / Mechanism and Machine Theory 87 (2015) 70–81

where
2 3
cos2ϕ sin2ϕ −2r p cosϕ
M21 ¼ 4 − sin2ϕ cos2ϕ 2rp sinϕ 5:
0 0 1

Operation of Eq. (13) yields the following:

2 3 2 3
x2 að cos ψ−1Þ cosðζ þ θ−2ϕÞ þ b sin ψ sinðζ þ θ−2ϕÞ þ r p ð cosðθ−2ϕÞ−2 cos ϕÞ
r2 ¼ 4 y2 5 ¼ 4 að cos ψ−1Þ sinðζ þ θ−2ϕÞ þ b sin ψ cosðζ þ θ−2ϕÞ þ rp ð sinðθ−2ϕÞ þ 2 sin ϕÞ 5: ð14Þ
1 1

Eqs. (8) and (14) can then be used to generate the conjugate curve tooth profile.

2.4. Undercutting equations

If a singular point exists on the tooth profile, undercutting will occur. In the theory of gearing [24], the mathematical definition of
tooth profile singularity is expressed as
 0   0 
x −Vx1  y −Vy1 
ð12Þ ð12Þ
 1  1
Λ 1 ¼  
dϕ  ¼ 0; Λ 2 ¼  
dϕ  ¼ 0: ð15Þ
 fθ −f ϕ   fθ −f ϕ 
dt dt

The angular velocity of the rotor to unit can thus be set at dϕ

dt
¼ 1 without losing generality. In Eq. (15), fθ means ∂∂θf :
h    i
02 02 ″ ″
f θ ¼ 2 x1 þ y1 þ x1 x1 −r p cos ϕ þ y1 y1 −r p sin ϕ ð16Þ

∂f
and fθ means ∂ϕ :

 0 0 
f ϕ ¼ 2r p x1 sin ϕ−y1 cos ϕ : ð17Þ

Operating Eq. (15) yields

 h    i
0 0 0  02 02 ″ ″
Λ 1 ¼ −2r p x x sinϕ−y cosϕ þ 2 2y−2rp sinϕ x þ y þ x x−rp cosϕ þ y y−r p sinϕ ð18Þ

 h    i
0 0 0  02 02 ″ ″
Λ 2 ¼ 2r p y −x sinϕ þ y cosϕ þ 2 −2x þ 2r p cosϕ x þ y þ x x−r p cosϕ þ y y−r p sinϕ ð19Þ

where
0 0
x ¼ −Ny1 ; y ¼ N x1 ð20Þ

h  i
″ 0 0 0 0 0 ″
x1 ¼ − r p cosθ þ af cosðζ þ θÞ þ ψ ζ cosðζ þ θÞ 2 þ ζ ψ þ ζ ψ sinðζ þ θÞ
h  i
0 ″ 0 0 0 02 0 02
þ ζ ψ sinðζ þ θÞg þ cosψf− cosðζ þ θÞ a−2bψ þ 2aζ ψ þ ψ a−2bζ þ aζ
02 ″ ″ 0 ð21Þ
− aψ ζ sinðζ þ θÞ þ ψ b−aζ sinðζ þ θÞg þ sinψf−b sinðζ þ θÞ
h    i
0 0 0 0 02 0 ″
þ ψ sinðζ þ θÞ 2a−2bζ −ψ b−2aζ þ bζ þ bζ ψ cosðζ þ θÞ
 0 ″
þ −a þ bζ ψ cosðζ þ θÞg

″ 0
y1 ¼ − r p sin θ þ ða−a cos ψÞ sinðζ þ θÞ þ b sin ψ cosðζ þ θÞ þ 2bψ cos ψ sinðζ þ θÞ
0 02 02
− 2aψ sin ψ cosðζ þ θÞ−ψ ζ ½að cos ψ−1Þ sinðζ þ θÞ−b sin ψ cosðζ þ θÞ
0 0  0  0 
þ 2ζ ψ a sinðζ þ θÞ þ b−aψ sin ψ cosðζ þ θÞ þ −a þ bψ cos ψ sinðζ þ θÞ ð22Þ
02 ″
þ ψ ½−a cos ψ sinðζ þ θÞ þ b sin ψ cosðζ þ θÞ þ ψ ½−b cos ψ cosðζ þ θÞ−a sin ψ sinðζ þ θÞ
 
″ 02 0 ″
þ ζ ψ þ ζ ψ ½a cosðζ þ θÞð cos ψ−1Þ þ b sin ψ sinðζ þ θÞ
 C.-F. Hsieh / Mechanism and Machine Theory 87 (2015) 70–81 75

″ −ψ0 s″ ðψÞ
ψ ¼ ð23Þ
s0 ðψÞ

h      i
″
2ab 2 b2 −a2 ψ02 sin2ψ þ ψ″ a2 þ b2 þ b2 −a2 cos2ψ
ζ ¼  2   2 : ð24Þ
a þ b2 þ b2 −a2 cos2ψ

Either Eq. (18) or Eq. (19) can be used to determine the absence or presence of undercutting in the rotor profile.

3. Design results and analysis

This study uses two-lobe rotors to exemplify the effects of the proposed new curve. Assuming an elliptical axial ratio parameter of
λ = b/a, then when λ = 1, the rotor is a traditional cycloid tooth, but when λ ≠ 1, it is an elliptic trajectory curve. The design
parameters are as listed in Table 1. Here, based on the mathematical models of the tooth profile and undercutting, the parameter λ
ranges from 0.9 to 0.3, producing the six results without undercutting shown in Fig. 3. When λ = 0.3, however, as Fig. 4 illustrates,
some undercutting values are zero (circled in the figure), meaning that profile undercutting occurs as shown in Fig. 5. To delineate
the effects of the proposed curve, therefore, these six new designs and one traditional design (λ = 1) are analyzed and their
differences compared. For comparative convenience, the chambers in all cases have the same cross-sectional area (200 cm2) and
same thickness (8 cm), so the analytical results reflect the effect of the various rotors profiles on flow characteristics with the same
chamber volume (1600 cm3).
If the rotor and chamber have the same thickness, then the simple index of volume efficiency can be calculated by the area
efficiency [4,13]:

 
2 πR2 −Arotor
η¼ ð25Þ
πR2 þ 4r p R

where Arotor denotes the cross-sectional area of the rotor, R is the radius of the chamber, and rp is the radius of the pitch circle (or base
circle).
As Fig. 6 shows, when the parameter λ number increases, the area efficiency decreases, a simple calculation from which an initial
estimate can be made of the trend of theoretical volumetric efficiency. To more accurately assess pumping performance for the
different rotor profiles, however, in the next section, fluid analysis model is constructed to analyze the flow characteristics in detail.

4. Fluid analysis and discussion

As already specified, this study analyzes the flow characteristics of a lobe pump using a two-lobe rotor as the example. Because the
new commercial CFD package PumpLinx has been successfully used to compare simulation and experimental results [21], it is used
here to construct the fluid analysis model. To conduct a CFD analysis of a Roots pump, a CAD model must first be created by stipulating
the rotor geometry design and then meshing the geometry with the program grid generator. PumpLinx solves conservation equations
of mass and momentum using a finite volume approach. In this paper, the fluid flow is assumed unsteady, compressible, and
turbulent.

4.1. Assumptions and boundary conditions

The input conditions are assumed to be as follows: the fluid type is air (dynamic viscosity = 1.853e−05 Pa s), which is compress-
ible and has a density dependent on the local pressure and temperature. For simple simulation, the gas is assumed as an ideal gas. So
the equation of state of the ideal gas is used in the fluid model. In CFD simulation, therefore, density is not a constant. The inlet pressure
is 101.325 kPa (i.e., one atmospheric pressure), and the outlet operation pressure is 150 kPa. The fluid is analyzed by using a turbu-
lence model (k-ε model) [22]. The rotor rotates at 3000 rpm, with the left rotor rotating clockwise and the right rotor rotating anti-
clockwise. The clearance value between the rotor and chamber is 0.1 mm, and the clearance value between the two rotors is 0.3 mm.

Table 1
Parameter design.

Cases λ = 0.4 λ = 0.5 λ = 0.6 λ = 0.7 λ = 0.8 λ = 0.9 λ=1

a 12.2104 11.7887 11.3646 10.9472 10.542 10.1524 9.78

rp 35.778 36.3555 36.93725 37.51085 38.0686 38.6059 39.12015
R 60.1988 59.9329 59.6665 59.4053 59.1527 58.9107 58.6803
76 C.-F. Hsieh / Mechanism and Machine Theory 87 (2015) 70–81

λ = 0.4 λ = 0.5 λ = 0.6

λ = 0.7 λ = 0.8 λ = 0.9

Fig. 3. Design results of new rotor profiles.

4.2. Moving grid generation and flow analysis

Here, the λ = 1 design is used to explain the flow calculation. First, it is important to construct a grid, whose 3D grid generation is
illustrated in Fig. 7. In this structure, the volume variation space between the two rotors and the chamber should have a higher
grid density, while the inlet and outlet spaces can have a lower grid density. This model uses a total of 121,016 cells, ensuring that
the calculation method produces more accurate results.
When two rotors rotate at certain angles, the volume begins to change. At the same time, the grid calculates automatically, solving
flow characteristics like flow velocity, pressure, and flow rate through iteration. Fig. 8 shows the calculation results for the dynamic
grid and pressure during rotation. As the figure clearly shows, as the rotation angle increases, the pressure changes, with its highest
distribution occurring in the outlet area. The pressure also rises inside the pump. These observations verify that the mode of analysis is
feasible.
The following results were derived for 2 to 4 cycles before flows reached a periodic steady state. The calculations were obtained
based on data from one revolution of the rotor after flows had reached a steady state. Fig. 9(a) illustrates the instantaneous flow
rate: a positive value signals fluid outflow from the outlet pipe while a negative one reflects flow in the opposite direction, which
means there may be a backflow (or reverse flow) phenomenon. The instantaneous flow rate also indicates that when the value of
λ is smaller, the flow rate fluctuation is greater, but when the λ value is larger, the flow rate fluctuation is relatively steady. In all
cases, the λ = 0.4 design produces the maximum positive flow rate and the minimum negative flow rate, and thus the best discharge
efficiency. Using the λ = 0.4 and λ = 0.5 designs as examples, although the peak values of the flow rate for both are large and close to
each other, the backflow in the λ = 0.5 design (− 0.049 m/s) is greater than in the λ = 0.4 design (− 0.016 m/s), which affects
discharge efficiency and makes the former slightly less efficient than the latter.
Fig. 9(b), on the other hand, clearly shows that the trend for the average flow rate does not mirror that from the area efficiency
prediction, in which area efficiency progressively decreases as the parameter λ number progressively increases. This observation
confirms that the rotor geometry significantly influences pumping performance. The higher average flow rates were obtained for
the λ = 0.5 and λ = 0.4 designs. However, the average flow rate ranking is as λ = 0.6 N λ = 0.8 N λ = 0.7. This result differs

Undercutting analysis (λ=0.3)

300

200

100

-100 0 10 20 30 40 50 60 70 80 90

-200

-300
Angle φ

Fig. 4. Undercutting analysis (λ = 0.3).

 C.-F. Hsieh / Mechanism and Machine Theory 87 (2015) 70–81 77

Fig. 5. Undercutting illustration (λ = 0.3).

from that in the area efficiency prediction, as is confirmed by the flow analysis, including the flow rate (Fig. 9(a)) and the flow velocity
in the Y direction (Fig. 10).
First, as regards a positive flow rate value, the λ = 0.6 design produced a waveform from small to large in one cycle, while the λ =
0.7 design generated mostly small waves, with a larger wave emerging only between 240° and 300°. The λ = 0.8 design, in contrast,
produced three similar waves, a waveform style also evident in the flow velocity pattern (see Fig. 10). In terms of a negative flow rate
value, all three designs suffer from a backflow phenomenon, with the smallest backflow occurring in the λ = 0.7 design. Nevertheless,
in all three cases, the area of positive flow rate is much larger than the area of negative flow rate. The largest positive flow rate area
occurred in the λ = 0.6 design, with the λ = 0.8 design coming second (λ = 0.7). Hence, the final average flow rate result of λ =
0.6 N λ = 0.8 N λ = 0.7. This indicates that a lager theoretical volumetric efficiency does not necessarily to get better flow character-
istics based on varied rotor geometric property. Just as the design in λ = 0.7 has a higher theoretical volumetric efficiency than the
design of λ = 0.8, however, the flow characteristics in λ = 0.8 design is better than the design in λ = 0.7. So, this new curve (elliptical
roulette) in a varied design parameter would lead to different effects on the fluid flow and leakage level.
The analysis of flow velocity also shows that when λ b 0.6, the area of positive flow velocity progressively increases as the param-
eter λ number progressively decreases. Specifically, it indicates that a smaller parameter λ has higher flow power and lower fluid
resistance, leading to more effective discharge. In order to show the difference from traditional design (λ = 1) and new design
(λ = 0.4–0.9), the increment can be determined by

ðλÞ ðλ¼1Þ
Q av −Q av
QI ¼ ðλ¼1Þ
 100% ð26Þ
Q av

where Q (λ
av
= 1)
and Q (λ)
av mean average flow rate of the traditional design and the new design, respectively. λ is from 0.4 to 0.9.
Compared to the traditional design, the new design in λ = 0.5 can increase the flow rate close to 300%. In addition, there is over
300% increment in the design of λ = 0.4 (see Fig. 9(c)).

Area efficiency
70
60
50
40
%
30
20
10
0
λ=0.4 λ=0.5 λ=0.6 λ=0.7 λ=0.8 λ=0.9 λ=1

Fig. 6. Area efficiency.

78 C.-F. Hsieh / Mechanism and Machine Theory 87 (2015) 70–81

Fig. 7. Grid calculation (λ = 1).

Furthermore, the flow rate fluctuation coefficient can also be used as a performance index related to vibration and noise, as defined
in [23]:

Q max −Q min
I¼ : ð27Þ
Q ave

As Fig. 9(d) shows, the maximum fluctuation coefficient occurs in the λ = 0.7 design. In designs λ = 0.4 and λ = 0.5, on the other
hand, where λ b 0.6, the fluctuation coefficient is relatively lower and the average flow rate is higher than in the traditional design
(λ = 1). The proposed new design, therefore, not only leads to higher discharge efficiency but also keeps the vibration and noise at
a relatively lower level.

0.0005 s 0.0025 s 0.00375 s

0.005 s 0.0075 s 0.01 s

Fig. 8. Dynamic grid calculation and pressure variation.
 C.-F. Hsieh / Mechanism and Machine Theory 87 (2015) 70–81 79

Instantaneous mass flux

0.8
0.7 λ=0.4
0.6 λ=0.5
0.5
λ=0.6
0.4

kg/s
λ=0.7
0.3
λ=0.8
0.2
λ=0.9
0.1
0 λ=1

-0.1
0 60 120 180 240 300 360
Rotation angle
(a) Instantaneous flow rate
Average flow rate
0.2

0.16

0.12
kg/s

0.08

0.04

0
λ=0.4 λ=0.5 λ=0.6 λ=0.7 λ=0.8 λ=0.9 λ=1

(b) Average flow rate

Average flow rate increment
350
300
250
200
%
150
100
50
0
λ=0.4 λ=0.5 λ=0.6 λ=0.7 λ=0.8 λ=0.9
(c) Increment of average flow rate (comparing to the
traditional design)

Flow rate fluctuation coefficient

9
8
7
6
5
4
3
2
1
0
λ=0.4 λ=0.5 λ=0.6 λ=0.7 λ=0.8 λ=0.9 λ=1
(d) Flow rate fluctuation coefficient
Fig. 9. Analysis of outlet flow rate.

Finally, Fig. 11 illustrates the outlet pressure, whose mean value progressively decreases as the parameter λ number progressively
increases. As Fig. 11(b) shows, the pressure difference between the highest and lowest points of the pressure curve shows the same
trend as above: the greater pressure difference occurs in the λ = 0.4 and λ = 0.5 designs. This result suggests that a smaller λ value
generates a larger pressure ripple. Yet it is also worth noting that the stability of the pressure waveform in the λ = 0.4 design is better
than that in the λ = 0.5 design, which matches the outcome suggested by the flow rate fluctuation coefficient. Overall, the flow
80 C.-F. Hsieh / Mechanism and Machine Theory 87 (2015) 70–81

Flow velocity
350
300 λ=0.4

250 λ=0.5

200 λ=0.6

m/s
150 λ=0.7

100 λ=0.8

50 λ=0.9

0 λ=1

-50
0 60 120 180 240 300 360
Rotation angle

Fig. 10. Analysis of outlet instantaneous flow velocity in Y direction.

analysis of the new curve's properties provide evidence of its value as a reference for designing pumps, compressors, or other fluid
machinery.

5. Conclusions

This study proposed a new elliptical roulette curve, applying it first to designing the rotor profile of rotary lobe pumps and then
deriving undercutting equations to check for undercutting on the tooth profile. Next, it used six new designs and one traditional de-
sign with the same chamber volume and clearance settings to investigate the effect of changing the ellipse parameter λ on fluid flow.
Although the results clearly indicate that all the new designs produce a higher average flow rate than the traditional design (λ = 1),
the magnitude of the average flow rate is significantly influenced by the rotor geometric design. Area efficiency, for example, can only
be used as a preliminary assessment of theoretical volumetric efficiency. The analysis of flow characteristics, on the other hand–
including flow rate, fluctuation coefficient, velocity, and pressure–clearly demonstrates that a new rotor profile whose λ value is
smaller (λ b 0.6) can not only achieve high discharge efficiency but can also control the level of vibration and noise. The geometric

Outlet pressure
220000 λ=0.4

λ=0.5
200000
λ=0.6
180000
Pa

λ=0.7

160000 λ=0.8

λ=0.9
140000
λ=1
120000
0 60 120 180 240 300 360
Rotation angle
(a) Instantaneous outlet pressure
Pressure difference
100000

80000

60000
Pa

40000

20000

0
λ=0.4 λ=0.5 λ=0.6 λ=0.7 λ=0.8 λ=0.9 λ=1

(b) Pressure difference ( Pmax − Pmin )

Fig. 11. Analysis of outlet pressure.
 C.-F. Hsieh / Mechanism and Machine Theory 87 (2015) 70–81 81

design of the rotor, therefore, is a very important factor in pumping performance, making this new curve a useful reference in
designing fluid machinery.

Acknowledgments

The author is grateful to the National Science Council of the R.O.C. for their partial grant (No. NSC 101-2221-E-150-023-) on the
package of PumpLinx and to Mr. Zhou, Q. J. for his partial assistance with the fluid simulation.

References

[1] F.L. Litvin, P.H. Feng, Computerized design and generation of cycloidal gearings, Mech. Mach. Theory 31 (7) (1996) 891–911.
[2] J.B. Shung, G.R. Pennock, Geometry for trochoidal-type machines with conjugate envelopes, Mech. Mach. Theory 29 (1) (1994) 25–42.
[3] A. Demenego, D. Vecchiato, F.L. Litvin, N. Nervegna, S. Manco, Design and simulation of meshing of a cycloidal pump, Mech. Mach. Theory 37 (3) (2002) 311–332.
[4] P.Y. Wang, Z.H. Fong, H.S. Fang, Design constraints of five-arc Roots vacuum pumps, Proc. Inst. Mech. Eng. C J. Mech. Eng. Sci. 216 (2) (2002) 225–234.
[5] G.C. Mimmi, P.E. Pennacchi, Dynamics effects of pressure loads in three screw pump rotors, ASME J. Mech. Des. 120 (4) (1998) 589–592.
[6] G.C. Mimmi, P.E. Pennacchi, Analytical model of a particular type of positive displacement blower, Proc. Inst. Mech. Eng. C J. Mech. Eng. Sci. 213 (5) (1999)
517–526.
[7] G.C. Mimmi, P.E. Pennacchi, Non-undercutting conditions in internal gears, Mech. Mach. Theory 35 (4) (2000) 477–490.
[8] G.C. Mimmi, P.E. Pennacchi, Compression load dynamics in a special helical blower: a modeling improvement, ASME J. Mech. Des. 123 (3) (2001) 402–407.
[9] H.C. Liu, S.H. Tone, D.C.H. Yang, Trapping-free rotors for high-sealing lobe pumps, ASME J. Mech. Des. 122 (4) (2000) 536–542.
[10] S.H. Tong, D.C.H. Yang, On the generation of new lobe pumps for higher pumping flow rate, Mech. Mach. Theory 35 (2000) 997–1012.
[11] D.C.H. Yang, S.H. Tong, The specific flowrate of deviation function based lobe pumps—derivation and analysis, Mech. Mach. Theory 37 (2002) 1025–1042.
[12] S.H. Tong, D.C.H. Yang, Rotor profiles synthesis for lobe pumps with given flow rate functions, ASME J. Mech. Des. 127 (2) (2005) 287–294.
[13] Y.W. Hwang, C.F. Hsieh, Study on high volumetric efficiency of the Roots rotor profile with variable trochoid ratio, Proc. Inst. Mech. Eng. C J. Mech. Eng. Sci. 220 (9)
(2006) 1375–1384.
[14] C.F. Hsieh, Y.W. Hwang, Study on the high-sealing of Roots rotor with variable trochoid ratio, ASME J. Mech. Des. 129 (2007) 1278–1284.
[15] L.C. Valdès, B. Barthod, Y.L. Perron, Accurate prediction of internal leaks in stationary dry Roots vacuum pumps, Vacuum 52 (4) (1999) 451–459.
[16] A. Burmistrov, L. Belyaev, P. Ossipov, M. Fomina, R. Khannanov, Combined experimental and calculation study of conductance of Roots pump channels, Vacuum
62 (4) (2001) 331–335.
[17] J.V. Voorde, J. Vierendeels, E. Dick, Flow simulations in rotary volumetric pumps and compressors with the fictitious domain method, Comput. Methods Appl.
Mech. Eng. 168 (2004) 491–499.
[18] G. Houzeaux, R. Codina, A finite element method for the solution of rotary pumps, Comput. Fluids 36 (4) (2007) 667–679.
[19] W. Strasser, CFD investigation of gear pump mixing using deforming/agglomerating mesh, ASME J. Fluids Eng. 129 (4) (2007) 476–484.
[20] Z.F. Huang, Z.X. Liu, Numerical study of a positive displacement blower, Proc. Inst. Mech. Eng. C J. Mech. Eng. Sci. 223 (10) (2009) 2309–2316.
[21] H. Ding, F.C. Visser, Y. Jiang, M. Furmanczyk, Demonstration and validation of a 3D CFD simulation tool predicting pump performance and cavitation for industrial
applications, ASME J. Fluids Eng. 133 (1) (2011) 011101.
[22] B.E. Launder, D.B. Spalding, The numerical computation of turbulent flows, Comput. Methods Appl. Mech. Eng. 3 (2) (1974) 269–289.
[23] K.J. Huang, C.L. Wun, Kinematic flow rate characteristics of external spur gear pumps using an exact closed solution, Mech. Mach. Theory 44 (6) (2009)
1121–1131.
[24] F.L. Litvin, Gear Geometry and Applied Theory, Prentice Hall, New York, 1994.